auto_explain and parallel queries issue

Previous Topic Next Topic
 
classic Classic list List threaded Threaded
3 messages Options
Reply | Threaded
Open this post in threaded view
|

auto_explain and parallel queries issue

Pavel Stehule
Hi

I am testing PostgreSQL in customer's environment and I found some unexpected behave.

When auto_explain is used, and there is slower parallel query, then auto_explain raises log for any slow process.

 arallel worker for PID 20089   : auto_explain database=NULL client=local appname=psql
 arallel worker for PID 20089   :          duration: 121.250 ms  plan:
 arallel worker for PID 20089   :          Query Text: select sum(a) from bigtable group by b;
 arallel worker for PID 20089   :          Partial HashAggregate  (cost=10675.00..10774.59 rows=9959 width=12)
 arallel worker for PID 20089   :            Group Key: b
 arallel worker for PID 20089   :            ->  Parallel Seq Scan on bigtable  (cost=0.00..8591.67 rows=416667 width=8)
 arallel worker for PID 20089   : auto_explain database=NULL client=local appname=psql
 arallel worker for PID 20089   :          duration: 128.247 ms  plan:
 arallel worker for PID 20089   :          Query Text: select sum(a) from bigtable group by b;
 arallel worker for PID 20089   :          Partial HashAggregate  (cost=10675.00..10774.59 rows=9959 width=12)
 arallel worker for PID 20089   :            Group Key: b
 arallel worker for PID 20089   :            ->  Parallel Seq Scan on bigtable  (cost=0.00..8591.67 rows=416667 width=8)
 res [local] SELECT: auto_explain database=postgres client=[local] appname=psql
 res [local] SELECT:          duration: 149.478 ms  plan:
 res [local] SELECT:          Query Text: select sum(a) from bigtable group by b;
 res [local] SELECT:          Finalize HashAggregate  (cost=13865.98..13965.57 rows=9959 width=12)
 res [local] SELECT:            Group Key: b
 res [local] SELECT:            ->  Gather  (cost=11675.00..13766.39 rows=19918 width=12)
 res [local] SELECT:                  Workers Planned: 2
 res [local] SELECT:                  ->  Partial HashAggregate  (cost=10675.00..10774.59 rows=9959 width=12)
 res [local] SELECT:                        Group Key: b
 res [local] SELECT:                        ->  Parallel Seq Scan on bigtable  (cost=0.00..8591.67 rows=416667 width=8)

It is expected behave? Minimally it is undocumented and in this case a global variable MyProcPort is not initialized.

Regards

Pavel


Reply | Threaded
Open this post in threaded view
|

Re: auto_explain and parallel queries issue

Tomas Vondra-4


On 04/16/2018 09:25 AM, Pavel Stehule wrote:
> Hi
>
> I am testing PostgreSQL in customer's environment and I found some
> unexpected behave.
>
> When auto_explain is used, and there is slower parallel query, then
> auto_explain raises log for any slow process.
>

FWIW this is the same issue that I reported in 2016:

https://www.postgresql.org/message-id/3f62f24e-51b3-175c-9222-95f25fd2a9d6%402ndquadrant.com

regards

--
Tomas Vondra                  http://www.2ndQuadrant.com
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services

Reply | Threaded
Open this post in threaded view
|

Re: auto_explain and parallel queries issue

Pavel Stehule


2018-04-16 10:33 GMT+02:00 Tomas Vondra <[hidden email]>:


On 04/16/2018 09:25 AM, Pavel Stehule wrote:
> Hi
>
> I am testing PostgreSQL in customer's environment and I found some
> unexpected behave.
>
> When auto_explain is used, and there is slower parallel query, then
> auto_explain raises log for any slow process.
>

FWIW this is the same issue that I reported in 2016:

https://www.postgresql.org/message-id/3f62f24e-51b3-175c-9222-95f25fd2a9d6%402ndquadrant.com

yes. The fix should not be too hard. The report can do only process who owns query text

Regards

Pavel
 


regards

--
Tomas Vondra                  http://www.2ndQuadrant.com
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services

Previous Thread Next Thread