Leader Election with Etcd

Create a backend server with a leader; if the leader goes down a new leader should be selected (use a central store like etcd to track the leader rather than peer-to-peer election). Asked in the onsite stage. The author had to install etcd and could not complete it.

Implement a practical service in which multiple backend instances coordinate through etcd. Only the instance holding a valid lease may act as leader. The service must detect leader failure, allow another instance to take ownership safely, and prevent stale leaders from continuing to process work.

Explain and demonstrate:

1. The etcd key, lease, keep-alive, and compare-and-swap operations.
2. Startup, leadership acquisition, renewal, resignation, and failover behavior.
3. How the application verifies leadership before handling leader-only work.
4. How you would test crashes, network partitions, delayed processes, and concurrent contenders.
5. Operational considerations, including observability and recovery.