High memory usage since v0.57.5

[jespercodes]

Even since upgrading to v0.57.5 I’ve experienced crazy high memory usage.

I’m running Metabase v0.57.5 self-hosted with Postgres 17 using Dokploy (template link: Dokploy Blueprints but updated the version numbers).

See the attached screenshots here:

Any ideas why this might be?

Thank you!

[dwhitemv]

Your charts show memory use has stayed flat, but there seems to be a stuck process somewhere in there that’s keeping CPU busy.

Have you tried restarting the container? Have you checked the Metabase log for any issues?

What version did you upgrade from on Dec 3?

[jespercodes]

Good old restart did the trick. Didn’t check the logs if anything was stuck. Will check if it happens again.

Thanks!

[TonyC]

You also might be impacted by this Metabase grinds to a halt after a week or so - Performance issue when upgrade from 56 to 57 · Issue #66385 · metabase/metabase · GitHub … Which will be solved by this [perf] Add metric for `lib.computed/weak-map` size; speculative fixes by bshepherdson · Pull Request #66404 · metabase/metabase · GitHub

[dwhitemv]

You know, I was able to trigger memory issues while developing a new dashboard yesterday. The main trigger was a map question with ~11k pins in it, metabase would start getting draggy loading the tiles and the GC threads would peg the CPU. If I get time today, I’ll see if I can reproduce and grab a heap dump.