From Bugzilla Helper: User-Agent: Mozilla/4.0 (compatible; MSIE 5.01; Windows NT 5.0) Description of problem: I am using nfs-utils-0.3.1-5 on the NFS server (Linux). The client is an HPUX 11 HP 9000/785. I get a 'No locks available' error on the client. The server logs a : "kernel: nfsd: request from insecure port (5900fee9:49437)!" in the messages file, even thought the exports file on the server specifies : "/home 89.0.254.233(rw,no_root_squash,insecure)" How reproducible: Always Steps to Reproduce: 1.share a directory from Linux 7.1 2.mount the share from HPUX 11 3.write a short program that uses fcntl to lock a file on the NFS share Actual Results: fcntl() returns -1, perror() prints "No locks available" Expected Results: Lock should have been aquired. When the client is a Solaris box this works fine. When the HP-UX connects to an NFS share on the Solaris box this work fine. Only client HP , server Linux seems to fail. Additional info: nfsd and lockd are loadable modules, /etc/init.d/nfs start , and /etc/init.d/nfslock were run
Sounds like lockd is dying on the server. Can you see the lockd process on the server after the failure? There is a later nfs-utils available (0.3.1-8). Can you try that? Also, lockd is start automagically in the kernel in kernels 2.2.18+, so you should need to run the lockd init script if you're running a recent enough kernel.
> ...so you should need to... Sorry, read "...so you shouldn't need to..."
OK, problem solved. After some time went by, (about 2 hours), I tried it again, and this time it worked. Looks like the lockd servers needed some time to get their act together. In retrospect I believe the bug originated from trying to use NFS locking, without starting the '/etc/init.d/nfslock start' server. The loadable kernel modules (nfs and lockd) were allways present, but I guess starting it this way does something good for it.