fix: do not voluntarily relinquish the replication lock

[bmuddha]

This PR disables automatic relinquishment of replication lock by primary node, which should allow the leader to perform quick restart without failover to other standby nodes.

Summary by CodeRabbit

• Chores
  • Increased lock and API timeout durations to 15 seconds to reduce premature timeouts and give operations more time to complete.
  • Simplified primary replication shutdown flow by removing explicit lock release on shutdown, yielding cleaner and more predictable termination behavior.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: ASSERTIVE

Plan: Pro

Run ID: c31dd456-f603-4f0b-96d7-49ff79a52a60

📥 Commits

Reviewing files that changed from the base of the PR and between 67b21c2 and 6f86983.

📒 Files selected for processing (1)

• magicblock-replicator/src/nats/mod.rs

---

📝 Walkthrough

Walkthrough

This PR increases three timeouts: NATS JetStream TTL_LOCK from 5s to 15s, NATS API_TIMEOUT from 2s to 5s, and the replicator LEADER_TIMEOUT from 5s to 15s. In the primary replication loop, the cancellation branch no longer calls producer.release().await; it logs that shutdown was received and returns Ok(None) to exit primary mode.

Suggested reviewers

• GabrielePicco
• snawaz

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch bmuddha/replicator/disable-voluntary-relinquishment

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[GabrielePicco]

LGTM!