maxmemory

[hermanbanken]

We deployed DragonflyDB using:

apiVersion: dragonflydb.io/v1alpha1
kind: Dragonfly
metadata:
  name: api-proxy-dragonfly
spec:
  replicas: 2
  resources:
    requests:
      cpu: 500m
      memory: 500Mi
    limits:
      cpu: 1000m
      memory: 1Gi

but this is output from the pod:

This usage of 22GiB is a bit problematic, because it would OOM at 1GiB already because of the Kubernetes memory limit. Is there a way to read this maximum memory either for the operator or for the Dragonfly instance itself?

Edit, apparently this exists: https://github.com/dragonflydb/dragonfly/blob/bc717a037d77f7b027de910c5ca5c64bf34e2af9/src/server/dfly_main.cc#L482-L501 from https://github.com/dragonflydb/dragonfly/pull/1109 which is part of:

• v1.3.0
• v1.2.1
• v1.2.0

But we are running v1.2.1 and still the detector detects 28GiB and not the 1GiB from the Kubernetes limit.

I think it is due to our LXC having only /sys/fs/cgroup/memory/memory.limit_in_bytes and upon closer inspection of https://github.com/dragonflydb/dragonfly/pull/1109 it seems like that file is not read at all.

[hermanbanken]

Opened a PR https://github.com/dragonflydb/dragonfly/pull/1293

[Pothulapati]

Hey @hermanbanken Thanks for not just raising the issue but also fixing it! Because the PR is merged, We can close the issue right?

Once a Dragonfly binary is released, You can expect an updated Operator release too! :)

[hermanbanken]

Cool, that is awesome, thanks!