We’ve been running our Discourse instance for a few months on forum.hobbydb.com now to support our growing collectible database with a community (replacing various Facebook groups). While everything is functionally stable, we’ve noticed some performance issues — particularly around Lighthouse scores, which are currently hovering around 60%.
Infrastructure: Entirely containerized, running on Kubernetes
Specs: MS Azure 2xlarge (so should be able to automatically expand to 6 cores, 12GB of RAM and up to 2GB for logs)
We’re trying to figure out where the performance bottlenecks might be coming from, and how we can better optimize our deployment — especially for frontend performance.
Has anyone faced similar issues when running Discourse over Kubernetes (especially via Bitnami)? Any recommendations, tips, or known pitfalls we should look into?
Appreciate any advice or suggestions — we’re keen to improve the experience for our users.
Yes your instance is very slow. That’s not typical.
Bitnami is a non-standard third party install. Have you raised these issues in their community?
Some of these online benches don’t make sense for a Single Page App (like Discourse) because each client downloads most of the JavaScript on first visit. After that most page transitions should be near instant as many things are cached and only some specific content json is being retrieved from the server on most transitions.
(I believe CDCK is working on spreading that download beyond first page pull but that’s kind of getting off topic and it’s not your issue)
My instances are self hosted with the standard install with very modest hardware (4GB VPS) and fly .
.