fix: allow to override default config/log_config.yaml by a custom one

[sami-racho-scalac]

Right now is not possible to replace the default config/log_config.yaml file.
So I think the value logConfigFileLocation should be used as path to render the configmap.

[tazarov]

@sami-racho-scalac, I think you are right, but even better would be to not store the logs on a potentially ephemeral pod filesystem. I'll be happy to update the chart to accommodate this or wait for you to reopen the PR .

[sami-racho-scalac]

Yeah, the pr needs more work. My objective is to being able to override the default log_config.yaml, but my current approach doesn't work.

[tazarov]

@sami-racho-scalac, I get you. The chart does create a configmap with the log config and then mounts that to the chroma pod. Would you like me to make it possible so that you can provide the path to the local machine file which could be used as logging config?

[sami-racho-scalac]

That's what I tried but if I'm not wrong helm doesn't allow to use files outside the chart.
I think the easier way would be to save the config inside a chart value

[tazarov]

@sami-racho-scalac in theory you can use a whole string containing the yaml config file and set that via something like set-file but I agree that this is not optimal.

I think it might make sense to expose logging configuration via variables and inject those into a template (one thing that helm is actually good at). What do yo think?

Which parts of the logging config do you want to update?

[sami-racho-scalac]

In my case I just need to change the log levels

[tazarov]

ok, that seems simple enough. I'll push the change today unless you want to contribute this as PR?

My idea is to treat the logging config as yaml file and then pass the logging params such as log level as one of the template params to fill in.

On a related note, I'm mostly done with an observability layer with OTEL, Prometheus, Grafana, Loki etc. I'll also add that to the chart over the next week.

[sami-racho-scalac]

Then I leave it in your hands. Thanks!

[tazarov]

@sami-racho-scalac, added logging level support via --set chromadb.logging.root/chroma/uvicorn flags. Let me know if that works for your use case.

[sami-racho-scalac]

Perfect! Thx so much.