Have you encountered issues collecting vmcores on kernel panic?

Latest response

 

Installing and configuring a mechanism to capture kernel memory dumps (vmcores) is critical when trying to solve kernel panic (oops) issues.
 
The knowledge solution documented here outlines the basics involved with ensuring that kdump is configured and functioning.
 
But our support engineers have seen many cases where users who follow these steps are still unable to successfully capture a vmcore on kernel panic. So, a few questions:
  • Have you encountered any situations in your environment that prevented your RHEL 5 or RHEL 6 system from capturing a vmcore using kdump?
  • Which method outlined in the article I linked do you prefer to use for collecting a vmcore?
  • Do you have any suggestions to improve the vmcore collection mechanism in RHEL?

Responses

Hi David,

 

I can definitely confirm this. Sometimes it is a hardware issue that causes kdump to not capture a vmcore successfully. For instance, a faulty disk controller could potentially cause a panic and then become unavailable. In this instance it is likely that you won't get a vmcore. An alternative here would to be use another capture method such as NFS, SSH or SAN.

 

Having said that there are still times when I am not sure why kdump simply won't capture a vmcore. If anyone has some instances of this to relate I would also be curious to hear about them.

Close

Welcome! Check out the Getting Started with Red Hat page for quick tours and guides for common tasks.