Hello,
  
  We have been wanting to use some of the latest features around chunked 
encoding support so we recently went down the path of compiling our own 
mod_wsgi for usage with the pre-compiled apache 2.4 that ships with amazon 
linux. We also use AWS Elastic Beanstalk for our deployments, and for some 
reason we are seeing a single segfault during the deployment process that 
goes away after the first most request. Subsequent restarts of the apache 
process do not re-exhibit this behavior, the only way to watch it is during 
a (re)deploy of code.

  Any hints as to what could be the issue, or better ways to debug than we 
have already employed, would be greatly appreciated.

  Some syslog context around the event:

Mar 12 20:19:46 prod-speech-2-85bb5279 httpd[15427]: [Thu Mar 12 
20:19:46.718654 2015] [mpm_prefork:notice] [pid 15418] AH00163: 
Apache/2.4.10 (Amazon) mod_wsgi/4.4.7 Python/2.7.8 OpenSSL/1.0.1k-fips 
configured -- resuming normal operations
Mar 12 20:19:46 prod-speech-2-85bb5279 httpd[15427]: [Thu Mar 12 
20:19:46.718679 2015] [core:notice] [pid 15418] AH00094: Command line: 
'/usr/sbin/httpd -D FOREGROUND'
Mar 12 20:19:49 prod-speech-2-85bb5279 kernel: [ 3585.331469] traps: 
httpd[15439] general protection ip:7f4c352195d8 sp:7f4c311beb10 error:0 in 
libpython2.7.so.1.0[7f4c3510e000+175000]
Mar 12 20:19:49 prod-speech-2-85bb5279 httpd[15427]: [Thu Mar 12 
20:19:49.319937 2015] [wsgi:error] [pid 15436] [client 127.0.0.1:57494] 
Truncated or oversized response headers received from daemon process 
'wsgi': /opt/python/current/app/application.py
Mar 12 20:19:49 prod-speech-2-85bb5279 httpd[15430]: 127.0.0.1 - - 
[12/Mar/2015:20:19:49 +0000] "GET / HTTP/1.1" 500 527 "-" 
"Python-urllib/2.6"
\Mar 12 20:19:49 prod-speech-2-85bb5279 httpd[15427]: [Thu Mar 12 
20:19:49.721978 2015] [core:notice] [pid 15418] AH00051: child pid 15431 
exit signal Segmentation fault (11), possible coredump in /tmp

  As mentioned in the syslog I was able to get a core dump during this 
issue, but unfortunately it isn't obvious what the issue is. Am I reading 
that there is a problem with select()? That doesn't sound right at all:

[Thread debugging using libthread_db enabled]
Using host libthread_db library "/lib64/libthread_db.so.1".
Core was generated by `(wsgi:wsgi)     -DFOREGROUND'.
Program terminated with signal 11, Segmentation fault.
#0  0x00007f4c3b599923 in select () at ../sysdeps/unix/syscall-template.S:81
81      T_PSEUDO (SYSCALL_SYMBOL, SYSCALL_NAME, SYSCALL_NARGS)

  Here is the script we are running to compile and install mod_wsgi (looks 
pretty straightforward, but maybe there is a hint in here?):

yum erase -y mod24_wsgi*
yum install -y httpd24-devel
wget https://github.com/GrahamDumpleton/mod_wsgi/archive/4.4.7.zip
unzip 4.4.7.zip
cd mod_wsgi-4.4.7
./configure --with-python=/usr/bin/python2.7
make
make install

The versions of apache we are seeing this with:

# rpm -qa | grep httpd
httpd24-tools-2.4.10-1.59.amzn1.x86_64
httpd24-2.4.10-1.59.amzn1.x86_64
httpd24-devel-2.4.10-1.59.amzn1.x86_64
httpd24-debuginfo-2.4.10-1.59.amzn1.x86_64
mod24_ssl-2.4.10-1.59.amzn1.x86_64

Thanks for any help you might have,
Chris Brinker

-- 
You received this message because you are subscribed to the Google Groups 
"modwsgi" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
To post to this group, send email to [email protected].
Visit this group at http://groups.google.com/group/modwsgi.
For more options, visit https://groups.google.com/d/optout.

Reply via email to