1.8 server child exit Segmentation Falt

John_Hayward 2010-09-23 13:06:03 UTC #1

After installing the 9-20 version of 1.8 I observe when browsing to various applications (admin, calendar etc) having the browser come back asking me what I want to do with a php file. In looking at the apache2…2.14 log I see:

[Thu Sep 23 07:49:38 2010] [notice] child pid 23180 exit signal Segmentation fault (11)
[Thu Sep 23 07:49:39 2010] [notice] child pid 23179 exit signal Segmentation fault (11)
[Thu Sep 23 07:49:54 2010] [notice] child pid 23181 exit signal Segmentation fault (11)
[Thu Sep 23 07:49:55 2010] [notice] child pid 23182 exit signal Segmentation fault (11)
[Thu Sep 23 07:49:55 2010] [notice] child pid 23184 exit signal Segmentation fault (11)

I have not seen this problem with 1.6.
I have seen this on two different virtual servers (xen).

I do have the issue with JPGraph when running the setup check installation.
Enviornment:
Ubuntu 10.04.1 LTS
PHP/5.3.2-1ubuntu4.5

Anyone seen this?
Any suggestions on how to fix?

johnh…

NoBody 2011-02-04 03:06:05 UTC #2

Yes, I have seen this.

Infact, I’m currently seeing this happen, with the exact same errormessages (segmentation faults in apache) when trying to open some eGW-modules, for example Timesheet or Filemanager…

Running a fresh installation of Ubuntu Server 10.04 with current latest versions of Apache2, PHP5(.3.3) and MySQL5, from the default aptitude-repositories. Installed eGW fresh from the opensuse.org-repository for my Ubuntu-version…

So, to reproduce, my guess would be to ‘just’ install Ubuntu 10.04 on some (virtual) PC (although the problem still seems to exist with 10.10, from what I understood whilst googling this), install default AMP and default eGW. Then just try to open one of those modules.

Too bad no-one has responded yet to this thread. A solution for this would be nice

Hans-Jurgen_Tappe 2011-02-04 09:42:45 UTC #3

Too bad no-one has responded yet to this thread. A solution for this would
be nice

PHP 5.3 is broken. A lot of threads address this, including instructions
for downgrade to PHP 5.2. Just search.

Regards,
Hans-Jürgen

The modern datacenter depends on network connectivity to access resources
and provide services. The best practices for maximizing a physical server’s
connectivity to a physical network are well understood - see how these
rules translate into the virtual world?
http://p.sf.net/sfu/oracle-sfdevnlfb

eGroupWare-users mailing list
eGroupWare-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/egroupware-users

rsteinmetz70112 2011-02-04 19:36:54 UTC #4

NoBody 2011-02-19 21:32:15 UTC #5

Removing APC did the trick for me. Apparently it’s installed by default by Ubuntu Server 10.04.

Thanks for that tip

rsteinmetz70112 2011-02-21 15:45:28 UTC #6

The ultimate all-in-one performance toolkit: Intel® Parallel Studio XE:
Pinpoint memory and threading errors before they happen.
Find and fix more than 250 security defects in the development cycle.
Locate bottlenecks in serial and parallel code that limit performance.
http://p.sf.net/sfu/intel-dev2devfeb

NoBody 2011-02-21 19:30:21 UTC #7

Nope, APC was the only one installed; xcache wasn’t. So there’s either a 3rd accelerator playing along, or APC on it’s own makes eGroupWare quit as well.

It’s however working fine now and since it doesn’t require the best performance, I’m fine with it

rsteinmetz70112 2011-02-21 20:40:37 UTC #8

Index, Search & Analyze Logs and other IT data in Real-Time with Splunk
Collect, index and harness all the fast moving IT data generated by your
applications, servers and devices whether physical, virtual or in the cloud.
Deliver compliance at lower cost and gain new business insights.
Free Software Download: http://p.sf.net/sfu/splunk-dev2dev

Xenopus 2011-02-23 10:29:33 UTC #9

There is more going on.
The segfaults were so severe on my stock 10.10 that it was impractical to use version 1.8 so I had to go back to 1.6. Removing all php caching improved things but was still not acceptable.
Then I saw this post and decided to disable the deflate module in apache. Lo! No more segfaults. I am now running server 1.8 on 64 bit Ubuntu 10.10 with php-5.3.5 and php-apc, have been fiddling with all applications for a few hours and all is well.
Confirm, anyone?

jqueiroz 2011-02-23 16:49:00 UTC #10

I’m having the same problem, Ubuntu server 10.04 32bits, PHP 5.3.2 from repository, eGroupware 1.8 from OpenSuse Deb repository.

Tried the solutions proposed here, nothing worked:

uninstalled php-apc; no other chaching module installed (afaik).
also tried installing php5-xcache, no result;
disabled mod_deflate, still doesn’t resolved the SEGV. But when this module is disabled, the output of the request is shown correctly, dispite the final SEGV error.

I’m accepting suggestions.

rsteinmetz70112 2011-02-23 21:23:05 UTC #11

Free Software Download: Index, Search & Analyze Logs and other IT data in
Real-Time with Splunk. Collect, index and harness all the fast moving IT data
generated by your applications, servers and devices whether physical, virtual
or in the cloud. Deliver compliance at lower cost and gain new business
insights. http://p.sf.net/sfu/splunk-dev2dev

NoBody 2011-02-23 23:46:33 UTC #12

Perhaps that’s the thing that makes the difference. I’m using eGW from the SuSE-repository too, instead of having it manually installed somewhere in /var/www/.

How did you guys install eGW?

NoBody 2011-02-23 23:48:01 UTC #13

Oh darn,

Never mind my last mail; read over it.

jqueiroz 2011-02-26 14:16:08 UTC #14

Well, I ran apache in debug mode over GDB and got this BT:

#0 0xb7893415 in ?? () from /usr/lib/apache2/modules/libphp5.so
#1 0xb7893dfd in gc_collect_cycles () from /usr/lib/apache2/modules/libphp5.so
#2 0xb7874ff5 in ?? () from /usr/lib/apache2/modules/libphp5.so
#3 0xb7819868 in php_request_shutdown () from /usr/lib/apache2/modules/libphp5.so
#4 0xb790681c in ?? () from /usr/lib/apache2/modules/libphp5.so
#5 0xb7fdb031 in ap_run_handler (r=0xb837e3c8) at /build/buildd/apache2-2.2.14/server/config.c:159
#6 0xb7fdee17 in ap_invoke_handler (r=0xb837e3c8) at /build/buildd/apache2-2.2.14/server/config.c:373
#7 0xb7fee728 in ap_process_request (r=0xb837e3c8) at /build/buildd/apache2-2.2.14/modules/http/http_request.c:282
#8 0xb7feb258 in ap_process_http_connection (c=0xb83762f0) at /build/buildd/apache2-2.2.14/modules/http/http_core.c:190
#9 0xb7fe38a1 in ap_run_process_connection (c=0xb83762f0) at /build/buildd/apache2-2.2.14/server/connection.c:43
#10 0xb7ff3bea in child_main (child_num_arg=) at /build/buildd/apache2-2.2.14/server/mpm/prefork/prefork.c:662
#11 0xb7ff3f21 in make_child (s=, slot=0) at /build/buildd/apache2-2.2.14/server/mpm/prefork/prefork.c:702
#12 0xb7ff46af in ap_mpm_run (_pconf=0xb800e0a8, plog=0xb8040170, s=0xb80128e8) at /build/buildd/apache2-2.2.14/server/mpm/prefork/prefork.c:978
#13 0xb7fc5a92 in main (argc=2, argv=0xbffffcc4) at /build/buildd/apache2-2.2.14/server/main.c:742

Seems that php is throwing the segfault after the end of the script. I searched all the code for calls to “gc_collect_cycles”, none found. I have no experience on PHP, but this seem to be a memory leak, or maybe a double-freed pointer.

I’ll try to report this on the PHP5 forums, see if someone can help us.

jqueiroz 2011-03-19 19:36:45 UTC #15

Just refreshing the question,

I opened a bug in Ubuntu about this problem:
https://bugs.launchpad.net/ubuntu/+source/php5/+bug/725812

Ralf_Becker_Stylite 2011-03-20 07:58:34 UTC #16

To my knowledge you need the latest patched version for at least php
5.3.3, afaik only available for Maverick, to get rid of the segfaults.

If you look at the changelog, you can see tons of fixes imported from
Debian 6 (which also works well by now):

http://changelogs.ubuntu.com/changelogs/pool/main/p/php5/php5_5.3.3-1ubuntu9.3/changelog

As long as Ubuntu is not pushing the 5.3.3 version to ealier versions,
the easiest option is probably to update your Ubuntu system.

Ralf

Just refreshing the question,

I opened a bug in Ubuntu about this problem:
https://bugs.launchpad.net/ubuntu/+source/php5/+bug/725812

–
View this message in context: http://egroupware.219119.n3.nabble.com/1-8-server-child-exit-Segmentation-Falt-tp1567870p2703228.html
Sent from the egroupware-users mailing list archive at Nabble.com.

Colocation vs. Managed Hosting
A question and answer guide to determining the best fit
for your organization - today and in the future.
http://p.sf.net/sfu/internap-sfd2d

eGroupWare-users mailing list
eGroupWare-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/egroupware-users

–
Ralf Becker
Director Software Development

Stylite GmbH
[open style of IT]

Morschheimer Strasse 15
67292 Kirchheimbolanden

fon +49 (0) 6352 70629-0
fax +49 (0) 6352 70629-30
mailto: rb@stylite.de

www.stylite.de
www.egroupware.org

Geschäftsführer Andre Keller, Ralf Becker
Registergericht Kaiserslautern HRB 30575
Umsatzsteuer-Id / VAT-Id: DE214280951

Colocation vs. Managed Hosting
A question and answer guide to determining the best fit
for your organization - today and in the future.
http://p.sf.net/sfu/internap-sfd2d

eGroupWare-users mailing list
eGroupWare-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/egroupware-users

jqueiroz 2011-03-20 13:39:40 UTC #17

No, upgrading to Maverick isn’t an option. I cannot afford leaving the security of using a LTS version, to enter a routine of upgrade the server each 6 months.

By the way, the problem is “under control”: disabling the “deflate” module of apache makes the SEGV to rise only after all the output is sent, so the users doesn’t see the problem, just me, when I consult the logs.

This gives me time to wait for a better solution, doesn’t matter if it comes in the form of a fix to PHP 5.3.2, or a backported new version.

rsteinmetz70112 2011-03-20 22:45:35 UTC #18

I think back porting 5.3.3 is appropriate for an LTS version. Perhaps we
can suggest that to Ubuntu. EGW is not the application suffering form
these issues.

would tOn 3/20/2011 8:39 AM, jqueiroz wrote:

No, upgrading to Maverick isn’t an option. I cannot afford leaving the
security of using a LTS version, to enter a routine of upgrade the server
each 6 months.

By the way, the problem is “under control”: disabling the “deflate” module
of apache makes the SEGV to rise only after all the output is sent, so the
users doesn’t see the problem, just me, when I consult the logs.

This gives me time to wait for a better solution, doesn’t matter if it comes
in the form of a fix to PHP 5.3.2, or a backported new version.

–
View this message in context: http://egroupware.219119.n3.nabble.com/1-8-server-child-exit-Segmentation-Falt-tp1567870p2705946.html
Sent from the egroupware-users mailing list archive at Nabble.com.

Colocation vs. Managed Hosting
A question and answer guide to determining the best fit
for your organization - today and in the future.
http://p.sf.net/sfu/internap-sfd2d

eGroupWare-users mailing list
eGroupWare-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/egroupware-users

–
Robert Steinmetz, AIA
Principal
Steinmetz & Associates

direnjie 2013-05-22 01:20:54 UTC #19

deleted -