[SEMW]

Definitely not just the console. We had hundreds of thousands of websocket connections to us-east-1 drop at 15:40, and new websocket connections to that region are still failing. (Luckily not a huge impact on our service cause we run in 6 other regions, but still).

[andrew_]

Side question: How happy are you with API Gateway's WebSocket service?

[SEMW]

No idea, we don't use it. These were websocket connections to processes on ec2, via NLB and cloudfront. Not sure exactly what part of that chain was broken yet.

[zedpm]

This whole time I've been seeing intermittent timeouts when checking a UDP service via NLB; I've been wondering if it's general networking trouble or something specifically with the NLB. EC2 hosts are all fine, as far as I can tell.