Newsgroups: php.internals
Path: news.php.net
Xref: news.php.net php.internals:43518
Mailing-List: contact internals-help@lists.php.net; run by ezmlm
Received-SPF: error (pb1.pair.com: domain yahoo.co.in from 202.43.219.78 cause and error)
DomainKey-Status: good
DomainKey-Signature:a=rsa-sha1; q=dns; c=nofws;
  s=s1024; d=yahoo.co.in;
  h=Message-ID:X-YMail-OSG:Received:X-Mailer:Date:From:Subject:To:MIME-Version:Content-Type;
  b=YfswwB6nHgQzIF2dgIEb8+fqhY1TtlS9OpT8RNGYWjyEr0usStMRyv3QmniWpiefWCy79U4k9wNCkvhgFXD5QRhKsqKYCrmLgfAb5zPY/ZMbgT2hy9C44HtTJR5pqa2oLTh06GX4bND2k3swyHBGUS8uAnornoiYaImQvez7Ag8=;
Message-ID: <1046.56078.qm@web8603.mail.in.yahoo.com>
Date: Fri, 27 Mar 2009 19:01:06 +0530 (IST)
To: internals@lists.php.net
MIME-Version: 1.0
Content-Type: multipart/alternative; boundary="0-1977906537-1238160666=:56078"
Subject: Fw: problem with apache segfault
From: rats1027-phpmaillinglist@yahoo.co.in

--0-1977906537-1238160666=:56078
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: quoted-printable

thank you so much all of you guys, but still we are facing the same problem=
.=0A=0A=0Awe recompiled new php source, upgraded httpd to 2.2.11 but still =
getting same =0Abacktrace. nothing more. we removed .htaccess rules and tes=
ted but no use.=0A=0Acan you guys suggest me any other ways of debugging pl=
s? actually i am not a=0Asystem administration guy. I am a php developer wi=
th little knowledge on httpd.=0ABut we have system administration group who=
 are working on this problem.=0A=0Ajust pls check out again the segfault st=
atement that we are getting:=0A=0Ahttpd[5250]: segfault at 00007fff84114ff8=
 rip 00002aaab1bd47ac rsp 00007fff84115000 error=0A 6=0A=0Athat bold line i=
s repeating many times randomly on our log file and always it is same on =
=0Aspecific server. we have 6 load balanced servers and all are giving the =
same error randomly.=0Aaddress locations are different from each server but=
 for a specific server it is same always =0Aand every day.=0A=0Ayou guys th=
ink any kind of attack thats happening over our systems? actually we do not=
 accept =0Amuch input from users other than a keyword(for search) which can=
 be an unicode and we have some=0APOST requests which are purely browser ge=
nerated data(no user data here).=0A=0Aim again pasting our gdb backtraces o=
f two different core dumps for reference:=0A=0A> #0  0x000055555557dfee in =
ap_merge_per_dir_configs () from /usr/sbin/httpd=0A> No symbol table info a=
vailable.=0A> #1  0x000055555557b121 in ap_directory_walk () from /usr/sbin=
/httpd=0A> No symbol table info=0A available.=0A> #2  0x00005555555765b9 in=
 ap_is_recursion_limit_exceeded () from /usr/sbin/httpd=0A> No symbol table=
 info available.=0A> #3  0x0000555555578b42 in ap_run_map_to_storage () fro=
m /usr/sbin/httpd=0A> No symbol table info available.=0A> #4  0x00005555555=
79cbc in ap_process_request_internal () from /usr/sbin/httpd=0A> No symbol =
table info available.=0A> #5  0x000055555558b668 in ap_process_request () f=
rom /usr/sbin/httpd=0A> No symbol table info available.=0A> #6  0x000055555=
5588900 in ap_register_input_filter () from /usr/sbin/httpd=0A> No symbol t=
able info available.=0A=0Aabove thing is happening many times say 10 to 15 =
times a minute(actualy=0Adepends on the traffic at that time)=0A=0A> #0  0x=
00002aaab1c46688 in ZEND_FETCH_DIM_RW_SPEC_VAR_UNUSED_HANDLER (=0A>     exe=
cute_data=3D0x5555714ea6c8)=0A>     at=0A /usr/local/src/php/php-5.2.9/Zend=
/zend_vm_execute.h:13204=0A>         opline =3D (zend_op *) 0x5555714e8798=
=0A>         free_op1 =3D {var =3D 0x2aaaac1454fd}=0A> #1  0x00005555714e86=
b8 in ?? ()=0A> No symbol table info available.=0A> #2  0x00002aaaac145afb =
in apr_pool_destroy () from /usr/lib64/libapr-1.so.0=0A> No symbol table in=
fo available.=0A> #3  0x000055555556a27b in suck_in_APR () from /usr/sbin/h=
ttpd=0A> No=0A>  symbol table info available.=0A> #4  0x000055555556adf6 in=
 main () from /usr/sbin/httpd=0A> No symbol table info available.=0A=0Aabov=
e thing is happening in very lesser frequency compared to the above.=0A=0A=
=0A--thank you so much guys, rats.=0A=0A=0Ajvlad wrote:=0A>>=0A>>> #2  0x00=
002aaaac145afb in apr_pool_destroy () from =0A>>> /usr/lib64/libapr-1.so.0=
=0A>>> No symbol table info available.=0A>> See http://httpd.apache.org/dev=
/debugging.html on how to grab the=0A>> -win32-symbols.zip package.  Unpack=
 it over httpd and you will probably=0A>> have more legible backtraces.  En=
sure you have all the debugging symbols=0A>> for php you'll have something =
completely legible.=0A> =0A> aha, and /usr/local/src/php/php-5.2.9/Zend/zen=
d_vm_execute.h is a very =0A> common path under Win32 :)=0A=0AHere?  Yea, i=
t is. =0A But looking at the libapr-1.so.0 module, agreed that=0AI've misre=
ad this :)=0A=0A> @Rathnakar Konda, is it possible that php was compiled ag=
ainst wrong Apache =0A> headers? Did you install something like OS-supplied=
 httpd-devel package to =0A> compile php or grabbed apache sources from the=
ir web site?=0A=0AI'm not used to seeing bad stack unwinds on unix, most pe=
ople don't optimize=0Athat aggressively.=0A=0AIf this is system-httpd, that=
's fine; if you installed an httpd package,=0Athere is likely an httpd-deve=
l package that also includes the debugging=0Asymbols you need (unstripped).=
=0A=0AIf not and you've built this, you can simply try -O0, or -g, or for h=
ttpd=0A--enable-maintainer-mode, but offhand I can't think of a trivial way=
 to=0Asimply avoid -strip, which is what you want for your build when thing=
s=0Ago wonky in your gdb where output.=0A=0ABill=0A=0AOriginal E-Mail:=0A> =
Hi Guys,=0A> =0A> We are have a problem with apache segfault on our product=
ion server. Please read bellow for description.=0A> =0A> Its=0A> a web appl=
ication written in php5 and implemented most of the oop=0A> concepts and lo=
t of regular expressions, curl, mcrypt, simplexml,=0A> mssql, exceptions an=
d user defined error handlers. When we run this app=0A> on our test server,=
 we had no problems, but when we moved it on to the=0A> production(here we =
used have big amount of traffic), initially we saw=0A> no problems from our=
 end user testing but from system log, we saw lots=0A> of 'segfaults' and t=
hus requests were being dropped(difference in=0A> traffic).=0A> =0A> Weird =
thing is that, on the=0A same apache httpd, there=0A> is an another applica=
tion running successfully which is having lesser=0A> oop concepts but with =
same libraries. We are running these two=0A> applications with virtual host=
 concept. We see 'segfaults' only when=0A> the traffic is very high on the =
first application.=0A> =0A> We have upgraded our php module from 5.2.6 to 5=
.2.9 but with no result. We have the core dump of the apache below:=0A> =0A=
> #0  0x000055555557dfee in ap_merge_per_dir_configs () from /usr/sbin/http=
d=0A> No symbol table info available.=0A> #1  0x000055555557b121 in ap_dire=
ctory_walk () from /usr/sbin/httpd=0A> No symbol table info available.=0A> =
#2  0x00005555555765b9 in ap_is_recursion_limit_exceeded () from /usr/sbin/=
httpd=0A> No symbol table info available.=0A> #3  0x0000555555578b42 in ap_=
run_map_to_storage () from /usr/sbin/httpd=0A> No symbol table info availab=
le.=0A>=0A #4  0x0000555555579cbc in ap_process_request_internal () from /u=
sr/sbin/httpd=0A> No symbol table info available.=0A> #5  0x000055555558b66=
8 in ap_process_request () from /usr/sbin/httpd=0A> No symbol table info av=
ailable.=0A> #6  0x0000555555588900 in ap_register_input_filter () from /us=
r/sbin/httpd=0A> No symbol table info available.=0A=0A_____________________=
___________=0A Add more friends to your messenger and enjoy! Invite them no=
w.=0A=0A=0A      Add more friends to your messenger and enjoy! Go to http:/=
/messenger.yahoo.com/invite/
--0-1977906537-1238160666=:56078--