Bob, I've no idea whether this is the right thing to do but ... our 8.1 "heartbeat" servers (www.linux-ha.org) needed to ditch the nfs server package and install the nfs-utils package in order to work again. The old servers were SuSE 7.0 machines and after the upgrade to 8.1 they couldn't get nfs locks when talking to Solaris servers. I installed nfs-utils-1.0.1-26 which includes lockd AND statd (the old package only had lockd). Now we don't get the lockd errors in syslog and we can mount the Solaris filesystems OK. Does that help ? Damian Bob Vickers wrote:
Dear All,
Today I upgraded a machine from SuSE 8.0 to SuSE 8.2 and galeon stopped working. Galeon tries to start gconfd-2 and this fails; gconfd-2 also fails if you try and start it by hand.
The message in the log is
gconfd (zmac089-13318): Failed to get lock for daemon, exiting: Failed to create or open '/home/ugrads/xxxxxxx/.gconfd/lock/ior'
If I run strace on gconfd-2 I see
open("/home/ugrads/xxxxxxx/.gconfd/lock/ior", O_RDWR) = 6 fcntl64(6, F_SETLK, {type=F_WRLCK, whence=SEEK_SET, start=0, len=0}) = -1 ENOLCK (No locks available)
I believe this an NFS locking problem. Galeon works if the home directory is on a local disc, but not if it is on an NFS file system. It fails with two different NFS servers: one of them SuSE Linux 8.0, one of them Tru64 Unix 4.0G. Neither server has changed recently.
Does anyone know of recent NFS client changes that might have affected locking? And are there any other tests I can try to confirm whether or not locking works at all or whether it is just gconfd that is affected?
rpcinfo reports (on the client) /usr/sbin/rpcinfo -p localhost | grep lock 100021 1 udp 32768 nlockmgr 100021 3 udp 32768 nlockmgr 100021 4 udp 32768 nlockmgr
I'm rather stumped: any ideas gratefully received.
Bob