Specifications

11-23
Cisco SN 5428-2 Storage Router Software Configuration Guide
78-15471-01
Chapter 11 Maintaining and Managing the SN 5428-2 Storage Router
Controlling SCSI Routing Instances in a Cluster
Stopping & Starting Instances
If the storage router is experiencing a problem with a specific set of IP hosts or storage resources, you
may wish to stop the associated SCSI routing instance from running anywhere in the cluster. The
no scsirouter enable command causes the specified SCSI routing instance to cease running on the
SN 5428-2, but does not cause a failover to another storage router in the cluster. This command
effectively stops an instance from running anywhere in the cluster.
Once a SCSI routing instance has been stopped, it can be re-activated by issuing the scsirouter enable
command. The scsirouter enable command must be issued from the same storage router as the
no scsirouter enable command.
When a SCSI routing instance is stopped, its configuration information is removed from the bootable
configuration of all cluster nodes, except for the storage router from which the command was issued. See
Chapter 12, “Command Line Interface Reference,” for more information about the no scsirouter enable
command.
Viewing Operational Statistics
Use the show scsirouter stats command to see access and connection information, such as the number
of active connections and the number of logins that have occurred since the storage router was last
restarted (or since statistics were last cleared).
For example, the show scsirouter stats command in Example 11-8 shows that SCSI routing instance,
foo, is currently active.
Example 11-8 Results of “show scsirouter stats” Command
[SN5428-2_PR1]# show scsirouter foo stats
Logins Logins Target Access Authentication
Router Started Accepted Active Failures Failures
---------- --------------- -------- ------ -------------- --------------
foo Aug 01 16:01:49 10 7 0 0
Handling Failover
In a cluster, storage routers continually exchange information as heartbeats to detect changes or failures
in the cluster. HA messages are sent using UDP over IP and, depending on the message type or situation,
may be sent as unicast or multicast messages. To make sure that HA information is exchanged reliably
between cluster nodes, the storage routers alternate transmission of heartbeats between the management
and the HA interfaces.
Each storage router in the cluster maintains and exchanges information about available resources.
Failover by eligibility is enabled by default; HA bases the decision to automatically fail over a SCSI
routing instance to another storage router in a cluster based on the Fibre Channel and other resources
available to that SCSI routing instance.
Failover occurs when:
• All mapped targets are unavailable or a critical resource for the SCSI routing instance is unavailable,
and some or all mapped targets would be available from another storage router in the cluster. A
critical resource can be a configured Gigabit Ethernet interface, a required Fibre Channel interface,
or an internal resource needed to run the SCSI routing instance.