Automatic Failover ensures immediate access to critical virtual machines (VMs) in the event of an outage:
- Continuously monitors VMs that have been configured for replication to a secondary site.
- Detects problems with VMs or network access for the primary site.
- Automatically fails over to the secondary site if the VMs in the primary site are disabled or unreachable.
- Brings destination VMs online immediately or after a specified delay.
This feature is available for the Virtual Server Agent with VMware.
Monitoring and Failover
To support automatic failover, you must identify three access nodes (VSA proxies) that reside in the primary site, to monitor VMs in the source datacenter.
- The first access node selected during configuration of automatic failover acts as a coordinator to monitor the source VMs.
- Other access nodes in the source datacenter act as workers.
Each worker monitors a list of VMs, with the coordinator making assignments as well as monitoring its own list of VMs.
- Each worker sends heartbeat packets continuously to the VMs on the list that the worker monitors, and checks responses to determine whether each VM is available or down.
- Workers communicate with the coordinator to report the states of the VMs they monitor.
- When necessary, the coordinator initiates an unplanned failover to bring destination VMs online.
- Configure a failover group for virtual machines that need to be monitored.
- Identify three machines in the source datacenter to be used for monitoring. Monitoring machines must meet the following conditions:
- Use machines that have at least 2 CPU cores.
- Verify that all monitoring machines are DNS resolvable and can communicate with the Command Center system, with the virtual machines they are monitoring, and with each other.
- Set the DNS entries for monitoring machines to the hostnames for the machines.
Note: If you define multiple failover groups to support automatic failover, assign a different set of access nodes for each failover group.
- Verify that all monitoring machines are DNS resolvable and can communicate with the CommServe system, with the virtual machines they are monitoring, and with each other.
- Install the following software on monitoring machines:
- Windows Server 2008 R2 or later
- Virtual Server Agent
- Java version 10
- Virtual machines being monitored must meet the following requirements:
- Each virtual machine must have a unique DNS name.
- VMware Tools must be installed on each VM.
Last modified: 10/28/2020 7:33:19 PM