Web update stuck at 50% since about 60 min

(Andrew Waugh) #1

I started a web update on our live site about an hour ago. It reached just less than 50% without any error messages, then as it changed from Compressing: locales to the first Compressing: application it just stopped.

My normal procedure is to run the updates on my sandbox, and see if they do anything odd, or generate error messages, then on the live site I wait until the daily backup has run, check that there is enough disk space, then initiate the web update while there aren’t many users online.

The web site is up, mails are going out, performance seems fine (So I don’t think we’re swapping, otherwise I’d see timeouts, right?).

What should I do, reset the web update and wait until someone with SSH access can do an app rebuild, or just leave it zombied until they can have a look at free?

EDIT: Going from v1.8.0.beta7 +77 (6ebddc4) to 9f299b6

(Joe Buhlig) #2

I run across this fairly regularly. I typically hit “Reset Upgrade” and then refresh the Versions page to see if it says I’m up to date or not. I can’t recall an instance that this didn’t work.

But then again, I’m comfortable doing that. I have SSH access to all Discourse instances I’m involved with. I can always rebuild if I need.

(Andrew Waugh) #3


I reset the upgrade.

/admin shows me still at v1.8.0.beta7 +77

and admin/docker shows me at 9f299b6

I did wait for/trigger the sidekiq jobs for stats. Now I’m not sure if the update completed or not.

(Sam Saffron) #4

I recommend you do a rebuild if anything is playing up.

(Andrew Waugh) #5

That’s the weird thing, I don’t see that anything is playing up. I’m going to look through the changes since 6ebddc4 and see if there is a way to check if any of them “went through”, then we’ll do a rebuild.

Thus far the web updates have worked flawlessly.

EDIT: I’d still like to see updates (and failed updates) in the GUI, similar to admin/logs/staff_action_logs