Fix: HAProxy Max Connection Limit Reached

!Symptoms

• New connections being refused or queued
• HAProxy stats showing maxconn reached
• Users experiencing connection timeouts
• Backend queue depth growing continuously

?Root Causes

• maxconn set too low for current traffic volume
• Backend servers responding slowly, holding connections open
• Connection leak — keep-alive connections not being released
• Sudden traffic spike (flash crowd, DDoS) exceeding capacity
• File descriptor limit (ulimit) lower than maxconn

#Diagnosis Steps

1. 1Check HAProxy stats page: `curl http://localhost:9000/stats`
2. 2Check current connections: `echo 'show info' | socat stdio /var/run/haproxy.sock`
3. 3Check system file descriptors: `cat /proc/sys/fs/file-max` and `ulimit -n`
4. 4Monitor connection rate: `echo 'show stat' | socat stdio /var/run/haproxy.sock`
5. 5Check backend response times in HAProxy logs

>Fix

1. 1Increase maxconn in haproxy.cfg global and frontend sections
2. 2Increase system file descriptor limit: `ulimit -n 65536` and `/etc/security/limits.conf`
3. 3Add more backend servers to distribute connections
4. 4Enable HTTP keep-alive timeout to release idle connections: `timeout http-keep-alive 5s`
5. 5Implement rate limiting to prevent connection exhaustion from abusive clients

*Prevention

• Set maxconn based on load testing, not guesses — test at 2x expected peak
• Monitor connection count and alert at 80% of maxconn
• Use connection queuing (server maxqueue) for graceful degradation
• Implement auto-scaling for backend servers
• Set appropriate timeouts (connect, client, server) to release stale connections

Related Error Messages