Bogus Physical Volumes

[Bart.Vermeiren]

Hi,

We're running an HPUX 11.00 (yes, we still have one due to openspool :( ) and I'm seeing some 'ghost disks'. When I read the syslog.log, a lot of messages like the next are spawning.

Nov 9 13:56:16 %system% EMS [12311]: ------ EMS Event Notification ------ Value: "CRITICAL (5)" for Resource: "/storage/events/disks/default/0_4_0_0.80.7.0.0.0.1" (Threshold: >= " 3") Execute the following command to obtain event details: /opt/resmon/bin/resdata -R 806813824 -r /storage/events/disks/default/0_4_0_0.80.7.0.0.0.1 -n 806813771 -a

When I execute the command, it always refers to a disk we had connected the server to on SAN. Today this connection is completely gone (no zoning anymore to the SAN).

When I use ioscan, I can still see the disks comming up. When I use pvdisplay on such a disk, I get this output:

# pvdisplay /dev/dsk/c48t0d0
pvdisplay: Couldn't find the volume group to which
physical volume "/dev/dsk/c48t0d0" belongs.
pvdisplay: Cannot display physical volume "/dev/dsk/c48t0d0".

A pvremove doesn't change a thing:
# pvremove /dev/rdsk/c48t0d0
Warning: Cannot determine block size of Physical Volume "/dev/rdsk/c48t0d0".
Assuming a default value of 1024 bytes. Continuing.
pvremove: Physical Volume "/dev/rdsk/c48t0d0" contains no LVM information

Normally we do a procedure to get luns visable we made on SAN:
ioscan
ioscan
insf -e
powermt config
powermt check

With powermt nothing is visible however.

Can I somehow remove these disks?

[Duncan Edmonstone · Accepted solution]

presumably these disks still show up as NO_HW in an ioscan?

A reboot will stop the messages (if you care enough to do one)

HTH

Duncan

[Bart.Vermeiren]

Very true Duncan and it is indeed a solution. But the problem is that we can't reboot the server soon. Is there perhaps another way?

[Duncan Edmonstone]

You could try adding the disks to

/var/stm/data/tools/monitor/disabled_instances

(IIRC the file is self documenting)

That may stop the EMS errors being reported, I'm aware of no way to get rid of the NO_HW entries on ioscan without reboot though...

HTH

Duncan

[Bart.Vermeiren]

I've done what you've asked and will now wait a few days to see if things get better.

[Wessam Aly]

You can also remove the ioconfig file (which contains a mapping of all your hardware addresses and special files) and reboot.

the ioconfig file resides in /etc/ioconfig and /stand/ioconfig.
you can rename the files and reboot. the system will stop while trying to boot with a prompt saying ioconfig file not found.
A that point you can run ioinit -c to recreate a new ioconfig file without all the stale entries from the previous disks.
ofcourse this is a bi risky since some hardware mapping may change. If yo have a lot of SAN disks connected at some point in time, it would be beneficial to do it, otherwise i wont recommend it.

[kevin_m]

You can run rmsf -H using the H/W paths for the NO_HW devices. It will remove the special files as well but Iâ m assuming you wonâ t be reconnecting those disks in the future. I used this cleanup technique on a 11.23 server after moving some LUNs and consequently stopped receiving EMS alerts.

- Kevin

[Bart.Vermeiren]

I haven't got any reports of bogus disks after I add them to that file - ok for me