Archives

All posts for the day December 10th, 2011

Linux core dumps and how to determine what caused.

Posted by phatdee on December 10, 2011
Posted in: Gdb. Tagged: , , , , . Leave a Comment

I am by no means an expert in this, but here is how to determine what caused the core dump.

A core dump looks like the following.

core.38240

The number usually is the PID of the program that crashed, however most of the time this hasn’t been very helpful..at least for me.

First figure out what program caused the issue, normally I just use some binary that most likely didn’t cause the issue, like here I use /bin/ps. Usually the last section will say, the binary your specifying doesn’t match the core file which is whatever program. =) In this example, /usr/sbin/ndsd.

Once we got the correct program, exit from this and run it again with the correct binary.

more to come…

Solaris /tmp is tmpfs, what happens when someone copies large file into /tmp and consumes all available memory,

Posted by phatdee on December 10, 2011
Posted in: Solaris. Tagged: , , . Leave a Comment

Found on Unix.com; Author Admin_XOR

This is a classic example of what happens when you run out of system resources like memory. The shell cannot create another process ( fork ) and thus you are restricted to the built-in commands provided by the shell you are using.

You copied the large file in /tmp which is tmpfs and resides on virtual memory. Dumping files in /tmp means consuming virtual memory. In turn, less room for other processes.

All of the commands are shell built-in (work in bourne and korn shells) and shell does not need to fork another process. So, cd lets you in the /tmp directory, echo lets you see the files in that and figure out the name of the file, > brings the size of the file down to zero solving your problem.