Failover

Introduction

The failover system is an optional feature that extends the master driver. Similarly configured Volttron instances collect from the same device(s) for easy horizontal scaling. Instances will take turns scraping and publishing data so that each has more free cycles between scrapes. If one dies, those remaining won't be interrupted. It can currently be found in the feature/failover branch.

                 _____________
                |             |
 ________       | Collector 0 |             ________
|        |----->|_____________|----------->|        |
| DEVICE |       _____________             | Target |
|________|----->|             |----------->|________|
                | Collector 1 |
                |_____________|

Configuration

Redundant Volttron instances will have almost identical configurations. The following variables tell the drivers how periodic scrape intervals should be adjusted:

• failover_array_size - The number of collecting instances.
• failover_instance_id - Zero is the first instance.

Below is an example of a master driver's config file using the failover feature. This will be one instance in a pair.

{
    "agentid": "master_driver",
    "failover_array_size": 2,
    "failover_instance_id": 0,
    # "failover_instance_id": 1,

    "driver_config_list": ["driver1.config", "driver2.config"]
}

To Do

• Automatically reschedule collection when an instance fails.
• Dynamically add and remove instances from a collection array. This should be available in VC.