Multiple Requesters

[wdbaruni]

The Problem

Today we only support a single requester node in a network for multiple reasons, including:

1. Job state not being replicated between requester nodes, and no support for an external data store. We only support local boltdb and in-memory store
2. No leader election, lease management or locking when scheduling or brokering concurrent evaluations for the same job
3. No lease management for background tasks that only one requester should execute, such as Housekeeping

Tasks

Beta Give feedback

1. Multiple Requester Shared State #3866
   th/resiliency +1
2. Lease Management #3865
   th/resiliency +1

[wdbaruni]

replaced by a linear project