On Tue, Sep 13, 2022 at 10:38:13AM +0500, Andrey Borodin wrote:
>
> And the other way is refactoring towards partitioned hashtable, namely
> dshash. But I don't see how partitioned locking can save us from a locking
> disaster. Problem is caused by reading all the pgss view colliding with
> reset(
> On 12 Sep 2022, at 23:01, Tom Lane wrote:
>
> Andrey Borodin writes:
>>> On 12 Sep 2022, at 18:18, Julien Rouhaud wrote:
>>> That being
>>> said I don't know if adding a timeout would be too expensive for the lwlock
>>> infrastructure.
>
> I want to object fiercely to loading down LWLock
On Mon, Sep 12, 2022 at 02:01:23PM -0400, Tom Lane wrote:
> Andrey Borodin writes:
> >> On 12 Sep 2022, at 18:18, Julien Rouhaud wrote:
> >> That being
> >> said I don't know if adding a timeout would be too expensive for the lwlock
> >> infrastructure.
>
> I want to object fiercely to loading d
Andrey Borodin writes:
>> On 12 Sep 2022, at 18:18, Julien Rouhaud wrote:
>> That being
>> said I don't know if adding a timeout would be too expensive for the lwlock
>> infrastructure.
I want to object fiercely to loading down LWLock with anything like
timeouts. It's supposed to be "lightweigh
> On 12 Sep 2022, at 18:18, Julien Rouhaud wrote:
>
> That being
> said I don't know if adding a timeout would be too expensive for the lwlock
> infrastructure.
Implementation itself is straightforward, but we need to add 3 impls of waiting
for semaphore with timeout.
POSIX have sem_timedwai
On Mon, Sep 12, 2022 at 05:32:55PM +0500, Andrey Borodin wrote:
>
> > On 12 Sep 2022, at 13:40, Julien Rouhaud wrote:
> >
> > I'm not a fan of that patch as it now silently ignores entries if the lwlock
> > can't be acquired *immediately*, without any way to avoid that if your
> > configuration an
> On 12 Sep 2022, at 13:40, Julien Rouhaud wrote:
>
> I'm not a fan of that patch as it now silently ignores entries if the lwlock
> can't be acquired *immediately*, without any way to avoid that if your
> configuration and/or workload doesn't lead to this problem, or any way to know
> that en
Hi,
On Mon, Sep 12, 2022 at 11:52:28AM +0500, Andrey Borodin wrote:
>
> === How to fix ===
> pgss uses LWLock to protect hashtable.
> When the hashtable is reset or new user query is inserted in pgss_store() -
> exclusive lock is used.
> When stats are updated for known query or pg_stat_statements
Hi hackers!
Recently I observed very peculiar incident.
=== Incident description ===
ETL database was operating fine for many months, regularly updated etc.
Workload was not changing much, but as far as it was ETL database - most of
queries were different all the time.
On the night of Septembe