Hi, I found the bug about archive_timeout parameter. There is the case archive_timeout parameter is ignored after recovery works.
[Problem]
When the value of archive_timeout is smaller than that of checkpoint_timeout
and recovery works, archive_timeout is ignored in the first WAL archiving.
Once WAL is archived, the archive_timeout seems to be valid after that.
I attached the simple script for reproducing this problem on version 12.
I also confirmed that PostgreSQL10, 11 and 12. I think other supported versions
have this problem.
[Investigation]
In the CheckpointerMain(), calculate the time (cur_timeout) to wait on
WaitLatch.
-----------------------------------------------------------------
now = (pg_time_t) time(NULL);
elapsed_secs = now - last_checkpoint_time;
if (elapsed_secs >= CheckPointTimeout)
continue; /* no sleep for us ... */
cur_timeout = CheckPointTimeout - elapsed_secs;
if (XLogArchiveTimeout > 0 && !RecoveryInProgress())
{
elapsed_secs = now - last_xlog_switch_time;
if (elapsed_secs >= XLogArchiveTimeout)
continue; /* no sleep for us ... */
cur_timeout = Min(cur_timeout, XLogArchiveTimeout - elapsed_secs);
}
(void) WaitLatch(MyLatch,
WL_LATCH_SET | WL_TIMEOUT | WL_EXIT_ON_PM_DEATH,
cur_timeout * 1000L /* convert to ms */ ,
WAIT_EVENT_CHECKPOINTER_MAIN);
-----------------------------------------------------------------
Currently, cur_timeout is set according to only checkpoint_timeout when it is
during recovery.
Even during recovery, the cur_timeout should be calculated including
archive_timeout as well as checkpoint_timeout, I think.
I attached the patch to solve this problem.
Regards,
Daisuke, Higuchi
archive_timeout_test.sh
Description: archive_timeout_test.sh
archive_timeout.patch
Description: archive_timeout.patch
