BladeCluster Solution Manual

3. Shut down or reroute traffic for any applications using the Expand-over-ServerNet connections
between the affected node and the rest of the BladeCluster. If your applications might be
affected and adequate alternate Expand lines are not available, shut down all applications
that depend on the Expand-over-ServerNet line-handler processes.
4. For each Expand-over-ServerNet line on the affected node, stop the Expand-over-ServerNet
lines to other nodes. For example:
STOP LINE $SC001
STOP LINE $SC012
STOP LINE $SC013
STOP LINE $SC014
STOP LINE $SC015
STOP LINE $SC016
STOP LINE $SC017
5. On the other nodes in the cluster, use SCF commands to stop the Expand-over-ServerNet lines
to the affected node.
6. If you do not plan to use an Expand-over-ServerNet line again, stop and delete the associated
line-handler process. For example:
STOP DEVICE $ZZWAN.#SC001
DELETE DEVICE $ZZWAN.#SC001
If you are moving the fiber-optic cable of a node to different Advanced Cluster Hub ports on the
same BladeCluster, you do not need to stop or delete the Expand-over-ServerNet line-handler
processes.
Starting the ServerNet Cluster Subsystem
At an SCF prompt, issue this command to start the ServerNet cluster subsystem:
START SUBSYS $ZZSCL
Starting the External ServerNet Fabric
NOTE: After a ServerNet switch replacement or after communications through an external
ServerNet fabric have been disrupted for any reason, direct ServerNet connectivity is automatically
restored after an interval of approximately 25 seconds times the number of nodes in the cluster.
You can use this command for faster recovery.
1. Use OSM to verify that the ServerNet switch is operational.
2. Follow the instructions in “Checking the External Fabric for All Nodes” (page 70) to determine
which nodes have ServerNet cluster connectivity problems.
3. On all affected nodes, use SCF to start the ServerNet fabric:
START SERVERNET $ZSNET.FABRIC.*
4. To avoid having to log in directly to each node:
START SERVERNET \remotenode.$ZSNET.FABRIC.*
Starting SANMAN
CAUTION: You should neither configure nor start the $ZZSMN (SANMAN) process unless the
BladeCluster will be connected to a 6700 series ServerNet cluster. The SANMAN process supports
6770 or 6780 ServerNet cluster switches and is not useful unless these switches are present.
Starting the ServerNet Cluster Subsystem 81