Removing EMC LUN from Solaris 10 host without a reboot

I have seen a few threads / posts on the forum's but nothing specific to what Iam after. So Iam hoping you can help.

Issue : Iam hoping to remove unused emc luns from my solaris servers , T2000 (5.10 Generic_142900-02). I have a list of the process I have followed for the removal of the luns. As it stands it looks like a reboot is required to remove the luns, but Iam hoping it just something I have missed. Please help if you can. I have tired the process below in different stages thinking that I have messed the sequence, but I was thinking that the removal should be as following??????

vxvm level,
dmp level
powerpath - Not applicable here
os - devfsadm level

Process

VXVM Level,

  1. Remove the disk from veritas control using 'vxdiskunsetup <device> and vxdisk rm <device>. At this stage I have also tried to clean up the devices from /dev/vx/dmp and /dev/vx/rdmp.
# vxdg -g testdg rmdisk emc0_ABC
# /etc/vx/bin/vxdiskunsetup emc0_ABC
# vxdisk rm emc0_ABC
  1. Remove the disk from storage side.
    Once the disk is removed from storage you will see the disk as <drive not available > in format. command inq does not see the disk either.

Then devfsadm -Cv - to scan for hardware changes

3) Check the dmp info, the devices should be disabled, which they are

# vxdmpadm listctlr all
# vxdmpadm getsubpaths ctlr=c2
# vxdmpadm getsubpaths ctlr=c3
c3t50060482D531F7C2d9s2 DISABLED - emc0_ABC EMC emc0 -

This is where we have issues,

4) cfgadm - At this point you should be able to remove the disk with " cfgadm" if the disk is in unusable state. If its in 'failing' state use 'luxadm' to offline the disk. You cannot unconfigure a disk unless it is in 'unusable' state and you have to unconfigure both the paths of your disk. But this is what Iam seeing,

c2::50060482d531f7ed,9 disk connected configured unusable
c3::50060482d531f7c2,9 disk connected configured unusable
# cfgadm -al -o show_SCSI_LUN |grep c3t50060482D531F7C2d9s2
# luxadm -e offline /dev/dsk/c3t50060482D531F7C2d9s2
# cfgadm -c unconfigure -o unusable_SCSI_LUN c2::50060482d531f7ed,9
 
# cfgadm -c unconfigure -o unusable_SCSI_LUN c2::50060482d531f7ed,9 cfgadm: Attachment point not found

Even a re-read of the sd.conf file doesn�t work

update_drv -vf sd

Even tried,

# cat /dev/null > /etc/vx/array.info
# cat /dev/null > /etc/vx/disk.info

We have tried removing the disk info from

5) Remove the references to the

# cd /dev/vx/*dmp/
# ls -la
# rm emc0_ABC* (from both rdmp and dmp)

:wall:

Try these instead one for c2 and then c3 :

# cfgadm -c unconfigure -o unusable_SCSI_LUN c2::50060482d531f7ed
 
# cfgadm -c unconfigure -o unusable_SCSI_LUN c3::50060482d531f7c2

Thank you for the reply. I have a question?

How would the unconfigure affect that PATH as there are other disks using that path. I dont want to cause any impact to the other disks currently in use?

Thats has worked!! But iam not sure how this affects the other Luns on the PATH. Doesnt look to have done anything to them. One of the reasons this didnt work before was that the disk were showing up as "failing" and not unusable.

# cfgadm -c unconfigure -o unusable_SCSI_LUN c2::50060482d531f7ed

# cfgadm -c unconfigure -o unusable_SCSI_LUN c3::50060482d531f7c2

BUT!!, there is always a but. I cannot remove the disk from vxdmpadm, its still showing up as disabled. Any help would be appreciated

c2t50060482D531F7EDd12s2 DISABLED - emc0_ABC EMC emc0 -