Question

2

Kernel dump using linux-crashdump under VMware?

rated 0 times [ 2] [ 0] / answers: 1 / hits: 1455 / 2 Years ago, sat, september 10, 2022, 11:47:59

I've followed the directions here on the Ubuntu help wiki for my 12.04.2 Ubuntu Server running in VMware Fusion 5, and everything looks good, so I do a echo c > /proc/sysrq-trigger as root. It panics the kernel as expected, and then hangs after printing out a bunch of stack traces.

I waited a few minutes to make sure any writes to disk (for the crash dump) finished, and I restarted the VM. When I log in again, ls /var/crash shows no files! Any idea how to get it to work, or whether it works at all?

Answers

Only authorized users can answer the question. Please sign in first, or register a free account.

oneeme

Add To Favorites

Follow

Total Points: 340

Total Questions: 110

Total Answers: 133

Location: Ghana

Member since Fri, Oct 21, 2022

2 Years ago

answered 2 Years ago uquelighted · Accepted Answer

The reason why this doesn't work properly has been well documented for over three or so years now: see this page of the Ubuntu wiki

On systems with <= 2G RAM we won't reserve enough memory for the crash kernel's initrd to fit. That will cause the crash-kernel to panic.

See also these bug reports for reference #785394 and #1235616

Workaround:

To fix this issue set

GRUB_CMDLINE_LINUX="crashkernel=128M"

in /etc/default/grub. n.b.: You cannot put this option into GRUB_CMDLINE_DEFAULT, because the the setting in /etc/grub.d/10_linux takes precedence over that.

To "permanently" fix the issue, edit `/etc/grub.d/10_linux and change the line

GRUB_CMDLINE_EXTRA="$GRUB_CMDLINE_EXTRA crashkernel=384M-2G:64M,2G-:128M"

to

GRUB_CMDLINE_EXTRA="$GRUB_CMDLINE_EXTRA crashkernel=128M"