NFS regression since 2.6.0-test3 (in -test4 and 4-mm1)

NFS regression since 2.6.0-test3 (in -test4 and 4-mm1)

[Voluspa]

Transfer time doubled and kernel log is swelling...

Remote is exported rw,no_root_squash and runs a Slackware 7.1.0
(Universal NFS Server 2.2beta47) kernel 2.2.16. Mounted here with a
simple "-t nfs". Both remote and local is ext2.

_2.6.0-test3_

"cp -a" the remote /etc directory takes 2 minutes. /var/log/kernel says:

Aug 27 03:59:59 loke kernel: eth0: media is TP.
Aug 27 04:02:34 loke kernel: nfs warning: mount version older than
kernel

Duh! I'm using your util-linux-2.12pre.tar.gz so stop complaining.

_2.6.0-test4 and -mm1_

The transfer takes 4 minutes and log is filled with:

Aug 27 04:20:47 loke kernel: eth0: media is TP.
Aug 27 04:22:34 loke kernel: nfs warning: mount version older than
kernel
Aug 27 04:23:21 loke kernel: nfs: server oden.fish.net not
responding, still trying
Aug 27 04:23:41 loke kernel: nfs: server oden.fish.net OK
Aug 27 04:24:00 loke kernel: nfs: server oden.fish.net not responding,
still trying

[Lots of lines deleted]

Aug 27 04:27:00 loke kernel: nfs: server oden.fish.net OK
Aug 27 04:27:01 loke kernel: nfs: server oden.fish.net not responding,
still trying
Aug 27 04:27:01 loke kernel: nfs: server oden.fish.net OK

Last night the complete backup took 9 hours (transfer of 1.2 gig).
That's why I noticed the issue.

Mvh
Mats Johannesson

Re: NFS regression since 2.6.0-test3 (in -test4 and 4-mm1)

[Frank Cusack]

On Wed, Aug 27, 2003 at 05:25:17AM +0200, Voluspa wrote:
> Aug 27 03:59:59 loke kernel: eth0: media is TP.
> Aug 27 04:02:34 loke kernel: nfs warning: mount version older than
> kernel
> 
> Duh! I'm using your util-linux-2.12pre.tar.gz so stop complaining.

BTW, I've submitted the util-linux patches needed for the mount update a
while back, I have no idea if they ever made it.  (My mail got put into
the "waiting for moderator approval" queue.)

/fc