Hi Team, We are planning to move from nfsv3 to nfs v4 and use dynamic pool for same, We are already on OneFS 8.*. I want to know since nfsv4 is stateful protocol , how oneFs manage node failure or IP move in isilon. How it maintain TCP session across all nodes.
NFSv4 is stateful but it does not rely on the TCP session for the state like SMB2 does so OneFS doesn't have to maintain that TCP session across all nodes. The work in 8.x for NFS v4 and dynamic pools is more around ensuring that lock states are maintained across the nodes of the cluster as that was the missing piece in earlier releases. An NFSv4 client can move with the IP address just like v3 did, but locking is quite different.
Thank You Adam for reply, so where these locks state are maintained, so that all nodes can able to access locks state.
I it in L1 cache of node, since L1 cache is coherent across nodes?
I'm not an engineer but as best I can tell there is an cluster-wide internal database that is accessible by all nodes. I doubt this is exposed to the end user. But we use this in conjunction with the standard for lock recovery in NFSv4.