Automatic Rollover and Resilience

Summary

The Automatic Rollover and Resilience functionality within TaskCentre provides for two TaskCentre Server machines to be setup as a resilient pair that will then mirror a single Store. The active Stores in use for each TaskCentre Server may be either a SQL Server Store or a TaskCentre Internal Store. They do not need to be both using the same Store type to be able to mirror Store contents.

During the server installation process, the default Master and Slave IP addresses are used to configure a primary connection between the two servers. However, provision is made to utilise a secondary connection to be used when the primary connection fails.

In practice this would require that both servers have dual network cards fitted to provide primary and secondary IP addresses for both. In addition to the LAN network connection it is recommended that a straight-through dedicated connection is made between the two servers using a cross-over cable. Providing a physical connection between the servers means that even if the LAN connection fails, the Servers can still communicate with each other.

Store Status Monitoring and Synchronisation

Once you have set up a resilient pair, then both servers monitor the status of each other to ensure that they are working correctly.

If a failure within the Master machine is detected, the Slave machine automatically assumes operation, continuing from the point where the Master Server failed and the Administrator will receive an email notification of the failure. Users will be informed through a message in the Status Bar at the bottom of the screen but the changeover will be seamless. In addition, if a Slave server failure is detected, then this will also be notified to the Administrator as although the server is not active, its Store cannot be updated.

Once the Slave is working again, its Store will automatically be updated from the Master server Store.

Maintenance Considerations

Maintenance is a process that optimises the TaskCentre store and is run based on a schedule defined in the Run Times and Maintenance tab of the Options dialog. Maintenance has to be scheduled carefully as the process in effect locks down the system and prevents Tasks being processed.

In a single server system this can be easily planned around times when a business is closed, but special consideration is required for running Maintenance when using a pair of resilient Servers as neither server will be available during this process.

Maintenance is always started by the currently active server; Master or Slave depending on their status. Once completed the maintenance is then started on the other server. When both servers have completed maintenance, the formerly active server becomes active again.

Designing Tasks for Resilience

Tasks should be designed using Memory to ensure that if a Server fails mid-execution, the paired Server can continue Task execution without repeating iterations already processed.