ASNODENAME vs. using the same NODENAME on different servers

[pjackson1]

We have two servers which share a clustered filesystem. The servers themselves are not in a clustered configuration--they just share a common directory. Currently both servers have a unique nodename: server1 and server2. The problem is they're both backing up our /shared directory so we're getting two copies. The solution which I went with (and what I think is probably best practice) was to create a new TARGET NODE on TSM (cluster1) with authority to two agent nodes cluster1-server1 and cluster1-server2 which are defined as asnodenames on each servers dsm.sys file. I then modified the dsm.sys files so that /shared is excluded on server1 and server2 node stanzas and is in a domain statement for cluster1-server1 and cluster1-server2 node stanzas. Through this setup, everything seems to be working fine.

My question is this: wouldn't it be easier just to create a new node called "cluster1" which is definied on both servers for just the /shared directory and still keeping each server's unique nodename as well. What kinds of problems could I encouter with this? Put another way, what's the point of ASNODENAME and agent/target nodes when you could just create an additional node which both servers use just for the cluster.

 

[GregE]

I have several servers clustered like that, and I don't use asnodename. To me it's much simpler to do the additional nodename to backup the shared disks.

 

[pjackson1]

Hmm. Maybe the only benefit is for troubleshooting. By using asnodename you might be able to see which client node was having issues?