perl-modperl mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Jie Gao <>
Subject RE: mod_perl Silent Failure, Very Mysterious
Date Wed, 16 Nov 2016 09:15:37 GMT
For the completeness of issue reporting, please run the following script and post the output:

use strict;
use warnings FATAL => 'all';

use ModPerl::TestReport;

From: William N. Braswell, Jr. []
Sent: Tuesday, November 15, 2016 8:01 PM
To: Jie Gao
Subject: Re: mod_perl Silent Failure, Very Mysterious

Hello Jie & Friends,

I got the core dump out of gdb and also some hopefully-helpful output.

Here is the output from gdb (not the core dump which is binary of course):

Then again maybe we have a problem because I'm missing debugging symbols inside /usr/sbin/apache2?

Program received signal SIGSEGV, Segmentation fault.    0x00007fffeed67a00 in ?? ()

Note the ?? part in the output above...

Where do I go from here?

~ Will

On 11/15/2016 at 12:14 AM, "William N. Braswell, Jr." <>
Hi Jie,

I have captured the output of `strace -f /usr/sbin/apache2 -e trace8 -k start`:

Looks like it is dying with a segfault, but I'm not sure what is causing this???

Thanks so much for your help!!!  :-)

~ Will


[pid  6550] stat("blib/lib/DynaLoader.pmc", 0x7ffdda87f280) = -1 ENOENT (No such file or directory)
[pid  6550] stat("blib/lib/", 0x7ffdda87f1b0) = -1 ENOENT (No such file or directory)
[pid  6550] stat("/etc/perl/DynaLoader.pmc", 0x7ffdda87f280) = -1 ENOENT (No such file or
[pid  6550] stat("/etc/perl/", 0x7ffdda87f1b0) = -1 ENOENT (No such file or directory)
[pid  6550] stat("/usr/local/lib/x86_64-linux-gnu/perl/5.22.1/DynaLoader.pmc", 0x7ffdda87f280)
= -1 ENOENT (No such file or directory)
[pid  6550] stat("/usr/local/lib/x86_64-linux-gnu/perl/5.22.1/", 0x7ffdda87f1b0)
= -1 ENOENT (No such file or directory)
[pid  6550] stat("/usr/local/share/perl/5.22.1/DynaLoader.pmc", 0x7ffdda87f280) = -1 ENOENT
(No such file or directory)
[pid  6550] stat("/usr/local/share/perl/5.22.1/", 0x7ffdda87f1b0) = -1 ENOENT
(No such file or directory)
[pid  6550] stat("/usr/lib/x86_64-linux-gnu/perl5/5.22/DynaLoader.pmc", 0x7ffdda87f280) =
-1 ENOENT (No such file or directory)
[pid  6550] stat("/usr/lib/x86_64-linux-gnu/perl5/5.22/", 0x7ffdda87f1b0) = -1
ENOENT (No such file or directory)
[pid  6550] stat("/usr/share/perl5/DynaLoader.pmc", 0x7ffdda87f280) = -1 ENOENT (No such file
or directory)
[pid  6550] stat("/usr/share/perl5/", 0x7ffdda87f1b0) = -1 ENOENT (No such file
or directory)
[pid  6550] stat("/usr/lib/x86_64-linux-gnu/perl/5.22/DynaLoader.pmc", 0x7ffdda87f280) = -1
ENOENT (No such file or directory)
[pid  6550] stat("/usr/lib/x86_64-linux-gnu/perl/5.22/", {st_mode=S_IFREG|0644,
st_size=10748, ...}) = 0
[pid  6550] open("/usr/lib/x86_64-linux-gnu/perl/5.22/", O_RDONLY) = 5
[pid  6550] ioctl(5, TCGETS, 0x7ffdda87ef50) = -1 ENOTTY (Inappropriate ioctl for device)
[pid  6550] lseek(5, 0, SEEK_CUR)       = 0
[pid  6550] read(5, "# Generated from DynaLoader_pm.P"..., 8192) = 8192
[pid  6550] --- SIGSEGV {si_signo=SIGSEGV, si_code=SEGV_MAPERR, si_addr=0x7f491064fa00} ---
[pid  6550] chdir("/etc/apache2")       = 0
[pid  6550] rt_sigaction(SIGSEGV, {SIG_DFL, [], SA_RESTORER|SA_INTERRUPT, 0x7f4918dea3d0},
{SIG_DFL, [], SA_RESTORER|SA_RESETHAND, 0x7f4918dea3d0}, 8) = 0
[pid  6550] kill(6550, SIGSEGV)         = 0
[pid  6550] rt_sigreturn({mask=[]})     = 139951873083648
[pid  6550] --- SIGSEGV {si_signo=SIGSEGV, si_code=SI_USER, si_pid=6550, si_uid=0} ---
[pid  6549] +++ exited with 0 +++
+++ killed by SIGSEGV (core dumped) +++


On 11/14/2016 at 11:54 PM, "Jie Gao" <> wrote:
Add the "-f" flag to strace and see what errors you get.



* William N. Braswell, Jr. <> wrote:

> Date: Mon, 14 Nov 2016 23:27:53 -0600
> From: "William N. Braswell, Jr." <>
> To:
> Subject: mod_perl Silent Failure, Very Mysterious
> Hello everyone, I'm trying to enable a web app (Catalyst ShinyCMS) to
> run under mod_perl, it is definitely supposed to work but it fails
> without any errors, and I've been trying for several days to get it
> working.
> My system info:
> * Perl 5, version 22, subversion 1 (v5.22.1) built for
> x86_64-linux-gnu-thread-multi
> * libapache2-mod-perl2 v2.0.9-4ubuntu1
> * Apache/2.4.18 (Ubuntu)
> * $Apache::Test::VERSION = 1.39
> * $Catalyst::VERSION = 5.90112
> * $ShinyCMS::VERSION = 0.007
> I know the app already functions properly when running via the
> stand-alone test server script (PSGI) or fastcgi, and I am able to run
> a trivial Catalyst test application successfully under mod_perl, so
> this issue seems to have something to do with the interaction between
> the ShinyCMS app and mod_perl.
> I have added print debug statements in my app, so when I call my app
> from within the Apache site conf file then I can see the debug
> statements prove that it is successfully going all the way through my
> app, and it even returns control to the embedded code w/in the Apache
> site conf file, but it still just silently fails anyway with the
> following non-descriptive error:
> "The apache2 instance did not start within 20 seconds. Please read the
> log files to discover problems"
> Unfortunately the log files do not provide any additional information
> whatsoever!
> Here are my Apache config files and the output of trying to start
> Apache manually, you can see my ">" statements which prove that my
> ShinyCMS app code is definitely executing all the way through and
> returning control to Apache:
> This is a really mysterious issue, when I run `apache2ctl -e debug -k
> start` as seen in the link above, it goes through the motions as if it
> is starting correctly, and it does not give any error messages, but
> when apache2ctl is done there simply are not any apache processes
> running...
> I only found 1 person on google with a possibly-similar issue, but he
> told me that he never found a solution:
> I also found 1 person on the mod_perl mailing list archives with
> another possibly-similar issue, but I have no way to follow up, and
> I'm not sure of the best way to totally rebuild everything from
> source:
> I have asked for help on IRC #httpd, user Unbeliever suggested I
> ascertain my real /usr/sbin/apache2 command line and run it through
> strace, then post to this mailing list.
> I used -x in apache2ctl which showed me I was simply sourcing
> /tmp/apache2/envvars and then running `/usr/sbin/apache2 -e trace8 -k
> start` which are the exact same args as accepted by the apache2ctl
> script.
> I have now run `strace /usr/sbin/apache2 -e trace8 -k start` which
> generates 2.5MB of text output, I have pasted the last few pages of it
> on hastebin, I can post the whole thing for download if needed:
> I am not sure how to interpret the results of the strace command
> output and debug this issue, help please?
> Any and all assistance is greatly appreciated! :-)
> Thanks,~ Will

View raw message