2

I'm trying to set up crash dumps in my Xen domU (pv). I've given the VM 3G memory, that should do. It has plenty of diskspace for / (no separate mounts for other directories).

root@t-51:~# kdump-config show
DUMP_MODE:        kdump
USE_KDUMP:        1
KDUMP_SYSCTL:     kernel.panic_on_oops=1
KDUMP_COREDIR:    /var/crash
crashkernel addr: 0x2f400000
current state:    Not ready to kdump # <--- this seems to be because /sys/kernel/kexec_crash_loaded is 0

kexec command:
  no kexec command recorded
root@t-51:~# cat /sys/kernel/kexec_crash_loaded 
0
root@t-51:~# dmesg | grep crash
[    0.000000] Command line: root=UUID=0d899955-d348-43a3-9341-527293093b5e ro  quiet splash crashkernel=128M@756M $vt_handoff
[    0.000000] Reserving 128MB of memory at 756MB for crashkernel (System RAM: 2999MB)
[    0.000000] Kernel command line: root=UUID=0d899955-d348-43a3-9341-527293093b5e ro  quiet splash crashkernel=128M@756M $vt_handoff

But as you can see from the dmesg log, it did actually reserve memory for crashkernel. Why does it show that it isn't ready?

Jille Timmermans
  • 31

2 Answers2

0

You just need the service to be enabled and then reboot:

systemctl enable kdump-tools

This would make the service to be started at boot time.

Bokrenok
  • 1
0

you should see the logs of the service kdump-tools. sudo service kdump-tools status Then you can identify the problem according to the logs.

I encounter this problem too. First i check the log, and i see the log like this:

Jul 15 16:38:15 generic-2-clone systemd[1]: Starting Kernel crash dump 
capture service...
Jul 15 16:38:16 generic-2-clone kdump-tools[1625]: Starting kdump-tools: 
Could not find a free area of memory of 0x6baf000 bytes...
Jul 15 16:38:16 generic-2-clone kdump-tools[1625]: locate_hole failed
Jul 15 16:38:16 generic-2-clone kdump-tools[1625]:  * failed to load kdump 
kernel.

from the logs, i can guess the reserved memory is not enough for kdump. then i modify the /etc/default/grub.d/kexec-tools.config, change the reserved memory from 128M to 256M. and then coldreboot.

now, it's ok.

peasantspring
  • 31
  • 2