4/24/03 9:28:49 AM, Tom Lane <tgl@sss.pgh.pa.us> wrote:
>Andrew Biagioni <andrew.biagioni@e-greek.net> writes:
>> My best estimate as to when they appear is when I abort a transaction in
>> PGAdminII by killing the corresponding process on the server ("kill -9
XXXX"),
>
>Yikes. You're not serious are you?
Sorry, forget the " -9 " bit. I only use that when I need to shut down
PostgreSQL and some process is not releasing at that time.
>That would certainly cause the backend to fail to clean up its temp
>tables... not to mention that the postmaster will treat this as a
>system crash and force all the other backends to punt too. Not the way
>I'd pick to cancel a misbehaving query.
With my "kill [procnum]" I don't seem to have any negative side effects, except
apparently for the temp tables (I _think_). Does that seem reasonable?
>Try something less invasive next time, like kill -INT.
>
> regards, tom lane
Thanks,
Andrew