[SELF HOST] "Metadata version not found for workspace" after failed version upgrade
I had Twenty v0.51.13 installed on a server. I tried to upgrade to v1.1 by following the procedure: going one major version at a time. However, it stopped working fine after v0.54. I tried a roll-back to v0.51.13 using my initial database backup, but now and after many tries, I cannot login anymore. I keep having the "Metadata version not found for workspace".
I already tried flushing the cache, syncing the metadata, etc... I'm lost now.
30 Replies
Hello at what version are you stuck right now ? Normally this a cache flush issue
Does restoring your initial backup results to a functional instance ?
Hi 👋🏻 ,
I'm at version v0.51.13. Restoring the initial backup didn't help.
I've tried multiple times with no luck.
Have you restarted your containers/pods ?
Yes I've done and too
Could you please share your cache flush logs ?
Just to be sure you're restoring the 0.51.13 dump on the same twenty instance isn't ?
Yes I do
Here are the logs:
No problem here
Please remove any remaining volumes using
docker compose down -v
be careful this is a destructive opération ( keep your dump file )So I should stop, remove and build again?
Adding the
--volumes
or -v
flag will delete any existing volumes
This flag has to be added to the down command
In a nutshell remove and stop at the same timeOk, should I pull data from the backup again now ?
After deleting the volumes yes you will to restore the backup
Ok I've done that. Now I have the "You're not a member of this workspace" error
Probably related:
Are you restoring a backup of 0.55 or your latest instance?
backup of latest instance
core.workspace.version is 0.51.13 in my backup
Unfortunately it seems like your dump is corrupted due to migration involved in upgrading to the latest version
Do you have any other backup available ?
No I don't.
Is there any chance to at least recover companies and contacts information I have in my backup? I can try to install a fresh V1.1
Up 🙌🏻 new ideas?
Yes totally
You can extract them using the csv export and re-import them in an latest instance
Please find similar dilemma https://github.com/twentyhq/twenty/issues/13174#event-18641829645
Thanks, will give it a try and keep you posted
I managed to get it back to work on a v0.51.14 fresh install.
I'll try to upgrade one version at a time again.
Congrats ! Please be sure to save a dump a dump before doing so
When trying to go from v0.51.14 to v0.52.11, I got the following error message after:
Cannot read properties of null (reading 'fromObjectMetadataId')
please share the whole logs
especially the line prefix
Here you go:
As you sggested in another post, I manually deleted the problematic relations. However, I have now this error:
It seems like a command is not idempotent
I would recommend restoring the backup, removing the problematic relation row, and then upgrade
Can you elaborate further please ? Which problematic relation should I remove ?
As you sggested in another post, I manually deleted the problematic relations. However, I have now this error:1/ Restore 0.52.14 backup in an 0.51.14 twenty instance 2/ " I manually deleted the problematic relations." 3/ Upgrade
I gave up. I installed v1.1.1 and will manually fill in back the data. RIP and thanks for your support. Hope it will never happen again
Sorry to hear that, you can import data as csv to fasten things up
Yep