We run into the same problem after installing updates around 5.0-1-errata193. After some time UCS reports that nextcloud isn’t installed anymore. Maybe it’s related to 5.0-1-errata196?
Even after removing all Nextcloud containers and images, reboot and reinstall, the error comes back after some time. This is the third try for us.
After installation univention-app info reports correctly:
UCS: 5.0-1 errata218
Installed: dhcp-server=12.0 mailserver=12.0 open-xchange-guard=2.10.5-ucs1 open-xchange-text=7.10.5-ucs1 oxseforucs=7.10.5-ucs4 samba4=4.13 4.4/nextcloud=21.0.7-0
Upgradable:
But after server is running for about 30 minutes it reports like this and in the app center you should install Nextcloud again:
UCS: 5.0-1 errata218
Installed: dhcp-server=12.0 mailserver=12.0 open-xchange-guard=2.10.5-ucs1 open-xchange-text=7.10.5-ucs1 oxseforucs=7.10.5-ucs4 samba4=4.13
Upgradable:
The Nextcloud container can then still be up, running and accessible via web. But when it isn’t accessible anymore, the output of journalctl -xb | grep docker then shows something like this:
Feb 12 16:53:45 testsrv named[1408]: listening on IPv4 interface docker0, 172.17.43.1#53
Feb 12 16:53:45 testsrv kernel: IPv6: ADDRCONF(NETDEV_UP): docker0: link is not ready
Feb 12 16:53:46 testsrv dhcpd[1803]: No subnet declaration for docker0 (172.17.43.1).
Feb 12 16:53:46 testsrv dhcpd[1803]: ** Ignoring requests on docker0. If this is not what
Feb 12 16:53:46 testsrv dhcpd[1803]: to which interface docker0 is attached. **
Feb 12 16:53:47 testsrv dockerd[1487]: time="2022-02-12T16:53:47.678965599+01:00" level=info msg="Removing stale sandbox ID1 (ID2)"
Feb 12 16:53:47 testsrv dockerd[1487]: time="2022-02-12T16:53:47.910935113+01:00" level=warning msg="Error (Unable to complete atomic operation, key modified) deleting object [endpoint ID1 ID2], retrying...."
Feb 12 16:53:47 testsrv systemd[1]: run-docker-netns-123456789012.mount: Succeeded.
-- The unit run-docker-netns-123456789012.mount has successfully entered the 'dead' state.
So the container stalls and get stopped, but until now we didn’t know why and only the Nextcloud container is affected. In our search for the error we also discovered that the output of univention-app is very inconstant.
Most times it’s correct:
nextcloud
Name: Nextcloud Hub
Latest version: 21.0.7-0
Installations: testsrv
But after some time, it shows this:
nextcloud
Name: Nextcloud Hub
Latest version: 21.0.3-0
Installations:
or that:
nextcloud
Name: Nextcloud Hub
Latest version: 21.0.5-0
Installations:
Removing cached data for 4.x and 5.0 like stated in this post doesn’t solve the problem. When running univention-app remove nextcloud it want to remove the version it thinks is the actual one, but this is not always the installed one.
After running univention-app update mostly points to the real current version, but even when it reports the server under installations, it isn’t listed in univention-app info after it is gone there.
@pixel
The container shouldn’t be “gone”. It should be just stopped or stalled.
Try docker container ls --all to show up also stopped container.
Sure, I do a rollback in Proxmox to the state before.
