m4_comment([$Id: fail.so,v 10.2 2005/10/19 21:10:31 bostic Exp $])

m4_ref_title(m4_tam Applications,
    Handling failure in Transactional Data Store applications,,
    transapp/term, transapp/app)

m4_p([dnl
When building Transactional Data Store applications, there are design
issues to consider whenever a thread of control with open m4_db handles
fails for any reason (where a thread of control may be either a true
thread or a process).])

m4_p([dnl
The first case is handling system failure: if the system fails, the
database environment and the databases may be left in a corrupted state.
In this case, recovery must be performed on the database environment
before any further action is taken, in order to:])

m4_bulletbegin
m4_bullet([recover the database environment resources,])
m4_bullet([dnl
release any locks or mutexes that may have been held to avoid starvation
as the remaining threads of control convoy behind the held locks, and])
m4_bullet([dnl
resolve any partially completed operations that may have left a database
in an inconsistent or corrupted state.])
m4_bulletend

m4_p([dnl
For details on performing recovery, see the m4_link(recovery, [Recovery
procedures]).])

m4_p([dnl
The second case is handling the failure of a thread of control.  There
are resources maintained in database environments that may be left
locked or corrupted if a thread of control exits unexpectedly.  These
resources include data structure mutexes, logical database locks and
unresolved transactions (that is, transactions which were never aborted
or committed).  While Transactional Data Store applications can treat
the failure of a thread of control in the same way as they do a system
failure, they have an alternative choice, the m4_refT(dbenv_failchk).])

m4_p([dnl
The m4_refT(dbenv_failchk) will return m4_ref(DB_RUNRECOVERY) if the
database environment is unusable as a result of the thread of control
failure.  (If a data structure mutex or a database write lock is left
held by thread of control failure, the application should not continue
to use the database environment, as subsequent use of the environment
is likely to result in threads of control convoying behind the held
locks.)  The m4_ref(dbenv_failchk) call will release any database read
locks that have been left held by the exit of a thread of control, and
abort any unresolved transactions.  In this case, the application can
continue to use the database environment.])

m4_p([dnl
A Transactional Data Store application recovering from a thread of
control failure should call m4_ref(dbenv_failchk), and, if it returns
success, the application can continue.  If m4_ref(dbenv_failchk) returns
m4_ref(DB_RUNRECOVERY), the application should proceed as described for
the case of system failure.])

m4_p([dnl
It greatly simplifies matters that recovery may be performed regardless
of whether recovery needs to be performed; that is, it is not an error
to recover a database environment for which recovery is not strictly
necessary.  For this reason, applications should not try to determine
if the database environment was active when the application or system
failed.  Instead, applications should run recovery any time the
m4_refT(dbenv_failchk) returns m4_ref(DB_RUNRECOVERY), or, if the
application is not calling m4_refT(dbenv_failchk), any time any thread
of control accessing the database environment fails, as well as any time
the system reboots.])

m4_page_footer