[bitnami/redis] Liveness Probes Failing on AKS 1.29.2 about charts

[Alwinius]

I have a similar issue with timeouts in the liveness/readiness probes. I can see that the containers cpu usage is considerably higher than in other clusters that are not affected. This leads to throtteling by Kubernetes with the default resource limits.

However, for me the difference between the two clusters is the other way around: EKS 1.29.2 works fine, but when installed one day later on the production system with EKS 1.28.7, the issue appears. Both use very simple configuration without persistence

[javsalgar]

Hi! I think that the issue may be related to properly configured resource limits/requests for your use case. Could you confirm that these are configured and it is not using the ones set by default in resourcesPreset? These are not meant for production and are configured to the minimum value working in our test suite

[Alwinius]

Thanks for reply @javsalgar. We are using indeed the default values:
resources:
limits:
cpu: 150m
ephemeral-storage: 1Gi
memory: 192Mi
requests:
cpu: 100m
ephemeral-storage: 50Mi
memory: 128Mi

But in one system the actual CPU usage is 0.02 (20% of request) and in the other it's 0.15 (150% of request and equal to the limit) where both instances are basically idle and contain maybe 10 keys.

[javsalgar]

Could you try increasing the CPU limits to see if that solves the issue? At least to pinpoint where the problem could be

[Alwinius]

Yes, this seems like a workaround. Initially the CPU usage is around 0.32 and then drops to 0.17. The instance contained 0 keys at that point.

[javsalgar]

It probably is related to how the database performs a first initialization. This is not happening in our tests with major k8s distributions (including AKS) but each deployment should adapt the resources section to their use case.

[Alwinius]

We updated the cluster with high CPU usage to EKS 1.29 now and the issue seems to be gone.

But I would still say that this is a bug if an empty redis consumes 10x the CPU than in another cluster.

[javsalgar]

Hi,

It could be, but probably it seems more of an upstream issue rather than an issue in the Bitnami packaging. Maybe it's worth reporting it upstream.

[github-actions]

This Issue has been automatically marked as "stale" because it has not had recent activity (for 15 days). It will be closed if no further activity occurs. Thanks for the feedback.

[github-actions]

Due to the lack of activity in the last 5 days since it was marked as "stale", we proceed to close this Issue. Do not hesitate to reopen it later if necessary.