JustKernel

Many times its not possible to reproduce a crash that happened in customer environment, talking specifically of kernel or hypervisor or other low level crashes. So few things to keep in mind:

1) Analyze the log trying to find traces that can hint at what may be happening :Core dumps, debug traces

2) Thinking what may be causing this. Software whether kernel or hypervisor is a giant state machine and things can go wrong even if a single bit is set wrongly and this wrong bit can be set by the hardware or some piece of software .

3 Trying to visualize the bigger picture of what may be happening here. Remember the exact point of crash may not show the true picture. It may have been triggered by some other issue.

4) Trying to understand is there something that customer may be trying to do which he is not supposed to do or is not supported.

5) Most importantly, try to reproduce/replicate it at your end. Then you can view the exact software and hardware state in front of your eyes.

Anshul Makkar
mailto: anshul_makkar@justkernel.com