Helm chart missing resource limits causing HPA issues and memory growth

Hey folks
Quick question about Casdoor’s Helm chart and HPA behavior:

We noticed the official chart doesn’t define any resources (CPU/memory requests or limits). That’s causing a few issues for us:

Pods can consume all node resources if no limits are set.

HPA doesn’t scale properly because there are no requests to base metrics on.

Our dashboards also lose visibility for CPU/mem utilization.

We’ve seen Casdoor start with low memory usage but grow up to ~900 MB in one cluster and 3–4 Gi in another.

Questions:

Does Casdoor have any internal caching or process that might explain the memory growth over time?

Would it make sense to add default (or example) resources to the Helm chart so HPA works properly out of the box?

Thanks a lot!

what's your usage pattern? What features you use in Casdoor? Have you used the syncer?

Our usage pattern is fairly light and standard: we’re using Casdoor via OAuth2, with password grant type, client credentials, user CRUD and listing permissions, user info and logout flows. We are not using the syncer feature (we have no external directory or database sync configured).

If useful: we also don’t have large user import/migration jobs or external IdP syncing currently.

Given this, do you think any internal Casdoor component (even unused syncer modules) could still contribute to memory/CPU growth over time? Or would you point me to where the growth might come from given this usage pattern?

no, you should profile the performance yourself