jbd2/dm-2-8 using 100% IO

Fairly regularly I'd see the system monitor in my panel go up to 100% "IOWait" usage. The iotop tool was showing that jbd2/dm-2-8 was the culprit. (This is on CentOS 6.4).

After a lot of digging around, and also looking at CPU usage, I found that gvfsd-metadata looked suspicious. Killing it finally fixed the problem (!)

It is not easy to find a clear answer to what that daemon, and gvfs, actually do. They seem to be related to the train wreck that is called nautilus.

Since I'm not using nautilus anyway (also see this tip) I decided to ditch it altogether:

rpm -e gvfs --nodeps rpm -e gvfs-fuse gvfs-archive gvfs-smb gvfs-obexftp

Performance is back to normal :-)

Update

The problem did reoccur, unfortunately. By checking the disk with smartctl:

smartctl -a /dev/sda

and:

smartctl -t short /dev/sda
smartctl -t long /dev/sda

I found that most likely the disk was fine. I now replaced the SATA cable (!)(the one that runs from the motherboard to the disk), and all seems well now.

In other words, the problem sometimes doesn't lie with the software or the disk itself.