I/O error when mounting advfs domain; REWARD for fixing the problem!

hnispel · *nix forums beginner Joined: 11 Apr 2006 Posts: 4

Hi,

the current problem occurred when shutting the ES45 down, there was
disk activity on all disks belonging to the domain, even after 40
minutes, so I did not shut down the MSA1000, but booted the ES45, then
one of the advfs domain did not mount...

Two weeks ago I posted a similar problem, now the same disaster
happened on another ES45. Both use Tru64 V5.1B (Rev. 2650), Firmware
revision: 7.1-2
Both use MSA1000 controller, Firmware version: 4.48 build 342
All these RAID5 Volumes are seen with status "VOLUME OK" from the
MSA1000, to the OS they should be mounted as one domain with one
fileset.
Differences:
One machine has it's unmountable domain in an extra shelf (Raid5
14disks 146GB 15K) and has two other (smaller) Volumes attached that
mount correctly.
The newer machine has the unmountable domain in the 'internal' shelf
(Raid5 7disks 300GB 10K) and has one other Volume attached, that mounts
correctly.
MSA1000 "show connections" the profile is "Default" (from post on
11Apr) on the other one correctly "Tru64".

fixfdmn and mount -dirty don't work.

salvage is only working with the -S option, but it's too slow (well
it's ~1,5TB) and can not restore everything from the last experience.

Suspicious is the recent patch of the OS and that both partitions are
pretty big in size and the amount of files (70 billion files)...
HP is investigating.
If you encountered a similar problem, please let me know.
You will get a reward if you fix the problem!

Thanks
Harald

Google