Hi
Attempting to upgrade from ch:amd64/focal/mattermost-k8s-25 to ch:amd64/focal/mattermost-k8s-26 resulted in the application status being stuck at "downloading charm":
mattermost .../mattermost-canonical-fl... maintenance 2 mattermost-k8s stable 26 10.x.y.z no downloading charm (ch:amd64/focal/mattermost-k8s-26)
After waiting for several minutes with no progress, I manually edited the deployment with kubectl to force new pods to be spawned, which seemed to "unstick" juju.
Via debug-log at the time of the upgrade:
application-mattermost: 06:24:55 INFO juju.worker.caasoperator.uniter.mattermost/16.charm downloading ch:amd64/focal/mattermost-k8s-26 from API server
application-mattermost: 06:24:55 INFO juju.downloader downloading from ch:amd64/focal/mattermost-k8s-26
application-mattermost: 06:24:55 INFO juju.worker.caasoperator.uniter.mattermost/15.charm downloading ch:amd64/focal/mattermost-k8s-26 from API server
application-mattermost: 06:24:55 INFO juju.downloader downloading from ch:amd64/focal/mattermost-k8s-26
application-mattermost: 06:24:55 INFO juju.worker.caasoperator.charm downloading ch:amd64/focal/mattermost-k8s-26 from API server
application-mattermost: 06:24:55 INFO juju.downloader downloading from ch:amd64/focal/mattermost-k8s-26
application-mattermost: 06:24:56 INFO juju.downloader download complete ("ch:amd64/focal/mattermost-k8s-26")
application-mattermost: 06:24:56 INFO juju.downloader download complete ("ch:amd64/focal/mattermost-k8s-26")
application-mattermost: 06:24:56 INFO juju.downloader download verified ("ch:amd64/focal/mattermost-k8s-26")
application-mattermost: 06:24:56 INFO juju.downloader download complete ("ch:amd64/focal/mattermost-k8s-26")
application-mattermost: 06:24:56 INFO juju.downloader download verified ("ch:amd64/focal/mattermost-k8s-26")
application-mattermost: 06:24:56 INFO juju.downloader download verified ("ch:amd64/focal/mattermost-k8s-26")
application-mattermost: 06:48:58 INFO juju.worker.caasoperator.uniter.mattermost/16 found queued "upgrade-charm" hook
application-mattermost: 06:48:59 INFO juju.worker.caasoperator.runner start "mattermost/17"
As you can see, nothing happened for about 24 minutes after the charm was apparently downloaded and verified, which lines up with the time I edited the deployment to modify an environment variable in order to force new pods.
Once I'd done that, everything proceeded as expected.
The behaviour is reminiscent of LP:1813884, but that was addressed years ago.
If this happens again it would be good to get a juju dump-db with yaml snippets for the affected application, unit, unittates documents.