March 9, 1996 Sybase Customers, The following represents problem reports that were resolved in System 11. This report is provided to aid you in identifying known resolved problems, should you encounter any during the operation of this product. The problems reported are ordered by importance (high to low), and then by problem identification. Our development process specifies that System 10 bug fixes be included in System 11 products. There are certain circumstances where this is not possible, specifically when: a) Problem fixes made in System 10 missed the cut-off date for inclusion. That would be problems resolved past the System 10.0.2.4 Rollup date of November 11, 1995; specifically those resolved after January 19, 1996, b) System 11 code implementations made specific problem fixes made in System 10 unnecessary, c) Problem fixes made in System 10 were believed necessary, but presented high risk of destabilization to System 11. Therefore, they weren't included in System 11. At present, there is only one problem that is associated with item c); 61628. Current plans call for these problems to be included in a subsequent System 11 maintenance release. Sybase System 11 Release Team Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 60187 In a VLDB with more GAM extents checking for a GAM bit is slow. 63996 Create database fails with error 1808 because it cannot find a 'default' device on which to build the database. 64101 Resolve the compilation warnings due to implict cast. e.g. "./generic/source/access/delete.c, line 419: warning: semantics of ">" change in ANSI C; use explict cast". 64813 Create procedure test21 as create database big_03; use master; execute big_02..test21; causes error msg 1809 with level 16. 64969 stress test causes DB-LIBRARY error: Unexpected EOF from SQL Server 65049 After "sp_configure 'user connections', 0", "reconfigure" fails with msg 5808 which requires "with override" option for reconfigure. 66291 ANSI warning that function main does not have return statement in the module probe.c. 68692 we get Msg 5813 Unable to run with specified memory but sp_configure shows that the config_value has been updated. To duplicate: reconfigure memory to 15000 (ok); reconfigure procedure cache to 97 (get Msg 5813) after reconfigure with override; 68902 bufgrab() incorrectly checks the size of the mass it will return when deciding to update counters. The result is that prefetch counters are not updated properly when resized to 2K. 70419 Rework incorrect subquery error message. 71630 Message 3625 needs to be changed to print out stats for doubles. Currently we get garbage. 71817 In hsm_destroy_slice we are calling ind_chgvalue after endupdate. 71954 SQL Server is limited to 2GB disk partitions. Larger partitions dont generate errors. They just product corrupted data. 72740 we're seeing sql error msg 5879 and dblib error 20018 in t_driver output when running langinstall in intllang test but the test cases succeeded. 72772 The reconfigure statement in upgrade program is giving an error message 5879 "reconfigure cmd no longer needed" after config. checkin. This is causing the upgrade to fail when going from 492 to cougar. 72918 When configuration parameters cnblkmax(1000) is not the same as cnblkio(500), SQL Server timeslices during checkpoint with the following stack trace. kpcheckio()->dcheck()->basis_dpoll()- >basis_dqdelete. Sybase Confidential Page: 1 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 73083 cm_writedes waits after MAXWRITEDES writes have been started to avoid flooding the system with I/Os. The count indicating when it is time to wait is being reset to zero too often causing checkpoint to impact running transactions. 74457 If statement erroneously changed in bugfix for 74029 74638 Cast of argument to cm_cid_from_name in cfg_mgr.c is incorrect. Should be BYTE * instead of char *. causes compilation warnings on solaris. 75582 Query plan for nested scalar aggregates has changed. Returns incorrect result. For example, select min(max(id)) from sysobjects returns min(id), not max(id). 75695 disk_startup() in disk.c uses scerrlog() instead of an internationalized error message. 75792 Current checkpoint may issue far more i/o's per second than the System 10. This causes considerable dip in the transaction while doing checkpoint, during benchmarks. Need to study and work minimizing the impact of checkpoint. 75848 no state number for ex_raise()s in lddb_fixdbosuid() line 1308, 1365. not really needed since ex_control, but violates protocol. 75859 un-neccesary execution of instructions. no need to check for sdes. It is already checked in dbt_rekeep routine. This is the else condition. 76185 "deadlock retries" DEFAULT value should be 5 instead of 5000. Also, "deadlock retries" is too generic a name and does not convey the correct meaning of this parameter. 76266 Backout 76068. 76338 Rec_init: Found that page 378 had non-log object id 36 while checking Syslogs allocation. This error is occurs during recovery of the master database. server Cannot recover the master database exiting. ueshutdown exiting. 76438 SQL Debug Regression:The following SQL is OK in system 10 but kills dbprocess in system 11: use pubs2 GO create table tmp (name varchar(30) not null, phone varchar(20) not null) GO 76441 SSM 10.3 REGRESSION: Temp table trouble: create table #bob (f int, f1 char(5)) | create unique clustered index bobidx on #bob(f) with ignore_dup_key | insert into #bob values (1,'aaaaa') | insert into #bob values (2,'bbbbb') | insert #b Sybase Confidential Page: 2 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 76442 SSM 11.0 DEPENDENCY:Cannot get oldest active transaction to show up. to repro: In connection 1, send in "begin tran xx". 76454 Kernel scheduling procedure generated bad status. 76751 The config values are lost if the cfg file is accidently removed. Boot server with 3 engines. Shut down server. Remove the cfg file. Reboot server. Server generates a new cfg file with the "default" value. sp_configure shows 3 on "Config Value" and... 76887 The server allocates 6% of the memory as slop. On large memory systems, like alpha, this is a significant amount. The buffer manager calculates how much memory it will need, incorrectly so the slop is currently required to boot, but is not used once 76968 ubmcalloc() called with single chunk, failing to allocate the memory should release the kbspinlock. 77082 select avg(avg(i2)) returns wrong results. Should be 1 instead of NULL. Found by running MATS(big). 77090 Compile error because type def appears in src_dcl.h. 77127 compilation error: added a parameter in nullerr but did not make changes to manager/text.c where nullerr was also called. 77325 dbcc buffer should have the cache name as an optional parameter 77326 dbcc delbuffr should have the cache name as an optional parameter 77331 dbcc bufcount should not require a cachename 77332 dbcc show_bucket should not require a cachename 77572 sp_unbindcache_all on an offline DB produces 3602 error; also, a somewhat related bug in cfg__unbindall_handler caused by accessing the variable-length datstat2 field with GETSHORT. 77698 Dbcc extentdump does not check its parameters appropriately. If given a bogus dbid, it stack traces (db_virtmap() tries to find a row in sysusages for this dbid, raises a 908 error, stack trace). 77754 A cursor on a join is supposed to be closed when the join column is updated. The code in cursfix_cleanup() will close the cursor if the current row is deleted. In other words, we will close the cursor on a deferred mode update to the row, even if ... 77782 Need to extend dbcc page func. to search for a given page in: 1) all caches (to find duplicates of page in all caches, if any), or 2) a given cache (not necessarily cache where page should be found ). Sybase Confidential Page: 3 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 77862 Compiler warning/error when compiling .../sequencer/protect.c, because forward port of 72994 introduced calls to kcerrlog() and ucbacktrace() whose prototypes are in uksrc_dcl.h (which is not included). 78027 sqlstatus does not seem to be reset when you reopen the cursor again. is this a correct behavior? 78336 The vstart parameter should be properly initialized if a value is not supplied. 78382 The maximum number of "number of devices" should be '256'. 78383 increase the number of hash buffers (to the next power of two) 78385 set a status bit(ESTMT_CURSORSTMT) at compile time in E_STMT if the statement is a cursor statement -- this will eliminate the call to the CURSOR_STMT macro at execution time. 78405 when a user is aliased to dbo in a database and the user tries to truncate the table created by the dbo in that database, the user is denied permission with the message 4710: "Cannot truncate table %.*s because you are not the owner or a user with SA ... 78417 Since checkin of 74611, terminate_process() typically prints *two* identical stacktraces. ucbacktrace() call was copied, not moved, from bottom of a code block to the top. Must remove duplicate call at bottom of that block. 78456 Regression from bugfix 77095 #ifdef SANITY needed to be added to function call. 78619 bugfix 78339 may have caused a regression.(not sure yet!). Backing out this fix so that testing can continue. 78880 Type propagation to queries with GROUP BY ALL is not applied to PROJECTION action resulting in a type mismatch, possibly giving wrong answers 79231 compilation failure on solaris and hp due to bug fix 77159 79742 When view trees are re-normalized and re-written to disk, we don't save the value of Pss->prowcount before calling delproc(). As a result, Pss->prowcount gets reset, and if the view is in a trigger, you can get wrong value of @@rowcount. 79762 There was a merge error in the port for the fix to bug 78230 from the beta stabilization code line to the mainline, causing compile errors in analyse.c. Sybase Confidential Page: 4 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 80057 Parameter missing in call to parserr in pars.c 80311 Log records for a subscribable function may not be correctly replicated because it is not recognized as subscribable. 80719 Solaris-specific: test scripts which employ punctuation symbols for variable substitutions or negative tests can return "Msg 3626:17:1 The transaction was aborted because it used too much stack space, ..."; 81061 server reports that the system segment is full when it in fact has more than enough free space to store the object being created. occurred several times while trying to compile stored procedures. 81581 The beta2 refresh server core dump on signal 11 when running: select name from sysobjects group by name having id > 0 with trace flag 302 switched on. 81641 There is a problem when we combine a join on both tables' item_no columns using equals (=) and another join on both tables' item_type column using not equals (!=). 81645 server dumps stack trace when front end open client 10.0.2 is executing a dynamic sql stmt (ct_dynamic) stmt againing the beta 11.0 server. 82335 pars.c gets compiler warnings on Solaris. qblock->qaggfirst is no longer a TREE * but is being assigned into a TREE *. It is now a CAGHEAD *. Warning is issued a number of times. 82596 On AXPOSF, Solaris, & RS6000 - backup/inational/srv.log complains /.../locales/japanese/common.loc missing, on HP srv.log complains /svrhpdly_drel1/release//locales/japanese/roman8/common.loc missing. See: /svribmsdly_tst2/tst_aux/960103/srv.log; 83189 Need to upgrade the version number of the server and master device to 11.0.1. 83409 this is not a bug. customer gave me wrong info... 85645 The SQL server incorrectly formats the log record for an insert on a table with replicated text/image column when the insert requires implicit conversion (e.g., from a char to text) or the inserted value's length for text/image is exactly 256. 85649 In replicated table with text/image column and a nonclustered index, if the key is updated and then a writetext is done, the server produces a stack trace. Sybase Confidential Page: 5 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 85677 create view v1 (id, name) as select id, name from sysobjects where id <> 1 and name <> 'sysobjects' group by id, name; select * from v1; returns different results then select outside of the view. 85722 The fix for bug 82664 is incomplete, and can cause cache and scan problems. It fails to unkeep buffers, and does not reset the scan arguments before restarting the scan. 85911 rollback trigger recovery test suite (rbtrec) - stack trace occur when server doing an "insert" a row into a table which has triggers cross-database w/o checkpoint. 86039 Peoplesoft uses GROUP BY without referencing an aggregate in a view definition and SQL server drops the WHERE clause due to incorrect processing of the CREATE VIEW statement... so the view written to disk is incorrect. 87223 Use 'sybinit' to upgrade a 11.0.0 server to 11.0.1. This failed with the preupgrade error "DB-LIBRARY error: 20019 Attempt to initiate a new SQL Server operation with results pending." 84417 0 Regression from backport 70572 to 11.0.1 line. Caused stacktrace sequencer->s_normalize->tmp_create->selectinto in dropcol/dropcol002/dcol101b.sql. Can be reproduced with script /remote/qa3/bugs/84417.sql 57636 1 The buffer pages are not calculated correctly. We don't account for hash tables, dummy buffers, etc. 57637 1 The B1 memory allocation functions need to be declared correctly. 57957 1 Correct ANSI compilation errors, specifically the error handling routine that expects an integer but receives a character string. 58998 1 Alter database while extending the database beyond 64GB will result in incorrect number of GAM pages and possible segfaults. 59180 1 Backout changes introduced by ID 56228. 59735 1 Config values used in kernel initialization are not always reflected in Resource->rconfig which the server uses for runtime config values. 59736 1 The server can seg fault if reconfigure is run on a multi-engine server. 60262 1 The s10cougar merge compilation errors need to be fixed across all platforms. Sybase Confidential Page: 6 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 60523 1 Server Seg. faults in routine idt_burnsize_init() during server boot. This is a merge problem. The Global Pss is not initialized properly during the server boot causing the problem. 61693 1 Optimize table closure functions to only close tables that were open. 62849 1 "bldmaster.c" compiled with the message "command line 7: SANITY redefined. 63061 1 Delete the function prototye for cdefault_upd() in "crconstr.c" since it is defined in "src_dcl.h". 63510 1 Some files have compilation errors and warnings on NT platform, which caused by missing typedefs, unconsistant PROTO and etc. 63773 1 SQL Server stack traces, when a " reconfigure with override" is submitted. 64005 1 Compilation warning on bcopy(): "Incompatible pointer type assignment" and multiple definition of MAXNAME. 64147 1 The function prototype and declaration of "prSTAT()" are inconsistent. 64161 1 Sort order returned by sp_helpsort using ISO-8859-1 character set differs from the chgsort test suite CANON. 64229 1 Need to cleanup the compilation errors and warnings caused by UN- ANSI code. 64288 1 In qryproc/execute.c, the parameter in function call prSTAT() was cast into int32 instead of uint32. 64348 1 Msg 2557 contains some extraneous binary characters. Also, the object name in the msg is changed from b.t1 to t1 and checktable is changed to CHECKTABLE. 64511 1 Need to add optimization level 3 option to GCFLAGS. The published TPC-C results say that 10.0.3 will be compiled with SC 3.0.1 compiler with optimization level 3 enabled. 64693 1 Messages 567 and 570 are generated when a non-SSO role user tries to change password for another user. Only the 570 error message should be generated. 64732 1 Load transaction causes 605 error. 64821 1 load iso1bin.srt caused msg 5824, load cset011.srt caused DB-LIBRARY error: DBPROCESS is dead and stack trace in error log file Sybase Confidential Page: 7 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 64971 1 "charset.c" compilation failed: line 604.29: 1506-193 (E) Function call argument cannot be assigned to corresponding parameter. ANSI-C problem. 65281 1 Need to have exception.h in mmsanity.c since the PROTO of ex_raise() is declared in it. 65847 1 Master bug for backports from System10. 65972 1 Error 7750 is producing reference to a wrong argument number. For example : on line 1044, the reference to argument 1 should be argument 2. Does error 241 makes sense in conjunction with error 7750 ? 66035 1 A 605 error followed by a stack trace results from invoking data_check(). SQL command causing this error is "dbcc checktable ("USER_TBO.authtab_1")" 66301 1 The cache_pad utility doesn't correctly pad a structure if elements preceeding the START_CACHELINE/PAD_TO_CACHELINE macros can cause hidden compiler paddings (eg. short or BYTE). 66598 1 uppause() and kpwaitalarm() need to be cleaned up. i) uppause() error msgs are not internationalised. ii) uppause() is missing return stmts for some failure cases, could lead to uaremove() being called twice for the same alarm. 66682 1 dbcc checktable command is generating stack trace. It is happening in test auth001a of auth test suite. 66848 1 Compilation of server source files on Solaris 2.3 produces lots of errors and warnings. 66981 1 3 errors in compilation with optimization of server files and many warning messages from C compiler during compliation of SQL Server using new compiler 3.0.1. The causes of warning and error messages during compilation should be fixed or unoptimized. 67126 1 'sp_configure "memory", 8192', 'reconfigure' fails with Msg 5808, which requires "with override" option for reconfigure. 67133 1 ref man says sp_dboption requires no role if used with no parameters, yet code always requires dbo or sa_role 67192 1 Segmentation violation due to incorrect NT port. 67368 1 Sequencer test allows 'guest' user access to a table that 'guest' should not be able to access. Sybase Confidential Page: 8 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 67478 1 Segments test suite generates reproducible 605 error. 67485 1 DBCC's NOT_OWNER error message is returning garbage characters at the end of its second string argument. 67763 1 Adjust test filters to filter out uppercase. 67917 1 upgrade.h requires definition of TEN022. 68054 1 In calculating memory, the SQL Server must take into account the number of engines configured for. 68063 1 Error 706 (process trying to remove a PROC_HDR that it doesn't hold in it's PSS) encountered when running cursor test suite. 68140 1 After configuring access to a remote server and adding the "local" server entry in sysservers using sp_addserver, remote logins will fail with error 7221 unless the server is shutdown and restarted. Problem: 'remote access' config is static, not dynamic. 68180 1 SQL Server uses the wrong offset in determining which network information structure to use. 68276 1 seqncr suite shows various problems with the server. Look at diffs and output files. 68284 1 Text conversion internal test suite generates 208 error. 68291 1 The server segfaults in (proc__maptree+0x48c)when running a tpcc benchmark on solaris. 68358 1 Open this bug for tracking for duplicate bug 61622 for System11. 68394 1 Function argument casts are incorrect in exec_dbcc.c. BYTE * is not always the same as char *. 68408 1 privileges are not properly worker. User A (in a schema) creates tables and grants privileges to user B. User B runs a program that contains dml operations on user_a.table_name. A -208 error is returned. this also happens when cursors are involved. 68428 1 restorem (rmas001b) is causing server to produce error 834 followed by a stack trace. 68429 1 Message 913 to printing twice. 68434 1 When executing an unexisting procedure, the error message, msg 913 displayed twice. Sybase Confidential Page: 9 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 68452 1 d_usedextents() doesn't properly set the bypiece flag to zero because we added an arg. The array index is off by one. 68500 1 Error 5813(Unable to run with ...) occured when reconfigure with override, the original error message was 259 (Ad-hoc update ...). Need to increase memory size because bufmgr checked in. 68512 1 A server with TCPIP_TLI defined will fail to boot with more than one engine reporting "kbralloc: no more room in region." The allocated kernel space is overflowed and a segmentation violation occurs. 68533 1 Threshold does not fire. 68553 1 Version string of files in $SYBASE/locales files must match version of SQL Server. 68631 1 while running against diagserver (mirror cmds3 runs against diagserver), setup fails with message 5813. 68673 1 dbcc iosize doesnt properly enable large I/O on system tables 68689 1 SQL Server generating compile errors during internal testing of individual routines. 68718 1 If two views are created in different databases with the same name, and data is selected across databases through views, we get 225's in cougar, and *wrong* results in 10.0.1. We found that we are resolving views by name instead of id's 68818 1 master device is probably getting corrupted and running out of space. Many testsuites (rbtrig, rbtrec, s2srpc, dropcol:dcol009) are producing 1803 error. 68835 1 The default memory configuration must be increased thus new buildmasters must be created on all platforms. 68850 1 Duplicate line in source code should be eliminated. 68859 1 The SQL Server does not compile on Solaris 2.4. 68899 1 tran and union testsuites produce msg 5115 when run in testlist. They run fine individually. May be this related to bugid 68818. 68910 1 Some showplan will display IO size of 0 Kbytes. In reality, SQL Server should and will always use default 2 Kbyte. 68937 1 208 errors raised for tables that exist in the database. Error is not raised if table is fully qualified with database.owner.name. Occurs in SPeeD benchmarking DB-LIB program. Disappeared when dblibs updated. Sybase Confidential Page: 10 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 68947 1 Stack trace in amx test with NUMUSERS=100 & RUNTIME=30. "SQL causing error: EXECUTE sp_amx_delete4 ...." 69050 1 Trying to upgrade a SQL server( version SQL Server/10.0.1/P/Sun_svr4/OS 5.2/1/OPT/Tue Jan 25 17:13:25 PST 1994) to System11 comes out with 925 error, state 1, severity 16: Maximum number of used databases for each query has been exceeded. 69051 1 Ran into bug 68280 while running s2srpc test on solaris cougar server. Problem could not be reproduced when tried again. 69079 1 Logscan output for insert gives object name and owner name as NULL; logscan output for delete gives object name as null and owner name as garbage. 69086 1 If the log for the master database is full then the server will boot but not accept connections. Due to the checkpoint() added in dsinit() by the Cougar buffer manager. 69104 1 stack trace occurs when running orderby: coldstart->dsinit->usedb- >dbt_get->dbt__fill->des_sysbuild->des__bootsystables- >des__bootobjscan 69123 1 Internal SQL Server header file is missing some definitions. 69214 1 The following c files in generic/diagnostic have "function expects to return value" warnings: extedit.c, pgedit.c, and dxedit.c. 69215 1 The following c files in generic/diagnostic have "Function has no return statement" warnings: diag.c, extedit.c, pgedit.c, and dxedit.c. 69217 1 generic/diagnostic/diag.c has 3 "implicitly declaring function to return int" warning messages. 69218 1 generic/diagnostic/dxedit.c has 2 "argument #6 is incompatible with prototype" warning messages. 69239 1 The diag.std utility fails to compile on sun_svr4 and ibmaix4.1/smp. 69320 1 charset cp437 does not get installed. charset tries to load file charset.loc and gets message 510, state1, severity20. Bad eop type 0x0. It is executing select id from sysobjects where name='sp_configure'. May be related to 66818. 69326 1 Performance degradation is seen if MRU buffer replacement strategy is used and desired pages are already in cache. Pages get flushed and have to be read into memory again. Sybase Confidential Page: 11 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 69361 1 When upgrading the SQL Server from 1002 to System11, the upgrade100 script doesnot recognize the 11.0 server. It fails by saying SQL Server is already a 10.0.2 SQL Server and No upgrade is required. 69364 1 Both SQLServer and Backupserver are limited to 2GB disk partitions. Larger partitions don't generate errors they just produce corrupted data. 69365 1 dbcc checktable uses page IO exclusively. When scanning a table it should be using prefetch to reduce io. 69374 1 logrec fails in logrec008/logrec009 when DBDEV1 is a raw device. Errors seen are 605, 834 and 3501. Database is expected to be off- line due to faked device failure, failure doesn't occur as device is raw & t_driver doesn't delete raw devices. 69384 1 User testing, when loading a performance test database using the System11 Server running on Solaris Version 2.4, the server stacktraces after loading 40 warehouses worth of data. 69386 1 server stacktraced with error 629 severity 21 state 1 while trying to delete some rows from bench database after a failed performance test load. 69417 1 Error 1105 followed by stack trace in s2srpc testsuite. The testsuite runs ok by itself, but fails when preceeded by rbtrig. The error message indicates the system segment too full to include syslogs in master database. 69437 1 The floating point value '4.94065645841246544e-324' is out of the range of machine representation (8 bytes). See bug 38261 69456 1 Currently yacc and yaccpar are required to be in /usr/bin and /usr/lib area of a designated machine. Only CS can install them there. This makes it impractical to make any changes to these and test before installing. Solaris has option (-p) for yaccpar. 69472 1 sp_spaceused syslogs returns invalid values (all zerros) 69473 1 In house performance testing of the optimizer generated MRU cache stategy when it was not the best choice. 69486 1 The test autocomp failed due to the size of the procedure did not stays the same after it was being renormalized. 69509 1 The use of a parameter in the SET CURSOR ROWS statement causes the parser to stack trace. [Happens on System 10 as well] 69524 1 Version strings for $SYBASE/locales/ should match the version string for $SYBASE/bin/dataserver. Sybase Confidential Page: 12 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 69534 1 ad_hoc_audit() built-in function erroneously requires SSO role to be used. The built-in is only usable from sp_addauditrecord, which can be DAC protected, thus it should not require the SSO role. 69555 1 Error 3702(cannot drop the database because it is currently in use) occured when dropping a database after which has been used and referenced. 69569 1 In an unupgraded database, the CREATE TABLE command creates the table in master, no matter which database the command is issued from. Causes upgrade to fail; installmaster also fails complaining of 'too many databases in query'. 69578 1 Stack overflow - stack guardword corrupt error message, and stack trace generated -- possibly trying to abort a transaction that split a data page. 69587 1 Intermittent instances of "Maximum number of used databases for each query has been exceeded. The maximum allowed is 8." when doing db cleanup. 69593 1 Correct source function declaration to parse all input parameters. 69768 1 Sybinit for System11 fails to boot SQL Server for the preupgrade test. 69801 1 Remove source calls accessing incorrect memory locations. 69814 1 Correct static function definition. 69815 1 Internal tools used to monitor buffer cache performance give inconsistent results. 69824 1 Attempt to generate a_messages file for system stored procedure messages from msgdb yields DB-LIBRARY error: 20049 - Data-conversion resulted in overflow. 69828 1 Correct compilation errors. 69830 1 Error 706, state 2 when a deallocate cursor accesses free memory. 69900 1 dbcc checkobjcache gives wrong cache binding information, which shows that an object or index has been bound to a log-only cache. 69911 1 We are running out of bits in the rgstat2 field of the VRANGE. The VRANGE has to be reorganized with all non-disk resident bits and fields moved into a new structure which will be allocated at Sybase Confidential Page: 13 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 69921 1 Adjusting a cache size and its corresponding buffer wash size can cause SQL Server to hang. 69972 1 For 2-phase commit transaction, if one of sub-transaction fails during prepared stage, the coordinator will get aborted, but not the subordinate transaction. 69981 1 ddlstres test (ddls002, ddls003) sometime fail with error 8201 - Keep count of descriptor was expected to be 1. Instead 0 was found. The process that runs into this error is a user process though ocassionally the checkpoint process dies with this error. 70025 1 Correctly check for NULL return values in source files. 70041 1 sp_movemass stored procedure is missing in the installmaster. Also,make sp_movemass executable only by SA, 70044 1 Stack overflow & corrupted stack guardword. Stack says: kpinitproc- >ex_cleanup->conn_hdlr->ex_cleanup->sequencer->s_execute->ex_cleanup ->[repeats]->s_execute->execute->ex_subhandle->ex_testhandle- >ucstackcrash->ucbacktrace. 70046 1 When using diagserver, getting "Unable to run with specified memory size of 7500" on doing sp_configure commands. Also getting Msg 259:Ad-hoc updates to system catalogs not enabled.." Tests:cmds3/4 70049 1 insert query in qrmp (qrmp105b) is stack tracing the server with 11.The problem is easily reproducible. 70057 1 The OAM pages corresponding to sysattributes table have incorrect page status (0x8000). 70067 1 Internal buffer monitors are not maintaining accurate information. 70071 1 sp_addlogin produces "Message empty" string in test mbna002. 70097 1 No permission granting is done for sysattributes. It should have select granted to public. 70102 1 Dbtable exerciser stack traces after error 821/20/1. Trace:dbt__xrop ->dbt_dropdb->bufdbclean->cm_dbclean->bufdestroy->cm_mass_destroy- >cm_bufunhash->ex_raise. 70164 1 SQL Server gets infected with 11 followed by a crash while executing. Looks like some leftover from previous subquery in buffer causes this to happen. 70171 1 dbcc showcache(): buffer pool changes not reflected until server reboot. According to the Buffer Manager spec, buffer pool configuration is dynamic. Sybase Confidential Page: 14 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 70180 1 Error 403 when running NIST tests. 70213 1 The changes made by sp_movemess not reflected in sp_cacheconfig output 70216 1 Server does a panic shutdown due to stack overflow on sunsvr4 due to stack size limitations. 70233 1 sp_movemess loses track of moved memory. Memory moved for one pool to another, cannot be moved back. 70239 1 While upgrading from 492 to System11, the upgrade110 script fails during the booting of the new 11.0 SQL Server. upgrade log says "Upgrade cannot connect to 11.0 SQL Server". errorlog has "kdinit_mempool: mpm_initpool failed", 70241 1 sp_cacheconfig reports: "Configured buffer pools exceed total cache size" when a 512k pool is created in a 512k cache 70247 1 During the upgrade from 1002 to System11, the upgrade binary fails with SQL Server message 701 - "There is not enough procedure cache to run this procedure, trigger ...". This happens after upgrade110 boots the new 11.0 SQL Server. 70290 1 dbcc checkobjcache does not include db name for non-existing db in the error message: Can't find database id for . Check that this database exists and re-run the command. 70293 1 Sprocs do not run after loading db from dump. Also, if sproc is renormalized and written back to disk in 11.0, gets a stack trace on subsequent read from disk. 70302 1 Test altsort/alts002 has many diffs of type "*** Error in sorted result....". Also key values returned are different. 70319 1 Clean up some unused variables and other nastiness which cause the C compiler on NT to complain. 70326 1 Resolve compiler warning of an implicit truncation of an int32 to a BYTE. 70338 1 Dataserver gets a bus error while booting. 70356 1 des_sysbuild() should handle tables with object ids between MAXUSER and FIRSTMASTER more gracefully (ignore them?). Entries in sysobjects with such ids will cause segvs/bus errors as random areas within the dbt or Resource are overwritten. 70364 1 Integrate Buffer Manager and Config Interface. Sybase Confidential Page: 15 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 70373 1 Internal performance testing of queries generated incorrect cache strategy and a stack trace. 70407 1 Testcase is failing with the message: can't allocate space for object syslogs in master database because system segment is full 70428 1 1) The ex_raise of DSK_DUP_PHYSNAME (5151) in disk.c passes no parameters but the message expects two (device name and length). Potential for segmentation errors. 2) Also Checking on existing database device names is required and currently missing. 70461 1 d_checkobjcache shows the index bound to a user defined cache is bound to another UDC, if you bind a database to a userdefined cache, and a table in that database to another user defined cache where the index on that table is bound to. 70464 1 Checkpoint during performance test brings the xact rate to grinding halt. It used to work nicely without impacting the xacts much in SYSTEM 10 server. This behaviour is noticied even with single buffer cache in System11 environment while running tests. 70465 1 During the process of upgrade from 492 to System11, the upgrade binary fails while remapping objects in a user database. The server errorlog has a stack trace. Happens while remapping triggers, on the p_hdrcrt field for a trigger.) 70474 1 Improve performance of 16 table joins. 70476 1 Improve optimizer performance when calling cache manager. 70528 1 New 10.0.2.2. SQL Server messages are not taken into account in the Cougar Nightly buildmaster binary. Codeconv test suite failed to report the new message 9506 in code001/code101[a-e].out. see consults for details. 70530 1 SQL Server failed to return 'Empty message' when error/warning message cannot be found within master..sysmessages. Just a NEW LINE character is returned from the server without warning. It also lead to some garbage characters being displayed. 70534 1 Reduce TRACEPRINT statements that are longer than 80 characters. 70556 1 When insert select with order by does not require a work table, set showplan reports the creation of one anyway. We don't appear to create the work table so it is a bug with showplan. 70615 1 Use of force index name (optimizer hint) results in a stack trace: s_compile_stmt->s_optimizecmds->decision_all->opt_prep_plan_step- >qut_cp_constr->chunks->signal. Sybase Confidential Page: 16 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 70623 1 If a new name is used for Backup server name in the sql server config, the new name should appear in the system table 'sysservers'; otherwise you can't shutdown the backup server from isql. You have to kill the backup server by killing its process ID 70628 1 If the server crashes when commiting the drop of an object, it may not come up when rebooting. Msg "Could not open controlling database in multi-db transaction" is seen when trying to recover master. 70634 1 sproc/adminsysprocs leaves the thread in master. Unfortunately, new sprocs for System 11 have been added after this file in makefile, as a result procs like bindcache etc. are added to master which is very low in space. 70650 1 The support routines in the cache manager (used by the optimizer) that return info about caches should not acquire spinlocks since they are called very often from the optimizer. 70679 1 Error message 2622 generated during load of transaction log during internal testing. 70684 1 Join processor unnecessarily checks for subqueries, as subq count is non-zero, but really there are no subqueries in ROOT/AGHEAD. Subquery count should be set to 0, when there are no subqueries. Subquery list should have non-NULL ptrs. at its head to ... 70705 1 Generate 813 errors when appropriate during page allocation. 70717 1 Signal 11 occurs when trying to configure 5000 user connections on Solaris 2.4. 70747 1 Stack Trace during amx test suite. This happened because the server booted up against existing devices and later 'delete $(DBDEV1)' in cmds.run cleans out one of the devices resulting in I/O Error. 70762 1 Need to added SYBCACHELINEALIGN_SERVER & SYBCACHELINEALIGN_KERNEL for cache pad strategy support. 70764 1 Error 834/20/1 during GROUP BY optimization with SANITY turned on. 70809 1 Assertion failures occur in subquery transformations code while running new subquery tests: sb504/sb54115.sql; sbadhoc/inlist1.sql. (Subq is being visited more than once for (subq) IN (value list) construct). Fix code, or fix assertion. 70820 1 SQL Server exits after printing first two lines of the startup messages when rebooted after 'sp_configure user connections, 1000', 'reconfigure with override'. No errors reported during reconfigure. Sybase Confidential Page: 17 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 70834 1 Correct ordering of error message input parameters. 70845 1 Diagnostic trace flag (CKPT,0) is not implemented in the SQL Server. Trace.h indicates it should disable checkpoints. At least one test (rbtrec) assumes that it's is implemented. 70854 1 Test suite on Solaris stack traces with 5 clients: Spinlock Resource ->rtablockspins[] info. seems to be inconsistent. Segfault at: lock__ins_logical(). Results in time slice error. Time slice was at default value of 100. Tests seem to run with. 70858 1 Correct MACRO usage that will context switch. 70864 1 Backport 64611 and 64615 for 1127 errors. 70896 1 Create a view in 1002. Upgrade to 11.0. Drop table on which view is based. Select from view. (View gets redefined, and written back to disk). First select works fine. Shutdown &restart server (so that view is read back from disk). On subsequent selects.. 70897 1 Segmentation violation when executing dbcc iosize(database_name, table_name). 70900 1 Trace output from 280 (to print query trees during transformations) does not print the correct tree if any subquery gets materialized. It only prints the last CMD subtree that is left in the tree. It should print the entire CMD list created after xforms. 70902 1 Segmentation violation when executing dbcc checkobjcache(dbname, objid, index_name). 70909 1 dbcc tune("bufwashsize", ... causes a time slice error. stack trace is kimemrange->conn_hdlr->sequencer->s_execute->exec_dbcc->d_tune- >cm_cid_from_name->_fini->terminate_process->ucbacktrace 70910 1 objid is replaced with the dbid in dbcc checkobjcache routine, if you choose to use dbcc checkobjcache(dbid, objid) format. 70913 1 Stacktrace during startup. Varies from startup to startup. All appear to be lock related. 70914 1 msg 1105, level 17, state 1, SERVER ERROR: "Can't allocate space for object 'sysprocedures' in database 'master' because the 'system' segment is full ... msg 2811, level 16, state 2, INTERNAL ERROR: "Cannot create procedure dbid 1, objid ... 70957 1 A cursor on a select with a subquery should not be updatable. 70958 1 declare cursor for update on a view consisting of subquery does not produce an error any more. The curs (curs012) is producing this regression. Sybase Confidential Page: 18 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 70967 1 When you bind a table in a database to a user defined cache, all the other tables in that databases are also bound to the same user defined cache. 70968 1 Stack trace Error: 1142, Severity: 22, State: 1, Invalid OAM Page 0. Found pstat=0,objid=0,indid=0... terminate_process, ... close_network, _coldstart, ... 70970 1 CT-Lib appln. with dynamic cursors returns wrong # of result rows. Problem (mysteriously!!) fixed in Cougar line but still exists in System 10 codeline. 70976 1 Reorder subquery cache optimization. 71004 1 When running access methods exerciser with multiple caches, a page is read into the wrong cache. 71022 1 Incorrect error message when declaring a cursor on a select for browse. 71092 1 Merger error by dirty reads project in config.h. The function declaration for initstruct() has returned from the dead. initstruct() doesn't exist anymore. 71116 1 Internal test pattern2/pat2003, and stakprot tests fail due to insufficient stack space. Also affected testsuites are codeconv, altdate, trgssprt, arith, sdblrpc, waitfor. 71146 1 Internal diagnostic routine test call in exerciser manager gives a seg fault if just the exerciser name is provided as parameter. Some exercisers (eg: buf mgr exerciser) pass just the exerciser name in dbcc exer_init_exer(). 71150 1 The new function for text replication txtimg_log_datarow() has no error checking when writing log records. Serious problems will probably occur if the log is full. 71160 1 Many unreferenced variables after dirty reads checkin. dbccdiag.c(1782) tabrow dbccdiag.c(1782) tabsdes dbccdiag.c(1782) tabptr ltintfc.c(4092) lock_status ltintfc.c(4552) obj_rep_stat ltsendfn.c(3036) log2 desmgr.c(4731) des_ident desmgr.c(6221) 71172 1 curs_updatable() - Walking of range tables only necessary if cursor is not already determined to be read only. We should also skip any null range entries. 71211 1 sp_movemass doesn't check to see whether the size of the destination pool is a power of 2, resulting in the return of a buffer pool as the destination pool. Sybase Confidential Page: 19 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 71212 1 Assertion violation raised while running aggdist suite aggd006.sql. This comes from subquery transformations. The subquery count appears as 0, but there is a non-NULL entry in subquery list off some ROOT node. 71215 1 Correct source conditional for iosize. 71216 1 dbcc iosize(dbname) and dbcc iosize(dbname, table_name) cause stack traces. 71231 1 When upgrade is run in qrmp/qrmpupgd test suite (upgrade from 10.0.1 to 11.0) without a checkpoint statement as the last statement in qrmp099b.sql we get the error message "Cannot upgrade pre-4.0 server" 71249 1 Incorporate the following error messages into sysmessages: 1729, 1730, and 7379. 71251 1 "sp_help sysconfigures" displays "varchar" as the "Type" of the column "status". This is incorrect since the "Type" should be "smallint". 71252 1 The asm function sparclock() does not have a 'nm type'. This makes it invisible to disasemblers and nm. Results in any seg fault in the spinlock code ... 71260 1 The version string in $SYBASE/locales/ for System11 should be updated from 10.0.2 to 11.0.0. This was done for other platforms, but not for HP. 71266 1 The minimum stack size (20480) is too small for the query that our test driver uses to verify database recovery. 71270 1 In sql.c, qryflags is declared as an unsigned long. A pointer to it is pass to functions which expect a uint32. An unsigned long is not the same as an uint32. 71271 1 stack trace seen in benchmark. signal 11. doing an update. stack trace is mangled but it appears to be in run(). 71284 1 set statistics subquerycache on, reports wrong values when subquery appears in a union. 71299 1 Review CAST assignments where a short is being assigned into a BYTE. 71301 1 Optimize cursors that have position changes as a result of any update/delete/inserts. 71302 1 Review CAST assignments of a short that is being assigned to a BYTE. Sybase Confidential Page: 20 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 71304 1 In the fix for bug 66598 some casts were taken out of the assignment on line 448 of proc.c. If the intention was to make the operations to be non-floating point then the constants need to be integral too. This line generates a compiler warning on NT. 71305 1 Msg 1243, Level 26, State 1: Couldn't find object to be released. DB -Library: DBPROCESS is dead or not enabled during access methods execiser testing. 71311 1 Msg 3626, level 17, state 1: The transaction was aborted because it used too much stack space... for more info, see /remote/spgqa/qa_110x/docs/QA_Failure_Rpts/Nightly/71311.arith 71315 1 Isolevel 0 scan forced to do a table scan will fail because we haven't allocated any space to save keys. 71330 1 The query type for deferred processing step for IEF is one of INSERT DELETE or UPDATE, but the e_streadvar and e_stresvar is set to -1. It should be a "real" value (in range of (0,15)). execute() code tries to access read and resvar SDES from Pss, and ... 71335 1 Compile warnings on NT after checkin of SLICES and friends. insert.c (1452) off, rowptr; checktable.c (783) slice; pagemgr.c (3222) oamcontext; sysslicesmgr.c (661) shortp [unreferenced variables] pagediag.c (1428) type mismatch 71342 1 select distinct c2, c3 from t for browse does not return distinct rows. There is a unique index on t(c1), which causes the distinct to be eliminated. s_addident() adds c1 to select list to make it unique. Distinct elimination logic should not consider ... 71387 1 Server produces stack trace while compiling a SQL statement having the 'force index' syntax (INDEX ). The pre-System11 style of force index () works. 71389 1 SQL Server performance testing generated incorrect index selection for some queries. Problem is in costing routine. 71390 1 Stack Trace found during in house testing. The STRAT_WriteTest client application of the "Stress Test" appears to cause the stack trace.> 71395 1 In house performance testing or optimizer selects inappropriate cache strategy for query involving a three table join. 71420 1 Make dbcc checktable() update extent OAM pages on validated objects. 71423 1 During page or extent deallocations, extent OAM page should be set if not already set. Sybase Confidential Page: 21 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 71442 1 create table notexists..#t1 (f1 int), where db notexists does not exist. Server will first print out "Database name 'notexists' ignored, creating table in Tempdb. It will then print out 2 instances of 258/20/2 - Database name 'notexists' missing during 71465 1 open_user_table() calls hsm_free_slgroup() without knowing the SLGROUP is in use. Causes seg fault and possible database corruption. 71466 1 Error 1621 generated during in house testing. This problem is induced by the 'Insert Test'. 71470 1 "usemigrate" error encountered during in house testing. This problem is induced by the "Insert Test" - usmigrate: the descriptor is not a read/write descriptor - cannot be migrated. 71475 1 Reorder function calls to log changed values. 71476 1 Booting SQL Server with the '-p' option gives the following warning in the errorlog: Warning: dataserver -p option fails. Please check for state '1' to see where it fails. Error: 1623. What is state '1'? 71508 1 Ensure current functions work with the new buffer management scheme. 71509 1 Slicing is disallowed on pre-System 11 databases. However, there is no easy way to check if the db is pre-System 11 until dump compatibility is done. 71520 1 Cursor declared with UPDATE clause on Distinct view appears to be updatable if the view qualifies for distinct elimination logic. In 10, you will get error 7725, but in 11, the declare goes through. (Update where current of anyway fails, as view is ... 71524 1 The calls to ncdelete and ncupdate in update() should be made after checking that there is atleast one unsafe index. 71536 1 DBCC after load db reveals table corruption: "Number of data pages counted by DBCC differs from count in the data OAM", "Msg 2525...Table Corrupt: Object id wrong... Page linkage failed for OBJID 5 INDID 0 ... Found 1 bad entry inthe sysindexes" 71558 1 SERVER ERROR: Issued "shutdown" to server 'qa71_svrsolst' but it is still up, after this sybinit configuration fails. 71566 1 Illegal row length during isolation level testing. 71614 1 Bus error in tdsrecv_language() during access methods execisor runs. Stack trace was _coldstart->_sigtramp->kisignal->terminate_process- >ucbacktrace. Sybase Confidential Page: 22 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 71615 1 Reserve ample space for OAM attributes in the OAM page. 71633 1 An exception must be raised if allocation hint space can't be created during create table or index command. 71643 1 Close the slice before raising the LOCK_INTERUPTED exception. 71649 1 Extend lock escalation threshold mechanism to support server-wide percentage scheme. 71655 1 dbcc listoam(1) seg faults - should give wrong params error msg. Stack trace:sequencer->s_execute->exec_dbcc->d_listoam->pg_list_oam- >open_user_table->open_byname->etext->_sigtramp. Also, an invalid index parameter leaves the table open. clean-up backout. 71657 1 Insert into a temp table stack traces: sequencer -> execproc -> s_execute -> s_mustrecompile -> seq_schema_check. 71663 1 Improperly set page status bit on pages with all fixed length rows is causing unnecessary page content searching. 71684 1 In the structure definition for the table "syscacheconfig", the declarations for the sensitivity and information column are missing. 71693 1 Query tree remapping is broken. dbcc remap() called to do Q_REMAP (option: 1) on a tree created in 10.0 fails in Cougar. Stack looks like: exec_dbcc->d_remap->qremap_main->qremap_prochdr-> qremap__tree. Problem is that in qremap__procread() we do not ... 71703 1 8402/21/3 error: Index not found: seq_schema_check() looks at the varct in the main estep (i.e. the first one in the current side of the UNION). Incorrect for deferred check case for RI where other user tables are added and show up in the varct of .. 71706 1 Clients are unknowningly killed by 4 engine server (NETMIGRATE feature turned on) when multiple clients are connecting and disconnecting simultaneously. 71722 1 pss_get() will return an invalid pointer if the passed in spid is beyond the range of valid spids. If spid is out of range (related to # of user connections) pss_get9) should return NULL. 71726 1 In cm_bind_cache(), while printing error 862 we are passing the 'dbid' where as the message doesn't seem to use the dbid at all. It would be useful to include the dbid in the printed message. 71727 1 In cm_grabmem() we raise 863 error when we run out of buffer resource during bufgrab(). When we raise the error the intention is to ask the user to reconfigure the buffer cache or rerun the query at a later time, but not aborting the user session itself. Sybase Confidential Page: 23 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 71744 1 SYB_RESOURCE structure is not being padded correctly on the diagservers for IBM and Solaris at least. 71753 1 lots of compilation warnings ... dbcc/dbccdiag.c, dbcc/checkalloc.c, access/xls_xxx.c, access/search.c ... the list is endless. 71765 1 If the scan index is unique and safe and a unique nonclustered index is unsafe, then the index update should be done in a deferred mode rather than in a direct mode. 71772 1 Process infected with SIG 11 when droping an index that is bound to a cache. Stack is: des_getval->tally_row->endupdate-> attr_delete_cache_binding->dropi->dropindex... 71774 1 In house performance testing routine generated error 1265 and a stack trace. 71789 1 The comments for hsm_alloc_page in utils/heapslices.c do not match the code. This occurred after the PLC checkin. The comments should reflect the code. 71801 1 Server error 8211: Mismatch found between the name and id descriptor hash table; stack trace: des__hashinst-> des__install-> des__fill-> des_get-> open__byid-> open_user_table... 71819 1 The optimizer should not choose a prefetch strategy if a non- clustered index is chosen and the query is not covered. 71833 1 Check page status after validating the page. 71835 1 Could not find row in systhresholds for database tpcc describing segment 4, free space 23288. FOLLOWED BY Failed to execute the threshold procedure for database 5, segment 4, free space 23288. 71847 1 The dbt_seqspin lock which is used to get the sequence number for buffers is initialized during boot to have SERVER_MAX_LEVEL. The level for this one should be reduced to SERVER_LEVEL_1 like all other server spinlocks. 71848 1 The test sysattr hits a stack trace in syat004.sql with stack trace saying kisignal: SQL Server encountered an invalid address (SIGSEGV), when running exec sp_unbindcache syat_db1. ksprsocktab -> Lku -> _tls_end 71851 1 When sp_unbindcache & sp_unbindcache_all are used to unbind a object from a cache, it returned misleading message and the object was locked up from any other transaction and exited with stack trace. 71857 1 [REPSRV] The log record for a writetext on a replicated text/image column is not correctly transferred by "dbcc logtransfer"; the log record appears to be correctly written to syslogs, however. Sybase Confidential Page: 24 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 71865 1 In fill_sysdepends(), getdbid() is called for sybsystemprocs, without checking the return value. If installmaster has not been run, sybsystemprocs does not exist and you get a garbage dbid. This leads to error 913. 71890 1 Compiler warnings on sun4/svr4 in access/search.c 71891 1 Compiler warnings on sun4/svr4 in bulk.h, checkalloc.c, dbccdiag.c and drop.c. 71892 1 Compiler warnings on sun4/svr4 in conn_hdlr.c. 71893 1 Compiler warnings on sun4/svr4 in lockmgr.c. 71911 1 Load transaction on failure doesn't produce "x transactions rolled back" message. 71937 1 Internal engineering diagnostic routines cause the server to die with segmentation fault. 71942 1 [REPSRV] DUMP DATABASE fails when the LTM truncation point is valid because replication requires the logging of an XSYNC. The logging og the XSYNC fails becuase the XLS session was opened as READONLY. 71947 1 Sybinit test, suite part siupgd5 failed to boot the new 11.00 SQL Server after PHASE 1 completes. This also occurred on RS6000, HP. 71961 1 Resolved compilation error. 72003 1 System 10 database need to be upgraded to create allocation hints space both in table and indices. 72016 1 The buffer manager error messages should be augmented to print the name of the cache where the offending buffer resides. 72023 1 In login(), PGLOCK bit has to be set for write_loginsdes before calling update(). 72035 1 sp_help table_name inside a transaction: Msg 226, SELECT INTO not allowed within a multi-statement transaction. Server terminates process.sp_help needs transaction test, Msg 17260 near start. 72056 1 Drop index calls curs_open_ind() to see if the user has any open cursors using the index being dropped. The code does not check to verify that the SDES is open before dereferencing sdes->sdesp->... 72071 1 Internal query map testing returns incorrect number of rows in result set. Sybase Confidential Page: 25 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 72119 1 xls.h needs to include a function prototype for xlm_ispermanentmarker(). 72143 1 There is an incompatibility in the name of monitor counters as specified in the functional spec to the names in the code. Also, couple of new monitor counters needs to be added to assist in configuring the server as specified in functional spec 72150 1 Several places in n_socket.c, n_tcp.c, n_tli.c and strmio.c are accessing npendingreads, nibardasockets, snrmask, snsmask, snrhdr, snshdr, snchdr and snmhdr etc. fields without the snspinlock held. 72182 1 ct_cursor(DELTE) on updated row did not generate expected error message from SERVER (Expected: The UPDATE?DELETE WHERE CURRENT OF to the cursor ... failed because the row has changed since it was fetched). 72183 1 Placing insert into ,tbname. values (:) and delete from where = : in one batch in esql/c program causes 3801 error: there is no host variable corresponding... 72201 1 s_execute() and tdssend_curinfo() can access freed memory on a deallocate cursor statement. The proc_hdr for the css can be freed on the call to css_deallocate(), but the css is passed to tdssend_curinfo(). 72204 1 sockp->soflags is modified in ksmaskspost() and ksmaskcpost() in n_tcp.c without holding the correct spinlock (Kernel- >kssocktab_spinlock). 72205 1 Unclear why affinity must be pushed to CONN_ENGINE during login sequence. Comment in conn_hdlr() seems to contradict prologue for usnomigration(). 72207 1 fast bcp always inserts into the first partition. bulk_fast() in utils/bulk.c directly inserts rows into the current last page of the first partition. A partition must be chosen for proper distribution across all partitions. 72209 1 After changing sortorder and default charset id thru' sp_configure, server hangs when brought up in single user mode to reconfigure and shutdown. Many test suites are affected including chgsort, pattern2, strfncs2, altsort, codeconv, etc. 72213 1 The sp_cachestrategy stored procedure is not working for prefetch and buffer replacement strategies. 72214 1 sp_cachestrategy does not parse "owner.object_name" or "db.owner.obj ect_name" parameter correctly, reports "There is no table named owner in the current database." Sybase Confidential Page: 26 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 72216 1 dropcol/dcol007: space on log segment fallen critically low in database dcol_db05 72231 1 Stack trace when sql server executed "initcntl.sql". Failed to commit transaction. OS signal 11 (seg fault) starting at xls_getnext and siginfo(code, address) = (1,0x0) unallocated memory. 72244 1 During upgrade from 100x to 11.0, the preupgrade doesnt check the space in all the existing dbs(master, model, user dbs, sybsystemprocs etc), which it does when upgrading from 49x to 11.0 72245 1 Configuration feature checkin discarded a fix for 70216 made on June 15th by not merging correctly with sun_svr4/cinclude/port.h 72248 1 need full support for cache configuration in the configuration manager. 72251 1 The data prefetch for text data is not working. sp_cachestrategy does not have effect on text retrieval. 72254 1 DIRTY_READS() expands each of its arguments several times and silently uses Pss. This is not good for performance, esp when args are of the form a->b or Pss->a->b. 72255 1 1) Server won't boot on hp800 with 1002 master device. 2) Fix warning on line 1262 in "sysrrows.c". 72256 1 Segfault while booting a server with a db in bypass recovery mode. 72260 1 signal 11 when show plan is ON. Create a table t1 (c1 int, c2 int unique), t2 (c1 int references t1(c2)). insert values (1,1) (2,2) in t1 and (2) in t2. Now open an update cursor on t1 and modify c2 from 2 to 3, it stack trace if show plan is on 72262 1 Make the changes required to support configuration parameter "page utilization". 72264 1 Fix compilation errors. 72265 1 When upgrading to cougar thru sybinit, it gives a SQL server msg 259 INTERNAL ERROR: "Ad-hoc updates to system catalogs not enabled. and also msg 2812 INTERNAL ERROR: "Stored procedure 'sp_instmsg' not found." 72268 1 Add configuration options to the configuration manager. 72269 1 entering 256 for 'number of devices' results in a negative 'user connection' setting. Sybase Confidential Page: 27 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 72272 1 CM_ALLOCPG_IN_RANGE macro in cachemgr.h can deny a 16K I/O on last extent of an allocation unit. The calculation in the macro need to adjust for the starting page also. 72291 1 dbcc dbrecover needs the changes required to off-line and on-line databases, because it could invoke recovery. Also, rebuildextents needs to be able to 'use' a database that is off-line since its also called from the diagnostic menu. 72292 1 Increase the file descriptor limit to 1024. Issue sp_configure "user connections", 1000. This returns with an error message Verification failed for parameter 'number of user connections allowed'. Expect "insufficient memory" as message. 72306 1 In ALTER TABLE, several messages regarding improper user DDL against partioned tables raise EX_CMDFATAL (severity 20) when they should raise EX_USER(severity 16). Severity 20 suggests an internal SQL Server error. 72324 1 MNE performance: Post login connection migration code is not MP safe, unused (apart from dbcc) and slows down the server. Recommend #ifdef'ing it out until it is fixed. 72327 1 Stored proc that uses 'like' clause will fail to use index on table if we are comparing a col (via like) with a parameter passed into the procedure. 72339 1 The routines cm_coalesce and cm_split need some fixes. They assume that a new pool is created. This is not always true. They (1)set a new value for the wash deficit and (2)do not update the wash size. 72341 1 css_find() will be called for each non-DECLARE cursor statement to find the matching css. This work can be done at compile time for most server cursors. 72342 1 sp_configure "maximum engines possible on-line", 4 does not appear work. A 586 message is returned saying that 4 is an illegal value. The NT machine this is being run on has 4 cpus. The message states that the legal values are between '0' and '0'. 72343 1 curs_validupd() is called at execution time for all positioned udpates and deletes. The check can be done at compile time for most types of server cursors. 72349 1 Error message 2007 encountered while running installmaster script. sp_configure is referring to itself recursively. This prevents users from installing/upgrading to Cougar server. 72352 1 Error 208 encountered while running installmaster script. It complains that master.dbo.syscacheconfig does not exist for sp_cacheconfig. Sybase Confidential Page: 28 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 72353 1 Tests on mirrored devices fail in some instances. 72372 1 Setting stack size to MAXINT value using sp_configure results in wrong value being taken. sp_configure "stack size", 2147483647 72402 1 ckpt_active_log_start calls xls_getnext with lohp=0 causing a segfault. stack trace is s_execute -> _ex_cleanup -> dmpx__truncatemain-> dmpx__find_truncpage -> ckpt_active_log_start - > xls_getnext -> logx_xlate_xrec. Found while testing dump- compatibility 72450 1 "Update where current of" generates a stacktrace: sequencer -> s_compile -> s_compile_stmt. Update references a distinct view in a subquery in the set clause. Distinct view gets materialized, but does not have R3T_WASINSUBQ bit. Code is trying to get the 72455 1 The function opencheck() [called from conn_hdlr()] causes a significant number of cpu stalls. This seriously affects MP scaling. Stalls are due to touching sdes->sdesp which brings unused SDES into cache from memory or another cpu's cache. 72465 1 When the config file contains an illegal value of anything greater than the number of available processors, the error message indicate the legal value range is between '1' and '1' when in fact the number of available processors may be greater. 72466 1 installmaster is producing errors. I built a master device, booted a server and ran 'installmaster'. Errors relating to sp_configure and sp_cacheconfig (bugid 70174) are encountered 72475 1 bug fix 71272 breaks the new sp_configure. subparams only check the parameter list for server cursors. need to check variable list as well. pull in 69877 instead. 72495 1 sp_configure fails with message: Configuration error: cannot write out file due to system error .. invalid pointer param no.2. See config test suite, conf009.out for examples. 72506 1 The housekeeper configuration parameter should be changed to use the new configuration interface. 72515 1 cfgdev has DIFF file which related to "number of devices" - i.e. confifuraton. Detail is located at" svrsold:/svrsoldl_dtst1/tst_sql/cfgdev/950626/ 72526 1 isolevel/restart:Msg 614,Level 21, Stat 1: Line 83 : A row on page 497 was accessed that has an illegal length of 5632 in database iso_db01. The SQL server is terminating this process. For detail see svrsoldl:/svrsoldl_dtst/tst_sql/isolevel/restart/950627 Sybase Confidential Page: 29 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 72527 1 isolevel/restart/isol045a: error 614 (illegal row length). Stack: exec_eop->getnext->rowsize->ex_raise. 72539 1 connect: USer 11 connects to ser successfully. This is a bug user 11 should NOT be able to connect to server. For detail plz see svrsoldl:/svrsoldl_dtst1/tst_sql/connect/950627/* 72542 1 The special databases (model, sybsystemprocs, sybsecurity) should not be on-line'd if they are marked to be in bypass-recovery. 72544 1 The config parameter 'enable debug engine (T1606)' must be removed. Having to document this would open up a nasty can of worms. 72546 1 Read only configuration options should not be written out to the config file. It makes no sense to have them in the config file. 72549 1 fptest: Floating point value '4.940656458414246544e-324 is out of range of machine representation(8 bytes) 72550 1 Preprocessor directives #ifdef B1 should be replaced with #if B1. sequencer/res_view.c lin 3045, useful/lookup.c lines 399,1665,1971. 72552 1 Replicated databases that require upgrade must not come online after recovery until Replication Server has drained the log. Add logic to check replication status, and mark those databases to come online automatically when replication completes. 72567 1 logrec: CANON needs to be uopdated. For detail plz see svrsoldl:/svrsoldl_dtst1/tst_sql/950625/logrec/* 72580 1 mirror/cmds8: diff file - can't open /dev/rdsk/cot1d0s5 vs No differences between the two devices. 72600 1 ddlstres: Multisql failed - Fatal error from server. See svrsoldl:/svrsoldl_dtst1/tst_sql/reruns/kyt/950627/ddlstres/DIFF/ 72627 1 fetch from cursor which is defined as select with order by causes stack trace (openxdes->begin__xact->soextalloc->addmerge->startsort- >exec_eop...) 72638 1 Add support for Display Levels in "sp_configure" and add new stored procedure "sp_displaylevel". Also, add support for displaying a parameter in each group where the parameter belongs to. This also includes the fix for bug 73889. 72639 1 Accumulating spinlock statistics (nump's, spins and waits ) by default, impacts SQL Server performance. Gathering Spinlock statistics should be turned on/off by a global flag. Sybase Confidential Page: 30 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 72641 1 Small cache configurations will cause buffer pools to be created incorrectly. 72648 1 The initial message produced by each internal upgrade step is not very customer-friendly. Change it to have less internal information and more customer-oriented text. 72651 1 Code cleanup: delete the definitons for configuration numbers since they are defined in 'cfg_def.h' instead. 72657 1 "buildmaster -r" should no longer needed and supported. As a result, the server should be changed to boot up successfully when the checksum is invalid. 72677 1 sp_configure "default sortorder" sets sortorder to wrong values that don't exist in charsets. Should give error when (after buildmaster, server is booted first time) setting to 0 or 256 or to 0, "default" as "default" is not a valid char set id. sp_config 72686 1 Function getrow() has a malformed ex_raise() call that concatenates the major and minor parts of the error number. This causes 'error 62421, severity 1' to be raised instead of 'error 624, severity 21'. 72689 1 The DES keep count that gets incremented in des_verify() routine should be done after testing if the DES is in an ACTIVE state and all of this under spinlock synchronization. This is to avoid incrementing the keep count of a 'des' being scavenged ! (..) 72692 1 The minimum for 'average time slice per process in milliseconds' should be lower, probably 1 ms. 72708 1 The Resource->rexer_ptr is being set to NULL in dsinit() inadvertently. It is being set when the symbol EXERCISOR is not defined. The symbol should be EXERCISER and not EXERCISOR! 72725 1 sp_configure "pre-read", 40000 gives msg: Illegal value '-25536' specified...(it gives 65535 minus (value supplied)). Any value above 32760 (or so) gets error, same for 'remote conn', 'remote logins', 'remote sites'. 72727 1 In accept/mirr006[f,g], test is diffing due to the master device changing, where as before nothing changed. 72729 1 Setting 'minimum engines required on-line' to 2 gets error msg: Illegal value 2 specified, legal values are between 1 and 4. If setting to 0 also gets error: legal values are between 1 and 32. 72736 1 950705 build stack trace dump. Running grgr004d.sql which generated 7 output files g[1-4, 6-7,9]004d.out files. In g1gr004d.out, it stated: The SQL Server is terminating this process. driver output & err log & related info is located at Sybase Confidential Page: 31 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 72764 1 Performance issues with bug fix 69901. This fix will cause an unnecessary call to upsleepgeneric in bufnewpage. 72765 1 When SQL executed " t grant permission" ""WITH GRANT OPTION" causing stack trace at "do_row" because of some memory allocation errors. kbdmemmap->connhdlr->sequencer->s_execute->stopxact->abort_all->undo ->do_row->_fini->kisignal->terminate_process 72780 1 s_chooselocks() only locks tables found in the main/subordinate E_STEP chain. The subquery project introduced subordinate query blocks, off of subquery nodes. These are not being checked for tables to lock. Tables only in subqueries don't get locked. 72785 1 Device compare between database device and mirror device shows diferrences between the two devices when none are expected. See cecil:/cecil1/testcleanup/nair/mirror/cmds8/DIFF/devc25.out for differences in the devices. 72789 1 Non-ANSI query causes timeslice when select i.id from sysobjects o, sysindexes i where i.id = o.id group by i.id having sum(i.id) = max(sum(o.id)). Problem occurs only if you have a nested aggregate in having clause. 72803 1 mcsp/clpv caused stack dump.SQL causing err:A. Valid table name only:'dbo_colpriv' exec sp_column_privileges 'dbo_colpriv'.sequencer>s_execute>execproc>sequencer>s_execute>execp roc>sequencer>s_compile>s_compile_stmt>s_optimizecmds>optimize_cmd>d ecision> 72822 1 Unexpected EOF from the SQL server during access method exerciser testing. 72826 1 Executing stored procedures 'sp_table_privileges' and 'sp_column_privileges' stack traces in sequencer->s_compile- >s_compile_stmt->s_optimizecmds->optimize_cmd->decision_all- >decision->opt_order->ord_needed->opt_considerFilter->mkj_cache_used 72827 1 Deadlocks encountered during access methods exerciser testing. 72828 1 Deadlocks encountered during access methods exerciser testing. 72851 1 Error 913 "Could not find row in sysdatabases with database id 0" and 8414 "Copy of sysindexes row failed". Stack: sequencer- >s_compile->s_compile_stmt->s_optimizecmds->optimize_cmd- >decision_all->decision->opt_order->ord_needed... 72881 1 The config parameter "engine adjust interval" and "current number online engines" should be hidden. i.e. they should not be diplayed or updated via "sp_configure". Sybase Confidential Page: 32 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 72885 1 Optimizer chooses the wrong join order in the inner block of a correlated subquery. Engineering has observed that there is an error in the estimation of join selectivity. 72887 1 mirror/cmds4/mirr006u.sql - "execute syscheck" retuns different value comparing to CANON. 950709's mirr006.sql returns:" mirroring disable on secondary"; CANON has "mirroring disabled on primary" 72898 1 Cannot force buffer replacement strategy on three system tables - 'sysobjects', 'sysindexes' and 'syscolumns'. This is because the indstatus2 field of indexrowp in q_mark_prefetch contains junk value and the corresponding bit can't be set. 72900 1 This bug has been entered to allow single batch (group) of bug fixes #72776, #72777, #72689 and #73315 for the DEScriptor Manager into Cougar. 72906 1 950709:mirror/accept/mirr006.sql. 'use mir; execute syscheck' returns: "The following devices have changed: master" whereas in the canon nothing has changed. 72912 1 We need a monitor event for page log allocations so that the cache analysis tool can recommend a size for the log cache. 72917 1 amxrun segmentation fault:Access methods error code 0 encountered but run does not know how to handle it. "Error: 539, Severity: 20, State: 1" Stack trace: kbdmemmap->conn_hdlr->sequencer->s_execute- >execute->exec_eop->run->ex_raise->s_handle->hdl_defaul- 72924 1 If SQL Server crashes while upgrading a loaded database, subsequent reboot will get a stack trace during recovery of that database. The stack trace will not recur on the next reboot. 72927 1 While running TPCC buffer chain gets corrupted and terminates the SQL Server. Server stack traces s_execute->curs_fetch->execute-> startscan->getpage->bufgrab->bufkeep Infected with 11 in bufkeep+0x13c. 72948 1 Wrong Outputs: select @row = count(rowid) from alts_data01,print "select vc30 != c30" select rowid from alts_data01, where vc30 != (select c30 from alts_data01 where rowid = @row) orderby rowid. should not have rowid 4,9 in the output. 72953 1 The text of server errorlog message 300088 contains a typo. It says 'total memroy' where one would expect to find 'total memory'. 72962 1 Max number of hash buckets per named cache is 256000. Just 512MB of memory in any one cache guarrantees hash collisions, let alone sizes like 1+ GB that larger tests are likely to use. Sybase Confidential Page: 33 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 72965 1 Provide a trace flag that will enable I/Os greater than 16k on data. 72967 1 Shuting down a server and re-booting it always seems to write a new version of the backup.sys config file and rename the previous version to backup.nnn. No config parameters were changed and the ... 72971 1 sp_configure shows cnbytio and cnbytmax as part of the 'Disk I/O' group. 72972 1 The default configuration for the PLC related to free lock movement between the engine free list and the server free list leads to high rates of transfers between the two lists. 72978 1 Msg 300073, "Page corruption noticed for page 7725 belonging to objectid 0 on virtual device 2, side = primary" in errorlog during mirror testing while restarting server with checkpointing disabled. 72979 1 An 1105 during drop index can cause an EX_CONTROL error to be raised - session terminates with no explaination to user. ...->s_execute- >dropindex->dropi->ind_updateindstat->ex_raise->s_handle- >hdl_default->close_network->... 72982 1 950710: auth/auth000a.sql. 'select * from sysloginroles' returns 1 additional row(8 rows) whereas canon return 7 rows. The diff is row suid 1, srid 5 status 0 72983 1 950710: auth/auth001a.sql. 'select show_role()' returns 1 additional role, oper_ts_role 72985 1 Changing cnblkio through sp_configure does not change/check the value of cnblkmax. It is thus possible to set cnblkio to be greater than cnblkmax. 72992 1 The "stack guard area" configuration parameter is missing from the sql server. The guard area is an important configuration parameter and should have been included or settable somehow. 73008 1 Error 557/16/2: Cursor 'cur1' can not be used as it could not be found. At execution time, while executing a sproc containing a declare cursor. Plan uses an index on table. Index is dropped forcing sproc to be recompiled. Sproc gets recompiled OK and ... 73040 1 In dropcol/dcol007, there are couple of occurences of 1105/17/4 and the message with this state (4) is printed in log__isalmostfull, whe n DBT2_LOGFULLABT and P2_NOLOGSUSPEND are set in dbt->dbt_stat2 & pss->p2stat respectively. 73046 1 The restree test can generate a query that does not get compiled into a plan, thus there is no E_STMT. In sequencer(), a check is made for client/language cursor by dereferencing the E_STMT. This causes a stack trace. Sybase Confidential Page: 34 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 73049 1 Bugfix 70176 added a new routine, j_setOuterRows(). This routine performs a calculation per decorated permutation. We could move this calculation into j_tables() itself so that we don't have to recompute rows/outerrows on a given table for (see notes)o 73059 1 Showplan output of 'select * from #temp table' does not display the index keys used in the index scan of this table. This is because the sdes->sqoff is not filled up appropriately in qualpage () of access/search.c 73074 1 join.c contains calls to preview_cost() - these should be calls to j_preview_cost(). 73075 1 At startup, SQL Server should fail the boot with an error msg if it determines that there are more devices in sysdevices than the current configuration of the "number of devices" parameter. This will prevent db's on these dev's from being marked suspect. 73077 1 Configuration parameter 'SQL Server asynch i/o (T1603)' has no effect on the server. When set to 0 it should disable async i/o for devices. Similar problem exists for TCP_NODELAY option as well. 73087 1 Unknown error occurs in backout of 1105 error when running backup bzautod test suite. This error occurs while running backup001.sql where a loop of page spiltting activity is being performed. 73088 1 File sproc/a_values is out of sync with sproc/a_messages for sp_helpdb message 'offline'. a_values has 'offline' = 17028, where a_messages has 17594. Correct a_values to agree with a_messages. 73109 1 Because the server sometimes dirties a buffer and links it to the des of a different object, we must look at all dirty chains in a des at checkpoint time. This is inefficient and causes a lot of spinlock traffic. This should be optimized for the log at 73110 1 DUMP TRAN fails in an offline database if the database's log version is not current, because DUMP TRAN must log dealloc records. Must prevent DUMP TRAN (except with no_log) in databases whose log version is not current. 73132 1 dump xaction does not flush the changes made to log page in dmpx__post_trunc() function.If the server is killed (without checkpoint) just after the dump transaction, log buffers may not have a chance to be flushed. Causes 605 when run dropcol & fips. 73137 1 Server allows "dump tran" from a database that has been successfully upgraded,even before the database is dumped. This should have failed with a "database not dumped since upgrade" error. Sybase Confidential Page: 35 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 73147 1 950712: qrmp000a.sql-Initialize virtual device 3 failed and 'create database 'qrmp_db1 on qrmp_dev1 = 10' failed and cause stack trace. See svrsoldl:/svrsoldl_dtst1/tst_sql/reruns/kyt/950712/driver.out.failed .1 line 798 73162 1 when a transaction is dumped in an "offline" database, ie in a database whose log has not been upgraded to 11.0, Msg 6401 occurs (Cannot rollback $dmpxact ..). The test output has been saved in /remote/qa3/bugs//offline.out 73165 1 generic/source/useful/mon_tab.c is not being generated correctly. As a result, checkins to file monitors show up moncount.h, but not in mon_tab.c. Results in wrong values for monitor counters. 73196 1 Create cluster index on a table bound to a cache that is not the default causes a 605 error. SANITY server also says 'Brought page into wrong cache!'. cri_bldindex->at_reserve->pg__lockoam- >getpage_noscan->getpage->wrongpage. 73230 1 Undefined symbol during nightly build while loading s_dataserver: sybgetptr - first referenced in sun_svr4/lib/libsrv.a(qremaputil.o). 73246 1 B1 related changes to the stored procedure sp_logiosize 73249 1 Setting 'user connections' to default gets warning: The OS setting for file descriptors, 1024, is not large enough to configure 'user con' to 1000, 'devices' to 10 and 'remote sites' to 10.' Verification fails, cannot reset to default. See conf010.out 73259 1 Large I/O and bcp don't work. Server hangs. Also brings page into wrong cache if table is bound to cache A and db is bound to cache B. 73264 1 This is a test bug 73266 1 While upgrading from 492 to cougar on solaris, the installmaster hangs after upgrade is succeeded. This happens only on solaris and only when upgrading from 492. 73283 1 Add the config parameter "shared memory starting address". Currenlty the field is defined as "cmemranges" in "cfg_options". 73284 1 When loading a 1002 sun_svr4/rs6000 dump, the server process generates a "timeslce -201" error and is "infected with 11. 73296 1 "online database" hangs when upgrading sybsecurity. When a dump of sybsecurity, created in 10.0 server is loaded back in 11.0 server, the process hangs while upgrading syspartitions [ID 1013]. Sybase Confidential Page: 36 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 73306 1 The boot msgs which state the size of each data cache refer to "buffer pools". All user docs refer to these entities as "memory pools". These msgs need to use "memory pools" as well for consistency and comprehension. 73309 1 Optimizer should apply transitive closure to search arguments before scoring search clauses. 73312 1 opentable.c:open_for_cacheid() can cause/force desmgr.c:des_get() to return a kept DES which is in a DROP state. This can cause the thread that is actually removing the des from cache to ex_raise because des_clean() will find a keep count greater than 1. 73313 1 desmgr.c:des_drop() calls and does NOT check/test des_markdrop() return status. This can be a pb if des_markdrop() finds a keep count for the DES different than 1, as it will not put the DES in a DROP state. Hence, des_drop() should not pursue execution. 73314 1 Signal 11 executing a query with an orderby on a char(30)null asc,char(255) desc, and varchar(255) asc;kpinitproc->ex_cleanup- >conn_hdlr->ex_cleanup->sequencer->ex_cleanup->sendformats->exec_eop ->getsorted->mergenext->soextdealloc-> 73327 1 New lockmgr config params: deadlock checking perios, max engine freelocks, freelock transfer block size should have their own notification routines instead of the default notification routine. 73352 1 Server failed to boot with the message "The total size of all configured caches (5200k) is greater than the memory available for caches (4876k).". It is found in "backup start" test suite. 73370 1 Bug fix 72204 released the kssocktab spinlock too soon in ksmaskrpost routine. Needs to be released at the end of the function so we add the request without an attention condition being set on the socket. 73378 1 Remove the config parameter "perform all disk i/o on engine 0" and "run user tasks on engine 0". 73381 1 create table with max_rows_per_page reports "Incorrect syntax error".Porting MAXROWS / PAGE from 10.1 to 11.0 has missing changes in sqlkgen.c. 73397 1 The changes made to cm_grabmem() under bugid 69901 change the calling protocol to cm__cache_wash(). This results in only one MASS being washed per call instead of two. 73398 1 Changes made since July 11th have resulted in the desupd spinlock being grabbed 23 times per transaction instead of 8. This will affect UP performance and scaling. Sybase Confidential Page: 37 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 73413 1 Error 1124, state 1 after 1205, State 1. os_backtrace->0x0- >_coldstart->sequencer->s_execute->putxact->end__xact->ex_raise- >s_handle->stopxact->undo->ex_raise->s_handle->hdl_default- >terminate_process. see /remote/spgqa/qa_110x/docs/Consults/Stab/ ... 73421 1 (In System 10 too) While altering/creating a database if a GAM extent needs to be allocated for the db, the allocation happens from pages in the db, but the allocation is logged in master! Can leave master unrecoverable if xacn aborts, in some cases. 73430 1 In amex testing, I am leaving a test case when count(*) sysprocesses =0, In other words all my client are processed.Even after count=0 some processes are running afterwards. Consult loc: /remote/spgqa/qa_110x/docs/QA_Failure_Rpts/Stab/73430.amex 73448 1 sp_configure fails to reject an invalid value for 'sql server clock tick length'. Upon rebooting the server, the change is (correctly) not made. No error msg is displayed. 73454 1 sp_configure problem, run stwophas s2setup to reproduce. first, sp_configure "user connections" to be 12 then "go" -> return status =0, then sp_configure "open database obj", 250, the server is complaing total memory 7500 is not enough for configuring 73466 1 lock__fillcache can ex_raise out-of-locks error even when lockaddress is called with LOCKRELEASE. This exception may be caught at a higher level causing xn rollback, which in turn tries to acquire lock (to undo the change) but fails with 1124 because it 73467 1 If a database, that has been loaded from a pre Systme11 dump, is brought "online", server crashes with Msg 706 (Process 1 tried to remove PROC_HDR ...". The script and its output has been saved. 73469 1 Execute a 'page split' update to database to background. and run dbcc on database. dbcc was then blocked by the update, both process hung and never went back. no error in errorlog, the dbcc process is lock sleep & update process is sleeping from sp_who. 73470 1 Mirroring on master device gets disabled upon a reboot of the server. 73471 1 dbcc/fixindex() is called by dbcc to fix broken indices. If this path is chosen, the allocation hints space is lost. This imapacts performance of VLDB allocation. 73501 1 version string and copyright date need corrections 73503 1 sp_configure "max network packet size", 1045 rounds the configuration parameter up to 1536. According to system 10 docs it should be rounded down to 1024. Sybase Confidential Page: 38 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 73505 1 In sortalloc() memcreate() could return a NULL and then eventually the server would throw a SEGV. The return value from memcreate should be handled properly. There could be other occurences in the server similar to this also (callers of memcreate). 73507 1 Create clustered index on a table bound to a non-default cache followed by an INSERT leads to a 605 error. Stack trace: execute- >exec_eop->getnext->wrongpage. Page belongs to objid '0' instead of the table. MaxB says this is NOT the same as 73196. 73508 1 While loading a System 10.0 user database using a System 11.0 SQL Server, the database shows inconsistencies with Msg 2559 "Data page number 369 is empty but is not the first page ..", and Msg 2583("Number of data pages...differs from...the data OAM"). 73510 1 sp_configure "additional network", 2147483646 causes the message Msg 5853, Level 19, State 1: Procedure 'sp_configure', Line 358: 'additional network memory' has been modified from 2147483646 to - 2147483648 by the verification routine. Then test hangs. 73513 1 Correct function declarations from void to int. The body declarations for dyadic_issue and monadic_issue should use ANSI declarations. 73521 1 des__fill_cache_ids needs to check for existence of sysattributes before continuing if booted against pre-cougar master device. 73526 1 Compilation warning "implicitly declaring function to return int: cm_attr_init_cached_bindings()" in "dataserver.c" (sym mode). 73534 1 DIRTY_READS() macro does not work correctly for execute cursors. The macro should also check for the open of an execute cursor. 73536 1 11.0 server allows loading the master database from a pre-11.0 dump; it should refuse the load. 73541 1 Binding the text column of a table results in sysattributes being tagged with a 'T' entry for table instead of 'I' for index. 73542 1 Nested aggregate query with subquery in HAVING clause returns different results than same query on System 10.E.g. select c1, avg(min(c2)) from t1 group by c1 having c1 = 1 should return only those groups with c1=1, but returns all groups. Related to bug.7 73548 1 db_fill() should not retrieve cache binding for database master as master can't be bound to a named cache. 73553 1 Code for SELECT INTO and IDENTITY in s_execute() should be extracted into a function call in order to make code more readable and make it easier for C compiler to optimize s_execute(). Sybase Confidential Page: 39 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 73554 1 The sp_unbindcache_all stored procedure doesn't work properly now that cache bindings are stored in the sysattributes in each database. 73557 1 Wrong argument passed to ex_raise() for case RIDONPAGE in function getrow() causing corruptions in the error messages. 73558 1 when run installmaster scripts, Msg 208, INTERNAL ERROR: master.dbo.syscacheconfig not found. It is a reopen of bug 72352, since the problems persists for a long time, in sybinit tests 73562 1 incorrect exit value for "bldmaster -v" exits with value of 1 instead of STDEXIT 73564 1 While doing an upgrade to cougar and setting the default language to japanese thru resource file, there is an INTERNAL ERROR "Incorrect syntax near ','." 73575 1 server get 806. dump tran master with nolog. simple test case is backupserver make_dbs. stack isucbacktrace, terminate_process, close_network, hdl_default s_handle ex_raise log_to_virt virtmap bufgrab getpage dmpx_scanlog dpdb_backout dmpx__truncatemai 73581 1 the function prototypes generated in cfg_ntfy.h have the wrong type in the first parameter (char * should be int). 73583 1 Internal search for an index row failed with a stack trace during sybinit. 73606 1 Buffer manager error messages need to display the name of the cache of the offending buffer. 73613 1 dbcc pglinkage() starting from a text/image page of a partitioned table seg faults. when end-of-chain is reached, pgd_showpglinks() calls slcache_findslice() which returns null 73622 1 cfg_awk generates the wrong prototypes for the notification routines while generating cfg_ntfy.h. It generates foo (char *, CFG_VALUE *) instead of foo (int, CFG_VALUE *). This would cause compiler warnings on almost all platforms and break compilation 73624 1 Ct-lib program with cursors with parameters fails with error: Procedure expects parameter @sql0_u, which was not supplied. 73652 1 upgrade failure during mkpages 73653 1 NIST tests - TEST0485 in sdl032 fails. Permissions problem. User FLATER is able to update the CITY and GRADE columns in HU.VSTAFF3 when they do not have UPDATE privilege. FLATTER only has update privilege on EMPNUM and EMPNAME. Sybase Confidential Page: 40 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 73655 1 OAM_MINFREE threshold value should be bumped up to 20. Under heavy load, a MP server running with more than 10 engines could actually reach the current value of 10 and not being able to allocate a new OAM page in the OAM if there is no FREE entries left. 73659 1 sp_cacheconfig log_cache1 also reports on a cache other than log_cache1 73660 1 sp_bindcache binds the sysattributes table of a user database sp_bindcache cache2, dbase01, sysattributes 73661 1 sp_bindcache_all returns 0 when the specified cache dows not exist! sp_bindcache_all "NoNexistent_cache" 73662 1 The Sysattributes SQL in upgrade/att_upin.h is not restartable. If upgrade fails, rows must be deleted from Sysattributes by hand before upgrade can be rerun. 73663 1 sp_cacheconfig creates a cache of negative size. sp_cacheconfig "t_log_cache4", "-512K", logonly 73664 1 A SQL Script containing a 'while (1=1) loop' invoked from 'isql' when interrupted keeps running on the Server end forever. Repeating this script and interrupting the 'isql' session will eventually cause the server to consume all 'pss'. 73665 1 Seg fault in xls_prealloc during internal sort. sendformats- >exec_eop->getsorted->mergenext->soextdealloc->_etext->kisignal. Happens after DUMP TRAN WITH NO_LOG ends. 73666 1 Error 605 in DBCC CHECKDB in dumpload testing. No stack trace. Buffer: bpageno=420. Page: ppageno=420, pobjid=16003088. SDES: scur.pageid=420, objid=240003886. Preverror=0. 73673 1 XDES in dpdb_determine_dump_instance() needs to be registered in the backout code so that it can be freed in the event of an exception. 73675 1 {Replication] Error handlers in textrepl.c and ltintfc.c have their parameters reversed. This code will probably seg fault if these error paths are ever executed. 73686 1 No exception handler installed for module MEMORYM in conn_hdlr() function. So, Any memory creation/memory allocation errors are not captured. Problem seen in ddlstres test using multisql, 20 users and pss_alloc_worktables() function returns no memory. 73694 1 uicalccnproc() does not account for deadlock_tune, housekeeper, and ksrequest tasks (for engines other than 0). Also, need to account for housekeeper in the extra PSS structures required. Sybase Confidential Page: 41 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 73708 1 after fast bcping into a partitioned table, dbcc checktable gives error 2503 - Table corrupt: page linkage is not consistent. see 73713 1 In q_mark_mru if the strategy decided by optimizer is not MRU then the LRU bit should also be set to help in printing debugging information 73720 1 Not able to boot SQL Server with system10 masterdevice. Server coredumps with a FPE error. 73721 1 The log i/o size set by sp_logiosize will not be correctly set in the DES after a load database or the database has been swapped back into cache (ie its DBTABLE was scavanged. 73724 1 The status bit DBT2_MIXED_LOG_DATA is not set a db_fill() time and so it not being set does not imply that the database has a dedicated log segment. XLS relies on the bit being valid whenever the dbt is active. 73726 1 Server doesn't not boot up successfully with pre-System11 master device: Arithmatic excemption. 73735 1 Bus error in bufawrite() during checkpoint in performance benchmark on 10 engines 73762 1 There are three occurances of error msg 559 in s_execute.c the instances on lines 1226 and line 1576 both have state 1. 73766 1 After upgrade to System11 from 1002, a select * from syslogshold returns Error:8204, Severity:20. 73778 1 950728 build - orderby/orby000i.sql: 'dump tran master with no_log' retunrs Msg 806, Level 21, State:1 Could not find virtual page for logical page 161136 in database 'master' The SQL server is terminating this process. Also in driver.out:stack trace 73779 1 950728-dbcc/dbcc002a.sql: corrupt database routine returns: no modifications for obj=240003886 indexid=2 error-2521 whereas canon returns: 2521 objid= index=2 extid= ex_alloc= ex_alloc= allocation reset 73780 1 950728-after running dbcc/dbcc001a.sql in driver.out:Error 7939, Severity: 22,State: 1.Table Corruption: The entry is missing from OAM for object id 1600308 indin 1 for allocation page 5376.See: svrsoldl:/svrsoldl_dtst1/tst_sql/reruns/kyt/950726/drive.out 73781 1 950728 build-dbccbuf/dbuf001.sql:'sp_cacheconfig dbuf_cache_1, 0, 2048' returns err msg: Msg 257, Level 16, State1: Implicit conversion from datatype 'INT' to 'VARCHAR' is not allowed. Use CONVERT function to run this query. Sybase Confidential Page: 42 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 73782 1 1105 error, state 4, occurs while running the backup bzautod test suite. Msg: Can't allocate space for object 'syslogs' in database 'backup_001' because the 'logsegment' is full. This occurs with the fix to 73087. 73784 1 bcp-in stack traced: spinlock being released improperly bulk_normal- >insert->ncupdate->insert_index_row->logchange->xls_prealloc- >plc__flush->log__newpage->pg__log_allocpg->pg__releasealloc- >lock_address->lock__fillcache->ulvspinlock->kepanic: 73807 1 Dynamic memory allocation is a scaling issue. Optimization and compilation time memory allocation should be minimized for commands such as execute procedure. 73810 1 s_free() calls css_deallocate() to deallocate all cursors at the current nesting level. This call is being done whether or not there are any cursors at all. It should only be done if there are any cursors (pss->pcss). 73812 1 Non-SA user cannot view cache config with sp_[cache|pool]config Assertion: The SA role is not required to check a cache status 73813 1 In generic/include/catalog.h, the macros MASTER_ONLY and PER_DB_CAT has been defined twice! Seems like a merge error. One entry for each of the macros needs to be deleted. 73818 1 The fix for 72176 disabled the XLS scanning log records from the PLC by forcing a flush at xls_startscan() time. This was a temp fix for beta-1. The problem (see XLSRESOLVE in xls_scan.c()) ... 73821 1 When executing "Insert Test", the server process becomes "infected with 11". The SQL Server cannot be reached ("SQL Server connection timed out") after this error occurs. Function List: nmskget 73824 1 Check of NIL parameters before calling memory routines. 73825 1 LOSTMERGERUN exception is raised with wrong arguments in somerge.c and sobuild.c. This existed prior to System 11.dbid rather than dbname is passed. 73831 1 sqlrsv11/bin_euc.sql: "loading file 'binary.srt'" rerurns - SQL msg 1105,state 3,severity 17: Can't allocate space for object 'syslogs' in database 'master' becasue the 'system' segment is full. 73833 1 codeconv/code001/code101g,code108a,code109a, sp_adduser,sp_getmessage,sp_serverinfo. Can't allocate space for aboject 'syslogs' in database 'sybsystemprocs'because the 'logsegment' segment is full Sybase Confidential Page: 43 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 73837 1 slice-related to changes to sort.c - 1) subtracting control pages from page count and 2) following the next slice's chain are missing. We somehow missed this file during the mega merge. 73839 1 msg 1105, level 17, state 3: "Can't allocate space for object 'sysprocedures' in database 'master' because the 'system' segment is full. msg 2811, level 16, state 2: Can't create procedure dbid 1, objid 720005596, with a group number of 1,/ 73842 1 dbcc checkobjcache() does not include correct cache name for the database bound to it. The message returned is always "Index 0 for object 0 in database X bound to cache default data cache (id 0). 73860 1 Update of an unique index column on a replicated table fails with error 8402: Index row for object 560005026 (index id 6) was not found in database. 73861 1 when installing language 'japanese' using langinstall thru sybinit, while processing some of the messages, it returns "Message 436 line 325: Error: '10.0.2.2' is not a proper version number." Lot of messages are encountering this error. 73864 1 When several users login simultaneously, login fails with 'nopen: No virtual sockets available' or 'upspawn: no more process slots available' errors. 73878 1 While upgrading from 10.0.2 to 11.0, error 1105/17/3 occurs. This occurs during the second step in upgrade, i.e addition of new messages. An "dump tran" is needed in upgrade.c somewhere. 73884 1 Downgraded to P1. Fixed in 73775. 73885 1 cm_grabmem() can potentially corrupt the buffer manager LRU/MRU chain by returning the LRU head as a free buffer if no free 2K buffes is available in the cache. 73889 1 sp_configure "password expiration interval" does not generate error even though the correct parameter is "systemwide password expiration" 73907 1 In exerbuff/buf001 - dbcc exer_init_exer(buffmanager) hangs. This was found while updating the test for the new buffer manager interface. To reproduce use the partially updated test in /remote/srel1/config_tests/tests/exerbuff. 73908 1 Fixed in 73775. The sdes is still required in des__fill_lkprom(). However modularity is maintained by calling sdes_hookup(). 73910 1 1105 error in master while executing a stored proc that filled up the log. Resulted in error 539.0 Sybase Confidential Page: 44 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 73918 1 In cm_init_cache_config(), cache_mem_ptr is incremented to align PAGE structures on a 2K boundary. The memory that is wasted to do the alignment is not subtracted from memory that is actually used for the cache. Possible boundary condition problems. 73919 1 The formula used to align PAGE structures on SYBPAGESIZE boundaries in cm_init_cache_config is incorrect if the pointer is already aligned. It will end up skipping ahead an entire PAGE. 73923 1 sp_bindcache binds syslogs when database is not in single user mode Assertion: sp_bindcache fails the entity to be bound is syslogs and the database is not in single user mode 73925 1 sp_bindcache binds a table with an open corsor on it. Assertion: sp_bindcache fails if there's an open cursor on entity. 73926 1 sp_bindcache binds a database that is in use. Assertion: sp_bindcache fails if database being bound is in use.. 73927 1 sp_cacheconfig allows user to configure server into unbootable state 73930 1 The number of freelist<->global list transfers of free lock structures has jumped from 12/sec to 283/sec between 95/07/19 and 95/08/02 during cached TPC-B with the default lock manager config. 73931 1 when a 10.0.2 database was upgraded many messages like "at_reserve: return value=.." was printed. This is probably due to the usage of the dbcc function object_atts() by upgd__do_alloc_hints() in utils/intupgd.c. See 72003. 73933 1 On recovery server gets a error 2601 attemp to insert duplicate key row in object sysconfigures with unique index csysconfigures. The server displays error 3307 Process 1 was expected to hold logical lock on page 968... Error while undoing log row...a 73934 1 StkTrace: current process (0xa000a) infected with Signal 11. Occurs in sp_poolconfig "cache1","4K", "wash = 10P" After creating cache with a name 150 characters long. 73941 1 server does not boot when memory is changed from DEFAULT, default data cache size also has to be changed. Can not configure 16k pool either from the config file. 73943 1 Total size of all configured caches (1104k) is greater than the memory available for caches(1020k) on svrhpdly: /svrhpdly_tst1/tst_sql/950801_rerun/sdblrpc/t_driver.out 73957 1 stack trace during startup after the server was shutdown with a database log full. _coldstart->dsinit->doinit->dorecover- >onl_online_db->buf_online_db->cm_validate_cachce_bindings->exraise. Server shutdown again and started dbcc checkdb(db) shows pg link. Sybase Confidential Page: 45 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 73961 1 During boot time recovery, the server generates error 3602, state 2 if the db being recovered previously got an 1105 error. Stack: cm_validate_cache_bindings ->buf_online_database ->onl_online_db - >dorecover 73965 1 when qrmpupgd tried to installmaster to upgrade 10.01 to 11.0, the installmaster failed. Stack trace with OS timeslice -201. SQL server causing error: update statistics sysreferences. 73975 1 mkjoindata() allocates a JOINDATA and its minions. Currently there are a number of smaller allocations which are based on MAXVAR. These smaller allocations should be based on the actual # of tables involved in the current ROOT/AGHEAD not on MAXVAR. 73989 1 Trying to drop an index when the logfile is full results in a stack trace. The error handler (droper) installed in dropi() does not catch the major error number, SYSIND, so the default handler kills all. To fix, add a case in droperr() for SYSIND. 73997 1 Incrementing monitor counters should be turned on/off by a flag. 74004 1 Conditionals of the form "if (@var1 RELOP @var2)" generate 5 EVALs and execute at least 4 of them - E_LD0, E_LD1, E_... (specific comparison operator), E_SIF and E_END. We should be able to reduce this to 4 generated and 3 executed EVALs. 74008 1 Attempt to log transfer of a XBEGIN (XBEGIN_EXT) with a username with length > 2, SEGV's in lt_get_userinfo(). Occurs on a database with a replicated table. 74021 1 Some method of correlating buffer counter groups with specific caches is required (as well as spinlock names). There is no way to get the cache id of a named cache in a way that can be manipulated in SQL or a script. A builtin should be provided.u 74023 1 Could not retrieve row from logical page 10512 via RID because the entry in the offset table for that RID is less than or equal to 0.SQL causing error : set done = 1 when running busy_reader.sql. Failed to getrow(). 74024 1 increase memory and increase 'pre-read' to 10000. Then cannot reset 'pre-read' back to default. See case 8 in conf010.out, config test suite. 74025 1 In exerlock test, server command was deadlocked with another process 74026 1 Running installmaster when sybsystemprocs already exists can result in an 1105 on the system segment while trying to write sp_logiosize into sysprocedures. Numerous errors occur after this. Sometimes the server will leave syslogs in sybsystemprocs Sybase Confidential Page: 46 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 74027 1 sp_poolconfig computes the default wash size based on the requested pool size and fails w/: "incorrectly configured wash" when less memory is transferred to the buffer pool than requested. 74029 1 The DIRTY_READS macro can be removed from s_lockstmt and s_unlockstmt. 74031 1 When more than 8 caches are configured the buffer manager becomes unstable. Most caches (including the default cache) are lost. The number of caches that show up seems random. 74033 1 The code in hk__upd_logflush() that obtains the current sequence number should use the DBT provided function dbt_curseq(). 74034 1 hk__upd_logflush() needs work. Comparision of variables of circ_long type have to use the CIRCLE_XX() macros. Subtracting one from a circ_long variable is also not supported. All HK code should be checked for similar mistakes. 74035 1 Dropping index while logfile full: dropi() calls ind_updateindstat(),which calls bufunpredirty(), then later raises error, causing dropi()'s backout code to call ind_rowrelease(), which calls bufunpredity() *again* --> DBFATAL (kills isql) 74036 1 Error 3461, couldn't install required upgrade item 1014, during ONLINE DATABASE after LOAD DB of 7+ gig dump. "Cannot open table" message is returned from object_atts. A table name "testing" is passed in object_atts, but not found. 74037 1 Error 3414, recovery failed, due to 6902, timestamp on page between old and new in log, in boot that followed failure to online a database loaded from System 10, that failure due to error 3461, couldn't install required upgrade item 1014 (bug 74036). 74038 1 sp_poolconfig cache1, "512K", "4K" Writes wrong wash in config file [4K I/O Buffer Pool]\n pool size = 512K\n wash size = 1024 K sp_cacheconfig shows the wash size to be 408 Kb 74039 1 When sp_cacheconfig is called on an Active cache with the same values as before, it: 1) fails returns 1. 2) sets invalid (low) in- mem wash size. 3) sets invalid (high) wash values in config file. 4) sp_configure ... "write" 74040 1 cm_validate_cache_bindings() raises 3602 when the log is full AND the cache bindings need to be updated. Since there is no handler on 3602 from there, stack trace and process termination result. 74041 1 Cleanup after drop database is not complete and can cause 605 errors. Sybase Confidential Page: 47 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 74046 1 Error 2501, state 4, in DBCC REPAIRINDEX, passes extra, unused parameter, resulting in 'Invalid pointer param number 2, pointer value 0x7' instead of missing table name that should be printed by message. 74072 1 With more than 8 caches sp_helpcache reports: Maximum number of used databases for each query has been exceeded. The maximum allowed is 8. 74102 1 When sp_cacheconfig is executed with a parameter which does not correspond to an existing cache, error 557 is returned. This should not happen; the proc should just return an unnumbered error msg. 74103 1 If 1) sp_cacheconfig is used to define a new cache, and 2) the server is not rebooted, and 3) sp_cacheconfig is used to delete the cache, THEN sp_cacheconfig should delete the cache immediately (and not show it in the "pend/del" state. 74104 1 sp_cacheconfig accepts negative cache sizes (i.e. sp_cacheconfig chris, "-1m"). 74105 1 sp_helpcache references caches that 1) were never allocated because the server wasn't rebooted and 2) were deleted. 74106 1 sp_helpcache displays an incorrect overhead amount for newly created caches (i.e. it gives units but no value "kb"). 74113 1 the sp_bindcache procedure does not accept substrings while all other cache related procedures do. (i.e. the substring 'def' should be associated with 'default data cache' provided that it is a unique substring). 74114 1 After sp_bindcache fails to bind a system table because the corresponding database is not in single-user mode, sp_helpcache still shows the binding as if it worked. 74115 1 After experimenting with sp_cacheconfig for a while, all attempts to run it fail with one or both of errors 2714 & 925. Rebooting the server seemed to clear this condition. 74119 1 sp_bindcache fails to bind an object to a cache since the user executing it is not in the database in which the object resides, the bind fails (as it should) but the msg states that it failed because the db is not in single user mode. 74120 1 The units in sp_helpcache are not consistent. Some are in Mb and others are in Kb. They should all be in Mb. 74121 1 When an attempt to unbind an object (via sp_unbindcache) fails, the msg states that the 'bind' failed, not the 'unbind'. Sybase Confidential Page: 48 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 74131 1 open_user_table() is called 20-30 times for a simple "select * from table where col="xxx". 74141 1 rec_complete() declares an unpadded XBEGIN_EXT an the stack into which it can copy an XBEGIN_EXT from the log, which would smash adjacent variables. This will happen when recovering a transaction that used two-phase commit. 74152 1 Do buildmaster to create a database on file system with size 10240. Boot Server on it and run Installmaster. Continue it to run and signal 11 is raised in function pg_log_alloc().Server built on Aug 7 1995. Case reproducible. 74174 1 In lock_sanity() the size of the scanarray bitmap is based on Resource->rconfig->cfglocks. It should be 3x larger to account for all allocated lock structures. May result in memory corruption when bits are set outside the range of the scanarray.But this 74181 1 members of the syb_buf structure could be rearranged such that hot fields and the cold fields are separated from each other. 74183 1 pg__updateoam() passes a DES ptr to des_updlkprom which expects an SDES ptr. A SEGV (Signal 11) then occurs in when des_updlkprom tries to dereference sdes->sdesp. Stack: des_updlkprom-> pg__updateoam-> pg__pgalloc->pg__set_allocbit_oam_loglock->...l 74185 1 Adding the buffer to kept chain while keeping the buffer and removing it from the kept chain while unkeeping it, makes kept chain a hot address. Currently the kept chain is used only for diagnostic purposes. Shouldn't optimize comparitively rare case. 74188 1 The cursor test suite has many problems when run with a diagserver 74207 1 The allocation and deallocation of memory for work sdeses on a per query basis has proven to be prohibitively expensive. 74214 1 Server infected with Signal 11 on reading the config file. i.e. sp_configure "config", 0, "read", "". trace: s_execute- >execute->exec_eop->run->cfg_main->cfg__loadconfig->cfg__updsysrows- >memmove_sanity. 74226 1 In exerdbt, the output file gets 1803, CREATE DATABASE failed due to no space, but errorlog shows 905 (couldn't alloc DBTABLE) after timeslice -201, and: dbt__fill->db_fill->des_sysbuild- >des__bootsystables->des__fill_cache_ids-> ... 74230 1 Neither create nor drop trigger currently bumps the schema count on the table. This causes us to have to call exectrig() on every execution of an update, delete or insert. If we bumped the schema count, then we could determine at compile time ... Sybase Confidential Page: 49 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 74260 1 "GRANT SELECT,UPDATE(EMPNUM,EMPNAME) ON STAFF3" resulted in error at (com_link_esteps+0x88). Tree *seq is invalid. 74268 1 Using symbolic libraries from /remote/cgr_sun4/engdirs/release/symlib, built 10 Aug, server generates error 8211 (mismatch between name and ID descriptor hash tables) when running the curs test suite, case curs002g. Stack trace gets timeslice error. 74279 1 950809 build canon of fips/fips:"from fips_tab1 where vc30 like "past ][" escape "[" order by i4, vc30" returns invalid pattern string. The char following the escape must be percent sign,underscore,left square bracket, right square bracket,or escape char. 74280 1 Obscure message when default cache is set to mixed in config file: Configuration Error: Configuration file "..." has an unknown format on line 0. 74281 1 Server doesn't report an error when the default cache type is log only in the config file. 74284 1 Server does not report error for duplicate cache entries in cfgfile. Sometimes when the default cache is specified twice, it reports: Invalid configuration for the default 2k pool in cache . The default 2k pool must be a minimum of 512k.2 74285 1 Server does not report error when: a user cache has the type "default data cache" 74287 1 Negative buffer pool size in config file results in a stream of Signal 11 messages but no valid stack trace: kernel current process (0x0) infected with Signal 11 kernel pc: 0x242f9c kechkfault (0xb,0xef9d6690,0xb,0x20 74292 1 cursfix_main() checks every SDES in a cursor to see if it is a dirty read scan. They either all are or none are. Check can be replaced by checking compilation level of statement in css->cstmt before looping through SDESs. 74302 1 During bfmgropt/bmo001 server gets a read error and the process gets infected with signal 11 in cm_bufunhash. This is not a hardware error as a similar problem occurs on hp. Please use test code in /remote/user5/generic/renovate/tests/bfmgropt to reprod 74306 1 When running benchmark tests server log has this error "kernel timeslice -201 current process infected---kernel SQL causig error: declare @status int exec @status = op_BenchNodes 240--Then stack trace, Sybase Confidential Page: 50 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 74330 1 dpdb_determine_dump_instant() assumes that failure from dmpld_logsync() means an out-of-log error, which should already have had its message printed, so it silently starts backout with no message of its own. 11.0's addition of xls API means ... 74331 1 Error 1105, State 4 while running backup server tests. 74339 1 Master bugid entered to bundle the following fixes: 71851, 73554, 73661, 73923, 73925, 73926 74344 1 When Server is booting, run dbcc iosize on database that has not been recovered yet. hdl_default->s_handle->ex_raise->closedb- >d_set_obj_iosize->exec_dbcc->s_execute->sequencer->conn_hdlr- >kernel 74346 1 Loading a configuration file with cache definition changes does not work. Also, during load,should check if the changes are to parameters which are dynamic. If the changes are to static parameters, then the load should fail. 74350 1 while upgrading from 49x to 11 interactively, after entering the old release directory, it returns "Could not obtain Hostname for Internet address '134.214.80.91'". after hitting few times we can continue the upgrade process which goes just fine. 74358 1 When a cursor fetch causes a user to be chosen as a deadlock victim, then a subsequent fetch on the same cursor will cause a seg fault in startscan(). - stack trace shows as sequencer->s_execute->curs_fetch ->execute->exec_eop->tdssend_column 74374 1 stack trace in 'mirror' test suite on Solaris (sioux). Error: time slice (-201) caused stack trace during installmaster. Seems to happen on a machine that is HEAVILY loaded. 74382 1 When creating a database if a GAM extent needs to be allocated for the db, the allocation is logged in master instead of in the database being created. Also itseems like the change to row for the dba is done in the sysusers table in master rather 74393 1 One table, one reader, one writer: deadlock! If the select uses a non-clustered index, the order in which it locks index and data is different than how an update( or insert or delete) 74398 1 TPC-C does a lot of string concatenations. The first line of code in concat() accesses Resource->rchar_attrib->cs_width_map. This code is only necessary if the concatenation does not entirely fit in the space allocated for the result ... 74418 1 timeslice error while performing some delete operations against the database. Infinite loop in res_view(). Sybase Confidential Page: 51 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 74419 1 The compiler (com_target()) generates an E_CPVAR EVAL for all updates, regardless of whether or not the table being update contains any variable length columns. None of the tables in TPC-C contain variable length columns. 74420 1 zetatest and yetatest failed with OS signal:timeslice -201. SQL causing error: if(@@error=1205) exec zetatest. 74430 1 SANITY error when a client or execute cursor that is not open is deallocated - C_DECLARED bit is expected to be set. 74431 1 The call to reset_all_sqcache() in estmt_loopend() is unnecessary for declare and open cursor as the subquery cache starts clean and gets cleaned up on a close and deallocate cursor. 74433 1 subquery cache statistics do not work correctly for cursors. Info is only printed out on a declare cursor where it is useless. It should be printed out on the open cursor and fetch statements as well. 74455 1 sp_unbindcache_all is returning 1 upon successful completion. It should return 0. 74458 1 When the wash size changes as a side effect of changes in number of MASSes in source and destination buffer pool, a message should be printed in the errorlog to avoid confusion to the customer. See cm__calculate_washvalues()/cachemgr.c for more details. 74484 1 950815 build on Solaris & IBM: curs/curs003.sql: "close table #specter;open SPECTER" returns Msg 1265,Level 20,State 1:A lock manager routine expected a lock record as an argument but was passed something else. SQL Server is terminating this process. 74485 1 950815 build:isolevel/ronlydb/isol071:a fetch from a cursor at isolation level 0 on a read-only db that does not have a unique index on it. Then call procedure p_rodb_proc from isolation level 0, make the read-only db read/write-got Msg 7792 Level 16 74486 1 When a database is swapped out of cache and brought back in, the buffer cache bindings are not updated from sysattributes correctly. In addition the logiosize is not initialized according to the values stored in sysattributes. 74493 1 q_bld_sublist() always allocates memory for an alignment buffer for each referenced column, even on non-aligned platforms and for datatypes that do not need alignment 74494 1 E_UPDATE, E_INSERT and E_DELETE EVALs are always immediately followed by an E_END. We should still generate the E_END, but we can optimize in run() by having the other 3 EVALS return RUN_SUCCESS. Sybase Confidential Page: 52 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 74505 1 73592 disabled some of the performance improvements made by 72287. 74513 1 The server will not start when configured to use a large amount of memory. No error message is given when the server boot fails. To reproduce sp_configure "total memory", 100000. Shutdown the server and restart... 74539 1 The code to walk the SDESs at the bottom of execute() appears to be a SANITY check except for the code added to fix 58641. The code should be removed. Instead, we can twiddle a bit in the CSS when a cursor contains a text/image column in the select list 74541 1 We are getting the wrong plans for tm_ah_sor_2 and tm_ah_sor_3 in the PC Mag benchmark. It looks like we are not realizing that the clustered index is on the ordering column and we are doing an unnecessary sort. 74544 1 Unnecessary ONCE aggregate is being generated for an expression subquery which returns a scalar aggregate. 74546 1 Not sure why we allow nested aggregates. Assume table t (a numeric(6,2)) has 3 rows: 2,4,6. select sum(a) from t gives 12, select sum(sum(a)) from t gives 20. Whereas select avg(a) from t give 4, select avg(avg(a)) from t gives NULL. 74547 1 New subquery tests have 3 failures - sbmisca.sql,sbmisca1.sql involve queries with nested aggregates and a subquery;sbmiscs2.sql is a simple nested aggregate w/no WHERE clause. 72789 changed some of the semantics of nested aggregates with HAVING clause.. 74548 1 Expression subqueries are not being correctly normalized. For "where int_col = (select tiny_int from table)",If any conversion is required for the result of tiny_int from the subquery, the CNVT node should be applied on top of the subquery and not under 74549 1 sp_cacheconfig "cache1", "1024" will stack trace the server if the number of engines configured is higher than 1. 74550 1 sp_cacheconfig stack trace: bmo000a.out hp_excleanup->conn_hdlr- >e_ins->r_i4Native->cfg__insertcache... on svrhpdly: /svrhpdly_tst1/tst_sql/950816/bfmgropt/t_driver.out 74571 1 when sp_configure 'total memory' is set to a high number (higher than memory in machine), it gets changed (there is no check made whether memory is available..??). Shutdown and restart server aborts start-up and there is NO error message stating reason 74572 1 If a usedb() causes a DBTABLE to be instantiated at runtime for at database, cache ids (bindings) and log I/O size info are not retrieved properly IF THE DBTABLE descriptor/struct is a RE-USED one (SCAVENGED). Also, the WRONG Sysattributes gets accessed. Sybase Confidential Page: 53 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 74582 1 Error 2624/20/1 (Insert into table DES pointer... failed becuase rowlength 8 is < minlen 10) followed by stack trace (insert<- bld_dynindex<-exec_eop<-exec_eop<-execute<-s_exec_subquery<- run <- exec_eop <- execute...). When you have a materialized view . 74584 1 Query processing code has a model that the last main ROOT in the outermost query block stores the counts for user and work tables (varct, workct). This needs to be reliably implemented in all places where work tables are generated - DISTINCT clauses, .. 74601 1 server_nspinlocks() allocates one spinlock per network TYPE per engine. ksrequest() initializes one spinlock per network per engine. If the number of networks exceeds the number of network types, then not enough spinlocks are allocated. 74604 1 Cache bindings for sysattributes are ignored. cm_save_syscat_caches is still looking for dbid in the sysattributes row. This means the binding isn't found when sysattributes is scanned. 74626 1 curs_close() currently always calls tdssend_curinfo(). It should only be called for ct-lib cursors. 74645 1 sp_unbindcache_all does not work. 74663 1 seqr073b.sql now gets error message repeated 3 times for the same error condition. It was not the case when running against yesterday's server. 74664 1 There is more work to be done in s_execute(). Auditing related code and stack variables should be broken out into separate functions. Same with most of schema create code and code to get identity_insert state. 74670 1 s_find_rg_entries() is being called too much at execution time. It appears to be called by both s_setuptables() and seq_schema_check() for most DML. It builds an array of pointers to the user range entries in the plan. 74680 1 When an unexpected config value is in sysconfigures (say the config value is removed from SQL Server but you boot off an older master device) SQL Server gives error 5859 on boot, stops processing sysconfigures, but boots anyway. 74712 1 This bug has been opened to track the closure of, 74119, 74120, 74203, 73812, 74106, 74113, 74285, 73658, 74062,74726, 74728, 74731, 74890. All these bugs are related to the stored procedures sp_poolconfig, sp_cacheconfig, sp_helpcache and sp_bindcache. 74717 1 Sp_syntax "set" still gives only {1|3} for "transaction isolation level". We now also have isolation level 0. Similarly, sp_syntax "select" does not display the AT ISOLATION clause. Sybase Confidential Page: 54 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 74728 1 sp_poolconfig gives an erroneous message if the buffer pool size is beyond 16K. Msg 5870 should not say "....Either change buffer size or change cache status to LOG ONLY" 74731 1 sp_poolconfig "cache1","16K","16K" issues a wrong error message "Less memory moved than requested in cache .....". The second parameter value "16K" should not be accepted, as the value is less than 512K. 74734 1 Create a temp or regular table in tempdb and building an index on the table will create 2546 errors. probably end up in running out of space in tempdb. Problem exists in System 10 too. 74747 1 Engineering diagnostic routine(net,showall) doesn't producing any output. 74751 1 Server configured for multiple engines each with 1024 file descriptors. The o/s file descriptors in sp_configure still contains 1024. This does not allow for "user connections" to be set higher than 1024. Msg 5849 Verification failed for parameter 74779 1 extra k in the description when you try to drop a 2k buffer pool 74780 1 customer types in "sp_syntax shutdown" to find the syntax for the shutdown command. What actually happened is that it shut down the SQL Server! This caused a customer system to stop and numerous other issues. 74782 1 One can overwrite the master device when issuing a disk mirror command. The command: disk mirror name="sysprocsdev,mirror="/dev/vg02/rmast10". "rmast10" is the master device. Bug 58192 was supposed to fix this and was shown as being fixed in ebf4429, 74786 1 nmskcpy could be speeded up by doing a MEMMOVE instead of a for (i = 1 to limit) *x++ = *y++. 74790 1 sp_poolconfig crashed the isql and closed the connection. 74811 1 configuration parameter, 'user connections', does not restrict the number of user connections. for instance, set 'user connections' to 25, and more than 40 connections can be made. 74817 1 Message 1279 prints garbage as the engine number. Message 1279 is SQL server run out of locks on engine . 74821 1 820, state 6 errors. Typical stack ends up as tally_fixrowcount- >pg_oamfixrowcount->bufdirty->xls_bufpindirty->bufpindirty- >ex_raise. Sybase Confidential Page: 55 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 74822 1 While running tests the server reported out of locks on engine error. We killed all the clients using the unix kill command and subsequently the server hung. sp_who showed one process was doing a commit tran and sleeping and the others were waiting. 74826 1 [ Replication ] Log Scanner cannot find XBEGIN if it is on the same page as the XCKPT record. This causes the LTM to get the wrong impression about the oldest open BEGIN. 74828 1 Using disk init with an already used vdevno results in an error message referring to a wrong PHYSNAME, but says nothing about vdevno. 74829 1 Once the monitor and spinlock counters have been turned on they cannot be turned off! 74832 1 Segmentation violation in pg__lockoam() while holding spinlock during insert into table without clustered index. ...->insert- >ncinsert->linkovflow->pg_allocate->pg__set_allocbit_oam_loglock- >pg__locklog_alloc->pg__oamspace->pg__lockoam 74835 1 The message text for msg 313 is empty after running buildmaster. To see problem - select * from t1 at isolation 0 (where there is no unique index on t1). Msg 313 added for 10.1. 74836 1 After more than 15 hours of running tests, server error "Error 605 has occurred. Attempt to fetch logical page 514 in database 'tempdb' belongs to id 8 not to object #letmeout. 74848 1 Latest build of 11.0 allows AT ISOLATION levels 1 or 3 with SELECT INTO, a SELECT in a CREATE VIEW statement, and a SELECT in an INSERT statement. 74854 1 950825 build:big/bigt022.sql: "select * from v3" return NULL whereas canon returns 1. See svrsoldl:/svrsoldl_dtst1/reruns/kyt/950825/big/* 74861 1 with one SQL Server 11 running, we are able to create a second SQL Server 11 on top of the first ones' mirrored device. 74864 1 buildmaster does not check whether the existing master device is being used by a SQL Server, before clobbering it. It appears that buildmaster assumes that it will be typically called only via sybinit after proper checks. 74865 1 Statements should be atomic. An update which fails due to an attempt to set a non-nullable column to null leaves some of the rows updated. The entire update should be rolled back. Sybase Confidential Page: 56 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 74877 1 Slow BCP into a table with a NC index hits log threshold and waits. Shutdown with nowait, reboot server, get error 3904 (can't unsplit page, more data than will fit on page). 74890 1 Error message 18143 has to be changed. Unit string "K" is displayed twice. 74895 1 Optimize the CM_FIND_BPOOL macro. It has a wasteful loop constructed with a while. Replace it with a simple arithmetic expression to derive the pool index. 74897 1 644/21/1 - index row entry for index id 2 of table in database raised while trying to insert into table via ct- lib program at customer site. Can easily reproduce problem using a small < 600 row table, with non-clustered indexes... 74919 1 Dump tran master with no_log fails with msgs 1105 and 4204. This may be related to the new efficient way to log page deallocations. dump truncate_only adds very few log entries to log deallocation of log pages. Not much left for no_log to clean up. 74942 1 cm_largest_mass() ranks 3rd on a profile of a 16 table join which returns only 1 row. (Test of compilation overhead.) It appears as it is being called twice for each decorated permutation. 74945 1 Server failed to boot with message that 2k cache was smaller than 512k, after a sp_poolconfig 'default data cache', '1M', '4k' was executed to add a 4k pool to the 35Mb default cache. 74946 1 sp_helprotect issues the following error at line 200: Arithmetic overflow during implicit conversion of INT value '508' to a TINYINT field. 74948 1 if "[" from config file or any other changes an error will be generated. However, the error message does not correctly identify the problem line. It returns back line 0. 74950 1 when there is a syntax error (missing "[" for example) in the configuration file then the server will not boot. The server should still proceed and boot even with this error in the file. There are default values. 74959 1 [ Replication ] When creating a replication definition, a duplicate oqid is detected by the repserver. 74961 1 Showplan output for a materialized view (which creates a work table) shows that worktable1 is created in the first step. Then in 2nd step it shows that the view is being selected from. This is not correct; viewname should be replaced with name of worktabl Sybase Confidential Page: 57 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 74969 1 Unable to load the master database following instructions in ch. 9 of the Sys Admin Guide. It appears that the server's checking for the existence of the wrong process (using ps -f rather than ps -ef) there are actually 2 processes. 74970 1 server is failing when doing a load transaction command. It appears that it is not checking for the proper processes (ps -ef). See bug 74969 for the similiar issue. 74976 1 The local variable 'min_logrows_for_truncate' introduced by the fix for 72697 in ckpt2() is likely to contain garbage if the loop through the databases raised an exception that then dropped back into the loop. 74989 1 Use of trace flag 1212 causes SEGV in lock_multiple(). Problem is use of rq->lkr_lockrec->lrspinlock without checking that rq- >lkr_lockrec is non-null. 75000 1 If a database containing stored procedures with insert into a temp tables with IDENTITY column is dumped to a device and is reloaded, the stored procedure fails to work after the load. It gives an ex_raise Msg 7743 An explicit value for the identity ...d 75001 1 If the I/O size is > 16k and the cache status is log only, weird errors would result since we are not supporting MASSes beyond 16k for either log or data. 75002 1 If cache status is "default data cache" for any cache, the server boot fails with a wrong message ".......... The default 2k pool must be a minimum of 512K". 75012 1 If the distribution of buffers in one cache is not uniform and there is idle time and the number of configured devices is large, the HK issues a large batch of IOs. This overloads the disk. 75013 1 If the distribution of buffers in one cache is not uniform and there is idle time and the number of configured devices is large, the HK issues a large batch of IOs. This overloads the disk. 75018 1 When 160 client connections try to execute the same stored procedure at the same time, SQL Server runs out of Procedure Cache. This test works on 4.9.2 SQL Server(even with 350 client connections). 75030 1 Improve performance on queries against empty tables using 2K buffer pools. 75038 1 If engineering mem/usage diagnostic was executed while SQL Server is loading a database, two time slice errors occur: 1 for checkpoint process another for the routine. They server shutsdown. Stacktrace for checkpoint procedure available. Sybase Confidential Page: 58 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 75041 1 The first execution of a proc causes process to become infected with 11. Increasing proc cache percent from default 20% to 50% did not help (total memory 90MB). This app was upgraded from 10.02 to 11. Worked on 10.0.2. Upgrade did not complain. 75045 1 Scanning foreign tables does not pick up cache strategy set up for the foreign tables. exec_ri_eop() should call mapsvs() as its first statement. 75057 1 Extent allocation for syslogs is not performed sequentially even though it should be. Sequential allocations will increase performance by allowing large i/o's to be more effective. It will also allow a mechanism to determine very useful usage information 75058 1 Do sp_poolconfig "default data cache", "1M","4K". Shutdown the server and reboot. Server will not boot and displays a wrong message "Invalid configuration for the default 2k pool in cache default data cache. The default 2k pool must be a minimum of.."2 75059 1 sp_cacheconfig does not warn the users to reboot after a valid cache configuration operation is completed. 75062 1 Do sp_poolconfig "default data cache","1M","4K". Shutdown the server. Edit the config file and remove the 4K buffer pool entry from config file that got created. Now reboot the server. Boots fine. Now do, sp_poolconfig "default data cache","1M","4K". 75081 1 Test case cust_subq10 in the dssperf suite took 22% longer on the 8/30 nightly server than it did on the Subquery project server prior to feature integration. 75082 1 Test case cust_subq17 in the dssperf suite took 54% longer on the 8/30 nightly server than it did on the Subquery project server prior to feature integration. 75083 1 Test case cust_subq2 in the dssperf suite took 24% longer on the 8/30 nightly server than it did on the Subquery project server prior to feature integration. 75084 1 Test case cust_subq2a in the dssperf suite took 25% longer on the 8/30 nightly server than it did on the Subquery project server prior to feature integration. 75085 1 Test case cust_subq3 in the dssperf suite took 113% longer on the 8/30 nightly server than it did on the Subquery project server prior to feature integration. 75086 1 Test case cust_subq6 in the dssperf suite took 22% longer on the 8/30 nightly server than it did on the Subquery project server prior to feature integration. Sybase Confidential Page: 59 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 75095 1 sp_logiosize should be executable by any user if the value parameter is NULL. 75101 1 Multiple bcp sessions randomly hang on MP servers. Seen on TPC-C loads and at beta1 customer sites. 75112 1 dbcc page() with wrong dbid stack traces ...exec_dbcc->d_callpage- >d_page->db_virtmap->ex_raise... should give error message but not stack trace. 75127 1 sp_bindcache "cache1", foo, foo reports "Specified named cache is not unique" if the a cache name like cache1 (eg cache10) exists. 75181 1 Error 2601 Attempt to insert duplicate key row in object 'sysconfigures' with unique index 'csysconfigures' when running sp_poolconfig. With a new config file, do 'sp_poolconfig "default data cache","1M","4k"' to create a 4K buffer pool. Then do ...o 75183 1 Server did not booted up successfully with the error message "Configuration Error: Two Named Caches with the 'cache status' defined as 'default data cache' is not allowed." although there is only one Named Cache with 'cache status=default data cache'...i 75189 1 Error 202/11/2: Internal error: Unable to open table at query execution time, while doing "insert into self_referential_table select * from another_table ORDER BY c1, c2". Error does not occur if you insert into a normal table, or insert into self-ref .. 75204 1 Error 2703 or seg fault (depending on platform) if you try to create a table with a primary key on a named segment, even if the segment exists. 75209 1 Need to check for duplicate buffer pools definitions in the config file. 75212 1 Server runs out of locks (errmsg 1204) and shutdown during BCP. BCP appears to obtain page locks indefinitely without upgrading to a table lock. According to the counters, it doesn't even try to upgrade. 75242 1 Server runs out of locks and the server process is terminated with sinlge error in log: Error 1204/19/3; kernel ueshutdown: exiting. User is running script which does: Begin tran; [insert into table select * from table ]; commit tran. 75243 1 Lock escalation is not kicking in correctly. User is running script which increases sizes of user tables by doing: Begin tran; [ INSERT into table SELECT * from another_table ] many times; commit tran; Sybase Confidential Page: 60 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 75244 1 While running batch pgm (Cobol program) which does inserts/updates etc, received an error: 3307/21/1 (Process 1 was expected to hold logical lock on page # in db for object index 4), while appl was doing rollback tran 75245 1 sp_unbindcache_all reports: "table still open - dbid 5 table id 8" table 8 is syslogs. The database is left in use so drop database fails. 7-8 more sp_unbindcache_all calls stack trace the server 75257 1 For level0 scan, the b1 security checks are being made without holding the scan buffer in predirty state. This can lead to in valid checks when lot of concurrent activity. The changes are to be made in getnext() 75263 1 sp_poolconfig "" recognizes unique substrings of the cache name if the procedure is called to print information only. If it is used to modify pool configurations, a substring for the cache name will result in an error indicating that the ...e 75265 1 For level-0 scan, in getnext(), when tiumeslicing out, the scan buffer may still be in bufpredirty state. In the worst case, if there is an error in upyield(), the buffer may never get bufunpredirty. 75266 1 For level 0 scan, the routine qualpage may get called on a page that may not even belong to the object. The object id should be verified on the page before calling qualpage(). This can happen when qualpage is called from the bottom of this routine 75267 1 If getxact() is used to start a nested xn in a different database than the outer xn, but nothing is actually logged in the inner xn, then putxact() will fail to close the log in the inner database. msg: table still open - dbid N table id 8. 75268 1 The sanity routine des__verify_bindings does not handle database bindings properly, or table bindings where the table is not explicitly bound, but the database is. run bufman 580 to reproduce. 75287 1 I/O error: 3225/21/1 occured on 2 separate instances (1)while trying to create db; could not recreate; (2) while extending db by 1000 MB on raw device of size 1000 MB. In 2nd case, w/a is to extend db by size 995MB. 75300 1 The log I/O size for master is not reported in the error log at boot time as it is for all other databases. Probably because this is done when a DB is onlined, and master is always online. 75308 1 Join between two tables with orderby selects wrong index resulting in bad performance. First table has clustered and non-clustered indexes & second table is a one row table with no indexes. The clustered index is on the joining column and should be used. Sybase Confidential Page: 61 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 75314 1 [REPLICATION] If txtimg_upd_table() fails when it is called by cri.c/cri_bldindex(), cri_bldindex() should also return FALSE.. 75316 1 [REPLICATION] In ltintfc.c, setrepstatus() and call_reptxtimgcol() are not lockin the table before calling txtimg_upd_table() and txtimg_set_replicate(). Failure to do so could result in text/image replication failure. 75317 1 [REPLICATION] In textrepl.c, getnext() loops in txtimg_upd_table() and txtimg_set_replicate() need to check for KILL_YOURSELF after the loops. Failure to do so, could result in partially updated replication information. 75318 1 [REPLICATION] In textrepl.c, txtimg_alloc_txrepdes() is not scanning SYSCOLUMNS using the clustered index. 75319 1 [REPLICATION] In desmgr.c, des_chgobjstat2(), the type of the bits parameter should be int, and also the casting of bits should be removed. The wrong parameter type and casting could cause the upper bits to be lost. 75320 1 [REPLICATION] In textrepl.c, txtimg_get_datarow() is acquiring the locks with lock_logical(), but releasing the locks with pageunlock(). This could cause consistency problems becuase the latter decrementes sdes->spglocks, but the former did not incremente 75331 1 Given t1(c1 int), t2(c2 int) and v1 defined as select * from t2, query select sum(sum(c1)) from t1 , v1 will return Msg 409 while select sum(sum(c1)) from t1 , t2 works fine. 75332 1 "C" source macros bit masks need reworking. 75338 1 after an index for a table is bound to a cache (default data cache), using sp_helpcache to display information resulted in SQL Server crashed with stack trace (no function names, only numbers). 75371 1 [REPLICATION] Transaction names that begin with a '$' need to have the '$' replaced by a '_' when the Log Scanner transfers xactnames to the LTM. 75383 1 When you run sp_bindcache on a cache which has been created but is not yet active, you get a message saying that the cache does not exist. This is confusing to the user who has just created the cache and can see it in sysconfigures. 75387 1 Stack trace in testing follows message "current process infected with 11". _coldstart->conn_hdlr->login->get_loginauth->startscan- >srchindex->getrootpage->lock_address->lock__fillcache->_fini_- >kisignal Sybase Confidential Page: 62 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 75473 1 This bug is opened to represent the fix for 75145, 75144, 75143, 75141. 75498 1 Compile of SQL Server failing on all UNIX platforms during symbolic builds. 75500 1 create database xxx on yyy = n generate "There is no disk named yyy. Checking other disk name" message on multiple test suites, e.g.multiprt, rules, update, vdis,etc causing diffs 75510 1 dssperf regr: qry95_subq26 query:Optimizer is generating a sort for ORDER BY even though the indexes selected scan the tables in the ORDER BY list. Pre-checkin subquery server was not generating a sort for ORDER BY, but nightly server is generating one. 75514 1 In exerbuff test, DIFF in create.out. Output file contains an extra line "There is no disk name bufexdev. Checking other disk names". This line occurs just before the line "CREATE DATABASE: allocating 2560 pages on disk bufexdbdev."Same for bufexdblog. 75515 1 create database xxx on master=10 prints 'There is no disk named master. Checking other disk names.' This is followed by 'CREATE DATABASE: allocating 5120 pages on disk master.' 75520 1 Create database on a user-created disk returned a misleading message "There is no disk named ''. Checking other disk names.", but the database was created anyway. 75532 1 Bugfix 72036 broke the monitor server. MPTP (memory pool table pointer) is overwritten with Resource->rmpt_ptr when monitor server calls MPM_INITCLASS causing monitor server to seg fault. 75539 1 [REPLICATION] When setting the LTM truncation point to the end of the log with settrunc (ltm, end), the truncation page gets set to 0 instead of to the last page of the log. 75541 1 timeslice -201 in multi-user multi-engine test. Customer tests run fine with 1 engine and with traceflag 7810 on multiple engines, but causes timeslice with 2 engines without 7810. Stack trace in email. 75566 1 Subqueries are attached many times per permutation, rather than once per permutation. This leads to wasted time and memory. 75568 1 Call to getnext() in crdb_disk() was incorrectly modified by vlm changes (checkin of bugfix 74263). The rearranging of parameters in the if statement which calls getnext() reversed the logic if statement. 75569 1 950912 build: vdis/vdis002.sql line 1008. select x,y from view11c order by x,y returns differnt results than the canon Sybase Confidential Page: 63 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 75575 1 select space(3) and select space(5-2) dont produce the same results.Further the you cannot use a variable as the argument. It produces wrong results. 75586 1 sp_cacheconfig deletes the default cache 75588 1 executing dbcc help with no other parameters causes the process to terminate with a 'process infected' error and a stack trace. The errorlog may be found on machine cougar35 at /work/rel11/install/errorlog.trace 75603 1 In ouput of test exerdbt: There is an extra row "5 exerdes" in sysdatabases. Causes diffs in dbt000.out and dbt004.out. 75609 1 sp_poolconfig "cachename". Add up all the individual values for config value and runvalue catogories . The numbers do not match with the total value that is displayed.. 75615 1 950913 build:login/logn010a,c,11a,11c,12a,12c,13a,13d,14a,14c,15a,15c,17,18 ,18c,sp_configure "passowrd options interval",100 returns, No matching configuration options, here is a listing of groups... Canon returns: Configuration option changed. The SQL 75617 1 Error: 679, Severity: 20, State: 1, Inconsistent update state found during expensive direct update of object '1792009415'. , followed by a stack trace. exec_eop->run->update->ex_raise->s_handle- >hdl_default->close_network->terminate_process->ucbacktrace- 75619 1 While adding info to location screen of the application this error comes up every time. Usually you will at least get the stored procedure name but this error doesn't specify. Msg 2805, Level 18, State 0, in SQL server public0, line 1: Bad pointer ... 75623 1 dump master; installpubs; overwrite master with buildmaster -m; load master; disk refit; now dbcc checkalloc(pubs2) shows 2540 errors. 75626 1 950914 build:login/logn011b,012b retrun login attempt was succeful, but canon returns Msg 4023, Leve 16, State 1, line 8: Your passwaord will expired in 25.0 days. logb013b,13c return passwrd expired & canon states using sp_password to change password 75635 1 upgrade is not idempotent with respect to inserting new rows in sysconfigures. If upgrade fails after inserting into sysconfigures, future attempts to upgrade will fail trying to insert into sysconfigures because there is a unique index. 75647 1 configuration file options in sp_configure assume file is in directory where isql session was first started. Change the directory and issue a 'write': file is written to previous directory, not current directory. Sybase Confidential Page: 64 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 75648 1 The optimizer does not select prefetch strategies for 4k buffer pools. 75662 1 when building 11.0 with 10.0.3 connectivity hdrs, "charset" fails to build. 75663 1 If the default cache definition is not the first named cache definition in the config file, then the server gets an error inserting the default cache definition into sysconfigures. Easy to reproduce. 75671 1 Drop column is only supported for specific customers, but other customers discover it and get themselves into trouble. We could put it under a trace flag and remove the trace flag when drop column is actually supported. 75672 1 [REPLICATION] If an multirow update is done in mixed direct delete/deferred insert and direct delete/direct insert, the log scanner passes incorrect updates to the RepSvr. 75674 1 Add better error message for when a cache has been configured too small. Also modify cache verification to ensure that the default cache is at least 512K large. 75744 1 test hangs for about 8 hours and then start to give the kernel message: Cannot send, host process disconnected:svrimbsdly 26300 spid: 9 75762 1 q_bld_sublist() does not allocate alignment space for INT1, but check_align() marks INT1s for alignment and subconst() tries to MEMMOVE to a nill-pointer. 75766 1 When an engine cache and the server global list are empty, the lock manager ex-raises. It should try to drain other engine caches before ex-raising. 75770 1 Page linkage corruption (error 2503 from dbcc checktable) occurs when bulk loading into a very large table (about 22 gig). Table was sliced 20 ways and 16k I/Os were used during the load. 75777 1 It takes about 4 times as much time to load & compile a stored procedure on System 11 compared to 4.9.2. This was measured using esql on a similarly configured databases and a stored procedure with 80 "select * from sysmessages..." statements. 75789 1 dbcc des() has become unreadable as cacheids for 255 indexes are printed regardless of whether a binding exists for each of those indids or not. Sybase Confidential Page: 65 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 75795 1 If a 2nd server is booted pointing to a device in use by another server, we properly print a msg saying the device is already in use. If we try to create a db on that device on the 2nd server, it properly fails. But, it doesn't clean up - model is locked. 75808 1 Query tpcd91_q2 takes 462% longer on System 11 than on the subquery project server. 75809 1 Query tpcd91_q3 takes 26% longer on System 11 than on the subquery project server. 75810 1 Query tpcd91_q5 takes 709% longer on System 11 than on the subquery project server. 75811 1 Query tpcd91_q8cs takes 151% longer on System 11 than on the subquery project server. 75812 1 Query tpcd91_q9 takes 2711% longer on System 11 than on the subquery project server. 75813 1 If a non-replicated procedure called inside a subscribable function updates a replicated table, the updates are logged as replicated. 75814 1 Adding up memory as shown by dbcc memusage shows a lot of unaccounted memory, especially for larger memory configurations. 75824 1 The stored procedure whseq5, in Air Products application ends up doing a table scan in 11.0 versus using a clustered index in system 10. The query involves an orderby. The query and the plan output is in vulcan/qa3/bugs/. Please contact howard 75842 1 SQL Server does not increment engine_0 clock_ticks monitor counter. Monitor server uses this counter to check the heart beat of the SQL Server. 75860 1 in diag/pagerbld.c function pg_extent_builder() calls getpage_init_scan_context(). The parameter tabid, to this function is obtained from an index row on sysindexes PAGE. We should hence use the GET.. Macros (GETLONG, GETSHORT, etc) to obtain the tabid. 75864 1 message 313 (DECISION, DEC_NODIRTYIND) seems to be deleted from msgdb. anyway, if this error is raised in a newly built server the message is empty. 9/20 build. 75871 1 sig 11 when running query with orderby. stack: sequencer-> s_execute->execute->sendformats. 75875 1 Signal 11 with no stacktrace when executing query with self join. Sybase Confidential Page: 66 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 75886 1 error in extra memory calculations will incorrectly set the default data cache size and the server will not boot under certain conditions. 75895 1 This bug represents the closure of 73306, 76651, 76652, 75727, 75807, 75708, 75803, 77872, 77873. Note that these are individual bugs and just for the purpose of tracking through eco all these bugs would be duped to a single bugid. 75902 1 The backout variable in reindex(), uses some local variables that are not treated with SYB_NOOPT. This can lead to some undesirable results. Please use SYB_NOOPT to keep these variables in memory. 75909 1 The flattening to an equijoin because of a uniqueness condition in subquery transformations does not currently work with work tables and it should. Problem is that index row is not created until later. 75924 1 [REPLICATION] Spurious SEGV's in writetext when textdata is received via network data. 75926 1 950920 build:HP,IBM&SUN,stack trace.sdblrpc/drpc912e:current proc infected with 11._coldstart+0x4()->conn_hdkr+0x2044->tdsrecv_rpc+0xc ->execrpc+0x320->execproc+0x470->sequencer+0xbac->s_recompile+0x1e8- >s_renormalize+0xc4->s_recrtproc+0x57c->n 75938 1 sp_spaceused has a high chance of deadlocking with create table/index or drop table/index. An 1150 error is followed by process termination and stacktrace:- 75945 1 Sometimes when sp_helpdb is done on a database while create database is in progress for another database, the DBCREATE goes in an infinite loop in prot_search() because cprot->e_cphnext is set to cprot itself. 75957 1 RESDOM has a linked list of NAMES struct which stores column and table name. These are added for processing UPDATE WHERE CURRENT OF using cursors. There should be no need for storing table name, as this should be obtained from the rgname off the RANGE.. 75962 1 Offshoot of 75018: Transient memory allocation for query compilation should _ALL_ come from compiler scratch space, so that the overhead of tree space memory is kept to a minimum during query compilation. For instance, reformat.c now allocates SCLAUSE .. 75964 1 send_tabname() uses varct from the first main E_STEP, which is wrong. It should pick this from the last subordinate E_STEP, and use the e_stvirtrg[] array. Otherwise, for browse mode queries with a distinct view, we will not return any tablenames for ... Sybase Confidential Page: 67 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 75980 1 s_compile() does tons of dereferencing of pss->phdr and pss->pplan. Stack variables should be used for both of these, noting that pss- >phdr changes at the end. 75983 1 There is no need to call findojoins() from decision() if the varct is < 2 as there cannot be a join, much less an outer join, unless there is at least 2 base tables. (No need to call pull_join() or cost_joins() either.) 75989 1 Stack trace from trace flag 302 when both ORDER BY and GROUP BY are used. traceprint->print_trace302_header->q_score_index->q_sarg_find- >decision->optimize_cmd... 75990 1 error msg 'total size of configured pools (17438k) in cache default data cache is greater than the total size (17437k) of the cache' does not make sense as default data cache was only 2048k. Server DID NOT BOOT. Always 1k difference in msg. No other 75991 1 Error message during exerdes test: kernel: kpsched: kpid 1376277 bad status 0x484. Also seen on OSF machine. 76032 1 While running diagserver message # 1601 causes server shutdown because of spinlock order violation. Stk trace: keipanic- >ulpspinlock->ind__searchlock->ind__fastsearch->ind__findrow- >ind_rowcopy->startscan->findtext->ex_print. The stack trace occurs 76042 1 Infected with 11; error at address 0x101996dc(build_res_row+0xa4);kernel SQL causing err:sp_bindrule r_ddls_01,"ddls_01.dt8".kisignal->terminate_process->ucbacktrace 76045 1 T1 and T2 are xacts that run concurrently - T1 fills the log and is suspended. Now a rollback tran on T2 also fails and is suspended. PLC log space reservation scheme has set aside log space. So there should be space to write the ENDXACT log record for T2 76053 1 2500 in-place updates of a single column in a 10000 row table is about 50% slower on System 11 compared to 4.9.2. Scripts to recreate the problem is in /remote/qa3/bugs/76053. 76064 1 Hard 605 error during dump tran with no_log. checktable results in 2525 with matching pages. 76067 1 Offshoot of 75018: Memory loss on Sun4 (with 8 byte alignment) on plan PROC_HDRs was found to be from 2 - 20%. With fix for 75955, this may be halved, but can be upto 10%. Investigate possibility of optimizing memalloc() to search for first page with .. 76070 1 The HK is not issuing batches of writes. This is because it thinks every buffer is pinned to the log. Sybase Confidential Page: 68 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 76112 1 "sp_setpglockpromote "server",NULL, 300, 300, NULL" prints message: [Please specify a non-NULL value for 'new_pct', since it has not been set previously with a non-NULL value.]even though the server wide values are always non-NULL and dont need (cont'd. 76113 1 Diagserver "SQL Server/11.0/B/Sun_svr4/OS 5.4/2/DEBUG/Tue Sep 26 03:39:15 PDT 1995" stack traces if you run "update statistics" on a table bound to its own cache. Error: "Brought page into wrong cache!" Stack: ucbacktrace>getpage>getpage_noscan>... 76132 1 The config values added as part of 74232 do not have the correct insert and upgrade versions in cfg_options. 76155 1 Running out of freelock structures may result in a seg fault, instead of a 1204 error. maxonlineengines is 1, "number of locks" is the default 5000. 76168 1 q_bld_sublist() allocates alignment space for every alignable column which appears anywhere in the query. Check_align() should allocate alignment space only for columns that might need it. Related to 75018. 76176 1 Problem message misleading when a full transaction log prevents a drop database (even after rebooting). 76177 1 Create a new master device. Use default configuration. sp_configure "total memory",20000. Do not reboot. sp_cacheconfig "newcache","12M". This gives an error message "Msg 5873, Level 16, State 1: Procedure 'sp_cacheconfig', Line 478:, Invalid configura 76180 1 Create a new master device. boot. sp_cacheconfig "newcache", "1M". reboot.sp_cacheconfig will display the new cache configuration. Shutdown. Delete the configuration file. reboot. sp_cacheconfig 76183 1 950925 Build: In Mircosoft Catalog Stored Procedure test suite: procedure sp_column_priviledge, sp_statistics, and sp_table_privileges do not return the same results as the canon; run mcsp/[clpv, stat, tbpv] test suite parts to reproduce diffs. 76192 1 Code routines needs to be updated since the 260 trace flag was removed. 76209 1 regen sql builds run 27-Sep-1995 all failing during sqltest.sh 76240 1 sp_setpglockpromote wrongly reports "The object_type column cannot be NULL" and returns 1, but updates the config file anyway. 76242 1 When memcreate() returns NULL in site.c and dataserver.c, it is not handled. Then eventually the server would throw a SEGV. Sybase Confidential Page: 69 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 76243 1 When memcreate() returns NULL in cursfix.c, it is not handled. Then eventually the server would throw a SEGV.This bug is discovered from bug 73505. There is a potential problem in the code. 76256 1 Memory corruption caused by an incorrect macro BYTESLEN, which is used only by the diagnostic tests. 76264 1 he routine tdsrecv_declare() in demo/tdsrecv.c should have an error handler for memory installed. It calls css_alloc() which calls memcreate() without checking the return status, assuming the error handler will catch it. i 76283 1 Compilation on "disk.c": line 1358: vdeno undefined. This is a regression of 75695. 76285 1 The server did not link successfully with the error: ld: Undefined symbol _cfg_notify_done_inproc. This is a regression of bug 76266. 76293 1 If a db, a table, and an index are bound to a cache, then using sp_unbindcache_all to unbind all objects from the cache from master ONLY unbinds the db from the cache. If in the db, nothing is unbound & msg says "Cannot bind", should be "Cannot unbind". 76294 1 dbcc memusage (preceeded by dbcc traceon (3604)) returns erroneous data. Reports Cache Memory = 0. Also, Configured Memory does not correspond to a valid total for Cache Memory plus the remaining 5 fields; Code Size, Kernal & Srv Struct, Proc Buf & Head 76311 1 Cache bindings does not accept owner name. I tried with an owner name server complained "the target object does not exist". Tried without the owner name it succeed. 76313 1 A table & its nonclustered unique index is configured to have its own named cache. Inserts are resulting in duplicate rows inspite of unique index and table corruptions with 605 errors are seen. Stack trace: ...ex_raise>wrongpage>getpage>getrow>getnext> 76319 1 sp_cachestrategy does not allow owner.tablename . Server complains there is no table named dbo in the current database. But it works fine without the owner name. 76330 1 As of the 9/28/95 build, all queries come back with a new summary format (x row(s) affected, return status = #) instead of just (return status = #) 76334 1 In sanity mode, the exception is raised when the index row is not found or when only one row is updated with unsafe unique clustered index. This is related to bug 75617. Sybase Confidential Page: 70 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 76341 1 sp_cacheconfig 'def' , '20M' and sp_poolconfig a 5M, 4K pool. Then, sp_configure 'def' '0'. This sets size of the 4K pool to 0, and also removes all cache configuration information from the configuration file. ALL of it is GONE, including the ... 76342 1 backout backported bug 53388 76346 1 if a database is in single-user mode, and in use, sp_unbindcache_all on a cache with a bound objct reports error message 924, and then 'Transaction count after EXECUTE' message, and leaves an open transaction. 76347 1 set db to 'read only'...sp_unbindcache_all produces 4 error messages (3906, 3602, 6401, and "Tranasction count after EXECUTE..." messages) and leaves an open transaction. 76351 1 kdrdconfig() does not return -1 in some cases of failure during config block checksum validation. Also, the device is not always closed before returning. 76352 1 alloc_proc MEMZERO MAX_MEMPG of sizeof P_MEMALLOC instead of MAX_MEMPG-1. This will cause serious corruption by zeroing out some other memory 76356 1 sp_cacheconfig 'default data cache', '20M' to lock in min size for default cache. Reboot. Shutdown, and edit config file entry to DEFAULT, start server. Attempt to config a 10M, 16K pool in ddc returns msg, stating size of default cache is 1024K... 76365 1 The sp_unbindcache_all does not work. It returns status = 0, indicating success but actually it does not unbind objects from the cache. sp_helpcache before and after sp_unbindcache_all will show the presence of cache binding information. 76367 1 Fallout from checkin of 75018: Extend memusage.c/prochdr_memusage() to add up the lost bytes count for alignment in the PROC_HDR and to report it as part of dbcc memusage. This change was made as part of 75018. 76376 1 compile error when compiling SQL Server. 76397 1 installing script: ins_syn_dblib raises the following error in procedure "insert_syntax": msg 213, level 16, state:4: Procedure 'insert_syntax', Line 12: Insert error: column name or number of supplied values does not match table definition. 76398 1 fix of 75038 breaks memusage on diag server. 76403 1 diag_get_cacheid has a path where the cache id is not returned properly. This will cause problems for page diagnostics. Sybase Confidential Page: 71 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 76412 1 inserting a row with duplicate key into a clustered index exclusively locks all pages in the overflow chain for that key. Should lock only the affected pages. 76414 1 config/conf007.out: sp-configure 'devices' to default gets error: You cannot set devices to less than value already defined in sysdevices. 76418 1 After deletion of a cache status bit in sysconfigures has been set to both 64 and 128. 76419 1 In decision(), we attempt to short-circuit join clause processing if there is < 2 tables participating in the step. The implication being that there is no join. Correlation conditions are treated as join clauses however, so this assumption is wrong. 76420 1 While backporting bug 53388, some irrelevant fixes were spotted such as TCB_SUID should be used instead of SA_SUID, and ex_callprint() should be used in place of ex_logprint() to deliver the warnings to users. 76421 1 If a table is the only object bound to a cache it is not being noted for optimizer calculations. In cases of indexes it works fine. As part of this fix zero out work_cost structure in q_score_index at every iteration of index scoring. 76423 1 if one attempts to start a server and some internal structures have changed and one does not run buildmaster, the server will stack trace in dsinit+0x3c5, usedb+0x69, ... The server should send a message to run buildmaster and not stack trace. 76431 1 When server is booted with a configuration file with only read permission to every body i.e 444 server gives a error message config file xxxx.cfg doesnt exist. This can be misleading. 76439 1 NEEDED FOR SSM DEVELOPMENT: Micro-help for sp_config variables is not available (Aug 18 build) 76440 1 SSM 10.3 REGRESSION: SQL Causes permission dialog to break: create table #bob (f int, f1 char(5)) | create unique clustered index bobidx on #bob(f) with ignore_dup_key | insert into #bob values (1,'aaaaa') | insert into #bob values (2,'bbbbb') | insert #b 76477 1 Error 1129: Update to OAM page could cause a negative page count. User1 inserts rows within a TX until log fills up and goes to sleep.user2 then kills the user1's process. Test case in /remote/qa3/bugs/76477 76482 1 sp_setpglockpromote "database", master, 100, 100, 10 reports "Incorrect position specifier in format string of PRINT command." but works otherwise. Sybase Confidential Page: 72 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 76486 1 After a crash occurs a page that has been deallocated in a transaction that has not commited may be subsequently allocated again even though it is still in use. Probable solution is to make DEALLOCs be Single Log Record Transactions. 76499 1 Using sp_poolconfig to create a pool in the default cache where the run size of the 2k pool shows enough memory fails with a message saying the total size of the default cache is only 1MB. 76506 1 Bcp-in resulted in timeslice followed by error 1205 (dead lock vctim) and error 1134 (Failed to get lock on OAM page). At this point server ran out of log space and spews out error 1105 & 4204. On reboot recovery hits error 3904 (Can't unsplit page...). 76509 1 Msg 1205, Your server command (process id xx) was deadlocked with another process ... issued when dump transaction running in a while loop, also Msg 2714 Procedure 'runner14', already an object named 'slcn2_14' in the database. 76514 1 951002 build:aggsubq/agsq020:line 72: select * from test returns 0x00d104, and Canon returns 0x00. Also seqncr/seqr051.out: returns NULL and canon returns 0x00. 76528 1 sp_unbindcache_all that includes a db with the "don't recover" status results in Msg 930, and the "Transaction count after EXECUTE..." message leaving an open transaction. Error handling problem. 76530 1 dbcc memusage reports "Cache Memory 0" which seems incorrect. In its System 10 incarnation it used to return Page Cache. We need a way to find out size of data cache for configured server. Per maxb /drubin this seems to be a bug. Need this functionality.. 76573 1 Looks like sp_dropmessage 50001 is not working properly. It says it deleted the message but immediately server is returning the details of this message 76593 1 Tried to create a buffer pool within isql and sql server terminates the process. 76597 1 In nist test suite the following tests fail: 114, 224, 267, 277, 404, 405, 406, 408, 443. 76609 1 access violation when materialized view has tables which are dropped and recreated with different datatypes 76621 1 Need to change the maximum value of "number of devices" back to '255' and change "sp_configure" to manipulate the negatvie number in "sysdevices.low" correctly. Sybase Confidential Page: 73 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 76649 1 If default data cache has 20M and then if you issue sp_cacheconfig "default data cache", "10M" and reboot, subsequently you will not be able to create a new named cache which is > 9.5M because the configured value for the default data cache is 10M. But 76652 1 If a named cache has 4M 4K pool and if you delete the 4K pool only a part of 4M is transferred to the 2K pool even if the pool is not in use. 76653 1 In exerdes test get "timeslice -501, current process infected.." followed by STACK TRACE: _coldstart->conn_hdlr->sequencer->s_execute ->dropdb->dropdb->dbt_dropdb->des_discardall->des_udesfree->_fini_. 76699 1 While trying to modify the cache type from mixed to logonly the proc checks to see that no non-log objects are bound to the cache. The fix for bugid 75262 skips databases which are offline. However, these databases may have non-log objects bound to the - 76718 1 Upgrade through sybinit can destroy entries on some OAM pages causing 7939 error when dbcc checkalloc is run after upgrade. 76722 1 without reboot: sp_cacheconfigure test, '5M', sp_cacheconfigure test, '0', sp_cacheconfigure test, '5M', status is still 'Pend/Del'. Should be 'Pend/Act'. procedure should test and clear the 'del' bit. 76730 1 Stack trace: : "current process (0x5b0025) infected with 11 -> Error at address 0x1013fae0 (save_eopstate+0x8) -> SQL causing error : ^A^Aa_curs1 -> curdb = 7 pstat = 0x10010000 lasterror = 0 -> preverror = 0 transtate = 0 -> curcmd = 46 76744 1 in mirror testsuite server is hanging after testcase mirr006e and the server process will not go away. Thus every time from there the server command fails with the message: kdconfig: unable to read primary master device etc 76750 1 sp_helpcache should display the status (Valid or Invalid) of the binding. The current output just displays the binding which is misleading. 76756 1 The 'Memory Used' column of sp_configure output always displays the value 0, regardless of the memory consumption of a particular setting. That column may rely on the same memory-calculation code that's spozed to be used for sp_helpconfig. 76761 1 Compilation errors on cursor test suite. 76762 1 Table Corruption:The entry is missing from OAM for object id xxxxx inid for allocation zzz. Sybase Confidential Page: 74 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 76786 1 "sp_configure 'number of devices', 0, "default"" got the error "The value of the 'number of devices' must be greater than the highest VDEVNO, '0', defined in sysdevices." 76802 1 Drop database - Revoke SSO role and try to drop the sybsecurity database. Though it gives an error message, it will drop the database. 76808 1 The column, memory_used, in syscurconfigs is set to 0 for all configuration parameters. The purpose of this column is to reflect the amount of memory used for a particular setting of a parameter. 76811 1 If the sysaudits table is dropped from sybsecurity database and then if the sso role is revoked and the sysaudits table recreated, the server creates the table though it displays a lot of error messagese 76827 1 No lock promotion when locks = HWM. Set LWM=HWM=PCT=2 The server should promote to table lock when the second lock is requested. 76864 1 Compiling dbccdiag.c results into warning "function expects to return value: d_page". 76881 1 605 error on SYSLOGS encountered when trying to load the master database. 76882 1 Msg 7930, Table Corrupt: keys in left child is not less than the parent key; check left page #nnn. Error in zeta, zeta4 and zeta18 "cougarised" suite parts - zeta11, zeta4_11, zeta18_11 respectively. These test named caches, table partitions and config 76906 1 The default value of 10 for the housekeeper free write percent seems to be too aggresive. Out of the box performance seems to be hampered when the customer tries to response time measurement with single user on the server. This has been seen by other 76921 1 Stack trace from s_compile() -> s_compile_stmt() when running Cobol batch application using dynamic SQL ct-lib cursors. Seems to be regression from fix for 76730. 76929 1 sp_cacheconfig shows the default data cache size is 9.23M. I checked this value for correctness by selecting the value from syscurconfigs. Assume current configured value for "total memory" is 20000. Now issue, sp_configure "total memory", 20000. 76938 1 sp_setpglockpromote "table", "master..sysobjects", 2, 2, 1 reports an error (FAILS) but returns 0. 76944 1 prJOINDATA() should not call prJPLAN() if the pcost field in the jplan is NULL. This field is not assigned until a cost structure has been associated with the jplan. These are debugging functions called from treeprint(). There is a window when ... Sybase Confidential Page: 75 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 76945 1 If there are two config files, one of which is cfg.001 and the other cfg.999, and the rename function is invoked, then it misinformis the user that the file has been renamed when in actual fact it has not. 76965 1 Error messages in errmsgid.c are out of date. For example BADID (3403) has the wrong text, starting 'Rec_init: Found that page' instead of 'During recovery initialization' which is the modern version. You can see that the wrong version is 76978 1 sp_logiosize "16" fails eventhough 16K pool exists. This is a regression caused by fix for 74572. 76985 1 Sp_cacheconfig can be used to change the cache type on 'default data cache' to 'mixed'. Without reboot, select status from master..sysconfigures where name = 'default data cache' displays 35 and 33. 76997 1 With sym server: Bind a database to a cache; dump and load the db; at the completion of load there will be a set of 40 some backtraces, of which there are a couple of forms, all ending in "Brought page into wrong cache!". From sanity code. See notes. 77005 1 sprocs sprc011.out diffs on IBM (possibly other platforms): available for caches (nnn) differs on solaris. sp_configure "procedure cache per", 80 gets different results for every run. 77020 1 In exerdbt get Error 8211:"mismatch found between the name and id descriptor hash table. Descriptor hashed by name =0x0 and hashed by id = 0xdd62985c". Followed by STACK TRACE 77021 1 One of the bcp process hangs while loading TPC-C database using multiple engines and multiple bcps going to to different tables. The hanging seems to be in plc__lock() where the plc_lockcount goes to negative and process puts to sleep forever with noonet 77022 1 exerbuff couldn't start dataserver. "kernel:kbcreate: couldn't allocate 68063232 bytes" 77026 1 IF SQL Server configured as japanese server and there are an database which separated data portion and log portion, preupgrade utlity failed by 2753 error. 77031 1 Internal beta site - Sproc created in 1002 and upgraded to Cougar fails when executed with 2805/18/0 - Bad pointer enountered while remapping. Sproc is.p_sel_prodrowsh_sp. Seemed to never have worked. 77033 1 Sproc created in 1002 and db upgraded to Cougar. Sproc fails when executed with 2805/18/0 - Bad pointer encountered while remapping sproc. Sproc.qoe_insert_one_sp_apc. Sybase Confidential Page: 76 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 77040 1 The cache binding for csysindexes is stored into the wrong field in the dbtable, overwriting the binding for sysindexes. 77053 1 The buffer manager should release the cache spinlock across a disk I/O. 77068 1 chache name parameter for dbcc page () command must be made optional, this required parameter will affect all the existing diagnostics and patching tools used by tech support. 77071 1 The height of the index tree is not being considered while generating prefetch strategies. This is a crucial part of the prefetch feature. Basically the cipages calculation in the function page_estimate () has been wiped out during one of the ... 77075 1 In zeta18 test error in zeta001bv.out. Msg 3703, Level 11, State 1: Cannot drop the table with object-id 544004969 in database 5, because it doesn't exist in the system catalogs. 77084 1 disk_init(): if udactivate() fails, exception is raised and opened system tables won't be closed. 77094 1 When SQL Server has named caches configured, "update statistics" gives "Error: 1108, Severity: 21, State: 3, Cannot deallocate extent..." and finally ends up with Error: 913, Severity: 22, State: 2, Could not find row in sysdatabases with database id 0.. 77108 1 when upgraded from 492 to 11.0, the syscomments table doesnt seem to have upgraded properly in user dbs. colid2 should be part of the primary key on this table, which shows up in master and model dbs after upgrade but doesnt show up in user dbs. 77140 1 Disk refit should check allocation pages to be sure they really are allocation pages (objid=99). Error in disk reinit could cause disk refit to read bad pages. 77159 1 Signal 11:bufawrite->cm_plcunpin->xls_bufunpin->plc__flush-> log__newpage ->pg_log_alloc->getsegname->initarg 77174 1 During lots of insert and update activity, SQL Server shutsdown with "Error: 859, Severity: 20, State: 1 Invalid Buffer Cache Access: Attempt to locate page '595314', database '4' in cache failed because the cache id '8224' in scan_context is invalid. 77175 1 Ensure the specified "pool size" and "wash size" in the config file are big enough for the specified buffer pool. If not, the server will failed on booting. 77194 1 Create database on 'nondefault' device generates error 1802 '... no such disk...' eventhough the device (disk) is correct and activated. Create database finishes sucessfully with appropriate '... allocating nnnn pages on ' message. Sybase Confidential Page: 77 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 77203 1 The command dbcc extentdump does not seem to be working even with cache name. I issue dbcc extentdump("default data cache", dbid, page) and get errors. 77205 1 During insert, got "Error: 2610, Severity: 22, State: 1, Could not find leaf row in nonclustered index 'ActTbIxC' that corresponds to data row from logical data page 359453, row offset 7 during update index attempt after data page split." Stack Trace: 77272 1 A sequence that corrupts the index :create table smtable(x int, y float) go create proc procsmtable @x int @y float as update smtable set y = @y where x = @x go insert smtable values (1,1.5) go procsmtable 1,2.5 77296 1 The command dbcc allocdump needs to be changed to not require cache name (see bug 77068). 77297 1 The command dbcc delete_row should be changed to not require cache name (see bug 77068). 77299 1 The command dbcc extentdump should be changed to not require cache name (see bug 77068). 77301 1 The command dbcc pglinkage should be changed to not require cache name (see bug 77068) 77302 1 The command dbcc usedextents should be changed to no longer require chache name (see bug 77068) 77303 1 The diagnostic menu used for patching requires a cache id. Currently there is no way to determine the cache id as this is generated when we init Resource->rcaches. (see bug 77068; either remove cacheid or provide a command to report it) 77305 1 Msg 515, returns wrong table name (returns source table instead of target table). 77307 1 Ran 'sp_auditsproc myproc, mydb, "both"' and got Msg 913, 22, 2 "Could not find row in sysdatabases with database id 0. 77309 1 Regression from fix for 76168. com_align_estep() incorrectly scans the subordinate EOP chain, missing EOPs deeper than 2 levels. This produces a stack trace for queries with 2 or more in-clauses. 77312 1 The configuration option "event buffers per engine" comes with min value of 0. If the user sets the "event buffers per engine" to 0 and then starts collecting events through monitor server clients, the server will seg.fault. The minimum value should be Sybase Confidential Page: 78 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 77376 1 Binding tempdb to a cache will corrupt the master database. Rebooting will fail with: Rec_init: Found that page %d had non-log object id %d while checking Syslogs allocation. 77416 1 server gets SIGSEGV in xpr_param->traceprint->_fini and closes client connection, when a 'dump database' is issued immediately after a 'dbcc traceon(8006)' command. Can reproduce using 'dump database model to '/tmp/model.dump''. Does not occur in sun4. 77420 1 Stack trace occurs in test suite when several users concurrently insert/update/delete replicated tables through replicated stored procedures. 77509 1 The query: grant execute on sp_procxmode to public, in installmaster gives error 710, 'Memory request for 0 bytes is too small." 77510 1 These files have one or more instance of passing NULL sdes to ind_dbinfoget(), resulting in default to currently open db (in API between non-dbmgr modules, db should not default): dmpxact.c, dumpdb.c, ldxact.c, loaddb.c, and recovery.c. 77552 1 Dbcc checkdb reports a 605 error after aborting four bcps running concurrently on a table with a varchar and image column that has no indexes. 77567 1 While running installmaster, mkvarchar() calls memalloc() with a request for "0" bytes. Because of this bug, bug#77166 cannot be fixed. Please refer to /remote/qa3/bugs/77166/README to see how to reproduce this problem. 77585 1 q_find_var() makes an assumption that less than 257 VAR nodes are in an expression, since critical decisions in JOIN and SUBQUERY transformations use this function, it needs to be fixed or wrong answers will result 77592 1 set db to 'read only'...sp_unbindcache_all (correctly) produces error 3906, but with erroneous db name. 77600 1 dbcc checktable on a table that has nonclustered index bound to a different cache can cache a index page into a different cache causing corruption. This is very important for diagnostics 77636 1 A MP window exists in the PLC log space reservation scheme, where a log record could exist in the PLC but we could fail trying to allocate space to the log to write out this record. This is because, if a record cannot fit into the PLC we unreserve ALL log 77637 1 sp_auditdatabase, sp_auditobject, and sp_auditsproc fail w/ message 913 'could not find dbid 0!'. looks like audit_option() is not func. properly. Sybase Confidential Page: 79 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 77643 1 In the PLC log space reservation scheme, a DBTABLE status is checked after reserving log space to see if the LCT was crossed. It is possible that this reservation succeeded but a different process crossed the LCT setting the status - this could lead to 77659 1 There is a merge error in srchindex. A couple of lines are repeated one after the other at line 359. search for "srchlock_info->state = &state;" 77660 1 The server sets sysdatabases.status2 to -32768 during recovery to indicate a database with no dedicated log segment. This value should be something more reasonable, so it is easier for users to interpret (current values are 1, 2, 4, 16, 32). 77692 1 Once a 16K buffer pool has been configured, one cannot get rid of the pool. its possible to move part of the 16K pool back, but not the entire amount. 77703 1 Holding the buffer manager spinlock while starting an I/O causes many extra spins unnecessarily. We can drop the spinlock when washing a buffer and during checkpoint. 77721 1 pgd_showpglinks in diag/pagediag.c assumes that a table's clustered index and its data pages always belong to the same cache. This could result in stale data being presented to users of dbcc pglinkage. 77725 1 Syntax errors when installing sp_monitor during installmaster. 77743 1 Server should not do a table scan for an update statement containing a 'where 1 = 0' clause. 77750 1 Installmaster fails at creating the sp_monitor stored procedure. 77753 1 sp_cacheconfig 'def', '20M' gives wrong error message. The server does not have enough memory for create a 20M cache pool. So the error should have been RECONF_CACHE_TOO_BIG. Instead, it generates RECONF_INV_DFLT_2K_POOL. 77763 1 Dbcc dbrepair (readpage/writepage...) read in a page using the old method of opening sysobjects, changing the resulting sdes's current pageid to the desired page, and calling getpage(). This can result in duplicate copies of pages being placed in the ... 77795 1 sp_helpcache produces errors when reporting binding information if a database is single-user and is in use by the user running sp_helpcache, even if the database has no binding to report. 77798 1 sp_unbindcache_all seems to do nothing. If you unbind a cache and then run sp_helpcache, it still shows the binding. Sybase Confidential Page: 80 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 77801 1 Increase memory to 12500 pages. Then sp_configure 'user connections', 239 says 'The total size of all configured caches is (1104k) is greater ... 77802 1 sp_configure 'user connections', 240 refers to 7 extra file descriptors. This doesn't agree with value (per process limit of file descriptors [256] - @@max_connections [248]), which is 8.' 77852 1 Benchmark found a regression for maximum number of configured users. After using sp_configure to configure 5000 users, shutdown, restart fails with "The sum of 'number of user connections' and 'number of devices' must be no greater than . 77853 1 Boot the se 5M 8k pool with a wash size of DEFAULT. Now issue sp_poolconfig "cachename","2M","8K". This command will fail with the following message. Msg 5867, Level 16, State 1: 77869 1 the monitor counters do not provide sufficient detail to tune multiple buffer pools. There needs to be an indication of whether pool sizes are correct, whether large I/Os are being used efficiently and whether MRU replacement is being used. 77873 1 When a pool is deleted there is no warning message about the objects which use that pool. Example, if 4k pool is deleted and if the log io size is set to 4k. 77877 1 In exerbuff test DIFF in bind.out:sp_bindcache 'cache9', bufferxdb, 'account4.index4' The target object does not exist.(return status = 1). 77880 1 lock unit tests(/remote/user5/qa_1100/tests/exerlock/SOURCE/lkunit.s ql) fail with a time slice error(on svrsole2). 77900 1 SQL Server now uses -C switch to designate the configuration file. This conflicts with prior use by NT which uses the -C switch to indicate that the SQL Server was started by the service control manager. 77901 1 config/conf009.out: 1. Cannot reset 'remote sites' back to it's default value. 2. It's value (set to 32767) prevents server from re- starting. If server cannot start with this value, it should not 78006 1 Message 5861 advises user to run sp_helpconfig. This needs to be changed, since we are not implementing sp_helpconfig. 78015 1 Stack trace during recovery of a read-only db with invalid cache bindings 78021 1 Catch 22: if db is read-only, and cache bindings need to be marked invalid when issuing 'online database', you get error 3906, and online database fails. and you can't change read-only status because the database is offline. Sybase Confidential Page: 81 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 78042 1 When a database dump is loaded on to another Server with different Buffer Cache configuration and "online database" is issued, "sysattributes" gets modified to invalidate cache binding. This will make it impossible to do any "load transaction" afterwards. 78050 1 Whenever installmaster is run, the columns suid, heur, and xid are removed from spt_committab. These columns are required for sites using XA server and will cause severe errors when starting XA server. 78073 1 Locks are not being released during certain backout situations. 78114 1 During stress/performance testing, we intentionally corrupt a database in order to prove we can reload it. The corruption caused an OAM page to be corrupt. Upon reboot, we discover that the oam page is not an oampage in pg__lockoam and raise an error. 78129 1 cache cannot ve viewed by non-SA user with sp_cacheconfig: "EXECUTE permission denied on object sp_cacheconfig, database sybsystemprocs" The sa-role is not needed to view a cache. 78210 1 Sys. Admin. Guid (page 10-17) says "When you upgrade an existing database, it does not automatically acquire a last-chance threshold. Use "select lct_admin ("lastchance", db_id()) to create one." Should not upgrade do this? 78221 1 select inttohex(-1) returns Msg 265, Level 16, State 1: Line1: Insufficient result space for explicit conversion of INT value '-1' to a VARCHAR field. A side effect of this problem is that inttohex returns only a 7 digit hex number; it used to return 8 78237 1 Error 574 occurs if one tries to change old password howdydo02 to new password howdydo03. If first 8 chars are the same, and the 9th char differs only in bit 0, this error will occur. 78241 1 If a transaction is sleeping on log full and you kill it with 'kill' or control-C it will not be aborted correctly. This is a regression caused by the fix to 76176. 78267 1 CRT_LOCK_FAIL (error message 2720) has text missing from sysmessages. Even the comments in cerr.h dont have the error message discription. 78291 1 Error 605 occurs when multiple scripts doing drop index, truncate table, create clustered index, and insert into table are run. However it does NOT corrupt data on disk! This is a soft 605 error 78294 1 Using cursor select at isolation read uncommited in embedded C (v 10.0.2) causing process infected with 11 and stack trace. Followed by error 6103/17/1. Sybase Confidential Page: 82 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 78338 1 A simple update statement like this "update item set i_price = i_price * 1.1 where i_id=" came out with wrong result while running the HP benchmark at Cupertino. for example i_price value before the update statement say 1145.0 and afte 78343 1 If "sp_thresholdaction" fails, suspended task fills-up the log resulting in "Error: 1105, Can't allocate space for object 'syslogs' in database ...". A thread suspended due to insufficient space in the log should be awakened only if sufficient space is 78345 1 Create clustered index i on dcol_05 (col8) with allow_dup_row failed and returned "Failed to reserve/initialize allocation hints space in index 'i' on table 'dcol_05'." 78347 1 xactman/tranman/tranmgr001: "set role "sa_role" off;sp_logiosize" returns: You must have the following role(s) to execute this command/proc:'sa_role'. In canon, the sp_logiosize is executed. 78355 1 Subqueries like : "select dbid from sysdatabases where dbid not in (select dbid from sysusages) for browse" stack trace end raise error 533 "Can't find a range table entry for range 3." 78397 1 When a new buffer pool is created sp_poolconfig does not correctly recompute the wash size of the 2k pool (it has fewer buffers now) Resulting Error: "The wash size for the 2k buffer pool in cache cache1 has been incorrectly configured."a 78428 1 Fix to 76045 would cause XACT_IN_ABORT bit to be set by transactions rolling back to a savepoint as well as those in abort. This will result in transactions that roll back to a savepoint bypassing the XLS log space reservation scheme. 78429 1 While running TPC-C on DEC-UNIX we got error 624. The requested update to the OAM for object -589840, dbid 2, index 1, oam page 624, would cause a negative page count. allocation page: 512, oamarray used: 3, oamarray unused: 13, deltaused: -2, delta ... 78431 1 Test case login/logn023[ab]: an all-privileged user attempts to lock the account of a user who is currently logged in; "sp_locklogin , lock" hangs, whereas it should have issued a warning but lock the account anyways. 78457 1 If udstartio() returns failure status to bufawrite(), bufawrite() tries to free the buffer() by calling udfree() before raising the fatal exception. But udfree() never frees the buffer because the dbstate in BLKIO structure is not set to DBCLIENT. 78477 1 UPGRADE failing with compile error "undeclared initializer name TEN025" in header file. Sybase Confidential Page: 83 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 78567 1 When there is a search clause on first column of index, and an OR clause on second column of index, and optimizer chooses OR strategy, showplan will say "positioning by key" for dynamic index, instead of "positioning by RID". 78620 1 If buf__writewait() raises an exception due to MASS_IOERR bit set, when it is being called through housekeeper()->hk__chores()- >hk__pool_wash()->buf__writewait(), atleast one cache spinlock is not released. Next time you try some write operation, ... 78621 1 when disk initing more than 128 devices, sp_helpdevice reports negative numbers for vdevno. The devices are actually created. When a create database is attempted on the devices which have negativce vdevno, it comes back with a message stating there isn't 78624 1 If a database is in single user mode and threshold stored procedure "sp_thresholdaction" is installed, we get Error 924 and bad messages when Server tries to activate the threshold stored procedure. 78654 1 server intermittently stack traces when raising error 820 while errorlog's device is full: ...log__newpage->pg_log_alloc- >pg__getpgbuf->bufdirty->bufpindirty->ex_raise->s_handle- >hdl_default->terminate_process->ucbacktrace 78667 1 Full functionality of dbcc log() is not available with dbcc xls(). There is no way to search all PLCs for records affecting a specific object, page, transaction. PLCs are searched by process spid. 78671 1 sp_helpcache output is not displayed properly. The values in the cache section are not aligned properly. 78697 1 The join optimizer does not always completely initialize the join plan. jplan->ppath is not filled in by j_nextvar and j_callsearch, leading to a segmentation fault in j_getmru_strategy. jplan->ppath should be set to range->rgsclause. 78716 1 An user owns a table. SA grants sa_role to that user. The table that was owned by that user is now unaccessible. Error 208 raised. 78724 1 sp_addlogin in master can only create new login but cannot set password correctly. I have the script to reproduce the problem. 78758 1 70943 did not address the case where the cursor is scanning outer table being updated via an NC index but the change is only to the data page. Cursor fix up code disconnects page, buffer, etc. from SDES and continues without checking and updating ... 78777 1 [ASID:alter_database4/alterdb_0330/1]When "alter database" is invoked to allocate some space to a database that has 128 devices fragments already allocated then command is aborted but sysusages contains 129 rows. Sybase Confidential Page: 84 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 78791 1 [ASID:disk_init3/diskinit_0270/8]When "disk init" fails with an invalid size parameter, and an attempt is made to use the same "vdevno" for another "disk init" without restarting the Server, the command does not fail. 78804 1 Alterdb: database has been filled up and extended; then several more long strings are inserted to the database; server disconnects. 78805 1 Backupserver autodump disk tests (bzautod, b/crwautod) all fail with Msg 930|14|1: DB cannot be opened because either an earlier system termination left LOAD DATABASE incomplete or the db is created with 'for load' option. 78806 1 beta 2-refresh: stress test. altsort/alts000 hangs indefinitely on all 3 pltaforms: Solaris, HP & RS6000 78807 1 beta 2-refresh: stress test. zeta18 hangs indefinitely on all 3 pltaforms: Solaris, HP & RS6000 78808 1 Beta2 refresh: exer lock stack trace after mesg:"A process to be waiting for the same object. 78815 1 Internal testing on text data hangs indefinitely on 3 platforms (HP, Sun, IBM). 78816 1 isolevel/busy_all hangs on all 3 platforms (HP, SUN, IBM) with the message "5 tasks are sleeping waiting for space to become available in log segment for database isol_db01" 78817 1 Regression found by the 'sort' test: 30MB db, 10K row table. Expected to handle insertion of 20 rows per unit time (total=500 unit time); the 11/5/95 build maxes out between 312 to 445 unit time, then hangs. Doesn't help to decrease rows/unit time 78841 1 Bugfix 77272 has taken a conservative approach of considering all the indexes to be unsafe when updating a table having a non- clustered index.This is causing a performance degradation while running the TPC-C benchmark.TPC-C has a table with non-clustered 78852 1 Stored procedure process infected w/ 11 memmove ucbacktrace terminate_process kisignal proc__fetchphdr proc_read ex_cleanup execproc s_execute sequencer ex_cleanup s_execute sequencer ex_cleanup conn_hdlr ex_cleanup kpinitproc upspawn spawn_conn_hdlr >>> 78865 1 There are some typos in the configuration error messages. Although this will be viewed as low priority, I believe we should avoid the embarrassment of having a customer point them out to us by fixing them ourselves first. Details in "Notes". Sybase Confidential Page: 85 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 78876 1 Union of char and varchar datatypes produces result of type char with length of unioned char column. Should promote result type to varchar and use length of longest column in union. 78904 1 sp_poolconfig reduces the default 2k pool to less than the configured value, deletes the 2k pool entry from the config file , and resets the configured value to 0. 78921 1 Transaction log got full due to large inserts/deletes to a table and then user does ^C or kill to isql session. Now the database cannot be dropped. sp_who shows no one currently using this database but drop database complains that database is in use. 78990 1 Flags in sysobjects for repserver 11 support need to be reassigned when upgrading from 4.9.2 and 10.0.2 78996 1 Customer is getting 605 errors when "select ... into #temptable..." Happened on a server built on Oct. 20, 95 with fix for bug 76313. Stack trace: ex_raise>wrongpage>getpage>getpage_noscan>pg__lockoam> pg__updateoam>pg__pgalloc>pg__set_allocbit_oam_loglo 79042 1 Subquery performance regression from 4.92 to 11.0, due to flat joins used in 4.92 and exists joins in 11.0. Since a distinct is placed in the main query, the exists joins can be translated into flat joins but currently they are not 79051 1 In include/oam.h file, the macro OAM_SPACE(p, n) does not take into account of attribute entries for calculating the amount of free space in the oam pages 79061 1 Undetected deadlock between bufawrite() having set MASS_WRITING during checkpoint and ROLLBACK TRANSACTION sleeping to set MASS_CHANGING, when ROLLBACK TRANSACTION (apparently) has PLC lock and bufawrite() is sleeping on the PLC lock. 79066 1 sp_poolconfig looses track of pool memory. cache1(4M){ 2k(3M), 4k(512k), 8k(512k) }. sp_poolconfig "512k", "16k" reports: Invalid pool size of 428k (214 buffers) encountered for the 2k pool 79099 1 Regression from 71524. A sequence that corrupts the index: create table smtable(x int, y float) go insert smtable values (1,1.5) go create index ind_smtable on smtable(y) (note NO go here) update smtable set y=3.5 where x=1 go dbcc checktable (smtable) go 79102 1 Change bitflag O2_PROC_SUBSCRIBABLE to reuse O_LOGGED in object.h. 79133 1 Boot the server with a 3M cache called "foo", 2k pool explicitly configured as "2M" and wash size as "512K". Then issue sp_poolconfig "foo","1M","4K". This will be successful. Then issue sp_poolconfig "512K","8K". This command fails with a ... Sybase Confidential Page: 86 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 79164 1 REGRESSION : A select using a view as the inner member of an outer join returns the wrong results. We seem to return matching rows from the inner table (the view) when there are none ! 79190 1 Boot the server with a name cache "testcahe" with 2KI/O pool=512K wash=DEFAULT,4KI/O pool=DEFAULT, wash=DEFAULT, 8KI/O pool=DEFAULT wash=DEFAULT; after booting issue the command sp_poolconfig "testcache", "512K", "16K". This command will be successful ..t 79232 1 Sort generates 1513 error (Reuse of Extent failed in Sort) while creating a index on a 600,000 rows table on tpcd benchmark. 79237 1 Updates performed against tables in three or more databases may disappear in some of the databases even though the transaction commits. 79255 1 In exerdes get diff in des999.out: Msg 3702, Level 16, State 4: Line 26: Cannot drop the database 'exerdes' because it is currently in use. 79262 1 sp_poolconfig comes out with 821 error conn_hdlr->sequencer- >s_execute->execproc->sequencer->s_execute->execute->exec_eop->run- >cfg_main->cfg__cache_pool_config->cm_coalesce->cm_bufunhash (error 821 severity 20 state 4. Attempt to unhash a buffer and the 79263 1 Server seg.faults when instrumented with diag__bhashcheck(0,1, (long 0) to print and check all the hash buckets in the cache in routine cm_coalesce() after cm__rehash_cache() is done. As per drubin this should work and it was working earlier. 79264 1 Nightly build server(Nov 9th 1995) timeslices when you execute sp_poolconfig "default data cache", "1M", "6K". To reproduce the problem, make sure you don't have any config file in your $SYBASE directory. Boot the nightly build server, do an installmaster 79291 1 This is not a bug. This is a new functionalility that has been added with bug 79061. 79294 1 In com_target.c, line 254 the expression (resvar = -1) is incorrect. It should be changed to (resvar == -1) or removed. 79328 1 Boot the server with a named cache with a 2KI/O pool=512 and wash=DEFAULT; Login and issue sp_poolconfig "namedcache", "2K", "wash = 0K"; Server sets the wash size to 0K which it shouldnt. 79344 1 During insert/update testServer built on Nov 9, 1995 hit Error 614 with stacktrace: ex_raise>rowsize>ind__rowfrombp>ind__fastsearch>ind__findrow>ind_row copy>startscan>exec_eop>execute>s_execute>sequencer>execproc>s_execu te>sequencer>tdsrecv_language>... Sybase Confidential Page: 87 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 79350 1 Error:#614:Whenever a table gets dropped, the Sysindexes MODIFY lock is not acquired within dropo() before to go and delete all Sysindexes rows for that table. That lock is required to synchronize with readers of Sysindexes that use the FAST Mechanism. 79352 1 trace_hang_on_fatal_error() must set its priority to MIDKPRI, else netinit task, always runnable and possessing a Pss, will keep the diagnostic process from ever being scheduled again, because it'l be on a queue never reached by the scheduler. 79354 1 Beta2 refesh: clex/xcurs999: drop database exec_curs_db, exec_curs_db2, clexec_curs_db1 returns: Cannot drop the database 'clexec_curs_db1' because it is currently in use. 79357 1 Running the bufman (beta) test suite, case buf901: Create unique clustered index, sp_bindcache (cache,db,table,index) => then server hangs on drop index 79390 1 This bug has been entered in order to back changes of bug #76634, which opened a window due to some Sysindexes Mgr Callers which do not get the MODYFY lock synchronization as required. (see also bug #79350). 79415 1 Error 821 on sp_poolconfig activity 79467 1 Change the sys10 or system11 password to "stilldogslow". You can still login to the server with "stilldogrlow". One less than 9th character. 79483 1 The bug fix 75046 introduced change to handle the buffer cache pool structures in the optimizer more efficiently. But this has resulted in a number of regressions in the optimizer's calculations. 79489 1 [REPSRV] For safety, the SQL/LTI thread locks the log before a log scan but quickly frees the lock if the starting log page occurs after the LTM truncation page. Unfortunately, this causes the SQL/LTI thread to block often if DUMP TRAN is also run often. 79525 1 In test datslc/dats011a.out a segmentation violation happens in bufu nkeep while doing insert. Stack:_coldstart->conn_hdlr->sequencer- >s_normalize->s_normquery->colnames_driver->colnames_recurse- >colnames->fill_resdom->startscan->ind_rowcopy->.... 79537 1 dbcc monitor() causes stack traces in inpupd test suite. Without this, this test suite cannot be run at all. Seems to work on other platforms (HP). 79601 1 Diagserver prints message 'PLC is not locked as expected:' to console and errorlog. This message would usually indicate a dangerous inconsistency in the XLS subsystem. Stack trace is undo- >xls_getmarker->xxs__lr_getmarker->plc__flush->plc__islocked. Sybase Confidential Page: 88 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 79642 1 sp_helpcache "default data cache" stack traces the server(signal 11) when entities are explicitly bound to default cache: dbase, table, index. 79648 1 When process is terminating, any attempt to kill that process should not result in setting P_KILLYOURSELF bit. This will result in recursive cleanup. One manifestation of this is 1129 error(76477). 79655 1 951114 build:on axposf&Solaris,clex/exec_curs_g.exe stacktrac with signal 11._fini-> kisignal->terminate_process->ucbacktrace. On axposf:kpinitproc->conn_hdlr->tdsrecv_open->sequencer->s_execute- >static procedure (no name)->curs_execute->sequencer-> 79718 1 Alter database on a database that is in "NOT recovered state" (hence marked as 'suspect') produces signal 11 stacktrace. extend()- >alterdb()->s_execute()->sequencer->tdsrecv_language(). 79723 1 Bugfix #72287 reduced number of evals for "if (@x=10)" from 7 to 4. But this needs to be backed out. jerryb suggested that a limited version of the bugfix - only for 0 table conditionals - is possible. That will help TPC-C performance. 79724 1 When deallocating an OAM page we could end up trashing the timestamp on the OAM page previous to the one being deallocated. This could cause recovery on this page to fail, 6902 problems during recovery etc. 79733 1 Backout of bugfix 75957 because it breaks one of the NIST tests and the "clex" test. 79737 1 After a successful completion of sp_cacheconfig , the following message is displayed by the sql server. "The change is completed. The SQL Server must be rebooted for the change to take effect". But the effect seems to be 79739 1 The cache status in the cache descriptor is not updated dynamically when cache status is modified using the sp_cacheconfig stored procedure. Any one relying on the cache status information in the cache descriptor will not see the latest information(for 79743 1 When an execute cursor is recompiled after finding the plan in the procedure cache a seg fault can occur on either the open or fetch on the cursor. The problem is that s_recompile() is not setting the SEQ_CURSEXEC bit in the new tree. This is only done 79755 1 Perform the following steps: 1) create database db1 on dev1 2) alter database db1 log on dev1. Step (2) should fail (and ask the user to specify the alter with a 'with override' clause in order for the alter to succeed) but doesn't. Sybase Confidential Page: 89 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 79760 1 Specifying conflicting option in the *.cfg file used during boot makes the server segfault. To produce the problem specify cache status=mixed cache and in the next line specify cache status=log only and try booting the server. 79761 1 Time slice in sysattr/syat002 test suitepart. The stack is _coldstart->conn_hdlr->tdsrecv_language->sequencer->s_execute- >execproc->sequencer->s_execute->execute->exec_eop->run- >attrib_valid->des_logiosize_notify->etext->_sigtramp->kaclintr->... 79765 1 Error 3225 reports bogus page and disk numbers. 79833 1 "cougarised" zeta18 test suite errors: 1. Table corrupt, entry missing ... 2. The # of pages used and unused do not match the counts in OAM entry.. 79862 1 Process doing a 'while (select count(*) on user table) begin waitfor delay end' is acquiring a lock on a syslogins page and holding it while sleeping on the waitfor alarm. This is demonstrated by the symptoms described in bug 78431 where... 79863 1 Error 1251, preallocated semawait already in use, in txtzeta. Not a simple self-deadlock: higher-level holder of log semaphore depends on unchanging state of end of log. 79867 1 config/conf001 test aborted after case conf008a(succeeded)- Test Driver: ERROR: Message 1895, Couldn't configure number of engines on-line. Happened on Solaris, axposf, Hp & RS6000. 79890 1 Attempt to insert duplicate key row into table with unique index gives Msg 2601 but also results in "The SQL Server is terminating this process." isql session terminates abruptly. 79891 1 Do sp_cacheconfig "foo","1M". Reboot. Check the cache size in syscurconfigs. It shows 1022k instead of 1024k. This is wrong and can lead to bad consequences. 79893 1 When a dump made from a replicated database is loaded to a database that should not be replicated, and the load is at a different revision than the target database, db can't be brought online because it can't (and shouldn't!) be replicated. 79962 1 Trace flag 1204 which prints out deadlock information does not print the new shared address lock type correctly. This needs to be fixed to avoid confusion while debugging deadlock issues in the field. 79973 1 rgnondiskres->rgdes is not set up correctly for certain queries. if a table is referenced more than once in the same statement (i.e. in 2 esteps) then rgdes is always null in the 2nd range table. the problem appears to be that the code that sets up... Sybase Confidential Page: 90 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 80014 1 951120 build:on NT&axposf:Run installmaster, hit Msg 1105 "Can't allocate space for object 'xxx' in database 'sybsystemprocs' because the 'system' segment is full; blocks many,many server and backup server tests. 80017 1 dbcc checkalloc returns Msg 7940, 'The counts in the OAM are incorrect...", on svrsundly /svrsundly_nttst1/tst_sql/951120_run/dropcol/dcol000/OUTPUT/dcol011 80030 1 getting an error 925 on a sproc that executes normally under system 10 and meets the rules concerning number of used db's. reproduceable in house. 80061 1 dbcc database command shows the following errors. Msg 7939 and Msg 7940. Detailed output is available in /remote/qa3/bugs/80061 80071 1 In the config file if the cachename has an extra space at the end before the "]" character , and if the server is booted weird things will happen. Symptoms:sp_cacheconfig wont show the cache, sp_bindcache will fail etc.,, Example: [Named Cache:PS_JOB ] 80080 1 951121 build:Stk trace-ief/ieft011: insert into multselfreftab values (30, 250, 100, NULL)". Also see bugid 79890.infected 80081 1 sp_logiosize "all" is broken. Error: 1265, Severity: 20, State: 1 A lock manager routine expected a lock record as an argument but was passed something else. 80121 1 The file cfg_ver.c does not compile due to invalid prototype definition 80132 1 Builds since 11/24/95: boot up new server & run installmaster -> creation of sp_instmsg fails - command to insert sysmessges gets Msg 259:16:2 "Ad-hoc updates to system catalogs not enabled." 80137 1 Occurence of error 806 with stack trace during normal processing. The error occured twice on aix and once on hp. The errorlogs containing the stack traces are in files errorlog.hp, errorlog.rs6000.1 and errorlog.rs6000.2 in /remote/qa3/bugs/80137 director 80141 1 The option "set table count" does not work for stored procedures. When a stored procedure is recompile the setting for the user executing the proc is used instead of the one set in the procedure. 80178 1 Server stack traces with sig 11 when executing a proc which requires a deferred referential check aftert the foreign key constraint is dropped. Regression from 71330.Bug script:/remote/qa3/bugs/80178 Sybase Confidential Page: 91 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 80191 1 In Swiss Bank test get Error 605 followed by STACK TRACE:dmpx_truncatemain->dmpx_scanlog->getpage->wrongpage->ex_raise 80199 1 timeslice -201, current process infected followed by Stack trace: xls_close->hp_excleanup->trunctab->putxact->end__xact->xls_close- >failed 80205 1 Server get 'timeslice -501' error while running miscellaneous sp_poolconfig commands. Server is doing a checkpoint. Stack trace: checkpoint->des_checkpoint->bufwritedes->cm_writedes->kaclkintr [fails]. 80206 1 Recovery FAILS. Restart after bug 80205. First get a 'non fatal' diagnostic from cm_bufunhash saying that the page number (57249) in the buffer header for the buffer being unhashed does not match the page number (56756) in the page. 80268 1 In cfg__cache_binding_check(), after a closedb(), we are not resetting the copy.dbt_opened variable to FALSE. In case of exception, this could lead to closing a database, we haven't left open and may result in stack trace. 80270 1 In esql, command 'exec myproc :s', where myproc is defined as 'create proc myproc(@p1 char(30)) as ...', gets error 201: Procedure expects parameter which was not supplied. 80273 1 10.02 subquery regression, which is allowed by ANSI, in that correlated aggregate references are allowed in the WHERE clause of a subquery if that subquery is in the HAVING clause of a parent subquery block 80303 1 Unable to Boot a server with more than 2 GB of memory configured when tried on a solaris platform. Encountered the following 2 problems. 1) The rounding of the passed in size(i.e MAXSHMSEGSIZE) i n os_create_region() makes the shmget call to fail due to 80305 1 sp_checkreswords has not been updated for 11.0. There are some new keywords, e.g. partition, online. 80306 1 Failed self-referential integrity statements could stack trace while aborting the command. The failure could be because of duplicate/incorrect data. This used to work correctly in System 10. Looks like a regression in Cougar. 80328 1 SQL server fails with 644 error when doing more than one row deletion for NOLOG update on sysusages at load master. The sqoff from sdes is not adjusted after each row is deleted. 80482 1 In cm__check_window_state() the cache spinlock may be dropped when buf__writewait() is called. A comment notes that the [window] pointer should be reset to the beginning of the window before continuing, however there is no code which does this. Sybase Confidential Page: 92 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 80525 1 Error 2628 can print the wrong database name in a multi-db transaction. 80697 1 In NIST test: The report.out file lists test 299 as nogo & test 231 is missing from the list. According to CANON both tests should be listed as PASS. 80704 1 In altsort/alts009:Error 2619:"Keys did not match overflow page when inserting row at end of page that has overflow page that is not disconnected" Followed by STACK TRACE:_coldstart->conn_hdlr- >sequencer->s_execute->execute->exec_eop->run->insert->cinsert 80708 1 Server issues misleading msg: "Some or all of the specified privileges were not revoked because they were not granted, or they were granted by some other user." when in fact a revoke operation may have succeeded. 80710 1 In altsort/alts009:Error 629"Fatal attempt to delete clustered index entry for page 12217 - index row contains page 12653 - referenced by index page 13094.STACK TRACE:_coldstart->conn_hdlr->sequencer- >s_execute->finish_cmd->endupdate->dodeferred->def_dele 80758 1 Assume we have a 4k pool in default data cache. When system boots model would be using 4k for the log io size. When somebody changes the binding of model and binds it to a new cache called "foo" which has only 2k pool. The sp_logiosize for model will . 80772 1 If a variable length column is updated and a key column of a unique index or index chosen by the optimizer to look up rows for update, we may get 2610 error and the update will be aborted. (2610 means that index is corrupt) 80777 1 sp_helpcache got Msg 5874: "The total size of all configured cache (21456k) is greater than the memory available for cache (3710k). 80854 1 if dropcol count changes on table referenced by dri, all subsequent queries that have that dri check attached to them will fail with 540/16/2 followed by 225/16/1. ri_save_counts() needs to set rgdelcol to be tabobjp->objostat.objcoldel. 80862 1 sp_helpcache produces extra output "V" at end of each line. eg: default data cache dbase01.dbo.t_table clustered index t_table_cl_index V 80868 1 The SSM Query to get database device properties does not work. To recreate the problem: Create a db device, Run the SSM Query (in an e -mail to nma). We discovered that if you allocate some space from the device to a database, then the query works. Sybase Confidential Page: 93 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 81020 1 [REPLICATION] Text/image replication during BCP is not translating columns in the datarow prior to logging the datarow for replication. Occurs when translations are required and the table is replicated, has text/image columns and is being bulk loaded. 81030 1 302 output is inconsistent. Printing of header and footer info should be moved into wrapper routine with call to q_score_index() to ensure that each type of costing prints at least some consistent and hopefully meaningful info. 81062 1 subquery regression - a_remapvar assumes subquery block 0, but in reformat.c the range entry for the reformat may be in a non-zero subquery block, causing remapping to be missed and wrong answers... 81145 1 Create stored procedure, then reboot server. Execute stored proc and get error 2805 - Bad pointer encountered whlie remapping stored procedure - 2812 stored procedure not found. 81146 1 Duplicate of bug 72037. If sp_auditoption is invoked to enable auditing, server gies a stack trace. 81169 1 Customer running multi-user test - previously hit bug 75541 and received fix (MNE problem), but now connections are hanging. single engine is fine; with multi-engine see "server site_hdlr: No LOGCONN for packet from site 2, channel 0". This causes the 81209 1 Server startup process infects with 11 and core dumps while initializing the master database. 81240 1 whenevr in a user databases created by a user without the sa role, there is a procedure A which calls proc B, first time it is executed it runs fine, but if the procs are dropped and recreated, get error 6701, it still references to old objct and wontrun 81247 1 Correlated expression subquery in SET clause of UPDATE statement is being flattened into outer query block if there is a unique index on the columns in the WHERE clause of the subquery. This is incorrect, and will lead to wrong results if the subquery.. 81249 1 In datasplit.c:ncrid_update() function, the index name displayed when NOLEAFRID error (2610) is raised is incorrect. The name of the clustered index is displayed instead of non-clustered index. 81259 1 RG_LEVEL1 bit is not propagated in view resolution... this could cause unexpected transaction behaviour 81303 1 killing bcp job spid in server while large bcp in was running caused an infected with 11 error, stack trace and servercrash. errlog in /remote/ts4/cases/10044000s/10044109/dseinfo/Errorlogs. Sybase Confidential Page: 94 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 81359 1 After create clustered index with fillfactor = 10, runing sp_spaceused gives 605 error. Object is OAM page, appears to be all zeros. getrespages->?->pg__oambuiltin->pg__lockoam->getpage_noscan- >getpage->wrongpage. 81363 1 Process infected with Signal 11 - server shuts down. Stack: address (xxs__pm_markers) s_execute exectrig trig_getloginfo xls_getmarker xxs_lr_getmarker plc__flush plc__lrtopage xxs__pm_markers. 81386 1 Checkpoint can hang forever after a table has been created in the database to be checkpointed. 81669 1 Prefetch is not being used in non-flattened subqueries. Peoplesoft makes extensive use of covering indexes in their subqueries. The values for the joining columns usually come in the same order as the covering index which makes prefetch on the 81677 1 Three-way join gets bad plan due to negative estimated cost of reformatting. Problem is that sortcost() is using integer arithmetic and returning int32 - change to doub 81767 1 When server is abnormally shut down such as with kill -9 or shutdown with nowait, the identity burning set factor does not take effect until the second time the server is shutdown abnormally. 81924 1 When one of the CIS applications is run on a diagserver, you get a timeslice in collocate(), when a clustered index is created. The wrong value of minlen is passed in, so collocate() gets the size of the row from the wrong place and loops forever.v 81982 1 sp_configure 'configuration file', 'read' gets msg on axposf platform: msg 5852 : "changing the value of '(cmastnet)' is not allowed since it is a static option". 82004 1 Disk unmirror fails & server becomes inaccessible (becoming a runaway process). To reproduce: mirror user db disk & log on primary & sec. devices, waitfor mirrorexit, remirror both pairs (ok up to this pt.), then attempt to drop primary disk permanently. 82040 1 sp_lock showed that the curor's update_page-blk not released after the commit. After ^C to kill the select that was blocked by the update, resubmitting the select worked. 82050 1 q_match_cols() could overflow the COST->ckeymap array trashing memory. This could happen if there are > 17 sargable predicates on the key columns. Right now there will be a separate entry in ckeymap for every predicate on a key column, it seems like ... 82197 1 after dbcc checkalloc (db1, fix), where "db1" is customer database, updates to db1..trade succeed but subsequent commands in batch are ignored! Sybase Confidential Page: 95 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 82204 1 Upgrade program when trying to add new configuration variables in sysconfigures should insert defaults and not some junks values that it gets from old sysconfigures. 82452 1 When customer tries to load a pre 11( 10.0) database dump it goes through fine but when bringing it on line it tries to upgrade and fails with msg 1732: Level 20 & state1 followed by 3451/20/1,3454/20/1. This happens on GA version . In beta it was fine. 82506 1 stack trace occurs after 'exec p_add_dcol_05' in dcolmbg.sql of dropcol test;kisignal->terminate_process->ucbacktrace 82664 1 timeslice error in isolevel/busy_all (a stress test performed at isolation 0). execute>exec_eop>exec_eop>exec_eop>exec_eop>startscan>qualpage>vspri ntf>kaclkintr>terminate_process. Consult: /remote/spgqa/qa_110x/docs/Consults/Nightly/.isolevel 82741 1 function order_costs() in correctly assuems that the max number of indices on a table is 32, so it declares a map:indexmap of size 32 bits. This can cause memory fault if the index id under consideration is greater than 32. 82798 1 The 'total data cache size' parameter does not display the correct amount of data cache displayed by 'sp_cacheconfig'. 82970 1 Some tokens in us_english server.loc are in the wrong format. For example, message #1525 has token %04x. This should be represented as %1!. Other types of tokens that don't get expanded are %u, %08lf, and so on. 82980 1 In the us_english server.loc message file, message #3307 has the version # 10.0.2.3retired. This causes problems during lang install. Is "10.0.2.3retired" a valid version #? 83009 1 stacktrace in the following routines: _coldstart>conn_hdlr>sequencer>s_execute>execute>exec_eop>_fini>kisi gnal>terminate_process>ucbacktrace> 83174 1 Cursor declared with dirty reads skips a row when base table is updated. Consider case where 1. "tab" has 2 col's a & b, 2. b has values 1,2,3,4,5, 3. unique nc index on b, 4. declare c cursor for select b from tab at isolation 0, 6. open c, (cont'd) 83190 1 Fetch on a cursor declared with dirty reads returns the same row twice (even when the row is not updated, or near updated rows). Details: unique nc ind on cols A, B, declare c cursor for select A,B from tab at isol 0, open c, fetch c, update tab such that Sybase Confidential Page: 96 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 83215 1 The builtin function attrib_valid() will timeslice when passed an < SYBPAGESIZE value for attribute int_value and class_id TRANSACTION_MANAGER. 83255 1 When a cursor is declared in a session that is at trans isolation level 0, and opened in a session at level 1, it should operate at level 1 (the level it was opened in), but it does not - it operates in level 0 which is incorrect. 83269 1 As "sa", create a sproc with a name starting with "sp_" in a user db and grant execute to "public". Login as a "public" user and execute the sproc in the user db. Select db_name() --> sybsystemprocs!! 83313 1 SMP server with 8 engines running. bcp data into a table partitioned to 20. After bcp completed, dbcc checktable shows error 2503, table corrupt: Page linkage is not consistent. It doens't show if one engine. 83323 1 The following cases fail under nist test: 13, 14, 15, 240, 241, 242, 275, 281, 282, 283, 435, 444, 485, 497, 502, 507. 83408 1 Two processes lock on tempdb..sysindexes. Blocking process is "sleep ing"; blocked process in "lock sleep". Both attempting INSERT.. Type of lock is shared page. 83421 1 A "select replicate(abc,5) from xyz" where abc is type datetime didn't display error "Msg 257, Level 16, State 1:Implicit conversion from datatype 'DATETIME' to 'VARBINARY' is not allowed. Use the CONVERT function to run this query." 83443 1 Case 3: 0 row affected compared to canon's 3 rows affected when "select * from sysattributes where char_value in ('sysattributes manager', 'valid attribute classes', 'valid attributes')". This case is checking master..sysattributes table gets populated 83520 1 Call of sp_setpglockpromote succeeds in changing lock promotion attributes, yet subsequent issue of sp_helpdb fails to return these columns from sysattributes: attribute, attribute_class, char_value, int_value, comments. 83547 1 The error handling code for ONLINE DATABASE restores the wrong log version ID to the dbinfo structure. 83580 1 When a stored procedure with cursors is recompiled, it can lead to a pointer to deallocated memory in the plan. This could cause many types of errors, including timeslices. 84376 1 Replace 1 line of code removed by mistake in fix 80141. Sybase Confidential Page: 97 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 84421 1 [ASID:dats_misc/datslc_03{3,4,5,6}0/2-5] sp_helpartition doesn't work as documentated: can actually be invoked on a partitioned table on a different db 84533 1 "*" is missing in Msg 5893. i.e. "(number of devices max online engines" should be "(number of devcies * max online engines)". 84750 1 SQL Server creates two shared memory segments instead of one. This is a regression caused by 80303. 85111 1 Within alttab.c:addcol() routine, the Sysindexes MODIFY lock should NOT be released once the insert of the Sysindexes TEXT entry has been processed. The modify lock should always be released at the end of the XACT (in closexdes()). 85599 1 dbcc des stark traces for objid 99. Stack is conn_hdlr- >tdsrecv_language->sequencer->s_execute->exec_dbcc->d_calldes->d_des ->prDES->ind_nextcopy->ind__findrow->ind__fullsearch->ex_raise- >s_handle->hdl_default. 85909 1 rbtrec got stack trace in rbtc016a.sql executing "insert db1..tb12 values (1,"this should stick"),stack trace:ex_cleanup->conn_hdlr- >kpcoldstart->sequencer->s_execute->tdsrecv_lang->sequencer- >s_execute->exectrig->sequencer->s_execute->exectrig->sequencer 87339 1 The ex_callprint() statement at line 193 of getxact.c does not correctly specify the error number - the EX_NUMBER macro must be used. 49349 2 The generic code for kaarlmintr() in alarm/sysclk.c would run a bit faster if this function were called with no arguments. 51139 2 Reduce the amount times we check internal structures for language/server cursors. 56043 2 Resolve compiler warnings for source conditional constructs added. 58018 2 Optimize *group bys* with a small number of groups. 58054 2 Correct the minor error number for abort command. 58873 2 Many statement not reached warnings on sun_svr4 builds (run.c search.c). Due to semi-colons after goto's/break's etc. 59124 2 It is possible to get the following type of message in the errorlog '00:94/08/10 22:53:36.72 server No such message', especially for errors during recovery of master. 59996 2 the ANSI C compiler warning 1549: warning: end-of-loop code not reached. This occured while compiling exerciser.c Sybase Confidential Page: 98 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 60001 2 Mixed mode operations in source file produce ANSI compiler warnings: line 684 & 739:"warning: semantics of ">" change in ANSI C; use explicit cast". 60331 2 Correct header definition that was changed after System10 merge. 60359 2 Now that we are using ANSI style function declarations the isighandler declaration in the kisignal struct will need to account for different styles of signals. 64099 2 Error code uses kcerrlog() instead of kcierrlog() hence no internationalization of error message. 64617 2 Resolve compile warnings on NT. 65528 2 Several of the page manager macros logic does useless calculation. For eg. NUMEXTENTS macro uses pindid and plevel fields from the pagehdr which has no meaning in an allocation page. 65726 2 Resolve ompiler warnings: constant 2147483647 is out of range of short comparison. 65893 2 CURSFIXUP_NEEDED() code in stopxact() should use the passed in abort_pss and not the global Pss pointer. 65940 2 Rework OAM functions. 66014 2 Resolve source compile warnings on diagnostic functions. 66017 2 The code added by 47067 in pg_objdeall() (pg_deal.c) dealing with the case of retainpg non-zero and not the same as the alloc page is asymmetrical with similar code in the same function. 66082 2 Mirror testing caused server stack trace: Error at address 0x100d1fc0 (appnd_log+0x2e8) --> current process (0x190019) infected with 11 --> Error at address 0x100db2a0 (closexdes+0x2c) -- > infect with signal 11) 66258 2 Undetected deadlocks while running concurrent select/insert/update queries. spid 17 (select) blocked on spid 36 (insert) blocked on spid 9 (insert) blocked on spid 30 (update) blocked on spid 17 (select). 66741 2 The 8203 error (severity 6) raised when des__fill() fails in des_get() with a unknown or identity error describes the wrong problem. 67136 2 SYB_STATIC is defined incorrectly in syb_std.h. When debugging it should be set to nothing, otherwise it should be set to static. Sybase Confidential Page: 99 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 67418 2 Random Write Update testing causes server to crash. 67566 2 Internal test generate incorrect sort order when testing sp_helpsort. 67718 2 SQL Server error code 233 needs to have the value '23000' assigned to it as its SQLSTATE value. 67746 2 According to the Ref. Manual, a user who originally created a message (using sp_addmessage) being dropped (using sp_dropmessage) can delete a message. But sp_dropmessage does not let you do that. 67988 2 Server generates a stack trace for "dbcc settrunc(ltm,'valid')" 68051 2 sp_placeobject doesn't allow object names that are qualified with the object owner (e.g. owner.table.index). 68157 2 Select with subquery w/aggregate in where clause returns incorrect results. 68256 2 A view created by select distinct * may cause a failure of the queries containing the view. 68361 2 Password is not being deciphered properly. 68362 2 Opened bug for Cougar tracking - duplicate of 63932 68387 2 delimeted identifiers no longer work. For example, the following is legal: create table "table" ("char" char(5)). The server must support this for nist compliance. 68458 2 The Server needs a SQLSTATE value for error 233. The value should be "23000". 68534 2 sp_bindcache allows non-SA users to [un]bind entities to/from cache 68634 2 deletetexte needs to set the cache id in the XALLOC record before the second call to pg_deallpage 68704 2 Master bug identification for System10 backports to System11. 68705 2 In several places, the code still checks for sanity using #ifdef SANITY instead of #if SANITY. Thus, the sanity code will get executed even if SANITY is #defined to 0 (or to any value, for that matter). 68731 2 sp_movemass doesn't calculate the size of the MASSes in the source pool correctly Sybase Confidential Page: 100 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 68751 2 All dbcc's that take cache id as a parameter need to be changed to take cache name instead. 68766 2 The engine-private variable Eresource is not declared as thread- local storage. This will break MP runs on platforms like NT where engines are lightweight threads. 68784 2 compilation error in files including clkprof.h on sun_svr4. The error message says: .../clkprof.h: line 54: undefined or not a type MSGDSC 68822 2 Internationalize kernel error messages. 68944 2 Message 18084 cannot be found in a_mesages, however, it is raised in sp_addthreshold and sp_modifythreshold. OTHER MESSAGES MAY BE IN AN INCONSISTENT STATE. 68976 2 The ex_raise of NOTARGETOBJ in getpage_init_scan_context() is missing a value for state. Leads to a misleading message. 69021 2 create a cache of size 4096 using sp_cacheconfig with the default memory. shutdown the server and reboot. server comes up, but returns a stack trace and the server doesnot allow to login. 69067 2 An attempt to create an update/insert/delete trigger when 1 already exists yields a 2110/20/4. 69068 2 Internal engineering memusage diagnostic doesn't display information about procedure cache - this is a regression from System 10 behaviour. 69105 2 Need to determine whether the call(s) to getpage() in the new log run deallocation code is correct with the new buffer manager. 69149 2 In house performance testing with 4K I/O log size generates error 2610. 69172 2 memallocd() always allocates 8 extra bytes, when only 4 extra bytes are needed to align the memory on an 8 byte boundary. It ought to only add 4 bytes to the size (DOUBLE_SIZE/2) rather than DOUBLE_SIZE. 69195 2 There are undefined C structures in a bulk copy file. 69228 2 The upgrade of SQL Server to cougar fails after the preupgrade eligibility test. It fails as INTERNAL ERROR: "'proc_role' is not a recognized built-in function name." and INTERNAL ERROR: "Invalid object name 'sp_checkreswords'." Sybase Confidential Page: 101 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 69300 2 str() returns garbage value unless real_val is first converted to real. 69312 2 If the 2gb limit still applies to disk init in system 11, disk init should complain if given a size larger than 2gb. Many 10.0 customers are getting bit by this. 69488 2 invalid statement within a batch that creates a view does not give syntax error message. e.g. create view v1 as select * from foo something go will go through. (allen,4/25)Oops, 'something' ends up being a table alias for foo. Not a bug. 69514 2 Correct source file function return values. 69901 2 The bugfix for 69244 is too restrictive and disallows allocation of large MASSes. This will prevent large I/Os on the log. 69925 2 Internal buffer monitors are not getting updated correctly. 70027 2 Server fails to detect a deadlock. Happens when (a) #user connec tion has been changed from default, and (b) the spid that completes the cycle is not the first one in the cycle to do a deadlock check. 70450 2 Correct function to pass number of variable length columns then rowsize. 70523 2 A user is allowed to group by a column of width 256, even though the maximum allowed width for a key column is 255. Index is built on 1st 255 bytes of column which can lead to incorrect results. 70536 2 SQL Server reports a truncation warning message when a password with a 30 bytes length is mentioned through sp_addlogin. It should only reports a truncation warning message if password length is actually Greater than 30 characters. 70591 2 Building a SQL Server binary without monitors (counters/events) fails. 70600 2 With bugfix 70419, name of clustered index is being reported. But for worktables, the clustered index has no name and "Index name not found" is being generated. This is misleading to the users. Should not need to generate name for worktable clust. index. 70734 2 When measuring cursor performance overhead, gprof shows that 2.8% of the time is being spent in reset_all_sqcache() called from estmt_loopend(). (There are no subqueries in test.) 70782 2 Reduce performance overhead (during query optimization) for adhoc queries. Sybase Confidential Page: 102 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 70821 2 Correct performance degradation in Buffer Manager diagnostics that call trace print while holding spinlocks. 70879 2 Procedure cache diagnostic routine returns error Msg 2812, Level 16, State 2, (and unrelated message) that procedure is not found when used with rules, defaults or triggers. It works for sprocs and views. . 71058 2 Correct search argument code to look for indexable arguments instead of search arguments. 71059 2 Incorrect message in sp_unbindcache_all. Expected a message of unbind; bug got Bbind Procedure 'sp_unbindcache_all' Command Failed: Cannot **bind** database '16790188' to cache. 71066 2 sp_bindcache c1, db, tab, "text" returns 0. tab has no text column 71091 2 Source file has array declarations for physname of device with size MAXNAME. It should be MAXPHYSNAME. 71099 2 Option "max_rows_per_page" is now a reserved word. It should be added in the routine that generates the reserved word check. 71214 2 In sp_movemass we don't check to see whether the source and destination mass sizes are same. We will unnecessarily move the masses. 71272 2 cannot optimize server cursors that use parameters. curs_copyvars() does not work as advertised -- has been broken since time immemorial. 71380 2 sl_id is introduced by slices project and should be renamed. There is slid already used by secure and it is confusing to have similar names. 71386 2 Internal monitors for buffers and not updated and produce invalid information as a result. 71432 2 The first TRACEPRINT() in cm_diag_showcache() should take cid as an arguement as the format string has a 'Cached id %d'. 71631 2 Makefile file of "generic/buildmaster" has hardcoded path "- I/usr/include/sys" in the INCLUDE list. This hardcode path should not be present. It spoils the OS sequence, which is first to pick files from "/usr/include". 71674 2 There is no need to call s_mustrecompile() for a fetch or fetch into. s_mustrecompile() and its subordinate functions show up as at least 3.3% of the overhead on a fetch. Sybase Confidential Page: 103 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 71685 2 If reresolution of a query is needed, an incorrect error message could be generate due to invalid information passed to the error handling routine. We should see error 1717 in this case. 71725 2 Internal routine that allocates memory to an table or index generates an incorrect errors message if it fails. This will make debugging the problem difficult. 71860 2 sort avert logic fails to properly consider indexes on temporary table created as part of a group by. if order by is on the first column in the group by we don't need a sort. 71901 2 In 11.0 trace flag 1205 prints the stack trace only for the victim in a deadlock, instead of each task involved in a deadlock. 71972 2 Msg 8201, Level 26, state 2: Line 26: Keep count of descriptor was expected to be 1. Instead 2 was found. 72098 2 Error 1105, state 3 during data slicing tests. 72173 2 sl_chgvalue() in sysslicesmgr.c gets lock on the page AFTER getpage(). Works fine, but to be consistent should lock BEFORE getpage(). Then, bufunkeep() can be removed from case stmts that handle errors from lock manager. 72202 2 legalinxact() is called at execution time for all E_STMTs. There is a large set of statements that are always legal in a transaction. This class should be checked for at compile time instead of at execution time. A bit set in the new ... 72211 2 Wrong error message given when sp_configure is used with wrong option name. 72237 2 sp_cachestrategy uses error message 17881 inappropriately, the message says "invalid role -- nothing changed", when the test is for object ownership, which is not a role. Suggest 17732: You do not own table 'tablename', or adding a nice generic... 72280 2 treergprint() does not print out the contents of the rgindname field in the range entry. If user forces an index by name, then rgindname will hold the name. Currently, neither treeprint() nor prPLAN() will display the name. 72307 2 reconfigure devices using negative value did not generate error. to duplicate: sp_configure "devices",-20. 72309 2 "user connections" is set to 10, but the user11 is able to connect to the server. Sybase Confidential Page: 104 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 72312 2 load db or load tran no longer produce the message "Remirroring the affected portions of the usage map that are on mirrored devices." 72323 2 Typo: "beween" should be "between". Error message: Illegal value '2' specified for configuration option 'current number online engines'. The legal values are beween '1' and '1'. 72370 2 Should not be possible to increase default nw packet size to greater than max nw packet size. 72371 2 sp_configure "remote logins", 2147483 ==> gets wrong error message. 72374 2 sp_configure "additional network memory", 2147483647 sets value to negative(MAXINT). 72377 2 "coamtrips" cannot be set to higher values, error message is inconsistent with usual message - does not give legal values or reason why verification failed. 72378 2 sp_configure "cindextrips" can be set to illegal values, sets to wrong values. Setting to 214748364 sets it to 52428. Setting to negative (-1) sets it to 65535. 72380 2 setting to illegal value of (-1) for default sortorder using sp_configure: wrong message (thinks it is being set to 255). Can set to illegal value of 256, sets it to 0. 72381 2 sp_configure "default character set id", -1 : This sets value of char set to 255. 72382 2 missing some config options from sp_configure output. verify that it is ok. not to have these. "memory range start point", "cnbytio", "cnbytmax" 72383 2 sp_configure "configuration file", "read" option does not change the value of dynamic options to their defaults if the value of the option in the config file being read-in is specified as "DEFAULT". 72385 2 The valid range is changed and it is from 1 to 127. 72386 2 sp_configure "configuration file", "verify" gets errors that are not valid - file exists, get error 'bad file number' etc. 72387 2 Can set "cmastnet" to negative value (-2) thru' sp_configure. On NT, gives wrong error message. 72389 2 sp_configure "configuration file", "write" to an invlaid file or path -- error "Invalid pointer param...". Sybase Confidential Page: 105 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 72393 2 When a invalid option name + value follows a sp_configure modify command of a valid option, server tries to change value of option changed previously. Eg: sp_configure "total mem", 8000. Next, sp_configure "junk", 2000. Tries to change total memory to 72449 2 Positioned updates (Update where current of cursor) cannot have tables specified in FROM list. Its illegal syntax as per ANSI and also our Command reference. Should raise a parser error. Currently, the following goes through: ... 72501 2 Incorrect results when using table name aliasing: Table tab1 has 2 rows; select t1.c1 from tab1 t1, tab1 should return 4 rows, but it returns only 2. The parser loses the table tab1 which is not aliased. But, select t1.c1 from tab1 t1, tab1 t2 returns the 72668 2 On Solaris: floating point '4.940656458412466544e-324' is out of machine representation(8 butes) which is msg 168,Level 15, State. ON NT platform error 3606, encounted. 72695 2 default char set id can be changed to 0 thru' sp_configure, though the default sortorder for this (ascii_8) does not exist in syscharsets. The server hangs when brought up in single user mode subsequently. Also setting to 256 sets config value to 0. 72716 2 Cannot reset "user connections" to 25. Sequence: set memory to 70000, set stack size to 2148352, set user conn. to 1000. Then, cannot reset user conn. to "default" or 25. Warning: os setting for file descriptors 1024 is not enough to set devixes to 10, . 72861 2 'cmastnet' should not be displayed in the config file or from sysconfigures table, should not be displayed in sp_configure. 73019 2 A threshold being defined will not get activated if the number of free pgs (threshold value) specified is greater than the current number of free pgs left in the segment onto which the threshold will operate. User must be notified with a Warning message. 73030 2 sp_cachestrategy should increment the schema count if the strategy is being changed. Use the schema_inc(objid) or indexdel count to do this. 73035 2 Showplan output should be updated to display the two new update modes (deferred_varcol and deferred_index). 73247 2 on NT, it is possible to set 'maximum engines' to 32. On Unix, error message says 'legal values are between 1 and 4'. See conf009.out for incorrect error message when setting to 32. 73251 2 'default network packet size' can be set to 100352 with 'memory' of 50000 on NT but fails on Unix (solaris). Validation is not consistent - is this a bug in validation in NT, or a bug on Solaris, or is it platform dependant? Test case conf010, see NTo Sybase Confidential Page: 106 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 73315 2 1) desmgr.c:des_udesfree() new 'flush' arg is not documented in the routine comments. 2) DEScriptor state in des_flush() should not be tested through bitwise 'and' operator, but rather with equality operator (==) . DES states are mutually exclusive. 73455 2 64-bit issue? The search value used by the optimizer to obtain selectivity appears different from the specified value. COnsequently the optimizer picks wrong index. 73480 2 A page which is in cache because of fetch-and-discard will be upgraded to the MRU end of the chain by a second fetch-and-discard access that finds it in cache. 73650 2 a table with a text column has an index entry in sysindexes (separate from the table entry). the first page (same as root) is of type PG_DATA. it should be of type text even though that page is never used for data/text - it is used as an allocation hint. 73654 2 sp_bindcache reports: Class 3 is not a valid attribute class. 73658 2 sp_bindcache: binds database to a logonly cache (returns 0) 73772 2 During startup, 'uaremove: alarm id 7 not active' is displayed in the errorlog. Stack trace shows that uaremove is called from deadlock_period_tune. 73886 2 dbcc des() prints the message "Index %d mapped to cache .." for all the 255 indices in an object, regardless of the indices existing or not. This leads to more than 250 lines of output for each object in dbcc des() and could be a minor irritant. 74028 2 'sort buffers' and 'sort page count' get set to 0 when setting to 65536 (greater than MaxI2). 74052 2 No error or warning when server is booted with invalid wash config: [4K I/O Buffer Pool]\n pool size = 512K\n wash size = 1024 K 74062 2 unbinding database using sp_unbindcache from a user database other than master returns misleading message "You must be in Master to bind a database.". The message should say "unbind", NOT "bind". 74144 2 Disable "dbcc tune" for parameters that can be configured via 'sp_configure'. i.e. 'schedspins', 'sortbufsize', 'sortpgcount', 'maxscheds', 'max_retries', 'bufwashsize', system wide 'oamtrips' and 'indextrips" to prevent confusion. 74233 2 We see lots of contention for address locks on root pages of indexes at high number of engines. Improving concurrency here will help scalability in TPC-C. The current method for locking root page can use some improvement. Sybase Confidential Page: 107 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 74258 2 dump database and dump transaction should not be grantable/revokable (disallowed in the parser). 74414 2 When there is no space to extend SYSGAMS, the error message does not indicate how to correct the problem. It should indicate more information about what can be done by the user to succeed in altering the database. 74460 2 the EXERCISER define is always true, hence exercisers are part of optimized server (is this good). Should the exerciser be under SANITY. Also, in dbtable exerciser, fake databases are used and theyhave id between 3600 and 4600. This means in a production 74726 2 sp_cacheconfig display values are not aligned properly. It is very difficult to read and is not a pleasing output. 74794 2 I tried sp_help titles and sql server threw me out of isql with the message Attempt to unhash buffer at 1xea6c1a20 with a buffer pageno of 5441 and database id 4 with HASHED status set failed - buffer was not found. Page pointer = 0xea4e4800, pageno = 544 74834 2 The AT ISOLATION clause is currently allowed in a create view statement unless the user specifies READ UNCOMMITTED or 0. The clause should not be allowed within a create view statement, no matter what the isolation level specified is. 74866 2 Certain run-time errors involving a temp table are displaying the internal name (#t1____....) instead of the external name (#t1). This happens with msg 515. 74871 2 The NETIO_POLL code in kpcheckio() is confusing. The call to os_schedcoma() is protected by an check to see if the engine is the NETENGINE. With MNE there is no NETENGINE. 74925 2 The define of NETENGINE to ENGINE_NUMBER when MNE is enabled leads to visually confusing code. E.g See kpcheckio's 'if (engine_id == NETENGINE)'. 74937 2 j_extra_sortcost() is called from j_cost() for every decorated permutation. The 1st thing that it does is check j->jsortavert to see if we need to consider sort ordering. If not, then it exits immediately. 74955 2 diag_get_cacheid() in dbccdiag.c, line 10512, returns nothing whereas it should return NULL_CACHEID. 75142 2 default cache has 2K and 16K pools; select avg(advance) from titles (index titles prefetch 8) reports I/O size 8Kbytes in showplan. Sybase Confidential Page: 108 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 75196 2 dbcc memusage always shows same value for "Proc Buffers:" and "Proc Headers:". Seems to be a bug in the calculation of "Proc Headers" in pr_alloc_usage() in memusage.c 75203 2 Back out the code "alter table sysindexes replace status2 default 0"in upgrde.c which is not needed. 75295 2 Assertion failure from cursor code: Assertion ((css->cstatus & C_DECLARED)) failed in file css.c line 440 when using Cobol appln. with debug server. Appln does dynamic SQL using cursors. Will try to get smaller program to reproduce problem. 75306 2 The size of server structures as shown by dbcc memusage is incorrect. The size seems to increase with the amount of configured memory. (In 10.x the size is a constant.) 75312 2 The macros MASTER_ONLY() and PER_DB_CAT() are defined identically TWICE in catalog.h 75313 2 [REPLICATION] When tracing a TEXTREP, textdiag.c should also display the txrepdes structures. 75321 2 [REPLICATION] new routines in textrepl.c do not have lock assumptions and 'called by' sections in their function headers. 75322 2 [REPLICATION] In textrepl.c and ltintfc.c, in error situations, if the name of the object (db/table/column) is not known, print the decimal id of the object instead of the string 'UNKNOWN'. 75580 2 The default network packet size should be rounded down as in bug 73503 else this results in a verification error where default network packet size is more than the max network packet size. 75787 2 When the test appl is run with "non-serial" mirroring and multi- engines it hangs without any msgs. The reason being that it runs out of "disk i/o structures". It runs fine when the "disk i/o structures" is bumped up or if serial mirroring is used. 75793 2 A disk mirror command will fail if a database on the target device is offline. Error 950: Databse 'db1' is currently offline. This is particularly bad if multiple databases are on the same device. 75834 2 DBCC PAGE on an invalid database ID produces error 908 with stack trace in errorlog and aborts user session. 75835 2 DBCC PAGE with invalid page number for DBID produces error 806 with stack trace in errorlog and aborts user session. 75956 2 When multiple permissions are granted in one grant command, only one range entry is used. The assertion at the end of s_find_rg_entries() fails with the command 'grant select, update(col1) from t1 to user1'. Sybase Confidential Page: 109 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 76044 2 Create index does not increase the schema count, so pre-compiled plans are not recompiled. Problem Scenario: Suppose that a pre- compiled plan for a stored procedure contains an update stmt, wh ich updates value for column c1 on a table with no indexes. 76095 2 after executing dbcc exer_init_exer(buffmanager) without proper setup buffer manager exerciser fails as expected but leaves some threads in limbo state. Later when shutdown is executed server keeps giving SHUTDOWN is waiting for 1 process(es) to completee 76104 2 Message 'WARNING: spid 1, with suid 1 and curdbid 5 has active XDES #0 with xactid (0,0)' in errorlog under the foll. conditions: User 1 inserts rows into a table until log fills up, threshold manager kicks in and suspends the task. User 2 kills ... 76119 2 Output from dbcc memusage for Buffer Cache memory details is garbled for cache name. This is especially so when you redirect output to a file. scan_lru_bufs() should memmove cache_name upto clen bytes only, and the print routine should print only as many 76151 2 With bug fix 72627, it should be possible to remove the system SDESs from the CSS structure as it appears that the system tables will not be held open past the open cursor. The allocations of these SDESs is a large chunk of the compilation overhead 76332 2 sp_unbindcache returns wrong error message. Try sp_unbindcache dbname. It returns, Procedure 'sp_bindcache', Line 254:, Cannot bind database '6138772'. Message should say unbind instead of bind. 76337 2 offline database is bound to a cache; sp_helpcache prints message 950 three times @ Line 196 76349 2 (In regr test case datslc/dats012) dump tran with no_log followed by an insertion to an existing db -> still get log segment full, but Msg 1105/3 is now printed anywhere between 2-4 times on different platforms. Please refer to history of server bug 72098 76382 2 the following command: sp_cacheconfig 4k_cache raises the following error: 'k_cache' is not a recognized parameter option. This is an incorrect message. The message should be: 'Incorrect syntax near '4', msg 102, Level 15, State 1. 76478 2 The logic in the loop at the bottom of q_match_order() is does not correctly handle order bys on joins where more than 1 table is represented in the order by clause. if ordertab[order] = 0 then only order, not keyind, should be incremented. 76495 2 sp_helptext does not handle "procedure versions", so that text of various procedure versions are intermixed if over 255 characters Sybase Confidential Page: 110 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 76651 2 Config values for the buffer pools in a named cache could become negative. 76670 2 At execution time, s_mustrecompile() only need to call open_range() on estep->e_stvirtrg[] instead of traversing all subquery block and calling s_opentables() recursivley.This can save us lots of time at execution time. 76723 2 sp_bindcache : cannot bind user tables in master database. Error - The Master database cannot be bound to a named cache . 76824 2 If a prefetch size 'x' is forced in a query and if the buffer pool for that size does not exist, then the highest available buffer pool size should be selected by the optimizer. 76969 2 after a disk init, when i create a database on the new device, i get an 1804 error "there is no disk named xxxx. checking other names..." works fine on solaris cougar server. the message may be BOGUS, the DB is created on the named device!! 76980 2 In function p_create_cols(), exception handler for UTILS is installed twice. 77200 2 The error message for 8203 need to be enhanced to print more information like (the object id etc.,). It is hard to debug with the information currently printed.... 77237 2 The buffer manager error messages should be augmented to print the name of the cache where the offending buffer resides. 77435 2 If both NOALIGN and NORMSHORT are set to 0 and GETSHORT is not defined in port.h, GETSHORT does not get defined in syb_std.h. It looks like the #if NORMSHORT and #if !defined(GETSHORT) have been exchanged. 77451 2 Config file used with multiple cache status (say like "mixed cache" and "HK ignore cache" are not getting written back with multiple cache status (to the config file) whenever any configuration is modified using the sp_configure. 77504 2 Message 3447 has a typographic error. It says "...SQL Server wil not bring it...". "wil" should be "will". 77562 2 The comment for getmidpg() describing the page splitting strategies has an error in the example: When describing the operation of ascinsert, row A3 is duplicated. The second A3 should be A4. 77565 2 Compilation warning on Solaris for endxact.c Problem is that prototype for hsm_free_slgroup() is not included. This is found in slicemgr.h which should be #include'ed. Sybase Confidential Page: 111 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 77580 2 Execution code calls mapsvs() which saves cache strategy info from E_OP into SDES. This is done every time we scan the inner tables of a join for each row from outer table. Also, its repeated for every execution of a subquery. This can be optimized out.. 77583 2 backing out changes for 77166 since the changes for bug # 77166 cannot be checked-in until the bugs listed under /remote/qa3/bugs/77166/README get fixed. 77787 2 q_var_find is declared as a function returning a rgid_t type in seq_dcl.h and as a function returning short in src_dcl.h. 77790 2 In source/access/do.c/do_sort(), objstatp is declared to be short *. It is assigned the value of &tabsdes->...objsysstat which is ultimately int16 *. Since objsysstat is explicitly declared as int16, pointers which refer to it should also be declared... 77792 2 source/access/endxact.c/end__xact() calls hsm_free_slgroup(), but does not include slicemgr.h which declares it. 77807 2 Optimizer uses reformatting strategy when it should choose nonclustered or clustered index. 77838 2 If the allocation page resulting from an allocation hint does not belong to the current segment for the object then the allocation hint is not good anymore and has to be reset. 78132 2 Beta feedback indicates that customers are not aware of having to run the online database command after loading a db. Both load database and load tran should display messages indicating that this is required after the load sequence has been performed. 78333 2 Boot server, add new cache thru sp_cacheconfig 'newcache', '1M' and shutdown the server. Remove $DSQUERY.cfg file. Reboot. Running sp_cacheconfig does not show any information about 'newcache'. This is correct. But, there are still entries left in ...... 78603 2 When using set rowcount with union all, the row count affects the number of rows returned from the first select in the union, and all subsequent selects in the union will return exactly one row. 78685 2 In dbccdiag.c/d_tune(), NULL is assigned to copy_par3[len_par3]; however, copy_par3 is an array of char. This causes a compiler warning message. 78761 2 dbt_newseq is used without being declared in or.c. It is being used via the bufdirty_nx macro. This is a compiler warning. 78780 2 [ASID:avg3/avg_0400/9]When "ansinull" option is set "on", "avg" returns the message "Warning- null value eliminated in set function" even though "avg" does not encounter a null value. Sybase Confidential Page: 112 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 78794 2 [ASID:sp_addlogin1/addlogin_0040/4]When "sp_addlogin" is invoked with a valid "deflanguage" argument, then the user will be assigned the specified language instead of the default language. The default language is getting assigned instead. 78836 2 [ASID:sp_addthreshold4/addthres_0160:010/6]When "alter database ... on with override" is used to place a data segment on a device that previously contained only the log segment, then the last chance threshold still exists. 78837 2 [ASID:sp_addthreshold4/addthres_0160:020/7]When "alter database ... on with override" is used to place a log segment on a device that previously contained only the data segment, then the last chance threshold still exists. 78838 2 sp_helpgroup "public" shows users who do not belong to any other group. It should also include all users, since every user is also a member of "public" group. 78862 2 When a GROUP BY ALL aggregate is used in a subquery then a SYB_ASSERT will fire erroneously in the diag server,... need to check for this case and avoid the SYB_ASSERT 78918 2 [ASID:create_database4/cdb_0330/3]When create database is invoked with the same device name for both "on" and "log on" using "with override" option , then Server is not issuing a warning message. A warning is expected after successful completion. 78919 2 Boot the server with a config file with an entry for named a cache and a 2k pool in the named cache, wash size as DEFAULT. use sp_poolconfig to create a new pool by taking memory out of 2K pool with valid values server writes config file with wash size as 79004 2 If dbcc page is asked to display a page in a database which is currently unusable (not recovered, etc.) and the user didn't specify a cache name (or -1), then dbcc page should raise an error requesting that the user provide a cachename (or specify an... 79065 2 When one allocates some caches, there is no message being display that the config file was updated. on other commands such as sp_poolconfig it notifies the user that the configuration file was updated, etc. 79117 2 the optimizer picks a table scan over an nc-index for different values of the column. even though both the values have the exact same number of rows. 79137 2 When trying to load database of previous version the server asseses it as old and tries to upgrade and then fails with 1105 but after increasing the size of device still it fails with 2601 as there are already entries in sysattributes.. Sybase Confidential Page: 113 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 79715 2 Undetected deadlock could happen in Sysindexes Mgr: ind__pagelock() should test for IT_NO_PREDIRTY bit set in itag->it_status before to unpredirty and re-predirty the buffer, if the lock could not be acquired. with IT_NO_PREDIRTY, predirty happens later. 79753 2 I ran buildmaster on a beta release and let it prompt me--I did not pass it any flags. It prompted me for configuration only which is no longer availabe. So, I picked that option (chose y). Got error Buildmaster: read error at block number 0 79991 2 sp_configure output is wider than 80 columns. In an 80-column window, this gets truncated on rs6000, wrapped on other platforms. 80204 2 PROC_HDR.p_lostcnt (used in memalloc()) is not initialized. 80407 2 Performance issue: For dynamic SQL applications, we go through cursifx_main() from (1) delproc()->endupate()->...->deldrow() and (2)from delcom(). These 2 calls are for deleting the rows for the temp stored procedure created for the dynamic SQL statement. 80725 2 In intlmsg, diffs in 8 files:The line "DELETE permission denied on object intm_tab1, database master, owner dbo" appears in English instead of other languages.The files are intm002[a-e] & intm005[a-c 81712 2 [REPLICATION] Exception handler code in bulk.c/bulk_normal() should not EX_DELETE() twice if the exception is SYSTEM because it will delete the handler installed by bulk_main() 81857 2 Table t1 (i int). Insert 0. If exists(select 1) and exists (select * from t1 where i<>0) returns TRUE but should return FALSE. Reversing the terms 'if exists(select * from t1 where i<>0) and exists(select i)' returns FALSE correctly. 82325 2 The new @@text* variables do not work as documented. Clients that used these to do text/image update w/ 10.1 (lolita) server will not work with 11.0 server. After a text/image update, @@textptr, @@textdbid, @@textcolid, @@textobjid are all NULL or 0. 83353 2 Trace flag 2901 will not catch fatal signals as it's required to, because signal-catchers (kisignal(), kaclkintr()) pass to terminate_process() a zero kpid instead of Runpid. Terminate_process() matches the passed-in kpid to the Pss's stored kpid, ... 83751 2 Several suspect parameters were found lodged inside server.loc message files, instead of the expected "%1!" format. These were found in the following message #'s 7944, 7945, 8211, 17973, 17974, 17975. Sybase Confidential Page: 114 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 84542 2 asterisk (*) missing in msg 5893 : The sum, .... number of devices '*' max online engines .... This is causing diffs in config test suite. 84638 2 The output of 'upgrade' is not in sequential order. i.e. the output of the 'online database' command displayed at the beginning instead of at the end. 57193 3 Function parameter mismatch for function des_getval(). 57959 3 Correct source functions to have exactly the same parameters as they are called through function pointers. 63365 3 During 10.0.1 Secure SQL Server dev, we added a proc_role() check to sp_checkreswords to limit its usage to the sa_role. However, this proc gets run by the *old* server during an upgrade (e.g. 4.x). This will cause a syntax error on that server. 65487 3 Need to port the MP exercisers to the cougar codeline. 66678 3 Change msg 2557 to display the owner and name of the object. 67488 3 generic/source/access/srchindex.c is not optimized during builds because it does not compile with optimization on Solaris 2.3 compiler (SUNWspro3.0.1). When compiler is fixed or problem in Sybase code fixed, makefile2 should be removed. 67490 3 generic/source/exer/exerciser.c is not optimized during server build because the optimizer fails. When the compiler is fixed or the offending code is changed, the file should be optimized again. 67492 3 generic/source/manager/sysindmgr.c is not optimized because the compiler fails when attempting to optimize it. When the compiler is fixed or the offending code is changed, the module should be optimized again. (SUNWspro3.0.1 compiler). 68149 3 showplan output for select...into, references to "worktable" are confusing, both the "Worktable created for SELECT_INTO" and the "TO TABLE Worktable" message. These messages should name the actual destination table. 68609 3 Remove build machine names from makefiles. 68848 3 Number of shift/reduce conflicts has increased from 16 to 17, with some amount of grammer rewrite for optimizer hints, it can be reduced by 1. 71122 3 Argument mismatch in ex_raise(SORTM, LOSTMERGERUN,..)in function startsort(). We are passing 'dbid'. But the message expects Database name length and name string. Sybase Confidential Page: 115 Closed Problem Report SYBASE 11.0.1 SQL Server - Mar 10 1996 Sun Solaris SVR4 2.4 CPR No. Pri Description ------- --- ------------------------------------------------------------------- 71356 3 Invalid cast of MAXI2. 72034 3 if dbcc/dbccdiag.c:d_getdbid() is passed a char it makes sure the db exists, but if passed a dbid it doesn't do any error checking. this can lead to nasty stack traces in the users of d_getdbid(). 72340 3 curs_copyvars() will copy the variable/parameter info for a server cursor from the proc_hdr to the css. It does a linear search through the proc_hdr's lists trying to match on name. 72379 3 q_check_rdonly() sets seqisolation with the intent of setting e_complevel in s_compile(). 72391 3 Change option name from "defaul sortord" to "default sortorder" as before. Very confusing to custonmers who are used to old name, new name neither intuitive nor change required. Fix will require qa suites to be also updated wherever used. 72394 3 Wrong error message for "current number online engines" when invalid value given (2 for example): legal values are between '1' and '1'. 72474 3 we should not have a complete function definition under SANITY as, if somebody wants to compile a single file in sym mode and if this file contains a call to a function, under SANITY in some other file, we will have to compile the other file in sym mode! 72767 3 The scope of a cursor should be limited to the nesting level that it appears in. A stored procedure should not be able to see any cursors declared outside of that procedure. This is not the case. 73881 3 SQL Server should have a function to print the addresses of different structures in the memory pool manager. 74006 3 sp_cacheconfig reports strange message if cache does not exist (The cursor 'sys_cursor' can not be used...) 74748 3 When you don't specify the unit (M or K) in the sp_cacheconfig command then it complains about the cache name. 71939 4 Review and change function comments.