From Linux NFS

Latest revision as of 22:28, 26 July 2011

the first things to check

showmounts

Run showmount -e <server> on the client. This will list list the available exports on the server:

[root@client1 ~]# showmount -e server.example.fake
Export list for server.example.fake:
/export *
/pub    192.168.56.*

In this example, server.example.fake has two mounts available: /export and /pub

tshark

XXX

krb5 problems

rpc.ggsd not running

Mount will hang if a sec= option other than sys (which is the default) is used:

Jul 26 18:12:29 client1 kernel: [485467.135178] RPC: AUTH_GSS upcall timed out.
Jul 26 18:12:29 client1 kernel: [485467.135182] Please check user daemon is running.

[root@client1 ~]# service rpcgssd restart
Restarting rpcgssd (via systemctl):                          [  OK  ]
[root@client1 ~]# ps wwuax | grep rpc.gssd
root       876  0.0  0.1  39376  2572 ?        Ss   Jul21   0:37 rpc.gssd -vvv
root      5271  0.0  0.0 109100   836 pts/0    S+   18:21   0:00 grep --color=auto rpc.gssd

couldn't connect to server's rpc.svcgssd

Like many errors, you get a permission denied

mount.nfs4: mount(2): Permission denied

But the syslog tells more (if rpc.gssd is running with -vvv):

Jul 26 18:25:45 client1 rpc.gssd[5325]: creating context with server nfs@server.example.fake
Jul 26 18:25:45 client1 rpc.gssd[5325]: WARNING: Failed to create krb5 context for user with uid 0 for server server.example.fake
Jul 26 18:25:45 client1 rpc.gssd[5325]: WARNING: Failed to create machine krb5 context with credentials cache FILE:/tmp/krb5cc_machine_EXAMPLE.FAKE for server server.example.fake
Jul 26 18:25:45 client1 rpc.gssd[5325]: WARNING: Machine cache is prematurely expired or corrupted trying to recreate cache for server server.example.fake

This error condition is detectable and we should probably have a better message - or maybe return a different error to mount (like EIO).

user not authenticated

XXX show weird mesg

no supported enc types

XXX