Scott Marlowe <scott.marlowe@gmail.com> writes:
> On Tue, Apr 20, 2010 at 7:31 AM, Tom Lane <tgl@sss.pgh.pa.us> wrote:
>> Scott Marlowe <scott.marlowe@gmail.com> writes:
>>> So, pg_cancel_backend(pid) is the 8.3 version, that may or may not
>>> always work then, right?
>>
>> No, I didn't say that.
> Well, my experience with it has been that it's similar and it doesn't
> always work.
pg_cancel_backend might or might not always succeed in canceling a
query, if that's what you mean. What I'm worried about with SIGTERM'ing
an individual backend is negative side-effects, like hanging the entire
database. Before 8.4 there was never any attempt to ensure that that
would work nicely, and if memory serves we did find some problems when
we tested it.
regards, tom lane