NetBSD Problem Report #48379
From www@NetBSD.org Wed Nov 13 17:27:23 2013
Return-Path: <www@NetBSD.org>
Received: from mail.netbsd.org (mail.netbsd.org [149.20.53.66])
(using TLSv1 with cipher DHE-RSA-AES256-SHA (256/256 bits))
(Client CN "mail.NetBSD.org", Issuer "Postmaster NetBSD.org" (verified OK))
by mollari.NetBSD.org (Postfix) with ESMTPS id 10424A6159
for <gnats-bugs@gnats.NetBSD.org>; Wed, 13 Nov 2013 17:27:23 +0000 (UTC)
Message-Id: <20131113172721.8404AA61BB@mollari.NetBSD.org>
Date: Wed, 13 Nov 2013 17:27:21 +0000 (UTC)
From: abs@absd.org
Reply-To: abs@absd.org
To: gnats-bugs@NetBSD.org
Subject: resolv.conf symlink changes not picked up by resolver
X-Send-Pr-Version: www-1.0
>Number: 48379
>Category: bin
>Synopsis: resolv.conf symlink changes not picked up by resolver
>Confidential: no
>Severity: non-critical
>Priority: medium
>Responsible: bin-bug-people
>State: open
>Class: sw-bug
>Submitter-Id: net
>Arrival-Date: Wed Nov 13 17:30:00 +0000 2013
>Last-Modified: Thu Nov 14 00:00:00 +0000 2013
>Originator: David Brownlee
>Release: NetBSD-6.1
>Organization:
>Environment:
NetBSD forsaken.absd.org 6.1_STABLE NetBSD 6.1_STABLE (GENERIC) #0: Thu Oct 24 13:50:23 BST 2013 abs@cerberus.social-events.net:/var/obj/nbbuild-6-amd64/objdir/sys/arch/amd64/compile/GENERIC amd64
>Description:
The system automatically detects changes to /etc/resolv.conf and reloads as required.
This works fine if /etc/resolv.conf is a file and changes, or if /etc/resolv.conf is a symlink and the contents of the file linked to changes, but fails if /etc/resolv.conf is a symlink to one file and changes to a symlink to another file.
>How-To-Repeat:
Assuming working nameserver at 192.168.1.1 and no nameserver at 192.168.99.99
cd /etc
echo 'nameserver 192.168.99.99' > resolv.1
ln -sf resolv.1 resolv.conf
[start app such as firefox, enter url with hostname, confirm it fails]
echo 'nameserver 192.168.1.1' > resolv.2
ln -sf resolv.2 resolv.conf
[try URL in firefox - it should now work, but fails]
touch resolv.1
[try URL in firefox - it now works, presumably as the system is checking for updates to resolv.1]
>Fix:
The system would need to handle the case of symlink specifically, and reload if readlink() changes, or the timestamp of the (final) linked to file changes.
>Audit-Trail:
From: Joerg Sonnenberger <joerg@britannica.bec.de>
To: gnats-bugs@NetBSD.org
Cc:
Subject: Re: bin/48379: resolv.conf symlink changes not picked up by resolver
Date: Wed, 13 Nov 2013 18:34:42 +0100
On Wed, Nov 13, 2013 at 05:30:00PM +0000, abs@absd.org wrote:
> >Synopsis: resolv.conf symlink changes not picked up by resolver
I don't think it is reasonable to expect that to work.
The system would have to monitor not only the symlink, but e.g. all
directories involved. Consider
/etc/resolv.conf -> /a/b
and someone renames a. This is a classic case of "Don't do this" for
me...
Joerg
From: David Brownlee <abs@absd.org>
To: "gnats-bugs@netbsd.org" <gnats-bugs@netbsd.org>
Cc: "gnats-admin@netbsd.org" <gnats-admin@netbsd.org>, netbsd-bugs@netbsd.org
Subject: Re: bin/48379: resolv.conf symlink changes not picked up by resolver
Date: Wed, 13 Nov 2013 17:44:21 +0000
On 13 November 2013 17:35, Joerg Sonnenberger <joerg@britannica.bec.de> wrote:
> The following reply was made to PR bin/48379; it has been noted by GNATS.
>
> > >Synopsis: resolv.conf symlink changes not picked up by resolver
>
> I don't think it is reasonable to expect that to work.
> The system would have to monitor not only the symlink, but e.g. all
> directories involved. Consider
>
> /etc/resolv.conf -> /a/b
I would have thought checking readlink() plus dev+ino would be pretty
comprehensive.
The PR was a followup to a question from Christos. I'd also be quite
happy if the current mechanism and limitations was documented in
resolver(3) or similar - right now from a user perspective its just
magic. Very cool magic, but undocumented and uncertain in the face of
symlinks.
From: Alan Barrett <apb@cequrux.com>
To: gnats-bugs@NetBSD.org
Cc:
Subject: Re: bin/48379: resolv.conf symlink changes not picked up by resolver
Date: Wed, 13 Nov 2013 21:42:05 +0200
On Wed, 13 Nov 2013, Joerg Sonnenberger wrote:
> I don't think it is reasonable to expect that to work.
> The system would have to monitor not only the symlink, but e.g. all
> directories involved. Consider
>
> /etc/resolv.conf -> /a/b
>
> and someone renames a. This is a classic case of "Don't do this" for
> me...
I agree that noticing such changes immediately is unreasonable,
but the resolver code could stat("/etc/resolv.conf") every now and
again and notice changes in st_dev or st_ino. It might as well
look at st_mtime and st_size at the same time, as a safety net in
case kqueue is broken or not available.
--apb (Alan Barrett)
From: "Christos Zoulas" <christos@netbsd.org>
To: gnats-bugs@gnats.NetBSD.org
Cc:
Subject: PR/48379 CVS commit: src/lib/libc/net
Date: Wed, 13 Nov 2013 18:58:51 -0500
Module Name: src
Committed By: christos
Date: Wed Nov 13 23:58:51 UTC 2013
Modified Files:
src/lib/libc/net: resolver.3
Log Message:
PR/48379: mention kqueue functionality
To generate a diff of this commit:
cvs rdiff -u -r1.26 -r1.27 src/lib/libc/net/resolver.3
Please note that diffs are not public domain; they are subject to the
copyright notices on the relevant files.
(Contact us)
$NetBSD: query-full-pr,v 1.39 2013/11/01 18:47:49 spz Exp $
$NetBSD: gnats_config.sh,v 1.8 2006/05/07 09:23:38 tsutsui Exp $
Copyright © 1994-2007
The NetBSD Foundation, Inc. ALL RIGHTS RESERVED.