An upgrade failed mid-way and the forum went offline. we used codex to fix the upgrade and it worked again.
Hosted on Digital Ocean
Somehow it’s only laggy from my admin panel
Findings:
discourse-question-answer is stale upstream. The copy you’re running is from 2022-08-17.
The live site was not failing because of it. categories.json, latest.json, and / were all serving after recovery.
The actual issue was narrower: during rebuild/bootstrap, Chat’s one-time category seeding hit the plugin’s BasicCategorySerializer hook, and that hook called object.send(key) too aggressively.
That produced Error seeding chat category - #<NoMethodError: undefined method ‘qa_enabled’ for an instance of Category> in the rebuild log.
What I changed:
Patched the running plugin serializer to use a safe fallback instead of blindly calling send.
Added the same compatibility patch to /var/discourse/containers/app.yml so future rebuilds re-apply it automatically.
Kept backups at /var/discourse/containers/app.yml.bak-2026-04-08 and /var/discourse/containers/app.yml.bak-2026-04-08-qa.
Validation:
https://forum.longevitybase.org/ returns 200.
https://forum.longevitybase.org/latest.json returns 200.
https://forum.longevitybase.org/categories.json is returning qa_enabled fields correctly.
One caveat:
I did not run another full rebuild after adding the persistent hook, because the live app is healthy and the equivalent patch was already applied successfully in the running container. The next
rebuild should keep it, but that exact path hasn’t been re-executed yet.
If you want, I can do one more pass on whether that plugin should eventually be forked and pinned instead of cloned directly from the abandoned upstream repo.