Fix: AWS ALB Returning 502 Bad Gateway

!Symptoms

• ALB returning 502 errors to clients
• CloudWatch showing elevated HTTP 502 count on the ALB
• Backend health checks passing but requests failing
• Intermittent 502s during deployments or scaling events

?Root Causes

• Target responding with invalid HTTP response or closing connection prematurely
• Target deregistering while in-flight requests are still processing
• Target taking too long to respond (exceeding idle timeout)
• Security group on targets not allowing health check traffic from ALB
• Target application crashing or running out of connections

#Diagnosis Steps

1. 1Check ALB access logs for the 502 entries — look at target_status_code
2. 2Check target group health: AWS Console → EC2 → Target Groups → Targets tab
3. 3Check CloudWatch: `TargetResponseTime` and `UnHealthyHostCount` metrics
4. 4Check target's application logs during the 502 window
5. 5Verify security groups: ALB SG → Target SG allow health check port

>Fix

1. 1Ensure targets return valid HTTP responses (proper status code and headers)
2. 2Increase deregistration delay: `deregistration_delay.timeout_seconds = 60`
3. 3Increase ALB idle timeout if targets need more processing time
4. 4Fix security groups to allow ALB health check traffic
5. 5Fix application connection handling — ensure graceful shutdown

*Prevention

• Set ALB idle timeout lower than application keep-alive timeout
• Enable access logging on ALB for debugging
• Use connection draining (deregistration delay) during deployments
• Monitor `HTTPCode_ELB_502_Count` and alert on spikes
• Implement graceful shutdown in application (handle SIGTERM, drain connections)

Related Error Messages