You are not logged in.
I have LVs containing /, /var, /home, and swap. If I have any snapshots of any of these, the 'Unmounting Filesystems' step during down/restart fails. It doesn't seem to cause any corruption; what I noticed initially was that the /var partition always needs journal recovery on the next boot. If I boot to the Arch installer and run dumpe2fs on it, it always reports clean. If I then run fsck from the installer, the journal recovery occurs there, so there's no recovery during boot of the installed system.
The only thing I found in the logs is a rearrangement of the /dev/dm-# entries, but that doesn't seem strange considering how snappshotting works. Plus it doesn't seem to happen until after reboot. I have absolutely no idea where to go from here, so any suggestions would be appreciated. Thanks
Last edited by alphaniner (2011-08-03 19:49:41)
But whether the Constitution really be one thing, or another, this much is certain - that it has either authorized such a government as we have had, or has been powerless to prevent it. In either case, it is unfit to exist.
-Lysander Spooner
Offline
I use snapshots to backup my root and often it will fail saying that the device is in use, even though my script first unmounts, then calls sync and sleep 2 before calling lvremove. A while later it will work.
Offline
I think your situation is very different from mine. I have never mounted or otherwise read from the snapshots. Just 'having them around' causes problems, even a snap of /home which is effectively idle (I made a user but I'm still logging in as root).
But whether the Constitution really be one thing, or another, this much is certain - that it has either authorized such a government as we have had, or has been powerless to prevent it. In either case, it is unfit to exist.
-Lysander Spooner
Offline
I've done some more poking about, but all I've come up with is the fact that this isn't an 'isolated incident'. I originally saw the error on my home box with Arch 64-bit, but I duplicated it on a 32-bit VBox VM. I have also tried both ext4 and reiser filesystems on /var, no difference.
Also, I forgot to mention before: once I get rid of the snapshot, the problems go away immediately. At the next down/restart the unmounting step does not fail, and there's no journal recovery at the next boot.
But whether the Constitution really be one thing, or another, this much is certain - that it has either authorized such a government as we have had, or has been powerless to prevent it. In either case, it is unfit to exist.
-Lysander Spooner
Offline
Per falconindy's suggestion on a potentially related thread, here's the output of ps aux and lsof from just before the unmount step in rc.shutdown:
I did some magic with the ps outputs and found these to be unique to the second case:
[ksnaphd]
[kcopyd]
[kworker/0:3]
/sbin/dmeventd
Last edited by alphaniner (2011-08-08 17:00:00)
But whether the Constitution really be one thing, or another, this much is certain - that it has either authorized such a government as we have had, or has been powerless to prevent it. In either case, it is unfit to exist.
-Lysander Spooner
Offline