Monday, May 13, 2013

Open Replicator Hot Pull Migration Simplified!


Microsoft Cluster Service (MSCS) with two Nodes:

*have powerpath latest version installed.

Step:1 Create a table similar to the one shown above and fill that information

Remote (source):
Device Information: inq –sym_wwn
FA’s Mapped: symdev list –dev 1A0 –sid 1000 –multiport
FA’ WWN: symcfg list –FA all –sid 1000

Contol (target):
1. Create a new device of same or greater size; map it to the FA’s.
2. Mask the newly create devices to the Control FA’s, this comes after zoning

Step:2 Zoning

For hotpull, all ports on control array that the target devices are mapped to must have access to the remote devices

In this case, zone as below:

Step:3 Mask the newly created device on target array to the source FA’s

Symmask –sid 2000 –wwn 5009035BAAA9BFE0 –dir 2A –p 0 add dev 1B0
Symmask –sid 2000 –wwn 5009035BAAA9BFE1 –dir 15A –p 1 add dev 1B0
Symmask –sid 2000 refresh

Step:4 Caution! Zone the host (windows cluster MSCS) to the control (target) symmetrix. DO NOT MASK new storage to the host. You will do this in step 9!

Which means the host still sees the old storage, we will let the host see new storage after the cutover process.

Step:5 Verify control and remote FA’s have visibility

Symsan list –sid 2000 –sanports –dir 2A –p0
Symsan list –sid 2000 –sanports –dir 15A –p1

Step:6 Verify if WWN’s of the remote devices are visible on the target FA’s

Symsan list –sid 2000 –sanluns –wwn 5009035BAAA9BFE0 –dir 2A –p 0
Symsan list –sid 2000 –sanluns –wwn 5009035BAAA9BFE1 –dir 15A –p 1

These commands should output the LUN WWN of the source device

Note: The device records might show incomplete due to SCSI reservations placed upon the devices by the cluster software

Step:7 Prepare the host for migration

1. Prior to begin the migration, the applications running on the cluster should be shutdown as the control (target) symm has no control on source IO activity, which might lead to data inconsistencies.
2. Work with windows admin to make sure the cluster resources are taken offline with at-most caution to move all resources to one node. In this instance I consider win_node2
3. Shut down the servers.

Step:8 Unassign the remote (source) devices from all cluster node HBA’s

Symmask –sid 1000 –wwn 100000000c9535011 –dir 1A –p 0 remove dev 1A0
Symmask –sid 1000 –wwn 100000000c9535012 –dir 16A –p 1 remove dev 1A0
Symmask –sid 1000 refresh

Note: This will prevent duplicate disks being visible to the cluster nodes when they are brought back online

Step:9 Mask cluster node HBA’s to the control (target) devices

Symmask –sid 2000 –wwn 100000000c9535011 –dir 2A –p 0 add dev 1A0
Symmask –sid 2000 –wwn 100000000c9535012 –dir 15A –p 1 add dev 1A0
Symmask –sid 2000 refresh

Step:10 Create device file that contains target (control) devices and remote (source) devices

hippo@unix ~]$ vi pair.txt

symdev=000180202000:1B0 60060160058714000090DE98F7DBDC11

Step:11 Create open replicator copy session

hippo@unix ~]$ symrcopy –f pair.txt create –copy –hut –pull –donor_update

-copy: Starts data transfer when the session is activated
-hot: Target (control) devices are online to the host during data transfer
-pull: Data being copied from remote array to the control array
-donor_update: All writes to the control device from the host will also be copied to the remote device. This protects against potential data loss that can be caused by failure of OR session during migration process.

Step:12 Query the session that all pairs are successfully created

hippo@unix ~]$ symrcopy –f pair.txt query

Check if status column says ‘created’

Step:13 Activate the Open Replicator session

hippo@unix ~]$ symrcopy –f pair.txt activate
hippo@unix ~]$ symrcopy –f pair.txt activate

Check if status column says ‘CopyInProg’

Step:14 Restart the cluster

1. Work with server team to power on the cluster nose that owned all the resources prior to the cluster being shutdown.
2. Confirm if the node sees all the devices, than restart the cluster service on the node.
3. Bring the cluster resources back online
4. Power on the other node in the cluster as well

Step:15 Monitor the open replicator session to ensure all the data is transferred successfully.

hippo@unix ~]$ symrcopy –f pair.txt query

Check if status column says ‘Copied’

Step:16 Terminate the open replicator session once the query shows data 100% copied for all the devices.

hippo@unix ~]$ symrcopy –f pair.txt terminate
hippo@unix ~]$ symrcopy –f pair.txt query


Post a Comment
Read more: