SUMMARY: Name Server problems

From: James Ashton (jaa101@barton.anu.edu.au)
Date: Wed Oct 28 1992 - 09:09:32 CST


Thanks to this list again - it's great.

THE QUESTION

Environment: mix of SPARC1, 2 and 10 CPUs running 4.1.3 with the bind
patches to /usr/lib/libc.so. There are 3 entries in /etc/resolv.conf
but our machines fail when just the primary nameserver is down.
Nslookup correctly times out and succeeds in using the secondary server
but nothing else does. I've been putting up with this problem for some
time. In the past I've got as far as finding that there was a Sun
patch (XXX465 from memory) but I was never able to find it via archie.
Can anyone confirm the applicability of this patch to my problem and/or
point me towards a method for obtaining it. As always I will
summarise.

THE ANSWER

The quick answer is yes, use 100465-01 and archie _DOES_ find sources
for it (at cs.uni-sb.de, metten.fenk.wau.nl and ugle.unit.no). I
should have looked harder. I've applied the patch and it tests out
well. One person noted that the patch will not help those statically
linked binaries which can't be recompiled.

THE GLORY

per@erix.ericsson.se (Per Hedeland)
montjoy@thor.ece.uc.EDU (Robert Montjoy)
ups!upstage!glenn@fourx.Aus.Sun.COM (Glenn Satchell)
ron@sarah.lerc.nasa.gov (Ron Gaug)
Christopher Davis <ckd@eff.org>
jdavis@noao.edu (Jim Davis)

THE RESPONSES

Date: Tue, 27 Oct 92 19:18:41 +0100
From: per@erix.ericsson.se (Per Hedeland)
To: jaa101@deakin.anu.edu.au
Subject: Re: Name Server problems

Assuming that you just have this problem when the *name server* is down,
but not the host running it (e.g. the named has died, or the host is in
single-user mode but with interfaces configured), then yes, this patch
is - partially - applicable. It consists of a new res_send.o that
replaces the one you picked up from libresolv.a and put in the shared
libc - however, ypserv (might not be of interest to you) and sendmail.mx
(might be) are already linked with (the static) libresolv.a, and so
won't be helped by this patch.

I found it (using archie) on metten.fenk.wau.nl (137.224.129.4),
directory /pub/sun-fixes.

______________________________________________________________________________

Date: 27 Oct 1992 13:05:06 -0500 (EST)
From: montjoy@thor.ece.uc.EDU (Robert Montjoy)
Subject: Re: Name Server problems
To: jaa101@deakin.anu.edu.au

What version of the "binf" patch did you install. Did you
use resolv or bind 4.8.3 or did you use the Sun resolver
routines out of libresolv.a.

If you are not using the resolv patch you may consider trying
it.

______________________________________________________________________________

Date: Tue, 27 Oct 92 19:48:12 EST
From: ups!upstage!glenn@fourx.Aus.Sun.COM (Glenn Satchell)
To: ups!fourx!barton.anu.edu.au!jaa101@fourx.Aus.Sun.COM
Subject: Re: Name Server problems

The readme file for the patch you mention is included below. As you
will see it is only for 4.1 or 4.1.1. However, looking through my files
I can't see it listed as fixed in either 4.1.2 or 4.1.3.

It is not recommended by Sun to install patches from earlier releases,
but as this is only a new .a file for the library you could try it and
see if it helps. Alternatively ask your local Sun answer centre to
build you a patch for 4.1.3 if it's not already fixed.

...

Patch-ID# 100465-01
Keywords: DNS res_send resolv.conf nameserver
Synopsis: SunOs 4.1;4.1.1: DNS res_send() fails if first nameserver in resolv.conf returns ECONNREFUSED
Date: 3/Jan/92
 
SunOS release: 4.1 4.1.1
 
Unbundled Product:
 
Unbundled Release:
 
Topic: DNS patch
 
BugID'd fixed for this patch: 1076977

Architectures for which this patch is available: sun3, sun4

Patches which may conflict with this patch: none

Obsoleted by:

Problem Description:

Bug 1078977
DNS lookup will fail if the first nameserver in /etc/resolv.conf is up but has
no nameserver daemon running. The ECONNREFUSED will be carried on down to the
other nameservers listed in resolv.conf even if they are up and their
nameserver daemons are running.

...

______________________________________________________________________________

From: ron@sarah.lerc.nasa.gov (Ron Gaug)
To: jaa101@deakin.anu.edu.au
Subject: Re: Name Server problems

This patch is available via anonymous ftp from cs.uni-sb.de in the file
/pub/misc/sun/sun-fixes/100400-100499/100465-01.tar.Z

One would think that this bug fix would have been included in 4.1.3.
But I don't see it in the list of fixed bugs.

Here's the readme file from the patch:

...

______________________________________________________________________________

Date: Tue, 27 Oct 1992 00:20:55 -0500
From: Christopher Davis <ckd@eff.org>
To: jaa101@deakin.anu.edu.au (James Ashton)
Subject: Name Server problems

You do want 100465-01. archie sez:

Host cs.uni-sb.de

    Location: /pub/misc/sun/sun-fixes/100400-100499
           FILE -rw-r--r-- 7365 Aug 10 11:18 100465-01.tar.Z

Host metten.fenk.wau.nl

    Location: /pub/sun-fixes
           FILE -rw-rw-r-- 7353 Jun 1 17:46 100465-01.tar.Z

Host ugle.unit.no

    Location: /pub/unix/sun-fixes
           FILE -rw-r--r-- 7353 Jun 1 13:35 100465-01.tar.Z

I think I got it from ugle.unit.no.

______________________________________________________________________________

Date: Mon, 26 Oct 92 19:28:11 MST
From: jdavis@noao.edu (Jim Davis)
To: jaa101@deakin.anu.edu.au
Subject: Re: Name Server problems

Not bad! From a note by Greg Earle on building the DNS-able libc:

   NOTE: You may wish to install Sun patch 100465-01 *first* before taking
   this step, as it provides a newer patched version of /usr/lib/libresolv.a.
   This fixes a serious bug that causes DNS nameserver lookups to fail if the
   first name server query yields ECONNREFUSED (Connection refused), even if
   the other nameservers specified in /etc/resolv.conf are both up & working:

Patch i.d. Bug i.d's SunOS Description
---------- --------- ----- -----------
100465-01 1076977
                        4.1 4.1 4.1.2 DNS res_send() fails if first nameserver
                        in /etc/resolv.conf returns ECONNREFUSED

>Can anyone confirm the applicability of this patch to my problem and/or
>point me towards a method for obtaining it. As always I will
>summarise.

(I know it doesn't say for 4.1.3, but it looks like not many of the 4.1.2
patches made it into 4.1.3 grumble mutter...)

But I believe the resolver code in the bind 4.8.3 distribution has this
failover already. Perhaps you could just build a new libresolv.a from those
bits? I think that's what I did back at the 4.1.1 level, and I don't recall
having any particular problems with building such a library.

______________________________________________________________________________
James Ashton System Administrator
                                             Department of Systems Engineering
Voice +61 6 249 0681 Research School of Physical Sciences and Engineering
FAX +61 6 249 2698 Australian National University
Email James.Ashton@syseng.anu.edu.au GPO Box 4 Canberra ACT 2601 Australia



This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:06:51 CDT