2012-03-09 06:04 AM
I'm not even close to Ohio, but I'd like information about Cluster-mode anyhow.
Is it possible to move CIFS sessions to other cluster nodes transparent to the CIFS clients?
In ONTAP 7.3 and SMB 1.1 all CIFS traffic will need to do a re-connect when failing over a HA-cluster. Depending of the application using CIFS this could be a big problem.
If it is possible what is the components making this happen? Where is the magic?
Is this a new feature in SMB 2.0 or 2.1?
Is it handled by the communication stack? (OSI model)
I haven't had too much time looking in to this, so if you could help me understand this better you would make my day!
2012-03-13 07:01 AM
In practice, many CIFS applications can handle a LIF failover to another node in a cluster. For applications that cannot, this might result in a “Lost Delayed Write” operation, but only (in my understanding) if those clients use Opportunistic Locking to cache the server’s file data locally, without updating it on the server as it’s manipulated locally by a client app.
2014-05-06 02:38 AM
In C-mode the the user will get disconnected if there is any activity/disruption occurs on the node on which the volume is created (which is used for share creation).Only the hosts (hypervisor) which are accessing the share with SMB 3.0 can access the share without any disruption.You have to enable the continous availabity option in the share to make the share available without disruption for Hypervisor.
Please comments on the response ..if it is not correct .