Large log files generated by watchdog

Help
1

I’m new to Flox and am giving it a spin to check it out. Even with a trivial environment, I’m seeing very large log files created under .flox/log, with many entries per second of the form :

2024-11-17T10:16:59.357869Z DEBUG watchdog{socket=“/root/.cache/flox/run/flox.0cf2a520.sock” log_dir=“/root/flox-test/.flox/log”}: flox_watchdog::process: still watching PIDs {AttachedPid { pid: 18, expiration: None }}

Is this expected behaviour, and is there a way to suppress these debug log messages?

In case it’s relevant, I’m running in an Ubuntu OCI image with podman in WSL2.

Thanks!

2

Thanks for flagging the problem! It seems we left debugging enabled in the release - I can confirm I’m also seeing this on aarch64 darwin and logged flox-watchdog is compute, storage-intensive · Issue #2383 · flox/flox · GitHub to document the issue.

I took a quick look at the rust tracing crate documentation but couldn’t figure out if there was an environment variable or flag that could be used to reduce the default log level down from debug to warn or error, but you may find these two commands useful while we work to release a hotfix:

pkill -STOP flox-watchdog
pkill -CONT flox-watchdog

We’ll get back to you ASAP with an ETA for the fix on this. Again, many thanks for bringing it to our attention!

3

That works. Thanks for the quick response!

4

And … yesterday we released version 1.3.5 that includes a fix for this issue. If you upgrade, perform one final pkill -CONT flox-watchdog, then exit and restart all Flox environments then you shouldn’t have any further issues. Please let us know if you find otherwise!

Again, many thanks for the feedback!

5

This works great, thanks. And this version also works in my WSL2 environment rather than me having to run in a container, so double-win!

2 Likes