I was doing some routine on my box and also decided upgrade Discourse via ./launcher rebuild. Surprisingly, it failed. Never seen this before. I am on stable branch, so the update was from 1.6.7 to 1.6.9. ./launcher rebuild app Ensuring launcher is up to date Fetching origin Launcher is up-to-date …

Hi, yes, exactly this error. What was happening for me was the following, and I believe the same is true in your case. I was on Scaleway VC1S. The stop timeout is not enough for this system, after 10 seconds PostgreSQL is still running. Now with -t 10 we are sending a kill signal to PostgreSQL whi…

OMG, some serious shit going on with one of the problematic sites https://cd-rw.org I restarted the App and site came back to life. Did a manual backup, but it is several megabytes smaller than the previous backup from early November. Did a rebuild and the site lost a lot of the settings and they w…

Reading the log, your postgresql was stopped abruptly, so it needs time to recover itself. The problem is our script doesn’t wait that much for a screwed up pg to recover so you will need some fiddling.

Yeah, I guessed something like that. Very interesting that this happened on both of my Scaleway instances, but never seen this before. Also, there might be room for improvement that before the 1st rebuild attempt there was no indication what so ever that something was wrong. Site basically worked, o…

Spotted something interesting in the cheap hosting thread : It is actually so slow that templates/postgres.template.yml had to be manually modified from sleep 5 to sleep 60 for postgres. Or could @zsero share us what kind of issues he experienced before making the modification?

We should increase our timeouts here @falco to make them safer.

I think @sam wants something like this: https://github.com/discourse/discourse_docker/pull/323

either that or the original proposal which made it dynamic if you can swing it. https://github.com/discourse/discourse_docker/issues/309

Hmm I thought this was a maximum amount of time to wait for Postgres to cleanly shut down, versus an absolute number.

https://github.com/discourse/discourse_docker/pull/324 looks like pg_ctl -w does all the magic for us. @ljpp said scaleway is slow, can you repro the problem consistently? I would like to test if this PR solves running in slow hardware

Upgrade failed due to unclean database shutdown

Product Support Self-hosting

sam (Sam Saffron) December 28, 2016, 10:34pm 9

Sleep 60 means everyone waits 60 seconds EVERY time a rebuild happens, just make it a param and the 1 in 1000 that has issues can bump it up in container yaml

Topic		Replies	Views
Database system was not properly shut down error when rebuilding Self-hosting	3	7087	September 19, 2019
Discourse update doesn't wait for Postgress DB to shut down Self-hosting	4	1242	December 29, 2021
2.6.0 Beta4 update failed with what looks like db errors Self-hosting	6	808	October 19, 2020
Discourse rebuild fails due to unclean shutdown Product Support	3	351	February 4, 2025
PostgreSQL Stuck During Rebuilding Self-hosting	51	2932	September 4, 2024

Upgrade failed due to unclean database shutdown

Related topics