Code enhancement to handle VIOS Path loss issues during Brocade haReboot - There have been multiple outages experienced in IBM VIO Server(VIOS) managed environment, during storage related events. The haReboot performed on Brocade SAN switches are supposed to be non-disruptive, but in reality they do take a finite amount of time. In VIO managed environments, this has sometimes resulted in impact and affected availability. During a haReboot the fabric services are not available and hence when the VIOS queries the Name server on the switch, it may not get the desired response due to the haReboot in progress. This results in the VIOS logging out and attempting to log back in again, resulting in path failures at the VIOS. To prevent the same from happening, the VIOS code can be updated to include an appropriate time interval that kicks in as and when the haReboot is in progress. The VIOS will then wait for this interval and pause any activity, before querying the name server again. The assumption is that the haReboot RSCN notification is sent correctly by the Brocade switches and received by the VIO servers. If not, then this can be worked out jointly by Brocade and IBM, to prevent outages in the future. The referenced Support Case has additional information.

• Comments 1