SLES11 SP1 can crash when a network with bonding devices is stopped

This document (7006170) is provided subject to the disclaimer at the end of this document.

Environment

SUSE Linux Enterprise Server 11 Service Pack 1

Situation

On SLES 11 SP1 if a bonding device is configured in balance-alb mode, a kernel oops can occur when the network service is stopped. If this occurs the system will crash and boot into the crash kernel. The oops can occur on the first network stop or after executing "service network start" and "service network stop" a number of times.

This issue is not limited to the balance-alb bonding mode. This same issue has been seen with the active backup bonding mode, and likely exists for other bonding modes as well.

Resolution

This issue has been fixed with kernel version 2.6.32.29-0.3.1 and later.

In case the kernel cannot be upgraded a workaround to avoid this crash is to set the miimon to zero before issuing a network stop command or shutting down the system. This can be done through the sysfs interface.

For example:
echo "0"> /sys/devices/virtual/net/bond0/bonding/miimon

This will set miimon to zero for the first network bond. A similar command would be used to set the miimon to zero for any additional network bonds that are setup, changing the bond0 in the path to bond1, bond2, etc.

Checking dmesg can verify that miimon was set to zero: "bonding: bond0: Setting MII monitoring interval to 0."

Disclaimer

This Support Knowledgebase provides a valuable tool for SUSE customers and parties interested in our products and solutions to acquire information, ideas and learn from one another. Materials are provided for informational, personal or non-commercial use within your organization and are presented "AS IS" WITHOUT WARRANTY OF ANY KIND.