Messages in this thread | | | Subject | Upon rebooting knfsd server: nfs_revalidate_inode fails. | Date | Tue, 17 Nov 1998 11:58:43 +0000 | From | David Woodhouse <> |
| |
I'm still having problems with knfsd when the server is rebooted.
We use NFS because it's supposed to be stateless - previously (with unfsd) you could reboot the server and it would just keep working.
By adding explicit entries for each host on our local net to /etc/exports, I've made sure that clients don't get 'permission denied' because of the dodgy wildcard handling. Not ideal, but it works.
Unfortunately, we still need to restart most of the NFS clients, as they complain of invalid nfs filehandles.
When we reboot the server, we get errors like this on the clients:
nfs_revalidate_inode: //.netscape getattr failed, ino=10241, error=-116 NFS: bad fh c6d9e04000002801000008010000081100000811000008010000000000000000 NFS: lookup failed, error=-116 nfs_revalidate_inode: //.netscape getattr failed, ino=10241, error=-116 NFS: bad fh c6d9e04000002801000008010000081100000811000008010000000000000000 NFS: lookup failed, error=-116 nfs_lookup: //xm_error failed, error=-116
(The server doesn't complain, except for: lockd: attempt to release unknown file! which, I suppose, is fair enough)
ISTR that once upon a time, the knfsd code would use the address of the dentry in memory as the filehandle, and hence there was no chance of this being reproducible across a reboot (and no guarantee even without a reboot if you were short of memory).
However, I thought that had been fixed. Does anyone remember this discussion, and the outcome of it?
I'm using 2.1.127-pre2 on the server, as I haven't had the courage to 'upgrade' it to the newer kernels that are crashing the other machines on the network. It's got all 6 of HJ's patches in it, along with a fix for the timeout problem with lockd.
I'm using knfsd-981022, and will be updating to the latest one next weekend - if I didn't have to go round and reboot all the workstations each time I restart the server, I'd have upgraded sooner.
Does anyone else see this problem, have a fix for it, or even a suggestion for where I should look?
---- ---- ---- David Woodhouse David.Woodhouse@mvhi.com Office: (+44) 1223 810302 Project Leader, Process Information Systems Mobile: (+44) 976 658355 Axiom (Cambridge) Ltd., Swaffham Bulbeck, Cambridge, CB5 0NA, UK. finger dwmw2@ferret.lmh.ox.ac.uk for PGP key.
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/
| |