Filesystem errors, how do I tell which volume its on from log errors?

anon091 · 03-09-2011, 08:56 AM

Looks like i need to do an e2fsck on some of my volumes. but how do I know which one md(9,0) is from the error below?

kernel: EXT2-fs error (device md(9,0)): ext2_check_page: bad entry in directory #659
61985: unaligned directory entry - offset=0, inode=3217201536, rec_len=14142, name_len=118

ButterflyMelissa · 03-09-2011, 11:36 AM

Hi,

Enter init 1 (single user if I'm not mistaken) unmount your volume anf run fsck without params, it should revert to the entries in inittab

Quote:

If no filesystems are specified on the command line, and the -A option is not specified, fsck will default to checking filesystems in /etc/fstab serial. This is equivalent to the -As options.

(Extract from this page http://www.linuxmanpages.com/man8/fsck.8.php )

Of course, looking into the /dev folder can help you find the devices in question as well...
This should happen when a fils sys check is forced (on my box it's done every so mounts) so a /dev/sdasomething should have been on the screen...

Good luck!

Hope (sincerely) to have helped a bit, if not, do ignore my post, thanks. ;-)

Thor

anon091 · 03-09-2011, 11:42 AM

Thanks. I do know how to run the fsck, but i'm just not sure how to tell which volume to run it on based on that error message.

ButterflyMelissa · 03-09-2011, 12:06 PM

oh, sorry about this...
Please allow me to read-in, as this is a way for me to learn.
Thanks

Thor

wpeckham · 03-09-2011, 12:46 PM

As long as you have an error detected, and you are shutting down for an fsck run anyway, why not run it against ALL of them.

In the newer versions, it will be VERY quick if the volume has no errors logged.

anon091 · 03-09-2011, 02:20 PM

Right, I planned on running it against all of them, but if i can minimize the downtime by taking a volume or two down before shutting the whole thing down and get those checked, that could save me hours, especiall if e2fsck needs to make multiple passes to fix it.

Multiple passes fixing errors vs. one pass because its already clean could save me a few hours of downtime when I do take the server down. I'm dealing with extremely large volumes here.