Working with MS Clusters

[KevHF]

I am looking into making my ARCserve 2K systems cluster aware and was wondering if anyone has actual achived this yet.
In essence I have several sets of MS clusters around the business, each one made up of two Compaq servers, external tape-drive, ARCserve 2k (SP2+bits) and backup jobs. Each node looks after it's own shared area of hard-disk for load waiting. However when one of the nodes fail (the shared disk area falls over to the other node gracefully) the backup fails (once the server re-boots) because it can no longer see the shared drive (system set up not to automatically fail back but wait for a manual fail-back in case of problems leading to the systems 'looping').
I have followed CA 'ARCserve Cluster Support' document and everything looks ok, but in my test lab if I fail a server during a backup job the other cluster doesn't automatically finish the job - and I do not see how it can.
Any ideas?
Many thanks
Kev

 

[webtoner]

My company currently run Arcserve2k on a win2k server with an exchange 2k cluster. (Active-Active) If failover occurs during the backup-job. It will have to be restarted. The job nolonger see the info store of the failed server. In addition, we currently have a issue with Arcserve 2k backing up any info store above 10 gig. Hve you seen this as well. Arcserve can't seem to fix it.

Clinton
ckinsler@webtonetech.com

 

[KevHF]

Clinton

Thanks for your reply.
If I read your response correctly then what is the difference in setting up ARcserve to run as a Cluster resource or running as standalone?
We do not use exchange so I am not sure what you mean by 'info store'. However we back servers up each one being 80Gb and some holding Access databases of very large sizes (5Gb plus).

Kev

 

[webtoner]

Sorry for the confusion. I assumed you were an exchange user. The Information store is what is you back up using exchange. In addition, When backing up exchange in a cluster environment, the exchange agent has to talk to the host server to complete. since the server has failed over the info store cannot be seen. I apologize for the confusion. Sorry I couldn't help.

Clinton

 

[Shawzer]

Back up using the virtual names and not the node names.

DR Shaw

 

[KevHF]

Thanks

Do you mean that in the source process I select through the network? Normally I just select the 'Current ARcserve Server' which obviosly isn't a virtual name. To backup via the network would increase the backup time window enormously.

Kev

 

[BillHicks]

Well, we tried this last year on a very similar setup. We spent a week and a half attempting to get this to work but had to give up. In the end we ditched 2k, and got a seperate server to back up the cluster using 6.61. The cluster we have is acyive/passive. We back up the C: drives of each of the clusters and the virtual node also. That way, if the cluster fails over, the backup does not bat an eyelid.