daveg writes:
> On Wed, Sep 07, 2011 at 09:02:04PM -0400, Tom Lane wrote:
>> daveg writes:
>>> The first version we saw it on was 8.4.7.
>> Yeah, you said that. I was wondering what you'd last run before 8.4.7.
> Sorry, misunderstood. We were previously running 8.4.4, but have been on 8.4.7
>
On Wed, Sep 07, 2011 at 09:02:04PM -0400, Tom Lane wrote:
> daveg writes:
> > On Wed, Sep 07, 2011 at 07:39:15PM -0400, Tom Lane wrote:
> >> BTW ... what were the last versions you were running on which you had
> >> *not* seen the problem? (Just wondering about the possibility that we
> >> back-p
On Wed, Sep 7, 2011 at 6:25 PM, Tom Lane wrote:
> Robert Haas writes:
>> I thought about an error exit from client authentication, and that's a
>> somewhat appealing explanation, but I can't quite see why we wouldn't
>> clean up there the same as anywhere else. The whole mechanism feels a
>> bit
daveg writes:
> On Wed, Sep 07, 2011 at 07:39:15PM -0400, Tom Lane wrote:
>> BTW ... what were the last versions you were running on which you had
>> *not* seen the problem? (Just wondering about the possibility that we
>> back-patched some "fix" that broke things. It would be good to have
>> a
On Wed, Sep 07, 2011 at 07:39:15PM -0400, Tom Lane wrote:
> daveg writes:
> > Also, this is very intermittant, we have seen it only in recent months
> > on both 8.4.7 and 9.0.4 after years of no problems. Lately we see it what
> > feels like a few times a month. Possibly some new application behav
daveg writes:
> Also, this is very intermittant, we have seen it only in recent months
> on both 8.4.7 and 9.0.4 after years of no problems. Lately we see it what
> feels like a few times a month. Possibly some new application behaviour
> is provoking it, but I have no guesses as to what.
BTW ...
daveg writes:
> On Wed, Sep 07, 2011 at 06:25:23PM -0400, Tom Lane wrote:
>> ... But maybe it'd be interesting for Dave to stick a
>> LockReleaseAll call into ProcKill() and see if that makes things better.
>> (Dave: test that before you put it in production, I'm not totally sure
>> it's safe.)
On Wed, Sep 07, 2011 at 06:25:23PM -0400, Tom Lane wrote:
> Robert Haas writes:
> > I thought about an error exit from client authentication, and that's a
> > somewhat appealing explanation, but I can't quite see why we wouldn't
> > clean up there the same as anywhere else. The whole mechanism fe
On Wed, Sep 07, 2011 at 06:35:08PM -0400, Tom Lane wrote:
> daveg writes:
> > It does not seem restricted to pg_authid:
> > 2011-08-24 18:35:57.445 24987 c23 apps ERROR: lock AccessShareLock on
> > object 16403/2615/0
> > And I think I've seen it on other tables too.
>
> Hmm. 2615 = pg_nam
daveg writes:
> It does not seem restricted to pg_authid:
> 2011-08-24 18:35:57.445 24987 c23 apps ERROR: lock AccessShareLock on
> object 16403/2615/0
> And I think I've seen it on other tables too.
Hmm. 2615 = pg_namespace, which most likely is the first catalog
accessed by just about an
Robert Haas writes:
> I thought about an error exit from client authentication, and that's a
> somewhat appealing explanation, but I can't quite see why we wouldn't
> clean up there the same as anywhere else. The whole mechanism feels a
> bit rickety to me - we don't actually release locks; we ju
On Wed, Sep 07, 2011 at 04:55:24PM -0400, Tom Lane wrote:
> Robert Haas writes:
> > Tom's right to be skeptical of my theory, because it would require a
> > CHECK_FOR_INTERRUPTS() outside of a transaction block in one of the
> > pathways that use session-level locks, and I can't find one.
>
> Mor
On Wed, Sep 7, 2011 at 4:55 PM, Tom Lane wrote:
> Yeah, and for that matter it seems to let VACUUM off the hook too.
> If we assume that the reported object ID is non-corrupt (and if it's
> always the same, that seems like the way to bet) then this is a lock
> on pg_authid.
>
> Hmmm ... could the
Robert Haas writes:
> Tom's right to be skeptical of my theory, because it would require a
> CHECK_FOR_INTERRUPTS() outside of a transaction block in one of the
> pathways that use session-level locks, and I can't find one.
More to the point: session-level locks are released on error. The only
w
On Wed, Sep 7, 2011 at 4:22 PM, daveg wrote:
> Yes, we make extensive use of advisory locks. That was my thought too when
> Robert mentioned session level locks.
>
> I'm happy to add any additional instrumentation, but my client would be
> happier to actually run it if there was a way to recover f
On Wed, Sep 07, 2011 at 10:20:24AM -0400, Tom Lane wrote:
> Robert Haas writes:
> > After spending some time staring at the code, I do have one idea as to
> > what might be going on here. When a backend is terminated,
> > ShutdownPostgres() calls AbortOutOfAnyTransaction() and then
> > LockReleas
Robert Haas writes:
> After spending some time staring at the code, I do have one idea as to
> what might be going on here. When a backend is terminated,
> ShutdownPostgres() calls AbortOutOfAnyTransaction() and then
> LockReleaseAll(USER_LOCKMETHOD, true). The second call releases all
> user lo
On Wed, Sep 7, 2011 at 5:16 AM, daveg wrote:
> On Tue, Aug 23, 2011 at 12:15:23PM -0400, Robert Haas wrote:
>> On Mon, Aug 22, 2011 at 3:31 AM, daveg wrote:
>> > So far I've got:
>> >
>> > - affects system tables
>> > - happens very soon after process startup
>> > - in 8.4.7 and 9.0.4
>> > -
On Tue, Aug 23, 2011 at 12:15:23PM -0400, Robert Haas wrote:
> On Mon, Aug 22, 2011 at 3:31 AM, daveg wrote:
> > So far I've got:
> >
> > - affects system tables
> > - happens very soon after process startup
> > - in 8.4.7 and 9.0.4
> > - not likely to be hardware or OS related
> > - happens
On Mon, Aug 22, 2011 at 3:31 AM, daveg wrote:
> So far I've got:
>
> - affects system tables
> - happens very soon after process startup
> - in 8.4.7 and 9.0.4
> - not likely to be hardware or OS related
> - happens in clusters for period of a few second to many minutes
>
> I'll work on print
On Fri, Aug 12, 2011 at 04:19:37PM -0700, daveg wrote:
>
> This seems to be bug month for my client. Now there are seeing periods
> where all new connections fail immediately with the error:
>
>FATAL: lock AccessShareLock on object 0/1260/0 is already held
>
> This happens on postgresql 8.
21 matches
Mail list logo