This is the mail archive of the
libc-help@sourceware.org
mailing list for the glibc project.
Re: Domains aliased to a pool of IP addresses doesn't get resolved
- From: Florian Weimer <fweimer at redhat dot com>
- To: Yuri Kanivetsky <yuri dot kanivetsky at gmail dot com>
- Cc: libc-help at sourceware dot org
- Date: Sun, 18 Oct 2015 13:06:25 +0200
- Subject: Re: Domains aliased to a pool of IP addresses doesn't get resolved
- Authentication-results: sourceware.org; auth=none
- References: <CAMhVC3YY6X+=_dhRbHkw+=m_exfD0yva1mLP78dq8wy47wT3tg at mail dot gmail dot com> <561B9C19 dot 2000906 at redhat dot com> <CAMhVC3bm7K7MXn-BsEPHq+SpVDyxrV5CJgbAVnfcC5nr7bYoqA at mail dot gmail dot com> <561F840B dot 2060802 at redhat dot com> <CAMhVC3ajOFhdA21F12mrLOgeJDtHx+DxzAixDF2OAEz6TSGz5g at mail dot gmail dot com> <5620194C dot 3090200 at redhat dot com> <CAMhVC3bKfeRmOSdKP3mz4Yu9NcyiLjwyz5ebWjibVka-UsY+3g at mail dot gmail dot com> <5620F6C4 dot 5090208 at redhat dot com> <CAMhVC3aoUT3-Lny0AAOWYDOJB7ZEiDS-Z+aJQcLTPG+EZYFkVw at mail dot gmail dot com>
On 10/17/2015 12:23 AM, Yuri Kanivetsky wrote:
>> But the resolver refuses the TCP connection. This is a resolver bug,
>> not a glibc bug, and even if we address the issue below, name resolution
>> would still fail.
>
> Can you explain what exactly is resolver.
The name server you configured for this system, with the IP address
10.0.2.3.
>> I will try to reproduce this locally and file a bug if necessary.
>
> Let me describe how to reproduce the issue. Install vagrant and
> virtualbox, run `vagrant init` in an empty dir, change in just created
> Vagrantfile file config.vm.box to ubuntu/vivid64. Then run `vagrant up
> && vagrant ssh`, and now you're in just created and started VM. After
> that, create the test program file (quoting it here):
It's a problem with your name server, I won't be able to reproduce this
with my local resolver because it does not show this behavior. The
failure is intermittent because in some cases, the initial UDP response
fits into 512 bytes, so no truncation occurs, no TCP fallback, and no
TCP failure, and the name resolution succeeds based on the UDP response
alone.
The search path issue I noted is mostly unrelated to your actual issue,
I just spotted it in the strace output. I failed this bug:
<https://sourceware.org/bugzilla/show_bug.cgi?id=19148>
Fixing it will *not* fix your original name resolution problem.
Florian