showmount: xxxxxxxx: RPC: Rpcbind failure - RPC: Timed out

[masoodsy]

i have ports open and rpcinfo -p works on both the client & Server on both Client and server.

/etc/dfs/dfstab also looks ok.

restarted nfs services on both nfs server and client.

but still i dont get why the showmount still show the rpc timed out !!

showmount: <nfsserver>: RPC: Rpcbind failure - RPC: Timed out
have read pretty much every forum discussion on this one but still i am able to make it work !!

anymore ideas guys..

[Bertram Stoeckler]

Can you please paste the complete command that is failing ?

[masoodsy]

here is the output.

# showmount -e infbc009
showmount: infbc009: RPC: Rpcbind failure - RPC: Timed out

[Bertram Stoeckler]

1. First check that "infbc009" is IP-resolvable, if not, try the IP-address instead

2. try to ping that host

3. run "rpcinfo -p" infbc009".  If this is not working, then your client cannot talk to port 100 of the server, check iptables, routers etc...)

4. run "rpcinfo -T tcp infbc009 100005". 

100005 is the mount-protocol and you should get a list of versions that are "waiting", e.g:

program 100005 version 1 ready and waiting
program 100005 version 2 ready and waiting
program 100005 version 3 ready and waiting

If you don´t get this, then your client cannot reach the server on the port that rpcinfo -p is reporting. 

So take the "rpcinfo -p" output and check the port for "mountd". The firewall need to allow that port for TCP.

5. run the same command for udp. "rpcinfo -T udp infbc009 100005". This is just a check. Some firewalls are misconfigured to drop UDP but allow TCP, and depending on the NFS-client OS and version the "showmount -e" may give up after a first attempt with UDP, or vice versa.

6. Next step of troubleshooting would be to collect a networking trace at client and server during failing mount.

Hope that helps.

BTW: On Linux "iptables -L" will print out the firewall rules.

[masoodsy]

thankyou so much for the steps...

I see all TCP ports that are listed under rpcinfo are accessible incluing the one for mountd and respond to telnet connections.

next step is to engage network guys with a network trace.

lets see what they say..

[Bertram Stoeckler]

What is the NFS-client OS-name and version? And did you get anything back when using "UDP" ?

If you are running in a mixed MTU-size environment, it may be possible that the server-reply is too large (and the router/switch is misconfigured)

In that case you can test with "ping -s 9000 <server-name>" and if that fails try lower numbers. 

[masoodsy]

both client and server are HP-UX machines.

MTU is 1500 on both of them.

from clinet, with udp this is what I get.

:/# rpcinfo -T udp infbc009 100005
rpcinfo: RPC: Rpcbind failure - RPC: Timed out

[Bertram Stoeckler]

Can you also please provide the HPUX version ?

"uname -a" will tell you.

There was(or maybe still is)  a known problem where the mount-command would use UDP for mounting, even if a TCP-NFS-mount were requested. I do not know if this also applies for the showmount-command, but should be able to test.

Thanks