> My understanding is you have 2 local disks. Node1 has diskA and Node2 has > diskB. That > means Node1 has no access to diskB and vice versa. Correct.
> So when you say "I need to make available a ZFS filesystem through NFS for > some clients" , >are you talking of the same ZFS file system spanning diskA and diskB or are >you talking of 2 >different filesystems, where each file system reside on one disk ? each file system reside on one disk. So, i have a nfs share on node1 (diskA), and everything is fine... when i fail occurs on diskA, the services failover to node2 with "another" ZFS filesystem on diskB (could/should be the same pool/zfs names). > In either case you can use the scdsbuilder and write your own custom agent to > monitor > the file system. Ok, that's the information i need to start, and the reason i post the first message here... i don't know how assemble the parts. Here is what i think i need to do: 1) Create a pool/zfs filesystems in diskA(node1) and diskB(node2). Let's call the pool "test". 2) Create the logicalhostname and NFS resources on SC 3.2 3) Configure the dependency of logicalhostname and NFS resources on pool "test". But, i need to build a custom agent to handle the mount/unmount of that pool/fs, because they are not HA. The assembly of point "3" is the issue for me... i'm looking for documentation, but i was trying to get some advices here too. Thanks a lot. -- This message posted from opensolaris.org
