@McFaul In reply to your post in the “Dev log for 3.8-15”
Answering here so as not to clutter the dev log post.
Sorry to not have replied earlier it’s just that I don’t have much to offer on your reported error. It looks like a simple database deadlock, except that these types of errors are not that often ‘simple’. As you say it could be related however there have been no other reports of this (as yet) and from the tests I’ve done import works exactly as was. The change to by-id was deep but didn’t really change much structurally.
I think this is a case of more info may help ie what is the nature of your pool and if you could paste the output of the following command here it may help others identify what the problem is.
btrfs fi show
Also a screen grab of your Disk page may help as well.
When I have seen db contention errors I have often simply tried again and had success as they are often very time dependant / transient. It very much depends what’s going on in the background when you attempt the pool import. I take it you have tried the import more than once?
Sorry but db contention is a little out of my area currently, but the slight change in pattern of db access by the by-id name change may have triggered something but quite a few imports have been done since this change and this db issue hasn’t arisen.
Another possible cause is if you have your Disks page on an auto refresh when you attempt the pool import that may increase the likely hood of this contention, due to the additional db pooling. It shouldn’t of course but there may be a bug where db actions are not atomic where they need to be or the like.
Also note that all your devices must not have serial issues as this may also cause problems.