SUMMARY: Migrate single node cluster from EMA storage to EVA storage without OS reinstall
From: Garsha, Adam (adam.garsha_at_marquette.edu)
Date: 07/07/05
- Previous message: Maglinger, Paul: "Looking for method to scan for bad blocks."
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Date: Thu, 07 Jul 2005 09:02:39 -0500 To: tru64-unix-managers@ornl.gov
Migrate OS disk for single node cluster from EMA to EVA storage, when
both storage arrays will be visible simultaneously.
########################################################################
###
0. Present necessary storage from EVA and adjust zones as appropriate
(so box can see EVA) and back everything up.
hwmgr -scan scsi
dsfmgr -v
# if output isn't good, then "dsfmgr -vVF" a couple times.
########################################################################
###
1. Use addvol/rmvol to update cluster_usr, cluster_var, cluster_root
domains with disk partitions from the EVA.
########################################################################
###
2. Delete, then Re-add quorum disk (where dskXX is an EVA based disk).
Currently, we have a single node and the
quorum disk isn't counting a vote, so we can do this with the system up
and online without any impact.
clu_quorum -f -d remove
clu_quorum -f -d add dskXX 0
########################################################################
###
3. Build copy of boot disk onto EVA disk
clu_bdmgr -c dskYY 2 # 2 is a non-existant cluster member
# Mount member2's root domain (now on dskYY) so you can edit member2's
/etc/sysconfigtab and restore the boot partitions:
mount root2_domain#root /mnt
# Restore the boot partition:
vdump -0 -f - /cluster/members/member1/boot_partition | (cd /mnt;
vrestore -x -f -)
# Adjust disk entries and swap disk entry in sysconfigtab
vi /mnt/etc/sysconfigtab # update sysconfigtab with information about
boot disk
...
vm:
swapdevice=/dev/disk/dskYYb
clubase:
cluster_seqdisk_major=19 # Get correct number from file
/dev/disk/dskYYh
cluster_seqdisk_minor=175 # Get correct number from file
/dev/disk/dskYYh
...
# Restore the h partition CNX information:
/usr/sbin/clu_bdmgr -h dskYY
The h partition information is copied from the cluster member where you
run the clu_bdmgr command to the h partition on dskYY.
# Unmount the new boot domain:
umount root2_domain#root /mnt
# Edit disklabel and change "label: clu_member2" to "label:
clu_member1":
disklabel -r -e dskYY
# Adjust /etc/fdmns links:
cd /etc/fdmns
mv root1_domain root1_domain_original
mv root2_domain root1_domain
# Use the consvar -s bootdef_dev disk_name command on member1 to set the
bootdef_dev variable to the new disk.
consvar -s auto_action HALT
consvar -l
consvar -a
# Shutdown
shutdown -h now
########################################################################
###
4. Adjust SRM's view of the world.
wwidmgr -show wwid # note that the old boot disk OS_Identifier (UDID)
was 20.
wwidmgr -clear all
init
wwidmgr -quickset -udid YY_UNIQUE_ID
wwidmgr -show wwid
init
show device
set bootdef_dev dga20,dgb20 (or whatever)
########################################################################
###
5. Hold onto your hat
boot
########################################################################
###
6. reset consvar auto_action
consvar -s auto_action BOOT
consvar -a
consvar -l
- Previous message: Maglinger, Paul: "Looking for method to scan for bad blocks."
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Relevant Pages
|
|
