System hangs on rm after upgrade to 10.1 with UFS

matheyden at gmail.com matheyden at gmail.com
Wed Apr 1 14:24:25 UTC 2015


Hi all,

After upgrading our off-site backup server from FreeBSD 10.0 to 10.1, we 
found the system would be unresponsive each morning and require a hard 
reboot. Assuming hardware failure, a new physical host was set up running 
10.1 but again, after a few days the same issue started occurring. After 
digging around a bit, I found the system was getting stuck while trying to 
remove the oldest snapshot folder from rsnapshots. All new processes seem 
to be stuck waiting on disk. If I leave it long enough, I'm no longer able 
to even ssh in to the system.

I've since set up a local virtual host and have been able to duplicate the 
problems on it as well. The virtual machine is running rsnapshot to back 
up the same data (in this case, the vmail folder from our mail server). 
While running FreeBSD 10.0, I had no issues. It was able to delete the 
oldest snapshots and everything worked as it should. After upgrading to 
FreeBSD 10.1, without making any additional changes, the system hangs when 
running rm to remove the oldest snapshot folder. 

Total backup size is around 150GB. Filesystem is single HDD with UFS and 
no mirrors/raid.
Manually running "rm -rf <path_to_oldest_snapshot" causes the system to 
hang too. Only way to correct the problem I've found is to force a hard 
reset.
I've tested it with a fresh 10.1 install and ran into the same issue.

FreeBSD 10.0:
http://pastebin.com/F1QPRW8u

FreeBSD 10.1:
http://pastebin.com/EfkXGhv6

PS, 2 hours after running rsnapshot on 10.1, just before losing access:
http://pastebin.com/yKAFu6Lw

Anyone have any idea what may be causing the problem? Is this a bug in 
FreeBSD 10.1?

Thanks.



More information about the freebsd-questions mailing list