Report information
The Basics
Id:
19430
Status:
resolved
Estimated:
16 hours (960 minutes)
Left:
16 hours (960 minutes)
Priority:
Low/Low
Queue:

BugTracker
Version Fixed:
(no value)
Version Found:
(no value)
Versions Affected:
(no value)
Versions Planned:
4.3.6
Priority:
P2 Normal
Severity:
S2 Normal
CVSS Score:
(no value)
CVE ID:
(no value)
Component:
(no value)
Area:
(no value)

Dates
Created:Wed, 04 Mar 2009 03:38:30 -0500
Updated:Wed, 28 Jun 2017 10:20:11 -0400
Closed:Wed, 28 Jun 2017 10:20:11 -0400



This bug tracker is no longer active.

Please go to our Gitlab to submit issues (both feature requests and bug reports) for active projects maintained by Internet Systems Consortium (ISC).

Due to security and confidentiality requirements, full access is limited to the primary maintainers.

CC: Andrey Borzenkov <arvidjaar@mail.ru>, oblin@mandriva.com
Subject: dhclient-script bug
Date: Wed, 4 Mar 2009 05:38:11 -0300
To: dhcp-bugs@isc.org
From: Gustavo De Nardin <gustavodn@mandriva.com>
Hello. dhclient-script (Linux version) puts the interface down on some circunstances, to flush caches and routes. At least some wireless drivers lose association when the interface is put down. Also, at least some (the same) drivers fail to signal deassociation when the interface is put down. So we get to Mandriva bug 43441. Please see comment 13 and 41 here: <https://qa.mandriva.com/show_bug.cgi?id=43441#c13>, <https://qa.mandriva.com/show_bug.cgi?id=43441#c41> As stated on #c41, Andrey Borzenkov pointed out "ifconfig inet 0" is enough. I also found Debian (+ Ubuntu) has this same fix since 2005 on their own script (<http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=323254>) which should demonstrate it works. If you don't see problems with this, I believe this should be fixed in your codebase. Little patch (so far without the "ifconfig up" fix, present in Debian's) here: <http://svn.mandriva.com/cgi-bin/viewvc.cgi/packages/cooker/dhcp/current/SOURCES/dhcp-4.1.0-prevent_wireless_deassociation.patch?view=log> Cc'ing our dhcp maintainer and Andrey, for refs. Please keep the Cc list on replies unless stated otherwise. Thanks.
Subject: Re: [ISC-Bugs #19430] dhclient-script bug
Date: Wed, 4 Mar 2009 09:39:50 -0800
To: Gustavo De Nardin via RT <dhcp-bugs@isc.org>
From: "David W. Hankins" <David_Hankins@isc.org>
On Wed, Mar 04, 2009 at 08:38:32AM +0000, Gustavo De Nardin via RT wrote: > As stated on #c41, Andrey Borzenkov pointed out "ifconfig inet 0" is enough. > I also found Debian (+ Ubuntu) has this same fix since 2005 on their own > script (<http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=323254>) which > should demonstrate it works. It might be a good time to rework the stock linux dhclient-script to use '/sbin/ip' for all interface configuration, and stop using ifconfig entirely (rather than both as we currently do for IPv6 and IPv4 separately). -- David W. Hankins "If you don't do it right the first time, Software Engineer you'll just have to do it again." Internet Systems Consortium, Inc. -- Jack T. Hankins
On Mon Jun 19 19:47:24 2017, tmark wrote: > The bug fixes look fine. Ideally, it would be better to do them > separately. I have moved the code relevant to 18183 to rt18183 branch. There is also rt19430a branch that contains only the fixes related to 19430 (+ 2 minor nits), but no test framework. The test framework will be handled in 45434. In summary: please review rt19430a and ignore (but not delete) rt19430. Test related commits from it will be cherry-picked later. > Did you do any manual testing the client under linux? Yes, but my tests were not extensive. I disabled ubuntu interface management and started the compiled dhclient with update linux script. It got IPv4 address properly and configured /etc/resolv.conf. I also ran the unit-tests that are available on rt19430. > Release notes: > > 1. When you add new entries, you append them to the end of the Changes > <version> section, not add them to the beginning, > It's opposite of the way we do it in Kea's ChangeLog. > > 2. You did thank Mr.Pollock, but normally we reply to the requester > explaining that we've addressed their issue and then ask them if they > want to be cited, and if so how do they want to appear. > > (BTW, both of these are discussed in the developer's survival guide ;) Where is the latest version of it available? I have mine from the last vodoo sessi^H^H^H^H^H dhcpd overview training you did couple months ago. > The rest of the comments have to do with the test framework. First, > thanks for doing this, it should come in handy. > > 1. How much work would it be to move the test framework under it's own > ticket and branch? Might be easier to merge into 4.3.6/4.1-ESV-R15 on > it's own and I am also undecided as to whether those branches should > get it. Not that much, as long as we maintain proper order of things. I have split the code. So far 18183 (branch rt18183) and 19430 (branch rt19430a) are ready for review. Once both are merged in, we could branch rt45434 and cherry-pick changes from the old rt19430. I have the list of commits ready, so should be easy as long as both fixes from 18183 and 19430 are on master. Rest of the comments ignored here as no longer relevant. They'll be addressed in 45434.