From: | Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> |
---|---|
To: | Alvaro Herrera <alvherre(at)alvh(dot)no-ip(dot)org> |
Cc: | Eugene <eugene1(at)sympatico(dot)ca>, pgsql-general(at)postgresql(dot)org |
Subject: | Re: total db lockup |
Date: | 2005-08-19 17:02:16 |
Message-ID: | 25943.1124470936@sss.pgh.pa.us |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-general |
Alvaro Herrera <alvherre(at)alvh(dot)no-ip(dot)org> writes:
> Ok, so it seems the lockup occured only with the hash indexes? Then it
> means we still have bugs in the locking code for those.
"Still" meaning "in the version he's using", which he hasn't told us
anywhere that I saw. (Internal evidence suggests it's 7.3 something)
> It doesn't surprise me.
A deadlock in hash indexes wouldn't be surprising in pre-7.4 code,
since 7.4 was the first version that even pretended to be deadlock free
in hash indexes. But it does seem pretty surprising that the deadlock
checker didn't fire and boot somebody out of the deadlock. We haven't
had any bugs reported in deadlock detection in a very long time (since
7.1 I think). This could mean there's still something wrong in there.
regards, tom lane
From | Date | Subject | |
---|---|---|---|
Next Message | Tom Lane | 2005-08-19 17:08:11 | Re: How to cancel a query if SIGINT does not work? |
Previous Message | Michael Fuhr | 2005-08-19 16:40:38 | Re: question about plpgsql replace function |