Re: BUG #15142: ERROR: MultiXactId nnnnn has not been created yet -- apparent wraparound in v9.5 - Mailing list pgsql-bugs

From Andrew Gierth
Subject Re: BUG #15142: ERROR: MultiXactId nnnnn has not been created yet -- apparent wraparound in v9.5
Date
Msg-id 87fu49k8k1.fsf@news-spur.riddles.org.uk
Whole thread Raw
In response to Re: BUG #15142: ERROR: MultiXactId nnnnn has not been created yet --apparent wraparound in v9.5  (molofeev <molofeev3@gmail.com>)
List pgsql-bugs
>>>>> "molofeev" == molofeev  <molofeev3@gmail.com> writes:

 >> You said a server crash occurred while you were creating indexes. Was
 >> that a hard crash (power failure, etc.) and is it possible that you had
 >> fsync disabled in postgresql.conf, or hardware or filesystems that don't
 >> respect fsync properly?
 
 molofeev> That was not hard crash, and fsync was disabled.
 
 molofeev> I was disconnected form remote terminal, which was performing
 molofeev> creating indexes from sql file.

Just being disconnected remotely (for some reason other than a server
crash) isn't enough to account for the corruption; fsync=off only results
in corrupt data if there is an actual OS-level crash or unclean shutdown.

So I wouldn't rule out the possibility of hardware or storage subsystem
problems.

-- 
Andrew (irc:RhodiumToad)


pgsql-bugs by date:

Previous
From: molofeev
Date:
Subject: Re: BUG #15142: ERROR: MultiXactId nnnnn has not been created yet --apparent wraparound in v9.5
Next
From: PG Bug reporting form
Date:
Subject: BUG #15144: *** glibc detected *** postgres: postgres smsconsole[local] SELECT: double free or corruption (!pre