Re: Wierd panic with 7.4.7 - Mailing list pgsql-hackers

From Tom Lane
Subject Re: Wierd panic with 7.4.7
Date
Msg-id 22290.1119981869@sss.pgh.pa.us
Whole thread Raw
In response to Wierd panic with 7.4.7  ("Joshua D. Drake" <jd@commandprompt.com>)
List pgsql-hackers
"Joshua D. Drake" <jd@commandprompt.com> writes:
>> Do you know what the dead client was doing?

> Unfortunately I don't. We didn't have PID logging turned on so I can't 
> tell which process it was. The only thing I was told was,
> "I am running a Full Vacuum, CRAP the server just died ;)"

Hmm ... VACUUM FULL is a tad weird because it marks itself committed
before it does the last stage of the operation (ie truncating the
relation).  I suppose it is possible that an error during that last
stage would have this symptom.  There's still the question of why
the postmaster log doesn't show the hypothetical initial error, though.

(Also, it's likely that there's some defense against this scenario in
VACUUM FULL, otherwise we'd hear this kind of report more often, I
should think.  I don't have time to go looking right now though.)
        regards, tom lane


pgsql-hackers by date:

Previous
From: Josh Berkus
Date:
Subject: Re: commit_delay, siblings
Next
From: Tom Lane
Date:
Subject: Re: Occupied port warning