Operating System - HP-UX
1850214 Members
2416 Online
104050 Solutions
New Discussion

Re: Force memory dump on a live system

 
SOLVED
Go to solution
Jerry Zhang
Frequent Advisor

Force memory dump on a live system

Is there a way to force a memory dump on a live system without rebooting?
(HP-UX 11i)
5 REPLIES 5
Jeff Schussele
Honored Contributor
Solution

Re: Force memory dump on a live system

Hi Jerry,

What are trying to dump?

A single process or the whole system?

If it's a process, you can core dump it with:

kill -3 PID

If you want a complete system dump then get into the GSP & issue the TC (Transfer Control) command. This will reboot the system & produce a crash dump. Make sure you have sufficient dump space to receive it & space to write it to on reboot. Run
crashconf -v
to review the crash dump configuration.
Crash dump options are configured in /etc/rc.config.d/crashconf, the dump device is defined in /etc/fstab & looks like
/dev/vg01/lvol1 - dump defaults 0 0

And no, there's no way to get a full crash dump w/o rebooting.

HTH,
Jeff

PERSEVERANCE -- Remember, whatever does not kill you only makes you stronger!
Jerry Zhang
Frequent Advisor

Re: Force memory dump on a live system

Thanks for your info. The dump I wanted is for whole system. Is there a way to do it without reboot?
doug mielke
Respected Contributor

Re: Force memory dump on a live system

Cauld you do a dd from /dev/mem /kmem etc?
doug mielke
Respected Contributor

Re: Force memory dump on a live system

And there are ways to induce a panic so that you can get a core dump, but there would be a reboot after.

With memory content constantly and quickly changing, it would be tough to get a snapshot of it without a system halt.

I'm curious what you need it for.
Kent Ostby
Honored Contributor

Re: Force memory dump on a live system

Jerry --

You can run some of the dump tools on a live system.

If you have q4 installed on your system (which is default for 11.x), you can do:

q4 -p /stand/vmunix /dev/kmem

At the q4 prompt you can use commands that are usually used for reading crash dump:

include analyze.pl
run Analyze AU > ana.out
include whathappened.pl
run WhatHappened > what.out
quit

Because you are running against something dynamic, it will not have 100% accurate information, but depending on what you are looking for, this could give you some valueable data while leaving the system up.

In general, the ana.out is going to help you more on a per-process point of view. The what.out would give you more of an idea on the system is running in general.
"Well, actually, she is a rocket scientist" -- Steve Martin in "Roxanne"