Cisco MDS: VSAN Isolation due to remote zone server not responding

• Added new zones and activated the zoneset

• An existing production VSAN became isolated after a recent zoneset activation

• Syslog shows the following errors from the core MDS switch:

  ZONE-2-ZS_MERGE_ADJ_NO_RESPONSE: %$VSAN 20%$ Adjacent switch not responding, isolating interface port-channel24
  
  PORT-5-IF_TRUNK_DOWN: %$VSAN 20%$ Interface port-channel24, vsan 20 is down (Isolation due to remote zone server not responding)  ISL

• `show zone internal event-history errors` output from the MDS-9124 Bladeswitch:

  `show zone internal event-history errors`
  12) Event:E_DEBUG, length:77, at 312878 usecs after Wed Jan 18 17:12:54 2017
   [134] zs_fc2_throttle_get_approval: size: 2083624 exceeds total_size: 2083624

• Fabric is mixed with 9500s running NX OS 6.2 and MDS-9124 Bladeswitches running NX OS 5.2.

• Bouncing the affected port channel and removing/re-adding the VSAN to the port channel trunk-allowed list did not resolve the issue

The VSAN isolation is due to the zone server not responding on the MDS-9124.

The problem is due to the zone DB size exceeding the limit of 2MB on the MDS-9124 Bladeswitch.

Cisco Documentation:
Cisco MDS NX-OS Release 5.2 Configuration Limits

1. Delete unused zones and zonesets

2. If deleting unused zones/zonesets does not dramatically reduce the size of the Full Zone DB, change the VSAN distribution setting from Full to Active Only.

   Commands:

   config t
   no zoneset distribute full vsan x
   exit
   copy run start

   Issue "show zone status vsan x" to verify the change.

Lastly, bounce (shut or no shut) the affected port channel interface to recover the VSAN from its isolated status.

Changing the VSAN distribution to Active Only resolve this but may cause the Full Zone DB to become out-of-sync. Ensure that zoneset activations are performed from only one switch in the fabric with the most up-to-date Full Zone DB information.