Bug#557003: libc6: DNS queries are extremely slowed by ipv6
On Thu, Nov 19, 2009 at 11:42 PM, Pedro Ribeiro <pedrib@gmail.com> wrote:
> On Thu, Nov 19, 2009 at 1:04 PM, Aurelien Jarno <aurelien@aurel32.net> wrote:
>> Pedro Ribeiro a écrit :
>>> On Thu, Nov 19, 2009 at 12:03 AM, Aurelien Jarno <aurelien@aurel32.net> wrote:
>>>> On Wed, Nov 18, 2009 at 09:40:03PM +0000, Pedro Ribeiro wrote:
>>>>> Package: libc6
>>>>> Version: 2.10.1-7
>>>>> Severity: important
>>>>>
>>>>> When using Firefox, Google Chrome, Opera, Synaptic and other applications which use DNS, I always have to wait a long time (2 to 10 seconds) for a host to resolve (i.e. messages like "Resolving host", "Looking up http://whatever";, etc).
>>>>>
>>>>> I always thought that this was a problem of my internet connection.
>>>>> I could never track the origin of it until finally like 2 hours ago I stumbled across this:
>>>>> https://bugs.launchpad.net/ubuntu/+source/glibc/+bug/417757
>>>>>
>>>>> To test if I had this problem, I opened up Firefox and got to about:config, entered ipv6 and set "network.dns.disableIPv6" to false. Immediately, every DNS query was extremely fast and browsing was up to par to my high speed internet connection.
>>>>> To do a definite test, I added "ipv6_disable=1" to my kernel command line and after that all the applications above were having much faster DNS queries.
>>>>>
>>>>> This is a rather serious bug that I have been experiencing ever since upgrading to testing (right after lenny's release), and always blamed on my ISP/wireless connection/configuration/etc, while finding it strange that every Mac or Windows PC near me using the same connection appeared to be much faster on DNS queries.
>>>>>
>>>>> I guess there must be much more users affected by this and like me, they have no clue why.
>>>>>
>>>>> It may be caused by a recent version of libc6, because the bug above is marked as "karmic regression" (the latest ubuntu is the karmic one, 9.10) and a friend of mine who is using lenny doesn't appear to have the same problem.
>>>>>
>>>>> Pardon me if this as already been reported, I searched around but could not find a Debian version of this bug.
>>>>>
>>>> Does adding "options single-request" to /etc/resolv.conf fixes your
>>>> problems?
>>>>
>>>> If so, it is due to a broken DNS server on your ISP side, and a bug
>>>> in Firefox and other application which explicitely ask to resolve
>>>> IPv6 adresses by not passing AI_ADDRCONFIG to getaddrinfo().
>>>>
>>>> --
>>>> Aurelien Jarno GPG: 1024D/F1BCDB73
>>>> aurelien@aurel32.net http://www.aurel32.net
>>>>
>>>
>>> Hi,
>>>
>>> you are right! It does solve my problem. But the strange thing is that
>>> I moved to another country in the last 2 months, but I've had this
>>> problem in my previous country also. So I guess its an application and
>>> not a DNS problem?
>>
>> Are you using the same modem/router and using it as a DNS server?
>>
>>> What do you think I should do, file a bug with all the respective
>>> applications, or just wait for the Ubuntu bug to go upstream / sort
>>> itself out?
>>
>> There is no real plan to fix it. Maybe implement one more workaround if
>> we found how broken is your DNS server.
>>
>> --
>> Aurelien Jarno GPG: 1024D/F1BCDB73
>> aurelien@aurel32.net http://www.aurel32.net
>>
>
> Well to be honest I don't think its my DNS server. I'm living right
> now on (lets call it that way) "Home 3", which is in country B. I'm
> living in a university campus (University Y), so I'm using a
> campus-wide network.
> In country A, about 2000km from here, I had the same problem on "Home
> 1", "Home 2" and in my previous university (University X), with
> different ISP's in each home and university. I may have had the same
> problem on different networks, but I'm not confident enough to affirm
> it.
>
> So if it is the DNS server, the each DNS server at each location was
> broken - which of course is not impossible.
>
> (apologies for the A,X,1 thing, its just for privacy)
>
> Pedro
>
Hi,
I reinstalled my system (upgraded to amd64) and the problem seems to be gone.
It may be that my old system was rotting, or the problem is i386
specific. Either way, since nobody else complained, I'm closing this
bug.
Thank you for your patience and help.
Reply to: