Implement zombie connection detection and cleanup in ConnectionManager; enhance tests for edge cases

This commit is contained in:
Juergen Kunz 2025-06-07 10:55:59 +00:00
parent 9094b76b1b
commit 47735adbf2
4 changed files with 564 additions and 2 deletions

View File

@ -373,3 +373,179 @@ The connection cleanup mechanisms have been significantly improved in v19.5.20:
3. Tests confirm no accumulation in standard scenarios with reachable backends
However, the missing connection establishment timeout causes accumulation when backends are unreachable or very slow to connect.
### Outer Proxy Sudden Accumulation After Hours
**User Report**: "The counter goes up suddenly after some hours on the outer proxy"
**Investigation Findings**:
1. **Cleanup Queue Mechanism**:
- Connections are cleaned up in batches of 100 via a queue
- If the cleanup timer gets stuck or cleared without restart, connections accumulate
- The timer is set with `setTimeout` and could be affected by event loop blocking
2. **Potential Causes for Sudden Spikes**:
a) **Cleanup Timer Failure**:
```typescript
// In ConnectionManager, if this timer gets cleared but not restarted:
this.cleanupTimer = this.setTimeout(() => {
this.processCleanupQueue();
}, 100);
```
b) **Memory Pressure**:
- After hours of operation, memory fragmentation or pressure could cause delays
- Garbage collection pauses might interfere with timer execution
c) **Event Listener Accumulation**:
- Socket event listeners might accumulate over time
- Server 'connection' event handlers are particularly important
d) **Keep-Alive Connection Cascades**:
- When many keep-alive connections timeout simultaneously
- Outer proxy has different timeout than inner proxy
- Mass disconnection events can overwhelm cleanup queue
e) **HttpProxy Component Issues**:
- If using `useHttpProxy`, the HttpProxy bridge might maintain connection pools
- These pools might not be properly cleaned after hours
3. **Why "Sudden" After Hours**:
- Not a gradual leak but triggered by specific conditions
- Likely related to periodic events or thresholds:
- Inactivity check runs every 30 seconds
- Keep-alive connections have extended timeouts (6x normal)
- Parity check has 30-minute timeout for half-closed connections
4. **Reproduction Scenarios**:
- Mass client disconnection/reconnection (network blip)
- Keep-alive timeout cascade when inner proxy times out first
- Cleanup timer getting stuck during high load
- Memory pressure causing event loop delays
### Additional Monitoring Recommendations
1. **Add Cleanup Queue Monitoring**:
```typescript
setInterval(() => {
const cm = proxy.connectionManager;
if (cm.cleanupQueue.size > 100 && !cm.cleanupTimer) {
logger.error('Cleanup queue stuck!', {
queueSize: cm.cleanupQueue.size,
hasTimer: !!cm.cleanupTimer
});
}
}, 60000);
```
2. **Track Timer Health**:
- Monitor if cleanup timer is running
- Check for event loop blocking
- Log when batch processing takes too long
3. **Memory Monitoring**:
- Track heap usage over time
- Monitor for memory leaks in long-running processes
- Force periodic garbage collection if needed
### Immediate Mitigations
1. **Restart Cleanup Timer**:
```typescript
// Emergency cleanup timer restart
if (!cm.cleanupTimer && cm.cleanupQueue.size > 0) {
cm.cleanupTimer = setTimeout(() => {
cm.processCleanupQueue();
}, 100);
}
```
2. **Force Periodic Cleanup**:
```typescript
setInterval(() => {
const cm = connectionManager;
if (cm.getConnectionCount() > threshold) {
cm.performOptimizedInactivityCheck();
// Force process cleanup queue
cm.processCleanupQueue();
}
}, 300000); // Every 5 minutes
```
3. **Connection Age Limits**:
- Set maximum connection lifetime
- Force close connections older than threshold
- More aggressive cleanup for proxy chains
## ✅ FIXED: Zombie Connection Detection (January 2025)
### Root Cause Identified
"Zombie connections" occur when sockets are destroyed without triggering their close/error event handlers. This causes connections to remain tracked with both sockets destroyed but `connectionClosed=false`. This is particularly problematic in proxy chains where the inner proxy might close connections in ways that don't trigger proper events on the outer proxy.
### Fix Implemented
Added zombie detection to the periodic inactivity check in ConnectionManager:
```typescript
// In performOptimizedInactivityCheck()
// Check ALL connections for zombie state
for (const [connectionId, record] of this.connectionRecords) {
if (!record.connectionClosed) {
const incomingDestroyed = record.incoming?.destroyed || false;
const outgoingDestroyed = record.outgoing?.destroyed || false;
// Check for zombie connections: both sockets destroyed but not cleaned up
if (incomingDestroyed && outgoingDestroyed) {
logger.log('warn', `Zombie connection detected: ${connectionId} - both sockets destroyed but not cleaned up`, {
connectionId,
remoteIP: record.remoteIP,
age: plugins.prettyMs(now - record.incomingStartTime),
component: 'connection-manager'
});
// Clean up immediately
this.cleanupConnection(record, 'zombie_cleanup');
continue;
}
// Check for half-zombie: one socket destroyed
if (incomingDestroyed || outgoingDestroyed) {
const age = now - record.incomingStartTime;
// Give it 30 seconds grace period for normal cleanup
if (age > 30000) {
logger.log('warn', `Half-zombie connection detected: ${connectionId} - ${incomingDestroyed ? 'incoming' : 'outgoing'} destroyed`, {
connectionId,
remoteIP: record.remoteIP,
age: plugins.prettyMs(age),
incomingDestroyed,
outgoingDestroyed,
component: 'connection-manager'
});
// Clean up
this.cleanupConnection(record, 'half_zombie_cleanup');
}
}
}
}
```
### How It Works
1. **Full Zombie Detection**: Detects when both incoming and outgoing sockets are destroyed but the connection hasn't been cleaned up
2. **Half-Zombie Detection**: Detects when only one socket is destroyed, with a 30-second grace period for normal cleanup to occur
3. **Automatic Cleanup**: Immediately cleans up zombie connections when detected
4. **Runs Periodically**: Integrated into the existing inactivity check that runs every 30 seconds
### Why This Fixes the Outer Proxy Accumulation
- When inner proxy closes connections abruptly (e.g., due to backend failure), the outer proxy's outgoing socket might be destroyed without firing close/error events
- These become zombie connections that previously accumulated indefinitely
- Now they are detected and cleaned up within 30 seconds
### Test Results
Debug scripts confirmed:
- Zombie connections can be created when sockets are destroyed directly without events
- The zombie detection successfully identifies and cleans up these connections
- Both full zombies (both sockets destroyed) and half-zombies (one socket destroyed) are handled
This fix addresses the specific issue where "connections that are closed on the inner proxy, always also close on the outer proxy" as requested by the user.

View File

@ -857,3 +857,41 @@ For detailed information about proxy protocol implementation and proxy chaining:
- **[Proxy Protocol Guide](./readme.proxy-protocol.md)** - Complete implementation details and configuration
- **[Proxy Protocol Examples](./readme.proxy-protocol-example.md)** - Code examples and conceptual implementation
- **[Proxy Chain Summary](./readme.proxy-chain-summary.md)** - Quick reference for proxy chaining setup
## Connection Cleanup Edge Cases Investigation (v19.5.20+)
### Issue Discovered
"Zombie connections" can occur when both sockets are destroyed but the connection record hasn't been cleaned up. This happens when sockets are destroyed without triggering their close/error event handlers.
### Root Cause
1. **Event Handler Bypass**: In edge cases (network failures, proxy chain failures, forced socket destruction), sockets can be destroyed without their event handlers being called
2. **Cleanup Queue Delay**: The `initiateCleanupOnce` method adds connections to a cleanup queue (batch of 100 every 100ms), which may not process fast enough
3. **Inactivity Check Limitation**: The periodic inactivity check only examines `lastActivity` timestamps, not actual socket states
### Test Results
Debug script (`connection-manager-direct-test.ts`) revealed:
- **Normal cleanup works**: When socket events fire normally, cleanup is reliable
- **Zombies ARE created**: Direct socket destruction creates zombies (destroyed sockets, connectionClosed=false)
- **Manual cleanup works**: Calling `initiateCleanupOnce` on a zombie does clean it up
- **Inactivity check misses zombies**: The check doesn't detect connections with destroyed sockets
### Potential Solutions
1. **Periodic Zombie Detection**: Add zombie detection to the inactivity check:
```typescript
// In performOptimizedInactivityCheck
if (record.incoming?.destroyed && record.outgoing?.destroyed && !record.connectionClosed) {
this.cleanupConnection(record, 'zombie_detected');
}
```
2. **Socket State Monitoring**: Check socket states during connection operations
3. **Defensive Socket Handling**: Always attach cleanup handlers before any operation that might destroy sockets
4. **Immediate Cleanup Option**: For critical paths, use `cleanupConnection` instead of `initiateCleanupOnce`
### Impact
- Memory leaks in edge cases (network failures, proxy chain issues)
- Connection count inaccuracy
- Potential resource exhaustion over time
### Test Files
- `.nogit/debug/connection-manager-direct-test.ts` - Direct ConnectionManager testing showing zombie creation

View File

@ -0,0 +1,306 @@
import { tap, expect } from '@git.zone/tstest/tapbundle';
import * as net from 'net';
import * as plugins from '../ts/plugins.js';
// Import SmartProxy
import { SmartProxy } from '../ts/index.js';
// Import types through type-only imports
import type { ConnectionManager } from '../ts/proxies/smart-proxy/connection-manager.js';
import type { IConnectionRecord } from '../ts/proxies/smart-proxy/models/interfaces.js';
tap.test('zombie connection cleanup - verify inactivity check detects and cleans destroyed sockets', async () => {
console.log('\n=== Zombie Connection Cleanup Test ===');
console.log('Purpose: Verify that connections with destroyed sockets are detected and cleaned up');
console.log('Setup: Client → OuterProxy (8590) → InnerProxy (8591) → Backend (9998)');
// Create backend server that can be controlled
let acceptConnections = true;
let destroyImmediately = false;
const backendConnections: net.Socket[] = [];
const backend = net.createServer((socket) => {
console.log('Backend: Connection received');
backendConnections.push(socket);
if (destroyImmediately) {
console.log('Backend: Destroying connection immediately');
socket.destroy();
} else {
socket.on('data', (data) => {
console.log('Backend: Received data, echoing back');
socket.write(data);
});
}
});
await new Promise<void>((resolve) => {
backend.listen(9998, () => {
console.log('✓ Backend server started on port 9998');
resolve();
});
});
// Create InnerProxy with faster inactivity check for testing
const innerProxy = new SmartProxy({
ports: [8591],
enableDetailedLogging: true,
inactivityTimeout: 5000, // 5 seconds for faster testing
inactivityCheckInterval: 1000, // Check every second
routes: [{
name: 'to-backend',
match: { ports: 8591 },
action: {
type: 'forward',
target: {
host: 'localhost',
port: 9998
}
}
}]
});
// Create OuterProxy with faster inactivity check
const outerProxy = new SmartProxy({
ports: [8590],
enableDetailedLogging: true,
inactivityTimeout: 5000, // 5 seconds for faster testing
inactivityCheckInterval: 1000, // Check every second
routes: [{
name: 'to-inner',
match: { ports: 8590 },
action: {
type: 'forward',
target: {
host: 'localhost',
port: 8591
}
}
}]
});
await innerProxy.start();
console.log('✓ InnerProxy started on port 8591');
await outerProxy.start();
console.log('✓ OuterProxy started on port 8590');
// Helper to get connection details
const getConnectionDetails = () => {
const outerConnMgr = (outerProxy as any).connectionManager as ConnectionManager;
const innerConnMgr = (innerProxy as any).connectionManager as ConnectionManager;
const outerRecords = Array.from((outerConnMgr as any).connectionRecords.values()) as IConnectionRecord[];
const innerRecords = Array.from((innerConnMgr as any).connectionRecords.values()) as IConnectionRecord[];
return {
outer: {
count: outerConnMgr.getConnectionCount(),
records: outerRecords,
zombies: outerRecords.filter(r =>
!r.connectionClosed &&
r.incoming?.destroyed &&
(r.outgoing?.destroyed ?? true)
),
halfZombies: outerRecords.filter(r =>
!r.connectionClosed &&
(r.incoming?.destroyed || r.outgoing?.destroyed) &&
!(r.incoming?.destroyed && (r.outgoing?.destroyed ?? true))
)
},
inner: {
count: innerConnMgr.getConnectionCount(),
records: innerRecords,
zombies: innerRecords.filter(r =>
!r.connectionClosed &&
r.incoming?.destroyed &&
(r.outgoing?.destroyed ?? true)
),
halfZombies: innerRecords.filter(r =>
!r.connectionClosed &&
(r.incoming?.destroyed || r.outgoing?.destroyed) &&
!(r.incoming?.destroyed && (r.outgoing?.destroyed ?? true))
)
}
};
};
console.log('\n--- Test 1: Create zombie by destroying sockets without events ---');
// Create a connection and forcefully destroy sockets to create zombies
const client1 = new net.Socket();
await new Promise<void>((resolve) => {
client1.connect(8590, 'localhost', () => {
console.log('Client1 connected to OuterProxy');
client1.write('GET / HTTP/1.1\r\nHost: test.com\r\n\r\n');
// Wait for connection to be established through the chain
setTimeout(() => {
console.log('Forcefully destroying backend connections to create zombies');
// Get connection details before destruction
const beforeDetails = getConnectionDetails();
console.log(`Before destruction: Outer=${beforeDetails.outer.count}, Inner=${beforeDetails.inner.count}`);
// Destroy all backend connections without proper close events
backendConnections.forEach(conn => {
if (!conn.destroyed) {
// Remove all listeners to prevent proper cleanup
conn.removeAllListeners();
conn.destroy();
}
});
// Also destroy the client socket abruptly
client1.removeAllListeners();
client1.destroy();
resolve();
}, 500);
});
});
// Check immediately after destruction
await new Promise(resolve => setTimeout(resolve, 100));
let details = getConnectionDetails();
console.log(`\nAfter destruction:`);
console.log(` Outer: ${details.outer.count} connections, ${details.outer.zombies.length} zombies, ${details.outer.halfZombies.length} half-zombies`);
console.log(` Inner: ${details.inner.count} connections, ${details.inner.zombies.length} zombies, ${details.inner.halfZombies.length} half-zombies`);
// Wait for inactivity check to run (should detect zombies)
console.log('\nWaiting for inactivity check to detect zombies...');
await new Promise(resolve => setTimeout(resolve, 2000));
details = getConnectionDetails();
console.log(`\nAfter first inactivity check:`);
console.log(` Outer: ${details.outer.count} connections, ${details.outer.zombies.length} zombies, ${details.outer.halfZombies.length} half-zombies`);
console.log(` Inner: ${details.inner.count} connections, ${details.inner.zombies.length} zombies, ${details.inner.halfZombies.length} half-zombies`);
console.log('\n--- Test 2: Create half-zombie by destroying only one socket ---');
// Clear backend connections array
backendConnections.length = 0;
const client2 = new net.Socket();
await new Promise<void>((resolve) => {
client2.connect(8590, 'localhost', () => {
console.log('Client2 connected to OuterProxy');
client2.write('GET / HTTP/1.1\r\nHost: test.com\r\n\r\n');
setTimeout(() => {
console.log('Creating half-zombie by destroying only outgoing socket on outer proxy');
// Access the connection records directly
const outerConnMgr = (outerProxy as any).connectionManager as ConnectionManager;
const outerRecords = Array.from((outerConnMgr as any).connectionRecords.values()) as IConnectionRecord[];
// Find the active connection and destroy only its outgoing socket
const activeRecord = outerRecords.find(r => !r.connectionClosed && r.outgoing && !r.outgoing.destroyed);
if (activeRecord && activeRecord.outgoing) {
console.log('Found active connection, destroying outgoing socket');
activeRecord.outgoing.removeAllListeners();
activeRecord.outgoing.destroy();
}
resolve();
}, 500);
});
});
// Check half-zombie state
await new Promise(resolve => setTimeout(resolve, 100));
details = getConnectionDetails();
console.log(`\nAfter creating half-zombie:`);
console.log(` Outer: ${details.outer.count} connections, ${details.outer.zombies.length} zombies, ${details.outer.halfZombies.length} half-zombies`);
console.log(` Inner: ${details.inner.count} connections, ${details.inner.zombies.length} zombies, ${details.inner.halfZombies.length} half-zombies`);
// Wait for 30-second grace period (simulated by multiple checks)
console.log('\nWaiting for half-zombie grace period (30 seconds simulated)...');
// Manually age the connection to trigger half-zombie cleanup
const outerConnMgr = (outerProxy as any).connectionManager as ConnectionManager;
const records = Array.from((outerConnMgr as any).connectionRecords.values()) as IConnectionRecord[];
records.forEach(record => {
if (!record.connectionClosed) {
// Age the connection by 35 seconds
record.incomingStartTime -= 35000;
}
});
// Trigger inactivity check
await new Promise(resolve => setTimeout(resolve, 2000));
details = getConnectionDetails();
console.log(`\nAfter half-zombie cleanup:`);
console.log(` Outer: ${details.outer.count} connections, ${details.outer.zombies.length} zombies, ${details.outer.halfZombies.length} half-zombies`);
console.log(` Inner: ${details.inner.count} connections, ${details.inner.zombies.length} zombies, ${details.inner.halfZombies.length} half-zombies`);
// Clean up client2 properly
if (!client2.destroyed) {
client2.destroy();
}
console.log('\n--- Test 3: Rapid zombie creation under load ---');
// Create multiple connections rapidly and destroy them
const rapidClients: net.Socket[] = [];
for (let i = 0; i < 5; i++) {
const client = new net.Socket();
rapidClients.push(client);
client.connect(8590, 'localhost', () => {
console.log(`Rapid client ${i} connected`);
client.write('GET / HTTP/1.1\r\nHost: test.com\r\n\r\n');
// Destroy after random delay
setTimeout(() => {
client.removeAllListeners();
client.destroy();
}, Math.random() * 500);
});
// Small delay between connections
await new Promise(resolve => setTimeout(resolve, 50));
}
// Wait a bit
await new Promise(resolve => setTimeout(resolve, 1000));
details = getConnectionDetails();
console.log(`\nAfter rapid connections:`);
console.log(` Outer: ${details.outer.count} connections, ${details.outer.zombies.length} zombies, ${details.outer.halfZombies.length} half-zombies`);
console.log(` Inner: ${details.inner.count} connections, ${details.inner.zombies.length} zombies, ${details.inner.halfZombies.length} half-zombies`);
// Wait for cleanup
console.log('\nWaiting for final cleanup...');
await new Promise(resolve => setTimeout(resolve, 3000));
details = getConnectionDetails();
console.log(`\nFinal state:`);
console.log(` Outer: ${details.outer.count} connections, ${details.outer.zombies.length} zombies, ${details.outer.halfZombies.length} half-zombies`);
console.log(` Inner: ${details.inner.count} connections, ${details.inner.zombies.length} zombies, ${details.inner.halfZombies.length} half-zombies`);
// Cleanup
await outerProxy.stop();
await innerProxy.stop();
backend.close();
// Verify all connections are cleaned up
console.log('\n--- Verification ---');
if (details.outer.count === 0 && details.inner.count === 0) {
console.log('✅ PASS: All zombie connections were cleaned up');
} else {
console.log('❌ FAIL: Some connections remain');
}
expect(details.outer.count).toEqual(0);
expect(details.inner.count).toEqual(0);
expect(details.outer.zombies.length).toEqual(0);
expect(details.inner.zombies.length).toEqual(0);
expect(details.outer.halfZombies.length).toEqual(0);
expect(details.inner.halfZombies.length).toEqual(0);
});
tap.start();

View File

@ -456,6 +456,48 @@ export class ConnectionManager extends LifecycleComponent {
}
}
// Also check ALL connections for zombie state (destroyed sockets but not cleaned up)
// This is critical for proxy chains where sockets can be destroyed without events
for (const [connectionId, record] of this.connectionRecords) {
if (!record.connectionClosed) {
const incomingDestroyed = record.incoming?.destroyed || false;
const outgoingDestroyed = record.outgoing?.destroyed || false;
// Check for zombie connections: both sockets destroyed but connection not cleaned up
if (incomingDestroyed && outgoingDestroyed) {
logger.log('warn', `Zombie connection detected: ${connectionId} - both sockets destroyed but not cleaned up`, {
connectionId,
remoteIP: record.remoteIP,
age: plugins.prettyMs(now - record.incomingStartTime),
component: 'connection-manager'
});
// Clean up immediately
this.cleanupConnection(record, 'zombie_cleanup');
continue;
}
// Check for half-zombie: one socket destroyed
if (incomingDestroyed || outgoingDestroyed) {
const age = now - record.incomingStartTime;
// Give it 30 seconds grace period for normal cleanup
if (age > 30000) {
logger.log('warn', `Half-zombie connection detected: ${connectionId} - ${incomingDestroyed ? 'incoming' : 'outgoing'} destroyed`, {
connectionId,
remoteIP: record.remoteIP,
age: plugins.prettyMs(age),
incomingDestroyed,
outgoingDestroyed,
component: 'connection-manager'
});
// Clean up
this.cleanupConnection(record, 'half_zombie_cleanup');
}
}
}
}
// Process only connections that need checking
for (const connectionId of connectionsToCheck) {
const record = this.connectionRecords.get(connectionId);