Re: [squid-users] Squid crash on OpenBSD 5.2

From: Tianyin Xu <tixu_at_cs.ucsd.edu>
Date: Tue, 8 Jan 2013 00:33:58 -0800

The patch looks good.

An inexistant DNS name definitely should not crash the system by
assertion failure. The good behavior is to tolerate the problem and
print out the error log to notify the users. It's better to print out
a message before simply return -1.

Best,
Tianyin

On Mon, Jan 7, 2013 at 11:06 PM, Lo�c BLOT <loic.blot_at_unix-experience.fr> wrote:
> In my case, it seems the ASSERT is thrown when GetAddrInfo look at
> inexistant DNS name. (in the backtrace the DNS name does'nt exists).
> Before there is 2 conditions for IPv4 and IPv6. This function is called
> before any helper i think (Comm::ConnOpener::start), squidGuard doesn't
> show anything about this case in the logs.
> One thing is sure, when i resolve this problem my squid doesn't crash
> anymore, and my clients can work without any problem.
> For asserts, i understand your view. In my case, i prefer to handle code
> rather than force crash program, that's better for stability. I know
> assert means the program comes to an unattended area.
> Since this patch, squid stability is now perfect
> 6104 _squid 2 0 3067M 3070M sleep/1 poll 20:53 0.73%
> squid
>
> 20h uptime, 0 crash, whereas before 2 min only.
> ~45000 requests were treated during uptime.
> --
> Cordialement,
> Lo�c BLOT, UNIX systems, security and network expert
> http://www.unix-experience.fr
>
> Le mardi 08 janvier 2013 � 00:58 +0200, Eliezer Croitoru a �crit :
>> I didn't had much time to look at the back-trace.
>> Also I dont know what is your knowledge is about assert or other things.
>> I dont know about this specific one but.. in most cases it's there for a
>> reason.
>> It's there to make an assessment about the state of the code.
>> Maybe the server is not crashing but just me throwing an idea about
>> maybe some of your clients getting the wrong content.
>>
>> So it's good..
>> If for most users there is no problem there might be one that you have
>> discovered but it might not be the case and you should try to start from
>> ground up and not to just remove it.
>>
>> Have you tried without squidguard at all?
>> maybe there is another bug that we should know about that this assert is
>> about?
>>
>> I hope you will get more guidance about it from one of the core developers.
>>
>> Best regards,
>> Eliezer
>>
>> On 1/7/2013 6:04 PM, Lo�c Blot wrote:
>> > Hello,
>> > at first time, only 10 squidGuard helpers are used. Next i increase the
>> > amount because i thought squid doesn't have so many helpers, to this
>> > limit 150/192.
>> > But it was a big crash. I have posted the crash datas and my fix on
>> > bugzilla (now no crash since the fix). Assert is a bad thing :(.
>> >
>> > http://bugs.squid-cache.org/show_bug.cgi?id=3732
>> >
>>
>>
>

--
Tianyin XU,
http://cseweb.ucsd.edu/~tixu/
Received on Tue Jan 08 2013 - 08:34:08 MST

This archive was generated by hypermail 2.2.0 : Tue Jan 08 2013 - 12:00:03 MST