Periodic connection loss

[asimpson]

What happened:
Every day the datasource throws a 500 error that is resolved by re-saving data source settings in Grafana. An error is logged as rpc error: code = Unavailable desc = error reading from server: EOF.

[asimpson]

One suggestion was to toggle the protocol selection in the config editor. We're waiting to see if that gets rid of the periodic error.

[Ryado]

This could be related to ideling in ClickHouse Cloud, wdyt @gingerwizard ?

[jkaflik]

@asimpson it's version v2.0.7?

@Ryado @asimpson

The main question is if this issue is reproduced on ClickHouse Cloud only or happens for ClickHouse in general.
For cloud, extending dial timeout would do the trick if service goes idle. It's 10s default on backend:

https://github.com/jkaflik/clickhouse-datasource/blob/13a95dd951983246c856b617fb8165fbb9b6ca1b/pkg/plugin/settings.go#L120

In UI we suggest 60s, but user can skip it easily:
https://github.com/jkaflik/clickhouse-datasource/blob/13a95dd951983246c856b617fb8165fbb9b6ca1b/src/selectors.ts#L64

[jkaflik]

@asimpson, this error message rpc error: code = Unavailable desc = error reading from server: EOF - isn't it an GRPC error, so Grafana could not talk to plugin process? I think we need to check what happens inside plugin. Is it something you can reproduce?

[jkaflik]

@asimpson do you have any updates on this topic?

[asimpson]

@jkaflik Unfortunately I was not able to reproduce the issue 😞

[aangelisc]

Is this still an issue @jkaflik?

[mshustov]

let's close until we have a report with more details