[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

NFS locking



Hi,

I'm having problems with a couple of applications and reading about the problem indicates its probably caused by locking not working correctly on NFS. The problem seems to lie with the server end of the set up - my guess is that it's not accepting the locks from the client machine. Both the client and the server machines are running the latest nfs kernel server and have statd and lockd running (according to ps). The relevant info from the server is:

sprocket:/data# rpcinfo -p
   program vers proto   port
    100000    2   tcp    111  portmapper
    100000    2   udp    111  portmapper
    100024    1   udp  37306  status
    100024    1   tcp  35125  status
    100003    2   udp   2049  nfs
    100003    3   udp   2049  nfs
    100003    4   udp   2049  nfs
    100003    2   tcp   2049  nfs
    100003    3   tcp   2049  nfs
    100003    4   tcp   2049  nfs
    100005    1   udp    866  mountd
    100005    1   tcp    869  mountd
    100005    2   udp    866  mountd
    100005    2   tcp    869  mountd
    100005    3   udp    866  mountd
    100005    3   tcp    869  mountd

sprocket:/data# ps axf | grep lockd
 3816 ?        S      0:26 [lockd]
 8361 pts/0    S+     0:00                      \_ grep lockd

sprocket:/data# ps axf | grep statd
 8394 pts/0    S+     0:00                      \_ grep statd
30319 ?        Ss     0:00 /sbin/rpc.statd

I have another machine that also acts as an NFS server with, as far as I can tell, exactly the same version of NFS running. When I run rpcinfo -p on that machine I see the report above and a number of lines like this:

100021    1   udp   1039  nlockmgr

which I presume are the lock managers. One of the applications that I am struggling with is Open Office but I have side stepped the locking problem there but turning off locking as per the well known hack. The other application I am struggling with is digikam which flatly refuses to work without locking.

If I move the digkam database to the machine that reports nlockmgr everything is fine. On sprocket digikam doesn't work so I assume that the lack of nlockmgr is the problem.

Further reading seems to say that lockd starts nlockmgr when it is needed but the server has never started nlockmgr to the best of my knowledge.

I'm really stuck at this point as I have tried everything I can think of and Googled till my fingers hurt. Some help would be really appreciated.

Graham



Reply to: