smf: Set clickhouse log level to 'information'#10484
Conversation
The default clickhouse log level is `trace`, which is exceptionally verbose. Set the log level to `information` to cut down on the amount of disk and CPU we're spending on logs.
|
Log level names taken from https://clickhouse.com/docs/knowledgebase/why_default_logging_verbose |
bnaecker
left a comment
bnaecker
left a comment
There was a problem hiding this comment.
THANK YOU
The main, and blocking, comment is below.
One more vague one is that we might want a knob to change this on demand, something like the retention-policy we added recently. ClickHouse watches its config files for changes and reloads it, so we could in theory add an API to the admin server which tweaks this on demand.
I don't know if we need to do that now. It's a lot more work. But if we don't, please file an issue, because I imagine that we might need it, especially when we get back to rolling out the replicated deployments.
| <clickhouse> | ||
| <logger> | ||
| <level>trace</level> | ||
| <level>information</level> |
There was a problem hiding this comment.
I think we need to change this a few places, which is very annoying. This one file here is only for the single-node case, and possibly only in development setups. The clickhouse-admin crates include code for emitting the entire config XML file for the servers they start. So we need to change it:
- Here
- the equivalent replicated XML file
- replicated admin server config generation
- single-node admin server config generation
2 participants
The default clickhouse log level is
trace, which is exceptionally verbose.Set the log level to
informationto cut down on the amount of disk and CPU we're spending on logs.Closes #7946