
These rules filter out web traffic based on conditions like HTTP headers, URIs, and IP addresses.

I am working on a decrypted packet capture on WAF, requires downtime so I need to schedule it - but figured I'd come here in the meantime.AWS Web Application Firewall (WAF) is a firewall security system that monitors incoming and outgoing traffic for applications and websites based on your pre-defined web security rules. After failing a few times, it succeeds and traffic moves on as normal - but this would explain why the TCP streams are also clean, nothing showing in TCP dump.

but experiencing some failures for some reason. This makes me think that maybe the WAF is trying to connect to the listener. AWS is absolutely certain nothing is wrong with TGW, and I agree. TCP dump on EKS at the ingress controller shows nothing arriving until the delay has passed, then instant reply - so the delay is prior to EKS, leaving WAF egress, TGW, and NLB.īeen working with AWS support, escalated to senior engineers with AWS and Barracuda and we're working on it - but even they are a bit stumped. TCP dump on WAF shows the packet arriving from client, then silence for 5 seconds until receiving the reply. Flow is client -> Public NLB -> WAF -> TGW -> Internal NLB (443) -> EKS.Īfter extensive testing, I have narrowed the delay down to the connection between WAF, TGW, and internal NLB. Normally everything works fine - but we have a few APIs which intermittently see a 5 second delay on requests.

We have a Barracuda WAF on EC2 (AMI) with a service connecting to a TLS listener on NLB. This has been a real bear of an issue, turning to the Reddit brain trust.
