ClickHouse closing pooled connections

[dizider]

Hi,
I'm struggling with HTTP connection pooling. I would like to reuse ClickHouse connections for multiple insert queries. I set the following client settings (related with pooling):

.enableConnectionPool(true)
.setSocketKeepAlive(true)
.setSocketTimeout(10, ChronoUnit.MINUTES)

Sometimes the duration between insert queries is few minutes, but the socket is closed after 30 seconds of inactivity by ClickHouse server due to timeout : ServerErrorHandler: Code: 209. DB::NetException: Timeout exceeded while reading from socket (peer: <ip>:49278, local: <ip>:8123, 30000 ms).

Is there something I missed, or misconfigured? Or is there any way to set larger socket keep alive timeout while creating the connection that ClickHouse will respect?

[mshustov]

ServerErrorHandler: Code: 209. DB::NetException: Timeout exceeded while reading from socket (peer: :49278, local: :8123, 30000 ms)

ClickHouse http_send_timeout setting should be adjusted. could you try to adjust it via serverSetting(String name, String value)

[dizider]

Thanks for an advice. I tried to setup the http_send_timeout property to hundreds of seconds, unfortunately without any results. When I checked system.query_log the only requests from client are inserts (which is fine) with 'http_send_timeout':'600' in settings. But ClickHouse server still closing connections after 3000 ms due to timeout.

Well, it is not related with connection pooling, since I am able to reproduce the same behavior when the enableConnectionPool option is disabled.

[mshustov]

This note in the docs makes me think you need to adjust it in the server config and reboot the server:

note
It's applicable only to the default profile. A server reboot is required for the changes to take effect.

[laeg]

@dizider did you test the recommendation of changing the server config?

[chernser]

Closing as no activity. Please open a new issue if problem persists.

[dizider]

Thanks for your help. I checked it again, and the correct parameter to tweak in the server config is keep_alive_timeout. After I increased it to 300 sec, everything works as I expected.

[chernser]

@dizider thank you for the response!