Board index Linux development

NFS client hangs

NFS client hangs

Post by Bodo Teichman » Fri, 19 Sep 1997 04:00:00



--Multipart_Thu_Sep_18_09:23:59_1997-1
Content-Type: text/plain; charset=US-ASCII

Lines: 23


NNTP-Posting-Host: iis.iis.fhg.de
X-Newsreader: Gnus v5.4.37/XEmacs 19.15
Xref: rznews.rrze.uni-erlangen.de comp.os.linux.networking:99152

--Multipart_Thu_Sep_18_09:23:59_1997-1
Content-Type: text/plain; charset=US-ASCII

hi ,
i have another nfs problem:
if a process reads a file over nfs the process sometimes hangs (it can't
even be stoped by Ctrl-c) after a few minutes (less then 2 minutes) the
process continues (or stops , if a Ctrl-c was pressed befor).

if a process read a very big file that problem occured about every 2 to
4 minutes .

(if i read the file from local drive, there is no problem)

i had the same  problem with the 2.0.18 kernel.

does someone know which kernel version does not have this problem with
nfs?

BTW: the server is not a linux system (netapp,solaris,irix) , so i'm shure that its not a
server problem and from other (not-linux) unix workstations i have no problem.

bodo teichmann

--Multipart_Thu_Sep_18_09:23:59_1997-1
Content-Type: text/plain; charset=US-ASCII


  Fraunhofer Institut fuer Integrierte Schaltungen (FhG/IIS)
Tel.: +49 9131 776367                  Fax.: +49 9131 776399 */

--Multipart_Thu_Sep_18_09:23:59_1997-1--

 
 
 
Top

1. NFS clients hang redhat linux 7.3 and limbo clients accessing auspex ns2000

We are using an Auspex ns2000 at the office for serving a lot
of different directories including our home directories.  The
mounting is controlled with amd getting maps from nis.  This
has worked reasonably well for all of our UNIX boxes and Linux
boxes... until I upgraded my desktop to RedHat 7.3 and an adv.
development box to Limbo (kernel version 2.4.18-5 and 2.4.18-5.58
respectively).

Processes on both of these boxes routinely hang when reading or
writing large files on the Auspex.  Things like ls work fine most
of the time.  Until I started to experience this problem we did
not mount with intr, that meant that when it happened, I couldn't
even kill the processes.  I've fixed that, but I can not get the
hangs to stop.  The mount parameters are:

    opts:=rw,nosuid,nodev,intr

I have messed around with using a smaller rsize and wsize, but that
results in the message

    NFS: short packet in readdir reply!

and does not fix the hang.  I read some discussion that the above
message indicates that the reply is packet is bigger than the buffer
the client has set aside to read it.  The adminstrators of the Auspex
say that they were instructed, by Auspex, to set the rsize and wsize
on the server to 32768.

Any suggestions on what will resolve this problem?  Do I need to tweak
my settings or get the Auspex admins to tweak theirs?  Or do some
surgery?

Thanks much!

--
Dick Schoeller

880 Winter St., Bldg 4 http://schoeller.ne.client2.attbi.com/
Waltham, MA 02454      781.663.4507

"Er ist ein Narr, der meint, es sei nicht schad, das Kind
auszuschtten mit dem Bad"         - Thomas Murner 1512

2. add support for PC-9800 architecture (PNP)

3. NFS client hangs when the server goes down

4. CDROM Problem

5. nfs clients hangs

6. test only

7. Linux NFS client hangs

8. How to NAT dependand on destination address?

9. PROBLEM: NFS client hangs when X is running (2.4.20)

10. NFS client locking hangs for period

11. solaris NFS hangs on linux client

12. NFS client locking hangs for period

13. NFS locks hang, FreeBSD 5.0 client, Linux server


All times are UTC
Board index