Re: fqdncache_purgelru causes fatal_dump

From: Amos Jeffries <squid3_at_treenet.co.nz>
Date: Thu, 28 Aug 2008 02:09:15 +1200

David Meyer wrote:
> Hi Amos
>
> Thanks for your reply.
> It's 2.6.Stable6 running on CentOS 5.2
> Just on a side note.When these core dumps occur the squid proccess never
> dies or crashes, but the core dumps is huge.
> Sometimes it gets to about 1.5 gig at times.
> Below is the output before the backtrace.Could u please explain what u
> mean by "trace by line"
> Thanks
> David

I said "trace with line numbers", look like so:

#2 0x006ef209 in abort () from /lib/tls/libc.so.6
#3 0x0806b987 in xassert (msg=Could not find the frame base for "xassert".
) at debug.c:514
#4 0x0808170b in httpBuildRequestHeader (request=0x10791f40,
orig_request=0x10791f40, entry=0xfc6ba30, hdr_out=0xbfed04f0, flags=
       {proxying = 0, keepalive = 1, only_if_cached = 0,
keepalive_broken = 0,
abuse_detected = 0, request_sent = 0, front_end_https = 0, originpeer =
0}) at
http.c:1195

... note the debug.cc:514 and http.c:1195, which tells us its the code
on that line breaking. And the fields saying what data Squid is playing
with that killed it.
Your trace only gets as far as the function name. Which is helpful, but
a step down from easily checkable.

PS. please reply-all, as squid-dev gets horribly cut from the reply line
by default.

Amos

>
> -----Original Message-----
> From: Amos Jeffries [mailto:squid3_at_treenet.co.nz]
> Sent: Wednesday, August 27, 2008 3:07 PM
> To: David Meyer
> Cc: squid-dev_at_squid-cache.org
> Subject: Re: fqdncache_purgelru causes fatal_dump
>
> puppy wrote:
>> Hi we keep getting huge core dumps in the squid spool directory.
>> Squid version is 2.6 Stable.
>> Did a backtrace of the core dumps using gdb ,below is the result from
>> the backtrace on all the core dumps.
>>
>> #0 0x0000003b33230155 in raise () from /lib64/libc.so.6
>> #1 0x0000003b33231bf0 in abort () from /lib64/libc.so.6
>> #2 0x0000000000473c2f in fatal_dump ()
>> #3 0x0000000000473f99 in fatal_dump ()
>> #4 0x000000000043efe6 in fqdncache_purgelru ()
>> #5 0x0000000000492ad7 in authSchemeSetup_ntlm ()
>> #6 0x000000000040d4a9 in ?? ()
>> #7 0x000000000042438c in drand48_at_plt ()
>> #8 0x000000000042bb6a in drand48_at_plt ()
>> #9 0x000000000042bf9f in drand48_at_plt () #10 0x000000000044f997 in
>> ipcache_purgelru ()
>> #11 0x0000003b3321d8b4 in __libc_start_main () from /lib64/libc.so.6
>> #12 0x00000000004098d9 in ?? ()
>> #13 0x00007ffff55aa398 in ?? ()
>> #14 0x0000000000000000 in ?? ()
>>
>> Can anyone please shed some light or explain how to fix this.
>> Thanks
>> David
>
> Which stable release of 2.6?
> Is there a trace with line numbers?
>
> Amos
> --
> Please use Squid 2.7.STABLE4 or 3.0.STABLE8
>

-- 
Please use Squid 2.7.STABLE4 or 3.0.STABLE8
Received on Wed Aug 27 2008 - 14:09:14 MDT

This archive was generated by hypermail 2.2.0 : Wed Aug 27 2008 - 12:00:06 MDT