Fix: MongoDB Connection Pool Exhausted

!Symptoms

• Application errors: 'MongoTimeoutError: connection timed out'
• MongoDB logs showing 'connection accepted' but no available connections
• Spike in application response times
• db.serverStatus().connections.current at or near the limit

?Root Causes

• Connection pool maxPoolSize too small for the workload
• Connection leak — connections opened but never returned to pool
• Slow queries holding connections for too long
• Application instances scaled up without adjusting MongoDB max connections
• MongoDB maxIncomingConnections reached (default 65536)

#Diagnosis Steps

1. 1Check current connections: `db.serverStatus().connections`
2. 2Check per-app pool stats: look at driver-level connection pool metrics
3. 3Find slow queries: `db.currentOp({'secs_running': {$gte: 5}})`
4. 4Check mongod log for connection-related messages
5. 5Count connections per source: `db.aggregate([{$currentOp: {allUsers: true}}, {$group: {_id: '$client', count: {$sum: 1}}}, {$sort: {count: -1}}])`

>Fix

1. 1Increase maxPoolSize in the connection string: `mongodb://host/db?maxPoolSize=100`
2. 2Fix connection leaks: ensure every connection is properly returned to the pool
3. 3Kill long-running operations: `db.killOp(<opId>)`
4. 4Add indexes to speed up slow queries holding connections
5. 5Increase MongoDB maxIncomingConnections if server resources allow

*Prevention

• Size connection pool based on: (app instances × maxPoolSize) < MongoDB max connections
• Monitor connection count with Prometheus mongodb_exporter
• Set connection timeout and socket timeout to release stuck connections
• Use connection pool monitoring events in the driver for observability
• Implement query timeouts: `maxTimeMS` on queries to prevent connection hogging

Related Error Messages