Filter
Top Products
Intel
®
NetStructure
TM
ZT 7102 Chassis Management Module
CMM Redundancy, Synchronization and Failover
Technical Product Specification 41
4.4 CMM Failover
Once information is synchronized between the redundant CMMs, the active CMM will constantly
monitor its own health as well as the health of the standby CMM. In the event of one of the
scenarios listed in the sections that follow, the active CMM will automatically fail over to the
standby CMM so that no management functionality is lost at any time.
4.4.1 Scenarios That Force a Failover
The following scenarios cause a failover as long as the standby CMM is operational:
• The active CMM is pulled out of the chassis.
• The active CMM’s HEALTHY# signal is deasserted.
• A reboot command is issued to the active CMM.
• The front panel reset button on the active CMM is pushed in for more than five seconds.
• The internal synchronization code version of the active CMM is less than that of the standby
CMM.
4.4.2 Scenarios That Prevent Failover
The following are reasons a failover cannot occur:
• The active CMM cannot communicate with the standby CMM via their IPMB bus.
• Not all data has been completely synchronized between the CMMs.
To determine the active CMM, use the CLI command:
cmmget –d redundancy
Eth1:1 will always point to the active CMM, which will allow you to access the active CMM
through the chassis at all times.
4.4.3 Scenarios That Failover to a Healthier Standby CMM
The scenarios listed below can only cause a failover if the standby CMM is in a healthier state than
the active CMM. The health of the CMM is determined by computing a CMM health score, which
is equal to the sum of the weights of the following active conditions. A CMM health score is
determined for each CMM whenever any of these conditions occur on the active CMM. The CMM
health score is composed of the sum of the weights of any of the three conditions listed below. Each
condition has a default weight of 1 assigned to it, causing all conditions to have equal importance
in causing failover.
To determine if a failover is necessary when one of these conditions occurs, the active CMM
computes its CMM health score, and requests the health score of the standby CMM. If the score of
the standby CMM is LESS than the score of the active CMM, a failover will occur. If a failover
does not occur, the Chassis SEL will contain an entry indicating the reason failover did not occur.
1. SNMPTrapAddress1 ping failure:
A condition is asserted and a health score is computed if the active CMM cannot ping its first
SNMP trap address as defined in the SNMPTrapAddress1 setting. Only a ping failure of the
first SNMP trap address (SNMPTrapAddress1) can cause the condition to assert.
SNMPtrapaddress2 through SNMPtrapaddress5 do not perform this ping test.