corefile Stack Dump

[Jasc]

I've been having frequent map server crashes, without anyway to pinpoint whats causing the error. I did some research and found that there is something called GDB stack dump using core files.

I am running a Linux VPS currently. I followed some earlier guides but they didn't explain clear enough on how I could generate the core file.

1. yum install gdb to install the container with gdb software

2. I had a file named bt in the root folder

3. Heres where I was stuck, I read in the guide that you have a file named the same as your executable but use extension core. So therefore, map-server_sql I made a copy of it and i renamed it to map-server_sql.core

4. When the map server did crash, I attempted to retrieve some info with it. I typed gdb map-server_sql map-server_sql.core then bt full

however it told me the map-server_sql.core was not valid core file.

100% honest, I wasn't sure where or how to obtain a .core file and Im not sure if I am doing the above steps right. I basically made a copy of map-server_sql in notepad, deleted everything in it, and renamed it using .core extension because I thought thats where it would save the crash.

If anyone can shed any light on where I made my mistake and help me out, that would be great!

Thank you!

Edited December 12, 2011 by Jasc

[quesoph]

sometimes the file is named as " core " with no extension. ( correct me if i'm wrong.)

1. use PuTTY to login to SSH

ulimit -c unlimited

2. Go to your server's folder.

cd trunk

3. go back to SSH and there should be a *core* file in the /trunk/ directory (or whatever your server folder is named)

Depending on your version of Linux, the file might be named <executable>.core or core.<process id>

Ex: map-server_sql.core OR core.30714

Here's how to list files with *core*

ls -al | grep core

4. run GDB and you should get some limited output of the problem

gdb map-server_sql <core filename>

5. If a valid dump was found, you should get a prompt

depending on Linux version it might look like "gdb>" or "(gdb)"

6. At that prompt, type:

gdb> bt full

And that will give you a full output of the problem area and a line number in which you should look for the crash. This just told you where the crash is located and what line you should look at to fix it. If you feel that is an eAthena development problem, feel free to submit a bug with that full report and one of the core developers can look into it.

Edited December 17, 2011 by quesoph