Redis MAXMEMORY should be clamped

[smlx]

Suggest 100Mi min to 2Gi max.

Making this value too large can cause site and cluster stability problems.

[tobybellwood]

Any prior art on how we could constrain this in an entrypoint - I can see lots of bash loops 🙈 - but maybe a tiny binary may work better and be reusable!

[smlx]

Maybe the logic should not be in an entrypoint but instead in the build-deploy tool? Bash for this kind of arithmetic is just painful.

[shreddedbacon]

Unfortunately, we can't control this value in a build if someone injects it via a .env or .lagoon.env based file, or if they code it in a Dockerfile

[seanhamlin]

I have seen a 10GB redis-persistent pod, this ends up causes loads of issues in the cluster, from node OOMs to the pod being evicted. Not to mention the 10GB PV backup.

Having a way to set limits on the amount of RAM a given redis pod can have would be ace. e.g. through spec.containers[].resources.limits.memory. https://kubernetes.io/docs/concepts/configuration/manage-resources-containers/#requests-and-limits