Kdump does not correctly recognize multiple port network interface

Solution Unverified - Updated -

Environment

  • Red Hat Enterprise Linux 6
  • Mellanox Technologies MT26448 [ConnectX EN 10GigE, PCIe 2.0 5GT/s] (rev b0)

Issue

I've configured my system to use kdump to write a vmcore to a network location. When the kernel crashes, there is no vmcore written to the network location, and no error messages.

Resolution

  • Upgrade to the kexec-tools package released by the following errata in 6.3.z: RHBA-2013:xxxx-x
  • After upgrading, add the following line to your /etc/kdump.conf file:
   extra_modules mlx4_core
   extra_modules mlx4_en
  • Then, make sure you have a correct HWADDR configuration line in /etc/sysconfig/network-scripts/ifcfg-$dev:
# grep HWADDR /etc/sysconfig/network-scripts/ifcfg-eth0

HWADDR="AA:BB:CC:DD:EE:FF"

Root Cause

Certain multiport network cards use one PCI bus address, and don't use the PCI function to differentiate the ports. Previously, when kdump maps the devices, it may select the wrong port for the network connection. The upgraded kexec-tools package adds logic that detects these cards, and configures them based on their hardware address.

This solution is part of Red Hat’s fast-track publication program, providing a huge library of solutions that Red Hat engineers have created while supporting our customers. To give you the knowledge you need the instant it becomes available, these articles may be presented in a raw and unedited form.

Close

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