SUMMARY: NFS Problem

From: Martin Greving (martin@informatik.trier.fh-rpl.de)
Date: Mon May 13 1996 - 03:14:01 CDT


Hi all,

My original question was:

> Hello,
>
> we are using a SS20 as a nfs and nis+ server for ~140 users and
> 12 nfs clients.
>
> Sometimes it occures that all clients hang with the message
>
> nfs server XXX not responding
>
> I can not find any network error or any message on the nfs server.
>
> But I found a very simple workaraound:
> After using 'ping somehost' on the nfs server the clients are working again.
>
> Following Patches are installed:
>
> 101571-03/ 101920-01/ 101969-04/ 102011-02/ 102079-01/ 102680-03/
> 101753-01/ 101921-04/ 101975-01/ 102020-02/ 102112-01/ 102711-01/
> 101758-01/ 101922-04/ 101979-03/ 102030-01/ 102137-01/ 102729-01/
> 101829-01/ 101923-03/ 101981-01/ 102035-01/ 102165-02/ 102820-01/
> 101878-01/ 101925-01/ 101983-01/ 102036-01/ 102216-05/ 102854-01/
> 101879-01/ 101933-01/ 102001-03/ 102037-01/ 102218-03/ 102922-03/
> 101880-03/ 101945-10/ 102001-09/ 102038-01/ 102286-01/
> 101902-01/ 101945-36/ 102002-01/ 102044-01/ 102292-02/
> 101905-01/ 101959-02/ 102003-01/ 102062-03/ 102332-01/
> 101907-02/ 101959-06/ 102007-01/ 102070-01/ 102664-01/
>
> Does somebody know a patch or bugfix for my problem?
>
> Thanks in advance. I will summarize.

I received 4 messages but no real solution.

Jens Fischer <jefi@kat.ina.de> stated, that there ist a newer version of
patch 101979, at least -06. A look at the SunSolve WWW-Server told me
that this patch is obsoleted by 102001 and the currect version is 10.
Version 09 is installed and I will try to upgrade in a few weeks.
Another idea was:
> Another workaround which we were using when we experienced the same problem is
> to start a ping -s -I 60 from an init script at boottime. This does a "wakeup"
> of the interface every 60 seconds. With this workaround we cut down the
> fault durations to a maximum of 60 seconds, this was ok for us.
> In the meantime we are using 2.5 on most of our servers, and we haven't seen
> this problem any more.

Herbert Wengatz <hwe@uebemc.siemens.de> wrote, that I schould take a look
on the paket size and the collision rate. It could be possible that our
station 20 is on the physical limit of NFS traffic. Herbert prefers a
second ethernet card.

Bob Reardon <bobr@hdss37.SUGAR-LAnd.WireLine.SLB.COM> wrote:
> We had a similar problem when we used to use the Sunlink DNI Decnet
> compatibility software product.
> Are you by any chance using dni ???
No, we are not using dni.

Darryl Gamble <darryl@lsil.com> wrote:
> Hello,
>
> I have almost exactly the same problem only the host that
> keeps "hanging" on me is my main mail server.
>
> It's very frustrating, but I haven't tracked down the problem yet.
> If you get any responses that actually help you out :-),
> I'd love to here them.
This info is very interesting. The haning machine is the mail server and not
a nfs server with a lot of traffic. So I assume that the traffic is not
the problem.

To solve our problem I will do the following:

- installing patch 102001-10
- starting ping -s -I 60 at startup
- reducing the net traffic

Thanks for all your help

  Bye
    Martin

-----------------------------------------------------------------
Martin Greving
Assistent
Fachbereich Angewandte Informatik, FH Rheinland-Pfalz, Abt. Trier
Tel.: 49+0651 8103 374 email: greving@informatik.trier.fh-rpl.de
Fax.: 49+0651 8103 454 http://fsai.trier.fh-rpl.de/~mgreving

Status: RO



This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:10:59 CDT