2.4.22-pre3 nfs-client write-problems

Erkki Seppala (flux@modeemi.fi)
Fri, 11 Jul 2003 23:51:47 +0300


I've experienced the problem with versions vanilla 2.4.20, vanilla
2.4.21 and 2.4.22-pre3+device-mapper.

The problem appears when I write big files relatively fast over
nfs. My client is Dual-Athlon 1900+, and the servers are Dual P3
500MHz (with 2.4.18) and Pentium 2 200MHz (with 2.4.18 debian bf-2.4
default kernel).

Both servers use kernel-nfsd. However, as things work perfectly with a
2.4.19-client and between the servers, it would hardly seem to be a
server-matter.

There is a gigabit-link between the Athlon and P3.

Reproducing: dd if=/dev/zero of=new-file-on-nfs-file-system -> dies on
I/O error.

If I dd if=/dev/zero bs=1024 count=1000000 of=existing-one-GB-file,
things fail similarly. (Last attempt failed at record 3073.)

It might be just luck, but it would seem to me that the Dual-P3 server
behind the gigabit-link fails faster than the slow P2-server (which is
infact network-topographically behind the P3).

Relevant lines from .config:

CONFIG_NFS_FS=m
CONFIG_NFS_V3=y
# CONFIG_ROOT_NFS is not set
CONFIG_NFSD=m
CONFIG_NFSD_V3=y
# CONFIG_NFSD_TCP is not set
CONFIG_SUNRPC=m
CONFIG_LOCKD=m
CONFIG_LOCKD_V4=y

I'm willing to try patches :).

-- 
  _____________________________________________________________________
     / __// /__ ____  __               http://www.modeemi.fi/~flux/\   \
    / /_ / // // /\ \/ /                                            \  /
   /_/  /_/ \___/ /_/\_\@modeemi.fi                                  \/
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/