跳到主要内容

🔒 Distributed Synchronization & Locks

In multi-instance microservice deployments, running background processes or mutating shared resources (e.g., inventory deduction or batch invoice generation) requires cross-process synchronization to prevent race conditions.

ferrox-sync provides distributed locking abstractions backed by Redis and in-memory synchronizers.

1. Using Distributed Locks​

ferrox-sync allows microservices to acquire mutual exclusion locks over Redis:

use ferrox_sync::DistributedLock;
use std::time::Duration;

let lock = DistributedLock::new("redis://127.0.0.1:6379");

// Attempt to acquire lock for resource "inventory_update_101" with a 10-second TTL
if let Ok(guard) = lock.acquire("inventory_update_101", Duration::from_secs(10)).await {
    println!("🔒 Lock acquired! Safely mutating shared inventory...");
    
    // Perform critical section logic...

    // Lock automatically releases when `guard` drops!
} else {
    println!("⚠️ Resource is locked by another instance. Skipping operation.");
}

2. Redlock Algorithm​

For high-availability clusters, ferrox-sync implements the Redlock Algorithm, acquiring consensus across multiple independent Redis master nodes to ensure safety even if individual Redis nodes fail.

3. ✅ Best Practices​

  • Always configure short TTLs on locks: Set lock auto-release timeouts to prevent deadlocks if a microservice process crashes unexpectedly during critical section execution.