Sorry, it's still behaving the same for me. Failed bac= k to 3.10.11-3 once more.

<= div>--
David P.

On Thu, Oct 10, 2013 at 8:01 PM, Dave Ta= ht <dave.taht@gmail.com> wrote:

3.10.15-4 is now out there, containing sufficient patches to get
dnsmasq to the current head of tree, and including the patch below.

http://snapon.lab.bufferbloat.net/~cero2/cerowrt/wndr/3= .10.15-4/

On Thu, Oct 10, 2013 at 1:23 PM, Simon Kelley <simon@thekelleys.org.uk> wrote:
> Having thought about this more, this patch is necessary
>
> http://the= kelleys.org.uk/gitweb/?p=3Ddnsmasq.git;a=3Dcommit;h=3D8584c502d37627d8abe18= 213771b5f4f98cb4aa3
>
> and should fix the bug iff
>
> 1) Dnsmasq is configured using --except-interface=3D<upstream inter= face> and
> there are no --interface=3D<interface we want to provide service on= > config
> lines.
>
> 2) Exactly one interface that dnsmasq should be listening on is around= when
> it starts, but others arrive later.
>
> I can't explain why it just broke though, this bug has been around= forever.
>
>
>
> Simon.
>
>
>
>
>
> On 10/10/13 19:30, Dave Taht wrote:
>>
>> On Thu, Oct 10, 2013 at 9:54 AM, Simon Kelley<simon@thekelleys.org.uk>
>> wrote:
>>>
>>>
>>> Does reverting
>>>
>>>
>>> h= ttp://thekelleys.org.uk/gitweb/?p=3Ddnsmasq.git;a=3Dcommit;h=3D397542b213ab= 4071734f1cdf4cc914d87100456f
>>>
>>> fix the issue? I fear it might.
>>
>>
>> Seems likely.
>>
>> I reverted that patch and put it in this build
>>
>> http://snapon.lab.bufferbloat.net/~cero2/cerow= rt/wndr/3.10.15-3/
>>
>> I won't be in a position to test stuff myself til sunday but c= ero's
>> devoted userbase seems to be hoovering over the reload button and = will
>> probably beat me to it....
>>
>>
>>>
>>> Cheers,
>>>
>>> Simon.
>>>
>>>
>>>
>>> On 10/10/13 15:43, Dave Taht wrote:
>>>>
>>>>
>>>> Dear Dr. Dnsmasq:
>>>>
>>>> When cerowrt made the jump between dnsmasq-2.67-test10 and=
>>>> dnsmasq-2.67-test17, detection of interfaces other than th= e first
>>>> started failing. It seems to be related to interfaces that= come up
>>>> after dnsmasq starts, as restarting it after the device is= fully
>>>> booted works. Have moved forward to 2.67-rc3 to no avail.<= br> >>>>
>>>> (along the way we migrated from kernel 3.10.11 to 3.10.13 = to 3.10.15
>>>> but I doubt that's the issue)
>>>>
>>>> Hot, fresh, firmware can be had at:
>>>>
>>>> http://snapon.lab.bufferbloat.net/~cero2/cerowrt= /wndr/
>>>>
>>>>
>>>> <knacky> =C2=A0 [10:39:25] has anyone had IPv4 DHCP = problems in the last two
>>>> 3.10.x
>>>> builds? =C2=A0currently running 3.10.11-3 and it works fla= wlessly.
>>>> upgraded to 3.10.13-2 and 3.10.15-1 and both present me wi= th an
>>>> identical issue. =C2=A0upon reboot after upgrading, DHCP l= eases are no
>>>> longer handed out on the wireless interfaces. =C2=A0disabl= ing and
>>>> re-enabling DHCP on the wireless interfaces will fix the p= roblem, but
>>>> the problem
>>>> <knacky> =C2=A0 [10:39:26] returns after a reboot. = =C2=A0disable/reenable DHCP on
>>>> the
>>>> interface will again temporarily fix it.
>>>> <knacky> =C2=A0 [10:42:03] also tried a fresh instal= l (using reset to
>>>> defaults
>>>> option) to avoid anything not properly interpreted from th= e config of
>>>> the previous version, but still get the same issue.
>>>>
>>>>
>>>> On Thu, Oct 10, 2013 at 5:30 AM, David Personette<dperson@gmail.com>
>>>> wrote:
>>>>>
>>>>>
>>>>> Just tested again with 3.10.15-2. My OSX (10.8.5) lapt= op worked,
>>>>> neither
>>>>> my
>>>>> Nexus 7 (2013 w/CM10.2) or my Fedora 19 laptop could r= esolve DNS over
>>>>> wireless. My wired Linux server (Ubuntu 12.04.3) was w= orking fine as
>>>>> well.
>>>>> Reverted to 3.10.11-3 once more.
>>>>>
>>>>> --
>>>>> David P.
>>>>>
>>>>>
>>>>> On Mon, Oct 7, 2013 at 7:40 PM, David Personette<dperson@gmail.com>
>>>>> wrote:
>>>>>>
>>>>>>
>>>>>>
>>>>>> I can confirm it as well. I wiped my config back t= o defaults, and it
>>>>>> wasn't fixed. Reinstalled the 3.10.11-3 build,= and restored my configs
>>>>>> and
>>>>>> all is well.
>>>>>>
>>>>>> --
>>>>>> David P.
>>>>>>
>>>>>>
>>>>>> On Mon, Oct 7, 2013 at 7:07 PM, Fred Stratton<f= redstratton@imap.cc>
>>>>>> wrote:
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> True for the last two builds. Wired works as e= xpected.
>>>>>>>
>>>>>>> Is this a problem with the development version= of DNSMasq? What is
>>>>>>> the
>>>>>>> recommended workaround?
>>>>>>>
>>>>>>> ______________________________________________= _
>>>>>>> Cerowrt-devel mailing list
>>>>>>> Cerowrt-devel@lists.bufferbloat.net
>>>>>>> https://lists.bufferbloat.net/listinfo= /cerowrt-devel
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>
>>>>>
>>>>> _______________________________________________
>>>>> Cerowrt-devel mailing list
>>>>> Cerowrt-devel@lists.bufferbloat.net
>>>>> https://lists.bufferbloat.net/listinfo/cerowrt= -devel
>>>>>
>>>>
>>>>
>>>>
>>>
>>>
>>> _______________________________________________
>>> Dnsmasq-discuss mailing list
>>> Dns= masq-discuss@lists.thekelleys.org.uk
>>> http://lists.thekelleys.org.uk/mailman/list= info/dnsmasq-discuss
>>
>>
>>
>>
>

--
Dave T=C3=A4ht

Fixing bufferbloat with cerowrt: http://www.teklibre.com/cerowrt/subscrib= e.html

_____________________________= __________________
Cerowrt-devel mailing list
Cerowrt-devel@lists.= bufferbloat.net
https://lists.bufferbloat.net/listinfo/cerowrt-devel