Messages in this thread | | | From | GEO <> | Subject | Questions and problems regarding kdump | Date | Sat, 19 Apr 2014 00:24:01 +0200 |
| |
Hi,
I have successfully set up kdump to handle oops/panics, however I have a few question regarding the whole process. First of all, I am currently under linux 3.13.0-24 and my /etc/default/kdump- tools contains the following options:
USE_KDUMP=1 KDUMP_SYSCTL="kernel.panic_on_oops=1" KDUMP_FAIL_CMD="reboot -f" KDUMP_CMDLINE_APPEND="maxcpus=1"
The default for KDUMP_CMDLINE_APPEND= seems to be "irqpoll maxcpus=1 nousb", but I removed nousb as I have a system running where the root device is connect over a usb port. What are the risks of removing irqpoll? I removed the option from the file because I seemingly got a kernel panic of the invoked crash kernel when it was called with irqpoll option. The output can be seen in the two attached images (I got to this screen after I invoked a panic with "echo c > /proc/sysrqhandler ", screen turned black for short time, which indicates the other kernel is starting and then received the output as seen).
When the crash kernel is invoked it boots to the Desktop, which seems very unnecessary because it will reboot anyway after dumping the memory. What option would you recommend for the kernel to just dump the memory (network not needed)?
Thanks
[unhandled content-type:image/jpeg][unhandled content-type:image/jpeg] | |