UNIX for Dummies Questions & Answers

We have an application that terminates with segmentation violation errors in the logs. No source code is available since this is a third party software that is way past its maintenance life cycle. Under these circumstances is there a way to force a core dump of the process for further analysis??

Please go easy on me, please... I'm a newbie...

Depending on the OS you are on, you may try to trace the program execution with utilities like truss (Solaris) or strace (Linux).

In this manner you can go deep in the analysis of the process system calls, open files, etc. that may cause segmentation fault.

According to your OS, have a look at man pages and try

SIGSEGV by default forces a core dump. If there is no core file the usual reason is that ulimit for the process coredump value is zero.

Code:

csadev:/home/jmcnama> ulimit -a
time(seconds)        unlimited
file(blocks)         unlimited
data(kbytes)         2015464
stack(kbytes)        256000
memory(kbytes)       unlimited
coredump(blocks)    0

A setuid process may also not want to dump core.

You probably should see a file named "core" in the cwd of the process, which is often the login directory. If this is a ulimit probelm, reset ulimit to allow coredump for the user that runs the process.

The problem is only happening in Prod. Tracing is out of the question.

I did the following:

$ ulimit -c 9999999999 ## set core file size to unlimited
$ kill -s SIGSEGV <PID> ## Is this how a signal is sent to a PID??

The process terminates and no core files are created.

Is it required that the application has to be built/coded specifically to accept signals?

TIA.