How to take a volume group offline that has a failed disk

[Gino Castoldi_2]

Hi,

Server: HPUX 11.11 MC/SG

We have a "test" volume group that has a failed disk. We want to take this volume group offline and replace the disk at a later date.
I tried to unmount the three filesystems but it hungs up on all three mount points. Is there a way to force an unmount?
When we run an ls -l command or lsof it hangs up. My guess is that when it tries to read that volume group it hungs because the disk has failed.

10 points to any answer.
Thank you
-Gino

[Pete Randall · Accepted solution]

Reboot the system, the VG will fail to activate and the file systems will fail to mount. To the best of my knowledge, there is no way to force an unmount (though there certainly ought to be).


Pete

[Pete Randall]

Gino,

Forgive the multiple replies, but further research has shown that 11i V3 has a -f option to force an unmount (not sure about V2) but that still leaves you (at V1) out of luck.


Pete

[chris huys_4]

Hi Gino,

The vxfs versions, 3.2 or 3.5, that
"came with" HP-UX 11.11, doesnt allow for forced unmounts in those circumstances.

So, as pete suggests, take the node, with the failed disk, out of the cluster, shutdown the node, pull out the failed disk, bring the node back in the cluster and insert the replacement disk at a later date.

NOTE:
From HP-UX 11.23 onwards and vxfs versions equal or greater then vxfs 4.1, the vxumount command would have been available, to do a unmount in those circumstances ; i.e. # vxumount -o force


Greetz,
Chris

[SoorajCleris]

Hi,

Just to value add

# uname -a

HP-UX sooraj B.11.31 U ia64 4118399276 unlimited-user license
[rx26-209]/
# vxumount -o force
vxfs vxumount: V-3-20147: Usage:
vxumount [-o [force]] {special | mount_point}
[rx26-209]/
# umount -f
usage: umount [-V] [-v] [-f] [-s] { special | directory }
umount [-F FStype] [-v] [-f] [-s] [-h host] -a


==========================

HP-UX rx76-1p0 B.11.23 U ia64 1087761808 unlimited-user license
[rx76-1p0]/
# umount -f
usage: umount [-V] [-v] [-f] [-s] { special | directory }
umount [-F FStype] [-v] [-f] [-s] [-h host] -a

[chris huys_4]

no points ;)

yakuza # swlist|grep -i vxfs
Base-VXFS 4.1.002 VERITAS File System Bundle 4.1 for HP-UX
EnableVxFS B.11.23.07 File-System library, commands enhancements for VxFS4.1 and 5.0
OnlineJFS B.11.23.0609 Online features of the VxFS File System

[root@yakuza]/
yakuza # which vxumount
no vxumount in /usr/sbin /usr/bin /usr/ccs/bin /usr/contrib/bin /usr/contrib/Q4/bin /opt/perl/bin /opt/ipf/bin /opt/hparray/bin /opt/nettladm/bin /opt/fcms/bin /opt/sas/bin /opt/wbem/bin /opt/wbem/sbin /usr/bin/X11 /opt/resmon/bin /opt/perf/bin /usr/contrib/kwdb/bin /opt/graphics/common/bin /opt/prm/bin /opt/sfm/bin /opt/hpsmh/bin /opt/upgrade/bin /opt/wlm/bin /opt/gvsd/bin /opt/sec_mgmt/bastille/bin /opt/drd/bin /opt/dsau/bin /opt/dsau/sbin /opt/firefox /opt/gnome/bin /opt/mozilla /opt/perl_32/bin /opt/perl_64/bin /opt/sec_mgmt/spc/bin /opt/ssh/bin /opt/swa/bin /opt/thunderbird /opt/gwlm/bin /usr/contrib/bin/X11 /opt/aCC/bin /opt/caliper/bin /opt/cadvise/bin /opt/ignite/bin /opt/sentinel/bin /opt/langtools/bin /usr/lib/vxvm/bin /opt/xpinfo /sbin /home/root

[root@yakuza]/
yakuza # cd /sbin/fs/vxfs

[root@yakuza]/sbin/fs/vxfs
yakuza # ls
bcheckrc fsdb mount umount vxumount
extendfs metasave newfs vxenablef
fsck mkfs subtype vxfsconvert

[root@yakuza]/sbin/fs/vxfs
yakuza # ./vxumount -?
vxfs vxumount: Usage:
vxumount [-F vxfs] [generic_options] [-o [force]]
{special | mount_point}

[root@yakuza]/sbin/fs/vxfs
yakuza # uname -a
HP-UX yakuza B.11.23 U ia64 3122855462 unlimited-user license

[S. Ney]

Gino,

Few questions, is Service Guard aware of your test vg? (if so what version of SG)

Have you tried: (for example)
umount /dev/vgtest/lvol?
fuser -cu dev/vgtest/lvol
ps -fp(process ID from above) -p(process ID)(If not service guard)
fuser -ku dev/vgtest/lvol
umount /dev/vgtest/lvol
vgchange -a n /dev/vgtest