2007-03-01 14:26:00
Today I faced the task of replacing a failing hard drive in one of our Tru64 boxen. The disk was part of a disk group being used to serve plain data (as opposed to being part of the boot mirror / rootdg), so the replacement should be rather simple.
After some poking about I came to the following procedure. Those in the know will recognize that it's very similar to how Veritas Volume Manager (VXVM) handles things. This is because Tru64 LSM is based on VXVM v2.
* voldiskadm -> option 4 -> list -> select the failing disk, this'll be used as $vmdisk below.
* voldisk list -> select the failing disk, this'll be used as $disk below.
* voldisk rm $disk
* Now replace the hard drive.
* hwmgr -show scsi -> take a note of your current set of disks.
* hwmgr -scan scsi
* hwmgr -show scsi -> the replaced disk should show up as a new disk at the bottom of the list. This'll be used as $newdisk below.
* dsfmgr -e $newdisk $disk
* disklabel -rw $disk
* voldisk list -> $disk should be labeled as "unknown" again.
* voldiskadm -> option 5 -> $vmdisk -> $disk -> y -> y -> your VM disk should now be replaced.
* volrecover -g $diskgroup -sb
The remirroring process will now start for all broken mirrors. Unfortunately there is no way of tracking the actual process. You can check whether the mirroring's still running with "volprint -ht -g $diskgroup | grep RECOV", but that's about it.
kilala.nl tags: unix, work, other tech, sysadmin,
View or add comments (curr. 2)
Posted by Thomas
Whoops! I meant to say "progress", not "process". Meaning, there is no equivalent to the "vxtask list" command, since that was introduced in VXVM v3.
All content, with exception of "borrowed" blogpost images, or unless otherwise indicated, is copyright of Tess Sluijter. The character Kilala the cat-demon is copyright of Rumiko Takahashi and used here without permission.
2007-03-01 18:21:00
Posted by DrFaulken (website)
Looks like we've both been dealing with <a href="http://journal.drfaulken.com/?p=516">naughty disks lately</a>. When you mention you can't track the process, do you mean the thread itself, or the progress of the rebuild?
Good luck -- I could compress a coal into diamond with the all the tension I've had since my parity drive died. :)