Verifying operation after a switch fabric failure

You can disable a switch fabric to show that data availability is not affected by the loss.

About this task

This test should take about 15 minutes.

The expected result of this procedure is that disabling a fabric results in all cluster interconnect and disk traffic flowing to the other fabric.

In the examples shown, switch fabric 1 is disabled. This fabric consists of two switches, one at each MetroCluster site:

Procedure

  1. Disable connectivity to one of the two switch fabrics in the MetroCluster configuration:
    1. Disable the first switch in the fabric: switchdisable
      FC_switch_A_1::> switchdisable
    2. Disable the second switch in the fabric: switchdisable
      FC_switch_B_1::> switchdisable
  2. Monitor the result on the console of the controller modules.
    You can use the following commands to check the cluster nodes to make sure that all data is still being served. The command output shows missing paths to disks. This is expected.
    • vserver show
    • network interface show
    • aggr show
    • system node run nodename -command storage show disk -p
    • storage disk error show
  3. Reenable connectivity to one of the two switch fabrics in the MetroCluster configuration:
    1. Reenable the first switch in the fabric: switchenable
      FC_switch_A_1::> switchenable
    2. Reenable the second switch in the fabric: switchenable
      FC_switch_B_1::> switchenable
  4. Wait at least 10 minutes and then repeat the above steps on the other switch fabric.