Parca server pod evicted on kubernetes

[Shreyas220]

I have deployed parca on AKS and it keeps getting evicted after exceeding the memory limit, possible memory leak

  Type     Reason     Age   From               Message
  ----     ------     ----  ----               -------
  Normal   Scheduled  26m   default-scheduler  Successfully assigned parca/parca-85bd8c4cd7-tj79g to aks-agentpool-15819346-vmss000000
  Normal   Pulled     26m   kubelet            Container image "ghcr.io/parca-dev/parca:v0.14.0" already present on machine
  Normal   Created    26m   kubelet            Created container parca
  Normal   Started    26m   kubelet            Started container parca
  Warning  Evicted    20m   kubelet            The node was low on resource: memory. Container parca was using 3090924Ki, which exceeds its request of 128Mi.
  Normal   Killing    20m   kubelet            Stopping container parca

[metalmatze]

This is probably to be expected when running in-memory only. After a certain while, the meta store will grow and grow and never shrink. On top of that comes the amount of data ingested, so it might very well be that it might be evicted eventually.

We've already made some big improvements to Parca's usage in #2220 again. That being said, the same principles from above are still true. Ultimately, we want to run Parca with persistence, so all data won't be held in memory anymore (especially the meta store).

[brancz]

@metalmatze I think you meant to link https://github.com/parca-dev/parca/pull/2202