Flytt av NFS resurs från en clusternod till en annan misslyckas.

I onsdags släppte Services for UNIX teamet på Microsoft en blogpost om att de har ett känt problem med att NFS resurser inte alltid vill flyttas snyggt och prydligt mellan clusternoder.

Jag fastnade för denna post just för att vi har haft ett liknande problem hos en av mina kunder. Detta gällde i första hand vid patchning av filservern eftersom att man få failade mellan noderna.

Eftersom att jag känner mig lite lat så kör jag lite Copy-paste här...

One of the common issue on NFS as a cluster resource is that during failover the resource fails to move from one node to other. The network name is moved but the NFS resource shows the status as “Failed”


This usually happen due registry corruption. Server for NFS service has to be started in certain sequence on a cluster node. We have seen multiple instances where the NFS service was restarted on one node only which caused the registry to be corrupted. The registry in question is:


"HKEY_LOCAL_MACHINE\Software\Microsoft\Server for NFS\CurrentVersion\Exports\"


Under this registry key ( Exports) all the NFS export are listed and should start from 0 and then on incremented by 1 for each new NSF share. If the numbers are not showing up in incrementing order (with a difference of 1) properly that means the registry is corrupted.


In case the registry is corrupted on any of the nodes, then the NFS resource will not failover or move to that node. To resolve the issue follow the steps below:


1. Browse to the registry location and delete all the folders listed under Exports on the passive node (that has issue).

2. Take all the NFS resource on the cluster offline.

3. Stop the server for NFs service one by one on both the nodes


4. Start the server for NFs service on both the nodes one by one


5. Get the NFS resource online


6. Move the resource to the passive node (which was having the issue)

Förhoppningsvis behövs inte detta tips, men man vet aldrig...