Using supervised failover

For Tarantool clusters that use supervised failover, Тarantool Cluster Manager offers tools for interaction with external failover coordinators from its web interface.

The tools for using supervised failover are located on the Failovers page available from the Actions menu on the cluster stateboard.

To view failover coordinators running on the cluster, go to the Failovers tab. On this tab, you can see the information about all Tarantool instances that the cluster uses as failover coordinators. The information includes:

• Current coordinator status – Active or Not active
• PID – process ID
• Hostname – the host on which the coordinator is running
• UUID – the coordinator ID
• Term – a value that defines the order in which coordinators become active (take the lock) over time.

To send a failover command to a coordinator, go to the Commands tab and click Add. Then, provide the command description in the YAML format. It can include the following fields:

• command – the command name. Possible value: switch – switch master in a replica set.
• new_master – the name of the instance to make the new master.
• timeout – the command execution timeout.

Example:

 
command: switchnew_master: instance-002timeout: 30

After entering the command, click Save to send the command for execution.

Tarantool assigns an id to the command and waits for the active coordinator to process the command.

All failover commands executed on the cluster are shown on the Commands tab with their ids and statuses. A command can have the following statuses:

• taken – a failover coordinator has started the command execution.
• success – the command has completed successfully.
• failed – an error occurred during the command execution. A short error description is shown in the Reason field.

To see the command execution details, click this command in the list.