Query execution time is growing!

[Vasily Laktionov <va...@gmail.com>]

Hello, Igniters.
We see some strange things happening: execution time is growing if we
execute same query concurrently in multiple threads.
And it doesn't matter if we query existing or not existing cache!
Very simple reproducer and it's execution results are attached.
We test it on prod node and even on empty node running in local machine.
Persistence is disabled.
Can anyone explain what it the reason for such Ignite behavior?
reproducer.java
<http://apache-ignite-users.70518.x6.nabble.com/file/t1538/reproducer.java>  
test-results.txt
<http://apache-ignite-users.70518.x6.nabble.com/file/t1538/test-results.txt>  



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: Query execution time is growing!

[Ilya Kasnacheev <il...@gmail.com>]

Hello!

You have 15 threads finishing in around 100ms each, the rest waiting much
longer: leading me to believe you have 16 vCPUs on your server node and
exhaust your SQL pool threads.

Try setting ClientConnectorConfiguration.threadPoolSize to 100 and
IgniteConfiguration.queriesThreadPoolSize to 100, restart and re-run.

Regards,
-- 
Ilya Kasnacheev


ср, 30 сент. 2020 г. в 21:18, Vasily Laktionov <va...@gmail.com>:

> Hello, Igniters.
> We see some strange things happening: execution time is growing if we
> execute same query concurrently in multiple threads.
> And it doesn't matter if we query existing or not existing cache!
> Very simple reproducer and it's execution results are attached.
> We test it on prod node and even on empty node running in local machine.
> Persistence is disabled.
> Can anyone explain what it the reason for such Ignite behavior?
> reproducer.java
> <http://apache-ignite-users.70518.x6.nabble.com/file/t1538/reproducer.java>
>
> test-results.txt
> <
> http://apache-ignite-users.70518.x6.nabble.com/file/t1538/test-results.txt>
>
>
>
>
> --
> Sent from: http://apache-ignite-users.70518.x6.nabble.com/
>