Network Status Monitoring Protocol (NSM)
Since NFS is a lot less stateless than many documents claim, reality requires NFS to keep track of state changes due to events like client or server restarts/reboots. The NSM protocol is used for NFS versions 2 and 3 to inform peers about reboots in particular so that file locks can be cleaned up.
The purpose of this protocol is to allow a client and /or a server to inform all its peers when the service has restarted so any hanging locks can be released or reclaimed. This is really important for NFS since locks held by a client will never time-out and never be automatically released othervise. (Good DOS attack for NFS is to capture a lot of filehandles and then send NLM requests to the server for them and just leave the locks hanging until someone reboots the server).
These notifications of a peer reboot is issued by the rebooting client when the service restarts based on information for whom to notify recorded on stable storage. Usually this protocol is using TCP as its transport and usually it will only send notifications to those peers where a file lock is being held using the NLM protocol.
Some implementations do not send notification to clients that only has locks in blocked state (not yet aquired/granted) since this state will anyway be resolved by the client reissuing the original (blocked) lock request after a timeout anyway. Other implementations send notification events to any and every peer it has that has talked NLM to regardless of whether any locks were currently held or not.
This protocol is not required if the NFS share is exported read-only or if file locking is not used.
The NSM protocol came after the original release of NFS when byte-range locking support was added in SunOS, as locking more obviously requires a stateful protocol. The purpose of the protocol is to notify peers of when they have restarted their service and thus all held NLM locks will have been purged.
The NSM dissector is fully functional. The Service Response Time statistics feature in both Wireshark and TShark can calculate response time statistics for this protocol.
There are no preference settings for the NSM protocol.
Example capture file
A complete list of NSM display filter fields can be found in the display filter reference
Show only the NSM protocol:
The real name for this protocol is NSM but in wireshark this protocol is called "stat" after the name of the service that usually implements this protocol "rpc.statd". This was a mistake.
XXX should we rename if to NSM ? ronnie
You cannot directly filter on the NSM protocol while capturing. Since NSM does not use a standard port you can not capture on a specific port either.
- rpcinfo can be used to "ping" the NSM service on a server.