Upgrading from 2.9.0.beta8 broke the site, unrecoverable

Hello

We have simply clicked the admin upgrade button.
Did not expect this:

Oops
The software powering this discussion forum encountered an unexpected problem. We apologize for the inconvenience.

Detailed information about the error was logged, and an automatic notification generated. We'll take a look at it.

No further action is necessary. However, if the error condition persists, you can provide additional detail, including steps to reproduce the error, by posting a discussion topic in the site's feedback category.

Now, while this is a bit discouraging (what if this happens when the site has a couple thousand users?)… we have no users yet. Lucky. But we can’t get it back up running. There is no discernible error, any step to take to recover what went wrong… unless the ominous automatic notification generated. Who received that notification? What does it say? Why does a simple upgrade of the software send the site into oblivion? How can we fix that?

3 Likes

Are you still able to access /logs on your forum? That is where the notification went.

The last two questions can only be answered after you have looked at it.

3 Likes

Nothing worked anymore unfortunately. So we had to re-instantiate a backup server side, and now we are a bit wary of updating again.

Is there any way to update in safe mode? I can hardly believe that every time the software updates we would have to run a risk of total downtime.

That is a real risk with every app, including Discourse. That’s why big guys are using a stage server and we smaller fishes are doing backups (site AND database) before every update, but at least before any upgrading.

3 Likes

Hello and welcome @smileBeda :slightly_smiling_face:

Could you give us some more detail about your set up? For instance, what version you’re running, whether it’s a standard install, and if you have any plugins (official or third-party).

Normally upgrades go very smoothly, though taking a backup of a live site just beforehand is always prudent just in case.

4 Likes

We are running 2.9.0.beta8
The update it asks us to do is docker_manager 45986ad4 (Comparing 45986ad40ec642ea617a083aecc75e0078878fb7...90a8a6c6cae0bba01de7fdf340da1ab878bff913 · discourse/docker_manager · GitHub)

It is a standard install from docker, however we run a couple plugins (but nothing unusual, really. No custom code so far)

At this point (without error log) I cannot really tell more, I guesss we’ve to bite the apple again and after a backup, run the update and if it fails again, submit the logs…

1 Like

Can you paste your app.yml file here? It will make it easier to see what was and what is.

3 Likes

There are also some good pointers on how to manually update from the server in this topic:

A ./launcher rebuild app can often be a good Plan B, and if you get any errors during that you can scroll up and have a read.

4 Likes

Did you do a

 ./launcher rebuild app

Or did you upgrade from the web interface? The full rebuild is sometimes required.

The logs you’re looking for are in

     /var/discourse/shared/log/rails/production.log

(if I typed it right on my phone)

5 Likes

close enough :grin:
/var/discourse/shared/standalone/log/rails/production.log
and maybe
/var/discourse/shared/standalone/log/rails/production_errors.log

5 Likes

Could you please elaborate on this? Which plugins?

3 Likes

Exactly my thoughts, that’s why I asked they paste/share their app.yml file

4 Likes