Exalogic: Unexpected reboot of Infiniband Switches due to a BUG present in firmware versions 2.2.14 and older (Doc ID 2720462.1)

Last updated on NOVEMBER 14, 2022

Applies to:

Symptoms

A brief lost of network connectivty in the Exalogic evnironment.  In an Exalogic Virtual enviornment, lost of network connectivity will be visible with all guest vServers running in the environment.  In a Exalogic physical, lost of network connectivity will be visible in all compute nodes in the environment. 

The lost of the network connectivty will be due to an unexpected reboot of both Infiniband switches at the same time the lost of network occurs.  This will be correlated with some management Ethernet network wiring or topology changes.

In an Exalogic virtual environment guest vServers or Exalogic physical environment compute nodes, the following log entries will be found during the time the network connectivity was lost:

From the /var/log/messages we can see that the MLX ports is brought down abruptly:

The reboot fo the Infiniband switches will be correlated with Management Ethernet network problems, wireing, or topology changes.

Changes

 Management Ethernet network problems, wireing, or topology changes in the Infiniband GW switches.

Cause

My Oracle Support provides customers with access to over a million knowledge articles and a vibrant support community of peers and Oracle experts.