Re: NFS server not responding, new in 6.0

2006-05-03 Thread Alex Zbyslaw

David Kirchner wrote:


We recently replaced FreeBSD 4.5 with 6.0-RELEASE on a pair of servers.
One of the servers runs rsync to copy its contents to the other server,
over a NFS mount. Everything worked just fine under 4.5, but with 6.0,
we're seeing dozens of these errors every rsync:

May  2 14:00:59 xxx1 kernel: nfs server xxx2:/usr: not responding


IIUC, why not rsync directly to the other server and bypass NFS;  see if 
that works better.


--Alex


___
freebsd-questions@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-questions
To unsubscribe, send any mail to "[EMAIL PROTECTED]"


Re: NFS server not responding, new in 6.0

2006-05-02 Thread Bill Moran
"David Kirchner" <[EMAIL PROTECTED]> wrote:

> We recently replaced FreeBSD 4.5 with 6.0-RELEASE on a pair of servers.
> One of the servers runs rsync to copy its contents to the other server,
> over a NFS mount. Everything worked just fine under 4.5, but with 6.0,
> we're seeing dozens of these errors every rsync:
> 
> May  2 14:00:59 xxx1 kernel: nfs server xxx2:/usr: not responding
> 
> The rsync does eventually complete successfully. The NFS client system
> uses the em0 driver on a gigabit port, and the NFS server system uses
> the fxp0 driver on a 100Mbit/full duplex port. The client system
> doesn't come close to 100Mbit during the rsync (or otherwise) -- more
> like 5Mbit. Neither server is what I'd consider "busy" -- they're actually
> basically idle unless this script or some crons are running.

Make absolutely sure that the em card is set to the correct speed/duplex
settings.  If not, manually bypass the autodetection and set the speed
and duplex.

We've been seeing a lot of em cards on gigabit misdetect the speed and
duplex.  The symptoms are lousy performance in some cases, and outright
failure in others.  In each case, manually setting the speed/duplex
fixes the problem and all is well.

> We're using NFSv3 soft, interruptable mounts. We've tried using TCP and
> UDP, and have tried different -r and -w sizes, up to 32768 each. We've tried
> it with and without nfsiod.

Looks like you've already tried a lot of things.

> We haven't tried changing the mount_nfs -D option, because that seems
> like it would only serve to mask the real problem, whatever that is.

I agree.

> a) Is this a real problem, or simply a reporting problem? What exactly is it
> reporting if it's not a real problem?

Sure is a real problem.  Unless you've got serious network congestion, in
which case it's still a problem, just not with NFS.

> b) If you've had this trouble before, what settings have you used to fix it?

Looks like you've already tried all the NFS tweaks I could think to
recommend.  Hopefully the media settings will help.

-- 
Bill Moran
Potential Technologies
http://www.potentialtech.com
___
freebsd-questions@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-questions
To unsubscribe, send any mail to "[EMAIL PROTECTED]"