failover single CIFS server on VDM

Thanks Richard that confirms what I suspected to be the case

the other issue my client had with the celerra is that he feels NFS should be supported on a VDM so that during a failover

it would seamlessly bring up the CIFS for windows and NFS for the unix environment at the target site

any thoughts on this? to me it doesn't seem like that big of a deal, the target NFS should already be exported so that during failover it just becomes RW enabled and your unix hosts would just need run a mount script

Can you export file system while VDM is not mounted on target Celerra ?

You can actually create NFS exports on VDM filesystems (useful for multiprotocol filesystems) but it has to be done at the command line.  So you could export the VDM owned filesystem on the target side, and use DNS for the NFS clients...  When the VDM fails over and updates DNS for the CIFS server/share, the NFS filesystem will be there as well (same FS as CIFS in a multiprotocol filesystem example), NFS clients using DNS will fail over to the new IP address on the target Celerra and mount the export.  It would take a little up front planning but can be done in a pretty automated fashion.

Okay, it doesn't appear that you can create an NFS Export for a non-existent file system but it does look like you could fail over the VDM to the target, create the NFS export, then fail back.  This would set the export for future failover operations.  Before a failover, the replication target has no shares, no mounts, and the VDM is set to unloaded.  I created a VDM, filesystem, and NFS export, then deleted the mount and unloaded the VDM, the export is still showing up in the config.

When using NFS exports in a VDM you will want to Alias the export, so use the -name option in the server_export command to give the export a short path.  The actual export path will look something like /root_vdm_#/filesystemname/...  You will want the alias to exclude the root_vdm_x portion at a minimum so that you can make the NFS export match on both sides of the replication.

The command I used (which did not include any client ACLs) was...

server_export server_2 -Protocol nfs -name testnfs /root_vdm_1/testfs

So the NFS export is just "testnfs" but that connects users to the /root_vdm_1/testfs filesystem root which is owned by my test VDM.

You can view the VDM paths with the "server_mount server_2" command in order to create the correct exports.

I haven't tried that, I'll see if I can attempt this on my Celerra VSA VM.

thanks for testing Richard, while this is good to know ..it's not practical in the real world. I am not going to failover and failback everytime i create new NFS exports on source.

Yes

You name it Richard ..stores for vSphere, digital image repository, data stores for Confluence (our web infrastructure), learning system (Blackboard). Some of our smaller clients like to use Celerra to dump oracle db exports.

Dynamox,

Out of curiosity, what are the NFS exports being used for?  Is this user data? VMWare VMs? or Oracle DNFS? other?

Heh,

I wish there was an automated feature to re-export NFS on failover in Celerra, maybe we can get a feature request for it...

You can use Replication Manager with some applications on Celerra NFS and that could help.  Also, for vSphere, if you use VMWare SRM, that handles NFS exports for vSphere as well.  Replication Manager also supports vSphere NFS exports.  That's about the extent of it without custom scripting.

something simple like NFS export list should be easy enough to replicate ..one would hope so but surprisingly it has not been incorporated into Celerra.

I'm checking into this internally, if I get anywhere I'll update you.