Problem

Listnode command is not showing the correct status on multi-node Transfer CFT

Listnode result differs depends from the physical node the command is issued

Resolution

You are advised to configure each NFS client in order it uses the same IP address to connect to the NFS server address.

1) callback path
The "clientaddr" value advertised by a NFS client to NFS server should allow the NFS server to "call back" the client in some situations. This reverse path must work (i.e.: the server be able connect back to the NFS server) for both NFS V3 and NFS V4.

2) server addresses
Another point related to NSF server addresses raises questions.
NFS client hosts not using the same IP address to communicate with the NFS server may led into issues.

Not all NFS server/client pair can work reliably when the server is multi-homed. This may involve how the user computed its client-id and/or its ability to detect multi-homed servers, etc...

Recommendations:

1) Use fixed IP addresses for the clients and make sure that they are reachable from the server TCPIP interface.

2) Avoid any kind of nat or firewall that would prevent the callback path from working.
Ensure that the advertised 'clientaddr' to the NFS server are correct for all hosts.

3) Make sure that hosts running CFT connect to the NFS server using the same IP address.


Useful information can be obtained on the WEB, using search patterns involving words like:
- nfsv4 nat multihomed callback delegation lock mount etc...