TSM TDP Exchange 2007 CCR cluster configuration?

[[email protected]]

Hi.
I've installed TDP 55 for exchange on one (active) of my 2 exchange2007 servers (CCR cluster).

I will install the other tomorrow.
My question is.
If the active node fails, how will The Exchange Scheduler move to passive node?
Because CCR cluster is not MSCS cluster where i can move services to the other node if one fails.

How do i configure this in corect way?

in "b_dpexc_userguide_win" its just says if you have an mscs or vcs cluster do this.....


BTW: I do legacy exchange backups,NOT VSS.

 

[koenneke]

TDP exchange in CCR 2007

Hello,

i'm have the same question. How to configure the failover between the servers.

greetz

koen

 

[moon-buddy]

As you know, CCR or Cluster Continuous Replication is a new feature in Exchange 2007.

TSM backup and TDP for Exchange should not be directly affected by this implementation as this is internal to Exchange.

What must be done to configure the TSM cluster backup is to set it up the way you do a normal cluster. The exception is that you will be using TDP for Exchange on the shared resource - the Mail Store drives - and configure the failover on this Shared resource via the MSCS Admin Manager.

 

[jnelson]

We are in the same boat....
Wondering if we need the clusteronly option on or off in dsm.opt

 

[moon-buddy]

We are in the same boat....
Wondering if we need the clusteronly option on or off in dsm.opt

The clusteronly option is not part of the tdp.opt file. The node's dsm.opt uses clusternode=no.

As I said, the TDP for Mail (Exchange) is configured to backup the shared Mail Store resource. The shared disk failover is taken cared of by the MSCS. The shared disks is not backed up the way a normal file type cluster is backed up.

TDP for Exchange is configured as shared resource (which follows failover rules) which does only one thing: backup the Exchange database that is on the Shared resource.

 

[tggilkey]

In MS Cluster Administrator, should the TSM Scheduler service resource be created in the "folder" that is the name of the virtual Exchange server? The IBM doc is not real clear. It says "Select the Exchange Server Cluster Group and rt click and create a new resource...

2003 MSCS 2007 Exchange starting with legacy backups first.

Thanks for any input!
Tom

 

[tsmlover]

moon buddy i guess there is no shared resource in CCR cluster. what did u mean by shared resource?

can i activate this task using schedule TDPExchange>tdpexc /excserv=mail

mail is the name of my cluster.. i see the redbooks but they didnt mention CCR cluster.

I asked my Exchange administrator he said there is no shared drive or resource in CCR cluster anymore.

 

[moon-buddy]

moon buddy i guess there is no shared resource in CCR cluster. what did u mean by shared resource?

can i activate this task using schedule TDPExchange>tdpexc /excserv=mail

mail is the name of my cluster.. i see the redbooks but they didnt mention CCR cluster.

I asked my Exchange administrator he said there is no shared drive or resource in CCR cluster anymore.

There is always a shared resource with any cluster - the drives that are seen by both nodes when a fail over happens.

 

[Mita201]

moon-buddy,
There is no shared resource in Exchange CCR, since it is not really a cluster.
It is something like DB2 HADR if you are from IBM world - a kind of log shipping.

Bjorn,

I think that there is no way to do it automatically - maybe with some scripting. There is no TSM utility that will do it for you - you will have to manage it yourself

 

[Rojer]

Mita201 i am agreed, i have the same issue i read TDP for MS Exchange server red book and browse thur the internet for automatic backup of Exchange CCR Cluster but they all talk about shared resource or share drive

since there is no shared resource and share drive in CCR cluster

so everytime i use this TDPExchange>tdpexc /excserv=clusterName for exchange backup manually

I guess it will be done using the script: does anybuddy have any idea to execute the above command using schedule

 

[bedac]

Take a look at this: http://www-01.ibm.com/support/docvi...cr&uid=swg21329825&loc=en_US&cs=utf-8&lang=en

- Bent

 

[tsmlover]

Dear Freinds,

I have studied and done lot of research to get the desired result, I spent 6 goddamn months..

now this is to inform you that .....

you can take backup only ur active node yes ! if the failover happen ur schedule will miss since there is no support of CCR Cluster in TSM .... yes!

you have to keep an eye on your Exchange failovers 24/7 ... in TSM all you have find recalmation , expiration, migration, fumigation, blah blah ... since this is only storage management software, machine controlling system ...

i am going for Backup Exec.. that provide wonderfull reporting, and all advance features that understand all new technologies....

 

[antwoneeter]

Good luck with Backup Exec, tsmlover...

For the original question of scheduling a legacy backup to run from whatever node is active, why not try a pre or postchedulecmd in the dsm.opt on the clustered disk to execute a script that contains your Exchange backup command? This of course assumes you have clustered services setup for the clustered disk resources. But in all the CCR clusters I have seen, non had any clustered disks except for the file share witness. In any case, in a CCR why wouldn't you want to use VSS to backup from the replica node? I could see if you ran into VSS problems, and I can relate, but this is the least intrusive backup method to an Exchange environment.

 

[Mita201]

I am not sure if you have seen link that Bedac has posted. I have never tried this, but as far as I understand, it is exactly what tsmlover wants: Backup from standby node in Exchange CCR cluster, and script running at both nodes, that will examine if the node is active or standby, and will run scheduled backup on standby node only. Has anyone tried this?

 

[moon-buddy]

moon-buddy,
There is no shared resource in Exchange CCR, since it is not really a cluster.
It is something like DB2 HADR if you are from IBM world - a kind of log shipping.

Bjorn,

I think that there is no way to do it automatically - maybe with some scripting. There is no TSM utility that will do it for you - you will have to manage it yourself

Don't automatically make the assumption that shared resource always equals shared disks drives. The nodes that handle the disk drives from a commonality viewpoint "shares" the resource. Why is this so? This is because you want a fail safe environment just in case one node fails with nothing handling the Exchange load.

To backup an Exchange environment, you really need to create a shared "resource" from a service viewpoint - that is, if node 1 fails, node 2 can take over.

 

[khawts]

I am in the same boat. We are using Exchange 2007 and Windows 2008. Both have their own clustering. I am testing the legacy backup before the VSS backup. If I cannot get the clustering to work on the legacy with CCR, it is not point to test the VSS.

So my question is, Exchange 2007 DO NOT HAVE shared drive neither to the Windows 2008 clustering.

So far, I have created the services and sheduler for TDP on my two nodes exchange but now sure what I need to do in creating the generic services for the cluster, since I have to TDP dsm.opt files on the C:\Program Files\tivoli\tsm\baclient.
There is no share drive to put the dsm.opt file.

If anyone has setup Exchange 2007 with Windows 2008, can you please provide the instruction on how configure TDP for that?

Your assistance is greatly appreciated.

 

[Mita201]

Don't automatically make the assumption that shared resource always equals shared disks drives. The nodes that handle the disk drives from a commonality viewpoint "shares" the resource. Why is this so? This is because you want a fail safe environment just in case one node fails with nothing handling the Exchange load.

To backup an Exchange environment, you really need to create a shared "resource" from a service viewpoint - that is, if node 1 fails, node 2 can take over.

moon-buddy,
I am 100% with you about the explanation of shared resource. But, if you read your own posts in this topic, you may see that you dind't talk about "shared resource" of any kind, but explicitly about shared disk(s) that contains mailbox store.

Khawts, please check bedac's post from this thread, and give it a try. I am really interested if this works, but I do not have enough resources to test it myself.

 

[khawts]

Mita:

So if not n wrong, I don't have to set up the generic services anymore. But I still need to put the script on both active and passive servers, correct?

 

[Mita201]

Yes, I think it is the way. You then associate both nodes with one schedule.

 

[khawts]

But the script is for the backup using VSS. I am testing with legacy backup first. Can I do that?
Also, do I have to have two tdp dsm.opt file on both C drive.