database/postgres
database/postgres
1
0
Fork 0
mirror of https://github.com/postgres/postgres.git synced 2025-05-20 05:13:53 +03:00
Code

Add more LOG messages when starting and ending recovery from a backup

Browse Source 
Three LOG messages are added in the recovery code paths, providing
information that can be useful to track corruption issues depending on
the state of the cluster, telling that:
- Recovery has started from a backup_label.
- Recovery is restarting from a backup start LSN, without a
backup_label.
- Recovery has completed from a backup.

This was originally applied on HEAD as of 1d35f705e191, and there is
consensus that this can be useful for older versions.  This applies
cleanly down to 15, so do it down to this version for now (older
versions have heavily refactored the WAL recovery paths, making the
change less straight-forward to do).

Author: Andres Freund
Reviewed-by: David Steele, Laurenz Albe, Michael Paquier
Discussion: https://postgr.es/m/20231117041811.vz4vgkthwjnwp2pp@awork3.anarazel.de
Backpatch-through: 15
This commit is contained in:
Michael Paquier 2024-01-25 17:07:56 +09:00
parent f57a580fd2
commit edbd1b41ab
1 changed files with 34 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

34
src/backend/access/transam/xlogrecovery.c
View File

@ -623,6 +623,22 @@ InitWalRecovery(ControlFileData *ControlFile, bool *wasShutdown_ptr,
if (StandbyModeRequested)
EnableStandbyMode();
/*
* Omitting backup_label when creating a new replica, PITR node etc.
* unfortunately is a common cause of corruption. Logging that
* backup_label was used makes it a bit easier to exclude that as the
* cause of observed corruption.
*
* Do so before we try to read the checkpoint record (which can fail),
* as otherwise it can be hard to understand why a checkpoint other
* than ControlFile->checkPoint is used.
*/
ereport(LOG,
(errmsg("starting backup recovery with redo LSN %X/%X, checkpoint LSN %X/%X, on timeline ID %u",
LSN_FORMAT_ARGS(RedoStartLSN),
LSN_FORMAT_ARGS(CheckPointLoc),
CheckPointTLI)));
/*
* When a backup_label file is present, we want to roll forward from
* the checkpoint it identifies, rather than using pg_control.
@ -761,6 +777,16 @@ InitWalRecovery(ControlFileData *ControlFile, bool *wasShutdown_ptr,
EnableStandbyMode();
}
/*
* For the same reason as when starting up with backup_label present,
* emit a log message when we continue initializing from a base
* backup.
*/
if (!XLogRecPtrIsInvalid(ControlFile->backupStartPoint))
ereport(LOG,
(errmsg("restarting backup recovery with redo LSN %X/%X",
LSN_FORMAT_ARGS(ControlFile->backupStartPoint))));
/* Get the last valid checkpoint record. */
CheckPointLoc = ControlFile->checkPoint;
CheckPointTLI = ControlFile->checkPointCopy.ThisTimeLineID;
@ -2123,6 +2149,9 @@ CheckRecoveryConsistency(void)
if (!XLogRecPtrIsInvalid(backupEndPoint) &&
backupEndPoint <= lastReplayedEndRecPtr)
{
XLogRecPtr saveBackupStartPoint = backupStartPoint;
XLogRecPtr saveBackupEndPoint = backupEndPoint;
elog(DEBUG1, "end of backup reached");
/*
@ -2133,6 +2162,11 @@ CheckRecoveryConsistency(void)
backupStartPoint = InvalidXLogRecPtr;
backupEndPoint = InvalidXLogRecPtr;
backupEndRequired = false;
ereport(LOG,
(errmsg("completed backup recovery with redo LSN %X/%X and end LSN %X/%X",
LSN_FORMAT_ARGS(saveBackupStartPoint),
LSN_FORMAT_ARGS(saveBackupEndPoint))));
}
/*