cancel
Showing results for 
Search instead for 
Did you mean: 

Clustered master upgrade from 7.5.0.6 to 7.6.02

Hello all,

 

I have a clustered master server (clustered using VCS global clustering option) in active-passive mode. The catalog(size is roughly 500 GB) is replicated to the passive node using SRDF/A.

OS is Windows Server 2008 R2 Enterprise. The whole environment runs on tape infrastructure only.

I am deciding on an approach to run the upgrade and hence need experts views on that.

Approach 1- Break the SRDF/A replication of catalog. Upgrade active node. Bring active down after upgrade. Make the 2nd node active. Upgrade passive node.

Approach 2- Do not break the replication. Take a clone of the catalog LUN before upgrade. Let the catalog replicate as and when the upgrade is in progress. After upgrade at active node, bring it down. Bring up the passive node. As upgraded database will already have replicated, upgrade here require only upgrade of binaries.

The full backup of the catalog takes about 6 hours to complete.

Can you suggest me the best approach among these? Or any other if you have a better approach. Also, how can I estimate time that will be taken for my environment?

 

-Regards,

Neetu

1 Solution

Accepted Solutions
Accepted Solution!

Hello   You should have

Hello

 

You should have SwapRoles = 0 + SplitTakeOver = 1 set. You can then do the remote failover, the cluster will make the devices in DR RW. During the R1 will remain in PROD and R2 will remain in DR. You can then perform the upgrade. Once its complete you'll need to offline the SG in DR, then re-establish the SRDF replication, then online in PROD.

View solution in original post

10 Replies

Looks like the installer has

Looks like the installer has a special process built-in to help you. Always perform backups before the upgrade. Cloning the drive is great way to protect yourself. For an environment this large I expect a very long downtime.

Follow the upgrade guide:

"To upgrade clustered Windows systems, log on to the active node (the
node with the shared disk)."

"For a clustered upgrade, the only option is Install a
clustered master server."

Guide is found here:

http://www.symantec.com/business/support/index?page=content&id=DOC6449

There are many steps and warnings. It's a good read.

 

 

Hi Neetu,   Your best option

Hi Neetu,

 

Your best option is to take a clone of the lun for backup / rollback. Once you upgrade the master in primary site (here it will update the EMM), you can then upgade the master in the secondary site.

You do not need to make the secondary active, just run the install there. It will see that its part of a cluster, is not online at the moment, and will simply do the installation of the binaries (and as you say, the catalog information would have been replicated so no change needed there). Afterwards you can online it there, and test.

Thank you for the inputs. But

Thank you for the inputs. But as http://www.symantec.com/business/support/index?page=content&id=HOWTO73064 (this link is provided in the 7.6 upgrade guide, also it refers about upgrade to 7.5 only, not sure if it is applicable to 7.6) below are the steps:


To upgrade a Windows primary master server cluster

1.Ensure that global cluster fail over is disabled for the duration of the upgrade.

2.Suspend replication between the primary and secondary sites

3.Upgrade NetBackup on the primary site master server cluster first, following the standard upgrade procedures (refer to the NetBackup 7.5 Upgrade Guide).

4.Ensure that the phase 1 and phase 2 image metadata migration processes have completed successfully before proceeding to upgrade the secondary cluster.

5.To update the secondary cluster, isolate the secondary master server cluster from the rest of the network.

6.Present the upgrade volume to the appropriate mount point. If no upgrade volume is available, see the note following this procedure.

7.Upgrade NetBackup using the standard upgrade procedures in the NetBackup installation guides.

8.Stop NetBackup.

9.Dismount the upgrade volume.

10.Reconnect the secondary master server cluster to the rest of the network.

11.Restart the replication between the primary and secondary sites and re-enable the global cluster fail over. Be sure to take extreme care when re-establishing the replication link to ensure that the synchronization occurs in the correct direction (from the primary site to the secondary site).


I am really confused about the approach that I should take now. I have a call open with Symantec and they say my upgrade might take at the most 2 hours (hard for me to believe!) and they asked me to follow above method.

Please help me.

Regards,

Neetu

Follow instructions given to

Follow instructions given to you by Symantec support. This is a public forum. Even information recieved from Employees (like me) in this forum are not official. The only official advice comes directly from your given support channel. Following advice given here during an open support case could seriously complicate your case.

Hi all, I went ahead with the

Hi all,

I went ahead with the steps as per Symantec suggested and have ended up in a difficult situation. The primary node upgrade to 7.6.0.1 from 7.5.0.6 was successful (had problem during installation due to WMI service though).

As I have mentioned this is a clustered environment using VCS GCO and SRDF replication, we had problems proceeding with the secondary node.

I had suspended replication before the start of upgrade which leaves the target volume in write disabled mode and hence unable to bring SRDF resource up in VCS to make it available for the upgrade procees (as upgrade needs IP and shared disk to be online).

Please advise. Has anyone come across such an environment and has upgraded successfully?

If I turn on the replication now and allow the target volume to have the updated database from source which is already at 7.6.0.1, will I be able to upgrade the secondary node NetBackup version to 7.6.0.1 successfully? Or will that fail?

 

Regards,

Neetu

About these steps: 5.To

About these steps:

5.To update the secondary cluster, isolate the secondary master server cluster from the rest of the network.

6.Present the upgrade volume to the appropriate mount point. If no upgrade volume is available, see the note following this procedure.

You will see that the steps are a copy&paste from: http://www.symantec.com/docs/HOWTO73064 

See the 'note' section:

Note: If no upgrade volume is available, consider using the following approach during the upgrade:

  • Before upgrading the primary cluster, stop the replication between the primary and secondary clusters so that the target volume can be mounted to the secondary cluster (refer to your replication documentation).

  • At step 5 of the upgrade procedure, present the replication target volume to the secondary cluster as the upgrade volume. 

 

 

Hello Marianne, I have

Hello Marianne,

I have referred to the note during my upgrade. If I stop the replication (currently only suspended), will I be able to bring up the SRDF resource in VCS?

Regards,

Neetu

The SRDF resource in VCS

The SRDF resource in VCS should probably be offline and Service Group(s) Frozen.
The TN says to stop the replication. Not suspend.

You will have to mount replicated volumes manually on both sides.

As per the TN:

  • Before upgrading the primary cluster, stop the replication between the primary and secondary clusters so that the target volume can be mounted to the secondary cluster (refer to your replication documentation).

Hi Marianne, All resources

Hi Marianne,

All resources except IP and shared disk resource need to be offline during the upgrade. And for shared disk(MountV) resource to be online, we need SRDF to be online.

Hence my query is if I stop the replication, will the SRDF resource come online- I am yet to get a confirmation for this? I have not gone ahead as I do not want to bring the primary node down when I already have one node down.

 

Regards,

Neetu

 

 

Accepted Solution!

Hello   You should have

Hello

 

You should have SwapRoles = 0 + SplitTakeOver = 1 set. You can then do the remote failover, the cluster will make the devices in DR RW. During the R1 will remain in PROD and R2 will remain in DR. You can then perform the upgrade. Once its complete you'll need to offline the SG in DR, then re-establish the SRDF replication, then online in PROD.

View solution in original post