During a lxc copy a b:a command snap tried to update lxd. The copy process was aborted and lxd/lxc was missing on the target server.
I had to do a snap refresh lxd to get it back up. I assume that should not happen. I am just not sure if this is a problem with snap or lxd or maybe both. Should I open a bug report somewhere or is it ok to report it just here?
Additional question: if the copy process gets killed half way through, can I just call it again without deleting the half finished container on the target server? Rsync should take care if it and copy just the missing files, right?
For the first part of your question, it is an issue with the snapd service that schedules the update. I do not think that the application (lxd in this case) can affect the refresh timing on snapd.
The default for snapd is to schedule the update at a random time within a day the update has appeared. They do this so that the Snap Store servers do not get inundated with too many updates at exactly the same time.
You can use the following keys for snapd to direct when it is convenient for you to get these updates,
For example, you can arrange any updates to take place during out-of-office hours, or outside the time that you are backing up containers.