A saturated connection pool means all available connections to the DB or Redis are in use. New requests wait or fail. It's often a symptom of leaks or slow queries.
Symptoms
- "Connection pool exhausted" errors
- Timeout on DB queries
- Performance that degrades under load
- Connections stuck in "Sleep"
Common Causes
- Unclosed connections: Code opens connections without releasing them.
- Long queries: Slow queries monopolize connections.
- Pool too small: Max connection count is insufficient for the load.
Diagnostic Steps
- Check active connections (SHOW PROCESSLIST)
- Identify connections in Sleep for a long time
- Examine long queries
- Check pool configuration in the application
Automate with MoniTao
MoniTao detects saturation effects:
- Alerts on degraded response times
- Detection of connection-related 500 errors
- History to correlate with traffic
Best Practices
- Use connections with timeout and idle timeout
- Always close connections (finally, using)
- Size the pool according to expected traffic
- Monitor active connection count
FAQ
What size for the pool?
Depends on your load. Start at 10-20 and adjust based on metrics.
How to detect connection leaks?
Connections in Sleep for a long time indicate leaks.
MySQL max_connections sufficient?
Must be >= sum of all your app pools + margin.
Can MoniTao monitor the pool?
Not directly, but response times reveal problems.
Ready to Sleep Soundly?
Start free, no credit card required.