NFSD problem

NFSD problem

Post by Hiro Sugawa » Tue, 07 Nov 1995 04:00:00

I have been struggling with the Linux NFS server configuration.
The most recent problem I've found is that whenever a new client mounts
the NFS volume exported by the Linux NFS server, the NFS server daemon
has to be killed and restarted. Otherwise, the client can see the
NFS volume as a read-only fiel system. After restarting nfsd, the client
can both read and write the NFS volume.

Has anyone experienced and solved this?


Oracle Corporation                           MPP Development
500 Oracle Parkway, Box 659107, Redwood Shores, CA 94065 USA


1. nfsd: Problem with corrupt files and dying rpc.nfsd with Redhat Linux 5.1 + Solaris 2.5


I'm running Red Hat Linux 5.1 on a PII 400 and am having some problems with
a dying nfsd process and corrupt files when I mount a linux file-system on a
solaris machine and try to copy/extract large files on the mounted system.

Basically every time I try to write a large file (~20MB)
from a Solaris machine to a Linux file-system mounted using NFS, my rpc.nfsd
process on the Linux server dies, and any files from the linux server opened
on any machine get corrupted.

My other HP machines or Windows NT machines do not seem to kill the nfsd process
when writing to the mounted file system.

Can anyone suggest what may be wrong here? I heard something about knfsd being
able to fix the problem. Can I use it without having to upgrade Linux to 6.0?
I also heard that there may be a Solaris patch that may fix my problem...

If it helps, I have some information that may indicate the problem.

My /var/log/messages on linux reports the following just before the process crashes:
Jun 17 17:16:10 server nfsd[837]: strange write req from <anon clnt>: len 53248
Jun 17 17:16:56 server nfsd[2274]: exports file has anon entries, but host
Jun 17 17:16:56 server nfsd[2274]: has non-private IP address ............!
Jun 17 17:17:01 server nfsd[2275]: Could not bind name to socket Address already in
Jun 17 17:17:01 server nfsd[2275]: could not make a udp socket

My current nfs RPM versions are the following that I updated using the latest RHL 5.2 updates.


My dmesg reports the following:

3c59x.c:v0.99L 5/28/99 Donald Becker

eth0: 3Com 3c905B Cyclone 100baseTx at 0xe400,  00:50:04:05:3a:7d, IRQ 10
  8K byte-wide RAM 5:3 Rx:Tx split, autoselect/Autonegotiate interface.
  MII transceiver found at address 24, status 786d.
  MII transceiver found at address 0, status 786d.
  Enabling bus-master transmits and whole-frame receives.

to the newsgroup.

Thank You,

Best Regards,

Alok Mathur            Off  : (513) 985-9877
TechnoSoft Inc.        Fax  : (513) 985-0522

Cincinnati OH 45242

2. OS/2 Backup of Linux Partition

3. NFSD - problems (nfsd version 2.0 and 2.2)

4. BUG: 2.5.40: sleeping function called from illegal context

5. 911 NFSD problem

6. help on upgrading gcc

7. nfsd problem

8. What is PGP ?

9. Mountd and nfsd problem

10. nfsd-problem - PLEASE HELP !!!

11. rpc.mountd/rcp.nfsd problems

12. NFSD-problems - PLEASE HELP!!!

13. NFSD problem : sendto (Out of memory)