1

I'm seeing lots of messages in my /var/log/messages file like this:

Sep  2 17:36:47 hostname kernel: lockd: couldn't create RPC handle for 192.168.0.1

This server is NFS mounting a few directories from 192.168.0.1.

I have no idea where to start diagnosing this, and I'm pretty sure it's related to this What does 'no locks available' mean?

Community
  • 1
Amandasaurus
  • 31,471
  • 65
  • 192
  • 253

2 Answers2

1

I can tell you that they are directly related. Once you resolve the issue with lockd / rpc.lockd, your Subversion lock issues will go away.

Can you verify that lockd / rpc.lockd started correctly on 192.168.0.1? From your client, can you do 'rpcinfo -u 192.168.0.1 100021' ?

This thread has some useful information on lockd issues

Tel Janin
  • 156
  • 3
  • # rpcinfo -u 192.168.0.1 100021 give this output: rpcinfo: RPC: Timed out program 100021 version 0 is not available That means it's not running? How do I restart it? – Amandasaurus Sep 02 '09 at 17:07
  • Is there anything in dmesg on 192.168.0.1 re: lockd? ie, 'lockd: new process, skipping host shutdown' or 'lockd_up: makesock failed, error=' Also, can you check /proc/sys/sunrpc/rpc_lockd and see what value that is? – Tel Janin Sep 02 '09 at 17:19
  • restarting the nfs server fixed it – Amandasaurus Sep 03 '09 at 10:13
1

Chances are that your NFS server is not running the rpc daemon. Try mounting your nfs with the "-o nolock" option. If that gets rid of your error messages, you have your reason. If there is only one client mounting that NFS share at a time, it should be safe enough to use the nolock option. This emulates the older NFS method of maintaining local locks only instead of remote locks.

sybreon
  • 7,405
  • 1
  • 21
  • 20