fix: Learner promotion stuck due to voter count and missing role transition

[JoshuaChi]

Problem

When scaling a single-node cluster to 3 nodes, the third node (Learner) gets promoted to Voter by the Leader but fails to transition to Follower role, remaining stuck as Learner indefinitely.

Impact:

• Learner cannot serve client read requests (returns "Learner can not process client read request")
• Cluster appears healthy to Leader but is functionally degraded
• Affects dynamic cluster scaling scenarios

Root Cause

Issue 1: Incorrect voter count calculation (d-engine-core/src/raft_role/leader_state.rs)

• membership.voters() excludes the Leader itself
• process_pending_promotions didn't account for Leader (+1)
• Result: calculate_safe_batch_size(1, 1) → 0, blocking promotion

Issue 2: Missing role transition detection (d-engine-core/src/raft_role/learner_state.rs)

• Learner received MembershipApplied event after promotion
• Did not check if its own role changed to Voter
• Failed to send BecomeFollower event to trigger transition

[coderabbitai]

📝 CodeRabbit Plan Mode

Generate an implementation plan and prompts that you can use with your favorite coding agent.

• Create Plan

Examples

• Example 1
• Example 2

---

🔗 Similar Issues

Related Issues

• Fix ClusterMembership API: Separate static role from dynamic leader info #201
• Handle self-removal in non-leader roles (Follower/Candidate/Learner) #207
• Fix remove_node() to Allow Leader Self-Removal #200

🔗 Related PRs

#183 - Feature/182 local client [merged]

👤 Suggested Assignees

• JoshuaChi

---

🧪 Issue enrichment is currently in open beta.

You can configure auto-planning by selecting labels in the issue_enrichment configuration.

To disable automatic issue enrichment, add the following to your .coderabbit.yaml:

issue_enrichment:
  auto_enrich:
    enabled: false

💬 Have feedback or questions? Drop into our discord!