[WIP] Active/passive failover support by worker manager clusters
Summary
This PR adds explicit failover strategy, with an initial support for active/passive failover based on clusters of managers that are assigned a priority by cluster. The highest priority cluster is then determined to be the currently active cluster of managers. Given active/passive failover is configured, a heart beat is initialized in each manager. This heart beat registers a given manager's cluster as active, evicts outdated active cluster, and if a given manager belongs to the highest priority active cluster, it will process queued work.
[WIP] Active/passive failover support by worker manager clusters
Summary This PR adds explicit failover strategy, with an initial support for active/passive failover based on clusters of managers that are assigned a priority by cluster. The highest priority cluster is then determined to be the currently active cluster of managers. Given active/passive failover is configured, a heart beat is initialized in each manager. This heart beat registers a given manager's cluster as active, evicts outdated active cluster, and if a given manager belongs to the highest priority active cluster, it will process queued work.