Movatterモバイル変換

[0]ホーム
URL:

Skip to content

Navigation Menu

Sign in
Appearance settings

Search code, repositories, users, issues, pull requests...

Provide feedback

We read every piece of feedback, and take your input very seriously.

Saved searches

Use saved searches to filter your results more quickly

 Sign up
Appearance settings

Commit2c8cff5

Browse files
committed
Ensure correct minimum consistent point on standbys
Startup process has improved its calculation of incorrect minimumconsistent point in8d68ee6, which ensures that all WAL available getsreplayed when doing crash recovery, and has introduced an incorrectcalculation of the minimum recovery point for non-startup processes,which can cause incorrect page references on a standby when for examplethe background writer flushed a couple of pages on-disk but was notupdating the control file to let a subsequent crash recovery replay towhere it should have.The only case where this has been reported to be a problem is when astandby needs to calculate the latest removed xid when replaying a btreedeletion record, so one would need connections on a standby that happenjust after recovery has thought it reached a consistent point. Using abackground worker which is started after the consistent point is reachedwould be the easiest way to get into problems if it connects to adatabase. Having clients which attempt to connect periodically couldalso be a problem, but the odds of seeing this problem are much lower.The fix used is pretty simple, as the idea is to give access to theminimum recovery point written in the control file to non-startupprocesses so as they use a reference, while the startup process stillinitializes its own references of the minimum consistent point so as theoriginal problem with incorrect page references happening post-promotionwith a crash do not show up.Reported-by: Alexander KukushkinDiagnosed-by: Alexander KukushkinAuthor: Michael PaquierReviewed-by: Kyotaro Horiguchi, Alexander KukushkinDiscussion:https://postgr.es/m/153492341830.1368.3936905691758473953@wrigleys.postgresql.orgBackpatch-through: 9.3
1 parent29e07cd commit2c8cff5

File tree

1 file changed

+25
-6
lines changed
  • src/backend/access/transam

1 file changed

+25
-6
lines changed

‎src/backend/access/transam/xlog.c

Lines changed: 25 additions & 6 deletions
Original file line numberDiff line numberDiff line change
@@ -2696,9 +2696,13 @@ UpdateMinRecoveryPoint(XLogRecPtr lsn, bool force)
26962696
* i.e., we're doing crash recovery. We never modify the control file's
26972697
* value in that case, so we can short-circuit future checks here too. The
26982698
* local values of minRecoveryPoint and minRecoveryPointTLI should not be
2699-
* updated until crash recovery finishes.
2699+
* updated until crash recovery finishes. We only do this for the startup
2700+
* process as it should not update its own reference of minRecoveryPoint
2701+
* until it has finished crash recovery to make sure that all WAL
2702+
* available is replayed in this case. This also saves from extra locks
2703+
* taken on the control file from the startup process.
27002704
*/
2701-
if (XLogRecPtrIsInvalid(minRecoveryPoint))
2705+
if (XLogRecPtrIsInvalid(minRecoveryPoint)&&InRecovery)
27022706
{
27032707
updateMinRecoveryPoint= false;
27042708
return;
@@ -2710,7 +2714,9 @@ UpdateMinRecoveryPoint(XLogRecPtr lsn, bool force)
27102714
minRecoveryPoint=ControlFile->minRecoveryPoint;
27112715
minRecoveryPointTLI=ControlFile->minRecoveryPointTLI;
27122716

2713-
if (force||minRecoveryPoint<lsn)
2717+
if (XLogRecPtrIsInvalid(minRecoveryPoint))
2718+
updateMinRecoveryPoint= false;
2719+
elseif (force||minRecoveryPoint<lsn)
27142720
{
27152721
XLogRecPtrnewMinRecoveryPoint;
27162722
TimeLineIDnewMinRecoveryPointTLI;
@@ -3099,9 +3105,11 @@ XLogNeedsFlush(XLogRecPtr record)
30993105
* An invalid minRecoveryPoint means that we need to recover all the
31003106
* WAL, i.e., we're doing crash recovery. We never modify the control
31013107
* file's value in that case, so we can short-circuit future checks
3102-
* here too.
3108+
* here too. This triggers a quick exit path for the startup process,
3109+
* which cannot update its local copy of minRecoveryPoint as long as
3110+
* it has not replayed all WAL available when doing crash recovery.
31033111
*/
3104-
if (XLogRecPtrIsInvalid(minRecoveryPoint))
3112+
if (XLogRecPtrIsInvalid(minRecoveryPoint)&&InRecovery)
31053113
updateMinRecoveryPoint= false;
31063114

31073115
/* Quick exit if already known to be updated or cannot be updated */
@@ -3118,8 +3126,19 @@ XLogNeedsFlush(XLogRecPtr record)
31183126
minRecoveryPointTLI=ControlFile->minRecoveryPointTLI;
31193127
LWLockRelease(ControlFileLock);
31203128

3129+
/*
3130+
* Check minRecoveryPoint for any other process than the startup
3131+
* process doing crash recovery, which should not update the control
3132+
* file value if crash recovery is still running.
3133+
*/
3134+
if (XLogRecPtrIsInvalid(minRecoveryPoint))
3135+
updateMinRecoveryPoint= false;
3136+
31213137
/* check again */
3122-
returnrecord>minRecoveryPoint;
3138+
if (record <=minRecoveryPoint|| !updateMinRecoveryPoint)
3139+
return false;
3140+
else
3141+
return true;
31233142
}
31243143

31253144
/* Quick exit if already known flushed */

0 commit comments

Comments
 (0)
[8]ページ先頭
©2009-2025 Movatter.jp