WebSocket connections being disconnected after 100 seconds

[darnfish]

NGINX Ingress controller version: 0.32.0

Kubernetes version (use kubectl version): v1.16.6

Environment: WebSocket server (Mega running on DigitalOcean Kubernetes behind a DO LoadBalancer

• Cloud provider or hardware configuration: 2 GB Memory / 50 GB Disk + 23 GB / NYC3 - Debian do-kube-1.16.6-do.2
• OS (e.g. from /etc/os-release): Debian
• Kernel (e.g. uname -a): Darwin Williams-MacBook-Pro.local 19.4.0 Darwin Kernel Version 19.4.0: Wed Mar 4 22:28:40 PST 2020; root:xnu-6153.101.6~15/RELEASE_X86_64 x86_64 (development)
• Install tools: kubectl apply -f https://raw.githubusercontent.com/kubernetes/ingress-nginx/controller-0.32.0/deploy/static/provider/do/deploy.yaml

What happened:
The WebSocket was closed with code 1006 after 100 seconds

What you expected to happen:
The WebSocket connection would stay open for longer than 100s

How to reproduce it:
Create a DigitalOcean cluster behind a DigitalOcean LoadBalancer (also behind Cloudflare proxy) and deploy a WebSocket server (specifically Mega) with the following Ingress config:

apiVersion: extensions/v1beta1
kind: Ingress
metadata:
  annotations:
    kubernetes.io/ingress.class: nginx

    nginx.org/websocket-services: gateway
    nginx.ingress.kubernetes.io/websocket-services: gateway

    ingress.kubernetes.io/proxy-read-timeout: "3600"
    ingress.kubernetes.io/proxy-send-timeout: "3600"
    ingress.kubernetes.io/proxy-connect-timeout: "3600"

    nginx.ingress.kubernetes.io/proxy-read-timeout: "3600"
    nginx.ingress.kubernetes.io/proxy-send-timeout: "3600"
    nginx.ingress.kubernetes.io/proxy-connect-timeout: "3600"
  name: gateway-redacted-chat
spec:
  rules:
    - host: gateway.redacted.chat
      http:
        paths:
          - path: /
            backend:
              serviceName: gateway
              servicePort: 80

Anything else we need to know:
Initially connections were timing out after 60s. This wasn't fixed by adding nginx.ingress.kubernetes.io/proxy-read-timeout and nginx.ingress.kubernetes.io/proxy-send-timeout, but weirdly it was fixed via ingress.kubernetes.io/proxy-read-timeout, ingress.kubernetes.io/proxy-send-timeout and ingress.kubernetes.io/proxy-connect-timeout. After this, connections started being dropped due to WS code 1006 after 100 seconds.

I have port-forwarded the WebSocket server and connected to it locally without any connection drop issues, so I'm confident the issue resides in either the DigitalOcean LoadBalancer or in ingress-nginx.

/kind bug

[darnfish]

After further testing it seems like it's an issue with CloudFlare, sorry for the hassle!