photoncloud-monorepo/chainfire/crates/chainfire-server/tests/cluster_integration.rs

//! Chainfire 3-Node Cluster Integration Test
//!
//! Verifies HA behavior: leader election, state replication, and node recovery.

use chainfire_client::Client;
use chainfire_server::{
    config::{ClusterConfig, NetworkConfig, NodeConfig, RaftConfig, ServerConfig, StorageConfig},
    server::Server,
};
use std::net::SocketAddr;
use std::time::Duration;
use tokio::time::sleep;

/// Create a 3-node cluster configuration with join flow
/// Node 1 bootstraps alone, nodes 2 & 3 join via member_add API
fn cluster_config_with_join(node_id: u64) -> (ServerConfig, tempfile::TempDir) {
    let base_port = match node_id {
        1 => 12379,
        2 => 22379,
        3 => 32379,
        _ => panic!("Invalid node_id"),
    };

    let api_addr: SocketAddr = format!("127.0.0.1:{}", base_port).parse().unwrap();
    let raft_addr: SocketAddr = format!("127.0.0.1:{}", base_port + 1).parse().unwrap();
    let gossip_addr: SocketAddr = format!("127.0.0.1:{}", base_port + 2).parse().unwrap();

    let temp_dir = tempfile::tempdir().unwrap();

    let config = ServerConfig {
        node: NodeConfig {
            id: node_id,
            name: format!("test-node-{}", node_id),
            role: "control_plane".to_string(),
        },
        cluster: ClusterConfig {
            id: 1,
            bootstrap: node_id == 1, // Only node 1 bootstraps
            initial_members: vec![], // Node 1 starts alone, others join via API
        },
        network: NetworkConfig {
            api_addr,
            raft_addr,
            gossip_addr,
            tls: None,
        },
        storage: StorageConfig {
            data_dir: temp_dir.path().to_path_buf(),
        },
        raft: RaftConfig::default(),
    };

    (config, temp_dir)
}

/// Alias for backwards compatibility (old tests use this)
fn cluster_config(node_id: u64) -> (ServerConfig, tempfile::TempDir) {
    cluster_config_with_join(node_id)
}

/// Create a single-node cluster configuration (for testing basic Raft functionality)
fn single_node_config() -> (ServerConfig, tempfile::TempDir) {
    let api_addr: SocketAddr = "127.0.0.1:12379".parse().unwrap();
    let raft_addr: SocketAddr = "127.0.0.1:12380".parse().unwrap();
    let gossip_addr: SocketAddr = "127.0.0.1:12381".parse().unwrap();

    let temp_dir = tempfile::tempdir().unwrap();

    let config = ServerConfig {
        node: NodeConfig {
            id: 1,
            name: "test-node-1".to_string(),
            role: "control_plane".to_string(),
        },
        cluster: ClusterConfig {
            id: 1,
            bootstrap: true, // Single-node bootstrap
            initial_members: vec![], // Empty = single node
        },
        network: NetworkConfig {
            api_addr,
            raft_addr,
            gossip_addr,
            tls: None,
        },
        storage: StorageConfig {
            data_dir: temp_dir.path().to_path_buf(),
        },
        raft: RaftConfig::default(),
    };

    (config, temp_dir)
}

#[tokio::test]
#[ignore] // Run with: cargo test --test cluster_integration -- --ignored
async fn test_single_node_raft_leader_election() {
    println!("\n=== Test: Single-Node Raft Leader Election ===");

    // Start single node
    let (config, _temp) = single_node_config();
    let api_addr = config.network.api_addr;
    println!("Creating single-node cluster...");
    let server = Server::new(config).await.unwrap();
    let handle = tokio::spawn(async move { server.run().await });
    println!("Node started: {}", api_addr);

    // Wait for leader election
    println!("Waiting for leader election...");
    sleep(Duration::from_secs(2)).await;

    // Verify leader elected
    let mut client = Client::connect(format!("http://{}", api_addr))
        .await
        .expect("Failed to connect");

    let status = client.status().await.expect("Failed to get status");
    println!(
        "Node status: leader={}, term={}",
        status.leader, status.raft_term
    );

    assert_eq!(status.leader, 1, "Node 1 should be leader in single-node cluster");
    assert!(status.raft_term > 0, "Raft term should be > 0");

    // Test basic KV operations
    println!("Testing KV operations...");
    client.put("test-key", "test-value").await.unwrap();
    let value = client.get("test-key").await.unwrap();
    assert_eq!(value, Some(b"test-value".to_vec()));

    println!("✓ Single-node Raft working correctly");

    // Cleanup
    handle.abort();
}

#[tokio::test]
#[ignore] // Run with: cargo test --test cluster_integration -- --ignored
async fn test_3node_leader_election_with_join() {
    println!("\n=== Test: 3-Node Leader Election with Join Flow ===");

    // Start Node 1 (bootstrap alone)
    let (config1, _temp1) = cluster_config_with_join(1);
    let api1 = config1.network.api_addr;
    let raft1 = config1.network.raft_addr;
    println!("Creating Node 1 (bootstrap)...");
    let server1 = Server::new(config1).await.unwrap();
    let handle1 = tokio::spawn(async move { server1.run().await });
    println!("Node 1 started: API={}, Raft={}", api1, raft1);

    // Wait for node 1 to become leader
    sleep(Duration::from_secs(2)).await;

    // Verify node 1 is leader
    let mut client1 = Client::connect(format!("http://{}", api1))
        .await
        .expect("Failed to connect to node 1");
    let status1 = client1.status().await.expect("Failed to get status");
    println!("Node 1 status: leader={}, term={}", status1.leader, status1.raft_term);
    assert_eq!(status1.leader, 1, "Node 1 should be leader");

    // Start Node 2 (no bootstrap)
    let (config2, _temp2) = cluster_config_with_join(2);
    let api2 = config2.network.api_addr;
    let raft2 = config2.network.raft_addr;
    println!("Creating Node 2...");
    let server2 = Server::new(config2).await.unwrap();
    let handle2 = tokio::spawn(async move { server2.run().await });
    println!("Node 2 started: API={}, Raft={}", api2, raft2);
    sleep(Duration::from_millis(500)).await;

    // Start Node 3 (no bootstrap)
    let (config3, _temp3) = cluster_config_with_join(3);
    let api3 = config3.network.api_addr;
    let raft3 = config3.network.raft_addr;
    println!("Creating Node 3...");
    let server3 = Server::new(config3).await.unwrap();
    let handle3 = tokio::spawn(async move { server3.run().await });
    println!("Node 3 started: API={}, Raft={}", api3, raft3);
    sleep(Duration::from_millis(500)).await;

    // Add node 2 to cluster via member_add API
    println!("Adding node 2 to cluster via member_add API...");
    let member2_id = client1
        .member_add(2, raft2.to_string(), false) // node_id=2, false=voter
        .await
        .expect("Failed to add node 2");
    println!("Node 2 added with ID: {}", member2_id);
    assert_eq!(member2_id, 2, "Node 2 should have ID 2");

    // Add node 3 to cluster via member_add API
    println!("Adding node 3 to cluster via member_add API...");
    let member3_id = client1
        .member_add(3, raft3.to_string(), false) // node_id=3, false=voter
        .await
        .expect("Failed to add node 3");
    println!("Node 3 added with ID: {}", member3_id);
    assert_eq!(member3_id, 3, "Node 3 should have ID 3");

    // Wait for cluster membership changes to propagate
    sleep(Duration::from_secs(3)).await;

    // Verify all nodes see the same leader
    let status1 = client1.status().await.expect("Failed to get status from node 1");
    println!("Node 1 final status: leader={}, term={}", status1.leader, status1.raft_term);

    let mut client2 = Client::connect(format!("http://{}", api2))
        .await
        .expect("Failed to connect to node 2");
    let status2 = client2.status().await.expect("Failed to get status from node 2");
    println!("Node 2 final status: leader={}, term={}", status2.leader, status2.raft_term);

    let mut client3 = Client::connect(format!("http://{}", api3))
        .await
        .expect("Failed to connect to node 3");
    let status3 = client3.status().await.expect("Failed to get status from node 3");
    println!("Node 3 final status: leader={}, term={}", status3.leader, status3.raft_term);

    // All nodes should agree on the leader
    assert_eq!(status1.leader, status2.leader, "Nodes 1 and 2 disagree on leader");
    assert_eq!(status1.leader, status3.leader, "Nodes 1 and 3 disagree on leader");
    assert!(status1.leader > 0, "No leader elected");

    println!("✓ 3-node cluster formed successfully with join flow");

    // Cleanup
    handle1.abort();
    handle2.abort();
    handle3.abort();
}

#[tokio::test]
#[ignore]
async fn test_3node_state_replication() {
    println!("\n=== Test: 3-Node State Replication ===");

    // Start cluster
    let (config1, _temp1) = cluster_config(1);
    let api1 = config1.network.api_addr;
    let server1 = Server::new(config1).await.unwrap();
    let handle1 = tokio::spawn(async move { server1.run().await });

    let (config2, _temp2) = cluster_config(2);
    let api2 = config2.network.api_addr;
    let server2 = Server::new(config2).await.unwrap();
    let handle2 = tokio::spawn(async move { server2.run().await });

    let (config3, _temp3) = cluster_config(3);
    let api3 = config3.network.api_addr;
    let server3 = Server::new(config3).await.unwrap();
    let handle3 = tokio::spawn(async move { server3.run().await });

    sleep(Duration::from_secs(2)).await;
    println!("Cluster started");

    // Write data to node 1 (leader)
    let mut client1 = Client::connect(format!("http://{}", api1))
        .await
        .unwrap();

    println!("Writing test data to node 1...");
    client1.put("test/key1", "value1").await.unwrap();
    client1.put("test/key2", "value2").await.unwrap();
    client1.put("test/key3", "value3").await.unwrap();

    // Wait for replication
    sleep(Duration::from_millis(500)).await;

    // Read from node 2 and node 3 (followers)
    println!("Reading from node 2...");
    let mut client2 = Client::connect(format!("http://{}", api2))
        .await
        .unwrap();
    let val2 = client2.get("test/key1").await.unwrap();
    assert_eq!(val2, Some(b"value1".to_vec()), "Data not replicated to node 2");

    println!("Reading from node 3...");
    let mut client3 = Client::connect(format!("http://{}", api3))
        .await
        .unwrap();
    let val3 = client3.get("test/key1").await.unwrap();
    assert_eq!(val3, Some(b"value1".to_vec()), "Data not replicated to node 3");

    println!("✓ State replication verified");

    // Cleanup
    handle1.abort();
    handle2.abort();
    handle3.abort();
}

#[tokio::test]
#[ignore]
async fn test_3node_follower_crash() {
    println!("\n=== Test: Follower Crash (Node Remains Available) ===");

    // Start cluster
    let (config1, _temp1) = cluster_config(1);
    let api1 = config1.network.api_addr;
    let server1 = Server::new(config1).await.unwrap();
    let handle1 = tokio::spawn(async move { server1.run().await });

    let (config2, _temp2) = cluster_config(2);
    let server2 = Server::new(config2).await.unwrap();
    let handle2 = tokio::spawn(async move { server2.run().await });

    let (config3, _temp3) = cluster_config(3);
    let api3 = config3.network.api_addr;
    let server3 = Server::new(config3).await.unwrap();
    let handle3 = tokio::spawn(async move { server3.run().await });

    sleep(Duration::from_secs(2)).await;
    println!("Cluster started");

    // Write initial data
    let mut client1 = Client::connect(format!("http://{}", api1))
        .await
        .unwrap();
    println!("Writing initial data...");
    client1.put("test/before-crash", "initial").await.unwrap();

    // Kill node 2 (follower)
    println!("Killing node 2 (follower)...");
    handle2.abort();
    sleep(Duration::from_millis(500)).await;

    // Cluster should still be operational (2/3 quorum)
    println!("Writing data after crash...");
    client1
        .put("test/after-crash", "still-working")
        .await
        .expect("Write should succeed with 2/3 quorum");

    // Read from node 3
    let mut client3 = Client::connect(format!("http://{}", api3))
        .await
        .unwrap();
    let val = client3.get("test/after-crash").await.unwrap();
    assert_eq!(val, Some(b"still-working".to_vec()));

    println!("✓ Cluster operational after follower crash");

    // Cleanup
    handle1.abort();
    handle3.abort();
}

#[tokio::test]
#[ignore]
async fn test_3node_leader_crash_reelection() {
    println!("\n=== Test: Leader Crash & Re-election ===");

    // Start cluster
    let (config1, _temp1) = cluster_config(1);
    let server1 = Server::new(config1).await.unwrap();
    let handle1 = tokio::spawn(async move { server1.run().await });

    let (config2, _temp2) = cluster_config(2);
    let api2 = config2.network.api_addr;
    let server2 = Server::new(config2).await.unwrap();
    let handle2 = tokio::spawn(async move { server2.run().await });

    let (config3, _temp3) = cluster_config(3);
    let api3 = config3.network.api_addr;
    let server3 = Server::new(config3).await.unwrap();
    let handle3 = tokio::spawn(async move { server3.run().await });

    sleep(Duration::from_secs(2)).await;
    println!("Cluster started");

    // Determine initial leader
    let mut client2 = Client::connect(format!("http://{}", api2))
        .await
        .unwrap();
    let initial_status = client2.status().await.unwrap();
    let initial_leader = initial_status.leader;
    println!("Initial leader: node {}", initial_leader);

    // Kill the leader (assume node 1)
    println!("Killing leader (node 1)...");
    handle1.abort();

    // Wait for re-election (should be < 1s per requirements)
    println!("Waiting for re-election...");
    sleep(Duration::from_secs(1)).await;

    // Verify new leader elected
    let new_status = client2.status().await.unwrap();
    println!(
        "New leader: node {}, term: {}",
        new_status.leader, new_status.raft_term
    );
    assert!(new_status.leader > 0, "No new leader elected");
    assert!(
        new_status.raft_term > initial_status.raft_term,
        "Raft term should increase after re-election"
    );

    println!("✓ Leader re-election successful within 1s");

    // Verify cluster still functional
    let mut client3 = Client::connect(format!("http://{}", api3))
        .await
        .unwrap();
    client3
        .put("test/post-reelection", "functional")
        .await
        .expect("Cluster should be functional after re-election");

    println!("✓ Cluster operational after re-election");

    // Cleanup
    handle2.abort();
    handle3.abort();
}