SMW Cluster Configuration

Both SMWs are connected to the boot RAID, and are connected to each other with heartbeat cables between the eth2 and eth4 ports on each SMW. The heartbeat connection monitors the health of the cluster. In addition, each SMW is connected to the boot RAID (through FC or SAS cards), to the site network through eth0, to the HSS network through eth1, and to the boot node through eth3. (For more information, see Network Connections for an SMW HA System.) An Integrated Dell™ Remote Access Controller (iDRAC) is required on both SMWs.

The following figure shows the major connections between components in an SMW HA system.

In a Cray SMW HA cluster, the two SMWs are configured in an active/passive configuration. This configuration lets the passive node take over the SMW functions if a software or hardware fault occurs on the active node. All HSS daemons run on the active SMW. (An additional stonith daemon, which monitors SMW health, runs on both SMWs.) At failover, all daemons move to the passive SMW, which then becomes the active one.

During initial installation, the first SMW that is installed and configured becomes the active SMW. The second SMW that is installed and configured becomes the passive SMW. However, either SMW can be active during normal operation. The cluster configuration does not remember which SMW was initially configured to be active.