----------------------------

Original Description

Description:

Election handoff is now enabled by default. This is configurable via a new server parameter, called "enableElectionHandoff", which accepts boolean values.

With this feature enabled, primaries that step down due to a replSetStepDown command (without force:true) will step up an eligible candidate. That node will call for an election immediately, as opposed to waiting for the election timeout to expire (which is 10 seconds by default). This is thus intended to speed up planned failovers. The primary that steps down picks the best candidate according to the following criteria (in order of importance):
1. It must be caught up and electable.
2. It has the highest priority (among all secondaries that meet #1)
3. Resolve ties by picking the member with the lowest index in the config

The user-facing behavior of the replSetStepDown command remains the same. It will not delay returning to the client. It is possible for the step-up to fail (e.g. due to network issues), in which case we would wait for an election timeout as before.

----------------------------

Description

Scope of changes (files that need work and how much)

• reference/method/rs.stepDown / reference/method/rs.stepDown
• rs configuration
• electionTimeoutMillis – parameters
• Not updating the following because the discussion is more auto failover vs rs.stepDown()
  • /replication
  • core/replica-set-elections

Impact to other docs outside of this product

none

MVP (work and date?)

Resources (e.g. Scope Docs, Invision)