Pyroscope retention period config

[rmn-lux]

Hello, thank you for the coroot. I found out about this great product in an article on Habr.

I have a default installation in kubernetes via helm chart. Faced a problem that in pyroscope either retention doesn't work or I don't understand something. The space is clogged with metrics very quickly and is not cleared.

The configuration was default, I reduced retention but eventually all the allocated space is clogged and everything stands at a standstill.

On the screenshot above is an example: in the morning at 9.00 a clean install was done, and by the evening at 18.00 it was already 3 GB. Can you please tell me how to configure retention correctly so that it works and the metrics take up less space? Thanks.

P.S. I realize that the problem is not in coroot itself, but in a third-party product, but I hope you can tell me how to properly work with metrics in pyroscope or suggest some documentation on this issue.

UPD. for 21 hours of work accumulated logs of 8 GB already. there are no production services, test microservices are installed from https://github.com/GoogleCloudPlatform/microservices-demo.

[apetruhin]

@rmn-lux, thank you for the report.
It appears that Pyroscope does not remove old data (grafana/pyroscope#1977).
We are currently exploring ways to fix or implement a workaround for this issue.

[def]

@rmn-lux, starting from version 0.27, Coroot uses ClickHouse to store profiling data. The default TTL is set to 7 days, but you can adjust it if necessary.