6.6

Since a failover cannot be performed during synchronization, it is important to review the queue information
prior to attempting a failover. If the queues are large, file operations on the active node are high and it may be
prudent to delay a failover due to the length of time required to completely clear the queue. Queue lengths can
be viewed from the Data: Traffic/Queues page of the vCenter Server Heartbeat Console.
Prior to performing work on the Primary node, a failover can be triggered by selecting the Secondary node and
clicking Make Active on the Server: Summary page.This changes the roles such that the active node becomes
passive and the passive node becomes active. This allows users to continue to work while the Primary node
is offline.
When the Primary node is back up and running, a failover can be triggered again so that the Primary node
becomes active and the previously active (Secondary) node becomes passive.
Note The failover process may be performed at any time with the proviso that the systems are fully synchronized
with respect to data files and registry replication. Failovers cannot be performed if either node is in an
unsynchronized or unknown state.
Recovering From an Automatic Failover
A manually initiated failover is a controlled switch (initiated manually from the vSphere Web Client or vCenter
Server Heartbeat Console) between the active and passive nodes. An auto-failover happens when any of the
following fail on the active node: power, hardware, VMware Channel communications, or when preconfigured,
a protected application or service.The passive node waits a preconfigured time after the first missed heartbeat
before initiating failover. When this period expires, the passive node automatically assumes the active role and
starts all protected applications.
The following recovery scenario assumes that initially the Primary node is active and the Secondary node is
passive. A failover occurs and the Secondary node becomes the active node.
Note When failover conditions, such as a power failure, cause failures in both active and passive nodes, a
condition may result that causes both nodes to restart in passive mode. In this situation, manual intervention
is required. See “Two Passive Nodes,” on page 81.
Procedure
1 Review event logs on both nodes to determine the cause of the failover. For assistance, use the Log
Collector (launched from the Taskbar icon) tool to collect information and send the output to VMware
Support.
If any of the following issues exist on the Primary node, performing a failback to the Primary node may
not be possible until other important actions are carried out. Do not restart vCenter Server Heartbeat until
the following issues have been resolved:
■
Hard Disk Failure – Replace the defective hard disk
■
Power Failure – Restore power to the Primary node
■
Malware – Clean the node of all malware
■
Communications – Replace or repair defective network hardware
■
Blue Screen – Determine the cause and resolve it. As required, submit the dump file to VMware
Support (www.vmware.com/support) for analysis
2 On the Primary node, launch the Configure Server wizard and verify the node Identity is set to Primary
and the Active Server is set to Secondary. Click Finish to accept the changes.
3 Disconnect the VMware Channel network cables or disable the virtual or physical NIC.
4 Resolve the list of possible failures.
52 VMware, Inc.
vCenter Server Heartbeat Administrator's Guide