bzr locking problems over NFS (solved)

[Ben Finney]

Howdy all,

I've been having trouble using bzr (version 0.7) on an NFS
filesystem. Any attempt to do bzr operations on a destination mounted
on NFS would result in::

    bzr: ERROR: bzrlib.errors.LockError: [Errno 37] No locks available

followed by further trace information.

Research on bzr forums (mailing lists, IRC) led to the common advice
"ensure rpc.lockd is running, both on the server and the client". This
was insufficient: lockd was running fine, but the problem occurred
nevertheless.

The change that has fixed the problem for me is: ensure rpc.lockd
*and* rpc.statd are both running on the server and client.

I'm no expert on NFS, but it seems many other operations can succeed
without rpc.statd running: my workstation had not been running it for
quite some time, but NFS filesystem operations continued to work. One
symptom was infrequent messages from the (running on the client)
lockd, saying "lockd: cannot monitor xxx.xxx.xxx.xxx" with the address
of the server.

Regardless, with rpc.statd and rpc.lockd both running on server and
client, bzr now works fine as far as I can tell.

I hope this is useful to others. (I also have hopes for the
new-fangled locking being developed for bzr 0.8, but at least now I
don't have to wait for that.)

-- 
 \     "If you ever catch on fire, try to avoid seeing yourself in the |
  `\        mirror, because I bet that's what REALLY throws you into a |
_o__)                                          panic."  -- Jack Handey |
Ben Finney