Phil Frost <indigo@bitglue.com> writes:
> Things were crusing along just fine as I can tell, and then all the
> postgresql processes terminated. This is all I see in the logs:
> LOG: received immediate shutdown request
Something sent the postmaster a SIGQUIT signal. You need to look into
what might have done that.
> I have a nightly cron job that runs "vacuumdb --all --full --analyze
> --quiet" which failed a few days ago with this:
> vacuumdb: vacuuming of database "dew" failed: ERROR: buffer 126 is not owned by resource owner TopTransaction
> PANIC: cannot abort transaction 478973, it was already committed
The PANIC is a known problem with VACUUM FULL: it commits its
transaction before it's really done processing, and so if it gets an
error after that point it's up the creek without a paddle. But I'm not
sure what could have caused the "not owned by" error in the first place.
Don't suppose you can reproduce the problem?
regards, tom lane