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

Commit852a5cd

Browse files
committed
Avoid problems with OID wraparound during WAL replay.
Fix a longstanding thinko in replay of NEXTOID and checkpoint records: wetried to advance nextOid only if it was behind the value in the WAL record,but the comparison would draw the wrong conclusion if OID wraparound hadoccurred since the previous value. Better to just unconditionally assignthe new value, since OID assignment shouldn't be happening during replayanyway.The consequences of a failure to update nextOid would be pretty minimal,since we have long had the code set up to obtain another OID and try againif the generated value is already in use. But in the worst case therecould be significant performance glitches while such loops iterate throughmany already-used OIDs before finding a free one.The odds of a wraparound happening during WAL replay would be small in acrash-recovery scenario, and the length of any ensuing OID-assignment stallquite limited anyway. But neither of these statements hold true for areplication slave that follows a WAL stream for a long period; its behaviorupon going live could be almost unboundedly bad. Hence it seems worthback-patching this fix into all supported branches.Already fixed in HEAD in commitc6d76d7.
1 parent94c5aa6 commit852a5cd

File tree

1 file changed

+12
-11
lines changed
  • src/backend/access/transam

1 file changed

+12
-11
lines changed

‎src/backend/access/transam/xlog.c

Lines changed: 12 additions & 11 deletions
Original file line numberDiff line numberDiff line change
@@ -7939,12 +7939,15 @@ xlog_redo(XLogRecPtr lsn, XLogRecord *record)
79397939
{
79407940
OidnextOid;
79417941

7942+
/*
7943+
* We used to try to take the maximum of ShmemVariableCache->nextOid
7944+
* and the recorded nextOid, but that fails if the OID counter wraps
7945+
* around. Since no OID allocation should be happening during replay
7946+
* anyway, better to just believe the record exactly.
7947+
*/
79427948
memcpy(&nextOid,XLogRecGetData(record),sizeof(Oid));
7943-
if (ShmemVariableCache->nextOid<nextOid)
7944-
{
7945-
ShmemVariableCache->nextOid=nextOid;
7946-
ShmemVariableCache->oidCount=0;
7947-
}
7949+
ShmemVariableCache->nextOid=nextOid;
7950+
ShmemVariableCache->oidCount=0;
79487951
}
79497952
elseif (info==XLOG_CHECKPOINT_SHUTDOWN)
79507953
{
@@ -8032,15 +8035,13 @@ xlog_redo(XLogRecPtr lsn, XLogRecord *record)
80328035
CheckPointcheckPoint;
80338036

80348037
memcpy(&checkPoint,XLogRecGetData(record),sizeof(CheckPoint));
8035-
/* In an ONLINE checkpoint, treat thecounters like NEXTOID */
8038+
/* In an ONLINE checkpoint, treat theXID counter as a minimum */
80368039
if (TransactionIdPrecedes(ShmemVariableCache->nextXid,
80378040
checkPoint.nextXid))
80388041
ShmemVariableCache->nextXid=checkPoint.nextXid;
8039-
if (ShmemVariableCache->nextOid<checkPoint.nextOid)
8040-
{
8041-
ShmemVariableCache->nextOid=checkPoint.nextOid;
8042-
ShmemVariableCache->oidCount=0;
8043-
}
8042+
/* ... but still treat OID counter as exact */
8043+
ShmemVariableCache->nextOid=checkPoint.nextOid;
8044+
ShmemVariableCache->oidCount=0;
80448045
MultiXactAdvanceNextMXact(checkPoint.nextMulti,
80458046
checkPoint.nextMultiOffset);
80468047
if (TransactionIdPrecedes(ShmemVariableCache->oldestXid,

0 commit comments

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