




















April 18, 2026, 12:27am 1
Sharing in case this helps anyone else that sees “Updating” in the admin panel although no update is actually in progress and the forum appears otherwise healthy.
While it never impacted docker manager updates, the past few versions I’ve had to do CLI updates as the admin panel always seemed to indicate discourse was updating.
My forum is small, no custom plugins.
The resolution to clear the issue was to flush the redis cache.
While I cannot share the underlying cause, it was super frustrating to hit a limit of my knowledge and understanding of the discourse update process (this is not a complaint, its an admission).
To this point, rebuild app had always been the reliable fix for pretty much any issue.
RGJ (Richard - Communiteq) April 18, 2026, 4:45am 2
Redis also contains user sessions and much more. Flushing Redis completely will log everyone out and delete all pending Sidekiq jobs. Scheduled jobs will only come back after the next restart.
This should only be a last resort and even then I encourage you to see which specific keys are the culprit instead of doing a flushall. This is like setting an entire building on fire because you want to get rid of a mouse.
Ed_S (Ed S) April 18, 2026, 7:15am 3
Could Redis be restarted instead? I think (Edit: wrong! see below) it’s an in-memory database so nothing would persist over, say, a server reboot. And a server reboot isn’t destructive (and could happen at any time.)
RGJ (Richard - Communiteq) April 18, 2026, 7:39am 4
That is incorrect. Redis does persist to disk.
fuse (Geoff Hughes) April 18, 2026, 11:35am 5
I appreciate your insight and expertise.
Things I tried:
Update from admin panel (failed/never started)
Followed the manual updating discourse steps (over the past 3 releases)
Did several launcher app rebuilds.
Searched the forum for a similar issue and solution
Finally resorted to a chatgpt query which uncovered redis options even though chatgpt itself said an unlikely fix (and suggested the things I had already tried).
This was the one action I took that resolved the issue.
I understand the analogy.
What else could I have done to better understand and resolve the problem as an alternative?
Ed_S (Ed S) April 18, 2026, 3:26pm 6
To be clear, you found
cd /var/discourse
git pull
./launcher rebuild app
did not work for you? (In what way did it not work? Is this a standard install?)
fuse (Geoff Hughes) April 18, 2026, 4:55pm 7
Did not do the git pull, did do the launcher rebuild app.
That process worked fine, just never cleared that the admin panel believed an update was in progress.
Ed_S (Ed S) April 18, 2026, 5:17pm 8
Thanks - so the observation is that the admin panel update isn’t working for you. Also, it hasn’t been working for me for a while! It would be better if it did work.
April 18, 2026, 8:42pm 9
i prefer the command line updating, but i just did a docker and discourse update from the UI on my iphone 15 and it worked quite well this time (standard install on DO).
i usually do it all from CLI on one of my computers, since it failed for me in UI last time i tried (a few months ago, so i had to ssh in anyways).
Ethsim2 (Ethan ) April 19, 2026, 6:46am 10
what deos free -h return?
what is UNICORN_WORKERS set to in app.yml?
what is db_shared_buffers set to in app.yml?
is db_work_mem commented out in app.yml?
how many (v)CPU’s deos your server have?
此内容由惯性聚合(RSS阅读器)自动聚合整理,仅供阅读参考。 原文来自 — 版权归原作者所有。