mount.nfs hangs forever - nfs4_discover_server_trunking unhandled error -512

[pwalden]

I have 3 newly upgraded F18 systems. 2 of them cannot make an NFS mount, while one does (slowly). I cannot determine what is wrong.

The command below

# mount.nfs walden4.local:/opt2 /opt2

hangs indefinitely. When I abort it with an interrupt, the message log adds the follow entries:

Aug 15 14:20:27 walden3 kernel: [ 9949.514708] NFS: nfs4_discover_server_trunking unhandled error -512. Exiting with error EIO
Aug 15 14:20:27 walden3 rpc.gssd[1046]: ERROR: can't open /var/lib/nfs/rpc_pipefs/nfs/clnt2/info: No such file or directory
Aug 15 14:20:27 walden3 rpc.gssd[1046]: ERROR: can't open /var/lib/nfs/rpc_pipefs/nfs/clnt2/info: No such file or directory

I tried googling for some answers with no results.

the /var/lib/nfs/rpc_pipefs/nfs directory exists but has no write permissions.

the machine that is working has the same directory permissions as the non-working machines:

Non-working machine:

[pwalden@walden3 ~]$ ll -Rd /var/lib/nfs/rpc_pipefs/nfs
dr-xr-xr-x. 2 root root 0 Aug 15 14:20 /var/lib/nfs/rpc_pipefs/nfs
[pwalden@walden3 ~]$ ll -R /var/lib/nfs/rpc_pipefs/nfs
/var/lib/nfs/rpc_pipefs/nfs:
total 0

Working machine:

[pwalden@walden6 ~]$ ll -Rd /var/lib/nfs/rpc_pipefs/nfs
dr-xr-xr-x. 5 root root 0 Aug 15 13:53 /var/lib/nfs/rpc_pipefs/nfs
[pwalden@walden6 ~]$ ll -R /var/lib/nfs/rpc_pipefs/nfs
/var/lib/nfs/rpc_pipefs/nfs:
total 0
dr-xr-xr-x. 2 root root 0 Aug 15 13:54 clntd
dr-xr-xr-x. 2 root root 0 Aug 15 13:54 clnte
dr-xr-xr-x. 2 root root 0 Aug 15 13:54 clntf

/var/lib/nfs/rpc_pipefs/nfs/clntd:
total 0
prw-------. 1 root root 0 Aug 15 13:54 gssd
prw-------. 1 root root 0 Aug 15 13:54 idmap
-r--------. 1 root root 0 Aug 15 13:54 info
prw-------. 1 root root 0 Aug 15 13:54 krb5

/var/lib/nfs/rpc_pipefs/nfs/clnte:
total 0
-r--------. 1 root root 0 Aug 15 13:54 info

/var/lib/nfs/rpc_pipefs/nfs/clntf:
total 0
-r--------. 1 root root 0 Aug 15 13:54 info

I determined that the rpc_pipefs appears to be mounted correctly on both systems:

# grep pipe /etc/mtab
sunrpc /var/lib/nfs/rpc_pipefs rpc_pipefs rw,relatime 0 0

[pwalden]

An update on this problem. After several yum updates, a new error appeared in the message log:

Oct 22 17:03:50 walden3 rpc.gssd[1082]: ERROR: unable to resolve 192.168.1.107 to hostname: Name or service not known
Oct 22 17:03:50 walden3 rpc.gssd[1082]: ERROR: failed to read service info
Oct 22 17:09:22 walden3 rpc.gssd[1082]: ERROR: can't open /var/lib/nfs/rpc_pipefs/nfs/clnt0/info: No such file or directory
Oct 22 17:09:22 walden3 rpc.gssd[1082]: ERROR: failed to read service info

I found a reference via google to add the missing entry into /etc/hosts. Adding the hostname and ip address to /etc/hosts has mount.nfs working again!!!!

For example adding:

walden4.local    192.168.1.107

does the trick.

The problem is that I use avahi and mdns_nss to dynamically manage *.local domain IP addresses. What do I need to do to get rpc.gssd to use this facility?

Here is my /etc/nsswitch.conf entry:

hosts:      files mdns4_minimal [NOTFOUND=return] dns myhostname

[sea]

Just out of curiosity, can you ping the ip (192.168.1.107 ??) directly from all the machines?

[pwalden]

Yes, ping and everything else seems to work fine.