Any way to reset a failed upgrade attempt for Mastodon? (Might apply to any app.)
-
SO, my Mastodon app STILL won't back up properly (latest was a failed upload of a pgsqldump), but I do want to upgrade it.
So I clicked the upgrade button and it started, but I forgot to select "Skip backup", and the backup started. Well, I know it's going to fail, and the last time I let it run, and fail, the app itself wouldn't restart, and I had to reinstall a saved SERVER IMAGE, at the host level, to get it running.
I stoppped the backup immediately, and for some reason the Mastodon is still running and available. BUT, I can't now rerun the Check for Updates, and the app is shown with an Error,
Error : Task Error - Task 15050 stopped
. But like I said, it is still running!So, the question is: Can I reset this failed backup alert/notice/status, so that I can Check for Updates, see the update, and this time select the "Skip Backup" so that all it does is update? Would this be some Docker-hidden flag in the Docker image? Or is there a CLI command to I can run to update without backing up?
Thank you!
-
@scooke That looks like a bug . Have you tried a refresh of the UI already ? Otherwise, you can always update with the CLI tool like
cloudron update --no-backup --appstore-id org.joinmastodon.cloudronapp@<package-version>
. See https://git.cloudron.io/cloudron/mastodon-app/-/tags for the next package version for your current package version. -
The latest mastodon package will purge more cached assets, which at least made the backups more stable for my instances. Also it allows to set for how long you want to keep cached content https://docs.cloudron.io/apps/mastodon/#cache-retention-days
-
Hm, I ran the command and got an error:
cloudron update --no-backup --appstore-id org.joinmastodon.cloudronapp@1.11.4
App update error: Could not determine app. Use --app to specify the app
Do I add --app, or replace --apstore-id with --app? Or append it after what you wrote?
EDIT: I did this:
cloudron update --app domain.example.com --no-backup --appstore-id org.joinmastodon.cloudronapp?version=1.11.4
where "domain.example.com" is the domain name of the Mastodon app. Then it seems to have started the upgrade process.=> Waiting for app to be updated => Queued . => Downloading image ........................... => Cleaning up old install ........... => Downloading icon ................
Oh oh - this result
App update error: Installation failed: ER_BAD_FIELD_ERROR: Unknown column 'userId' in 'field list'
FWIW< I was trying to update from 1.10.0 to 1.11.4... aer there intermediate steps I needed to take?
Well, like last time, since there is no reliable recent backup, I am counting on SSDNodes' full disk backup to restore the ENTIRE VM so that I have a running Mastodon again. Let's hope and pray...
-
@scooke said in Any way to reset a failed upgrade attempt for Mastodon? (Might apply to any app.):
FWIW< I was trying to update from 1.10.0 to 1.11.4... aer there intermediate steps I needed to take?
Yes, indeed. You must always only update the packages one version after the another. You cannot skip releases. It will cause issues in migrations (like above). Most apps are rolling releases, they are not designed to upgrade from one arbitrary version to another.
-
Welp, I tried upgrading from 1.10.0 to the next one, 1.11.0, but the same error came up. Seems to me to be something that must have come up with someone else, or elsewhere. Any ideas?
App update error: Installation failed: ER_BAD_FIELD_ERROR: Unknown column 'userId' in 'field list'
-
I have a sinking feeling I'm going to need to just start from scratch, and install a totally new Mastodon using the latest from Cloudron and then get it aALLLLLLLlll set up again. Unless someone has an idea. I've search for the error and it returns nothing. I'm the only one in the fediverse with this problem. Shame.
-
@scooke said in Any way to reset a failed upgrade attempt for Mastodon? (Might apply to any app.):
App update error: Installation failed: ER_BAD_FIELD_ERROR: Unknown column 'userId' in 'field list'
mmmm, reading this error message again. This actually looks like an error in the box code and not from Mastodon itself. Can you please check what is in
/home/yellowtent/platformdata/logs/<appid>/apptask.log
? I suspect there is a larger backtrace somewhere in that file. -
Hi @girish. Unfortunately I don't have that anymore because the only way I can get my Mastodon running is to make use of the full disk backup that SSDNodes offers. The last successful backup of my Mastodon wa back in January, but even then (you may recall) that backup for some reason, though listed, was not present.
Any other way I can check?
FWIW, the previous fail, several days ago, had something to do with a pgdump upload failure. -
@scooke I am out of ideas then. If you like, you can reach out on support@cloudron.io and we can debug further to see if we can resurrect the app.