NFS lockd problems
NFS server: Linux dwarf 2.2.12 #2 Fri Oct 8 20:52:57 EDT 1999 i686 unknown
ii nfs-common 1.4.7-2 Common NFS Tools
ii nfs-server 2.2beta46-2 User space NFS server.
NFS client: Linux jekyll 2.2.12 #5 Sat Sep 4 11:28:01 EDT 1999 i586 unknown
ii nfs-common 1.4.7-2 Common NFS Tools
ii nfs-server 2.2beta46-2 User space NFS server.
(Actually, they're both clients and servers to each other. I'm mounting
file systems in both directions. But for the purposes of this problem,
dwarf is the server.)
The problem is that file locking does not work. If I upgrade mutt
to one of the newer versions (e.g., 1.0pre3-1.2), mutt attempts to
lock my mailbox even though I'm using Maildir and not mbox format.
(See bug #44386.) I've also experienced a "hang" when Angband (variants)
attempt to lock the high score file, which I believe is the same problem
(no locking over NFS).
So I thought I'd be clever and fix the problem myself by upgrading the
NFS server box to 2.2.12. That went well enough, but the file locking
*still* doesn't work.
If I reboot the system, there's a [lockd] process, but locking doesn't
work.
If I manually stop NFS services (/etc/init.d/nfs-server stop;
/etc/init.d/nfs-common stop), and kill (-9) the lockd process, then
I cannot restart lockd. Any attempt to do so gives me this:
dwarf:/home/greg# rpc.lockd
lockdsvc: Function not implemented
In "strace rpc.lockd":
SYS_169(0x10000, 0xbffff23c, 0, 0x2fe, 0x699) = -1 ENOSYS (Function not implemented)
(or, after upgrading strace:)
nfsservctl(0x10000, 0xbffff25c, 0) = -1 ENOSYS (Function not implemented)
I'm going to try knfs next....
--
Greg Wooledge | Distributed.NET http://www.distributed.net/
wooledge@kellnet.com | because a CPU is a terrible thing to waste.
http://www.kellnet.com/wooledge/ |
Reply to: