Raja a "plain and simple" MC/SG cluster with shared storage can be easily migrated to a new storage with OR
The main "gotcha" is that hpux stores LVM configuration in /etc/lvmtab file that you can't edit by hand. Thus you have to import volumegroups after migration by hand (since hpux won't ever recognise new disks with different hardware path as valid disks for your existing VGs).
1) create OR -hot -push -differential sessions from source storage to target. 2) activate sessions 3) wait 'till all devices are in Copied state 4) recreate OR sessions (that's why you used -differential and -push at step 1) 5) export VG definition using -s option (preserving VGID in export file) 6) shut down the whole cluster (you can keep hosts up and running but cluster needs to be down) 7) activate OR sessions (thus copy remaining tracks) 8) while OR copies tracks from SRC to TGT, change zoning and masking 9) discover new devices from host point of view (ioscan / insf) 10) scan drives for volumegroups on new storage (using VGID / -s option). 11) find new CLUSTER_LOCK device and change config file 12) rebuild MC/SG cluster with new cluster lock 13) bring everything up again
Thank you very much and I have one more question I am planning to do migration from clarrion to DMX for solarsis server using SVM using open replicator hot pull.My question is in HP-UX or AIX we will export and import volume groups but for solarsis what we need to do to export and import the metainit? Is there are any other extra care do we need to take in this migration?
xe2sdc
4 Operator
•
2.8K Posts
0
August 5th, 2009 06:00
The main "gotcha" is that hpux stores LVM configuration in /etc/lvmtab file that you can't edit by hand. Thus you have to import volumegroups after migration by hand (since hpux won't ever recognise new disks with different hardware path as valid disks for your existing VGs).
1) create OR -hot -push -differential sessions from source storage to target.
2) activate sessions
3) wait 'till all devices are in Copied state
4) recreate OR sessions (that's why you used -differential and -push at step 1)
5) export VG definition using -s option (preserving VGID in export file)
6) shut down the whole cluster (you can keep hosts up and running but cluster needs to be down)
7) activate OR sessions (thus copy remaining tracks)
8) while OR copies tracks from SRC to TGT, change zoning and masking
9) discover new devices from host point of view (ioscan / insf)
10) scan drives for volumegroups on new storage (using VGID / -s option).
11) find new CLUSTER_LOCK device and change config file
12) rebuild MC/SG cluster with new cluster lock
13) bring everything up again
I'm pretty sure there are a lot of other options.
raja4
45 Posts
0
August 6th, 2009 11:00