Oh, indeed, there is a backup in Gladys, I had forgotten (well done indeed!!)
Otherwise, another solution would be to simply delete all your Matter nodes in the interface by clicking on « decommission, » that should also work:
I took advantage of this opportunity to restart a build by setting the « Matter.js » integration to « debug » log level to get as many logs as possible. The build is currently running: Build Gladys dev images · GladysAssistant/Gladys@03cd3b1 · GitHub
Ok, that’s the operation I was doing while decommissioning 
Let me know when I can pull the new version
@prohand you can already test re-commissioning your devices on the old version.
The new version will be ready in approximately 5-20 minutes. I will post a message here!
1 Like
In fact, it’s already ready, the build was super fast thanks to the cache!
1 Like
I have a functional instance with the old version.
I will test with the latest version of the image:
gladysassistant/gladys:upgrade-matter-latest
1 Like
Thanks for all your tests! Looking forward to seeing the logs and if it works, remember to let Gladys run for a few minutes to properly let the migration execute without interrupting it.
I still have the devices disconnected:
And I see 2 Nodes:
I did see logs regarding a matter migration.
In any case, there’s something more compared to the last time, I suppose thanks to the debug log.
I’ll share the log file with you in DM
1 Like
Thank you very much! It’s better with the debug logs
It’s interesting, I have the impression that the migration is going well, however I do see the disconnections in the logs:
Failed to resume connection to @1:ae7e1c89870a2cbe with udp://[fdde:adbe:ef12:5678:be24:11ff:fe88:3755]:5540, discovering the node now: (Failure (1) / NoSharedTrustRoots (1)) Received general error status for protocol 0
I have forwarded this to Appolon77
Okay, thank you
In addition to that, I made a PR to fix the « temporary » migration bug that causes a « short disconnection » to be normal according to him:
But apparently, he won’t do a backport just for that, for him it can go into 0.17 but not before
Appollon77 told me:
- NoSharedTrustRoots means that the fabric was fucked up
Would you be able to start this test with a fresh instance of Gladys (new folder, and Gladys completely fresh from scratch)? You can use another folder on the disk for the volume for this.
You install Gladys production « stable », and you configure your Matter devices.
Then, you zip the « matter » folder so we have an archive of your Matter.js 0.13 state, and you send me this zip.
Then, you re-test the migration and you send me the logs!
I’ll pass everything on to Appollon77 
I’ll do this for you in the evening, no problem
1 Like
I’m using a Matterbridge on the same subnetwork as Gladys, but it’s not on the same VM.
Do I need to start from scratch on my Matterbridge server as well?
No, no, you only need to create a new instance on Gladys.
Make sure to do this in a separate environment, by modifying the Docker volume in the Docker run to do this in another folder:
Concretely, this line:
-v /var/lib/gladysassistant:/var/lib/gladysassistant \
Should become for example:
-v /var/lib/gladysassistant_test_2:/var/lib/gladysassistant \
On the left of the :, it’s the path on the host, and on the right, it’s the path in the container.
I have some good news! It’s working fine by starting with a clean version of Gladys stable and migrating as you indicated
Do you still want the folders and logs, or is it no longer needed?
Small question, how can I delete the backup that restores itself automatically?
1 Like
Oh great!
I think I’ll add a button in the interface to start over.
Yes that would be great 
Ah, and for the logs and the folder, no need if everything works again in the end!
1 Like
1 Like