tsaarni

Running Kind cluster

Preconditions

Install docker using install instructions at https://docs.docker.com/engine/install/ubuntu/:

$ sudo apt-get install -y \
    ca-certificates \
    curl \

tsaarni

#!/usr/bin/env python3

import os
import subprocess
import requests
import json
import gzip
import hashlib

SOURCE_IMAGE = 'docker://alpine:latest'

tsaarni

proto_print /tmp/tmpoyvg8bdx/b/envoy/api/v2/ratelimit/ratelimit.proto
proto_print /tmp/tmpoyvg8bdx/b/envoy/api/v2/core/health_check.proto
proto_print /tmp/tmpoyvg8bdx/b/envoy/api/v2/cluster/outlier_detection.proto
proto_print /tmp/tmpoyvg8bdx/b/envoy/admin/v2alpha/tap.proto
proto_print /tmp/tmpoyvg8bdx/b/envoy/api/v2/rds.proto
proto_print /tmp/tmpoyvg8bdx/b/envoy/api/v2/listener/udp_listener_config.proto
proto_print /tmp/tmpoyvg8bdx/b/envoy/api/v2/cluster/filter.proto
proto_print /tmp/tmpoyvg8bdx/b/envoy/admin/v2alpha/server_info.proto
proto_print /tmp/tmpoyvg8bdx/b/envoy/api/v2/listener/quic_config.proto
Please apply following patch to directory './api'

tsaarni

How to use debug containers

Following K8s feature-gate must be enabled in kind (kubeadm) config file to access the feature

featureGates:
  EphemeralContainers: true

First we start debug / emphemeral container and attach it to contour shutdown-manager container.

tsaarni

#!/bin/env python3
#
# This script can be used as a server when you need to test the handling of
# TCP connection establishment timeouts of your client.
#
# The protocol can be HTTP, HTTPS or just about anything else, since connection
# will never be established.  It will hang in TCP handshake.
#

import socket

tsaarni

Manual test procedure for slow start mode

This document describes manual test procedure for PR projectcontour/contour#4772

The test is executed by using k6. Additionally, it uses InfluxDB to record the performance test results and Grafana to visualize them.

Preparations

Spin up influxdb and grafana using docker-compose:

tsaarni

Step-by-step development tutorial: making a code change to Contour and seeing live results

This tutorial is a step-by-step guide to making a small code change to Contour. It shows how to run Contour locally on your laptop and have it control Envoy(s) running in a Kind cluster. It allows for a very fast feedback cycle and easy debugging.

Preparation

Create a Kind cluster, for example by running:

tsaarni

apiVersion: apps/v1
kind: Deployment
metadata:
  name: echoserver
spec:
  selector:
    matchLabels:
      app: echoserver
  template:
    metadata:

tsaarni

Fetching OIDC discovery endpoint from https://kubernetes.default.svc/.well-known/openid-configuration...
Fetching public keys from https://172.20.0.3:6443/openid/v1/jwks...
Validating the default service account token in the pod...
Token is valid!
Claims: {
    "aud": [
        "https://kubernetes.default.svc.cluster.local"
    ],
    "exp": 1749207355,
    "iat": 1717671355,

tsaarni

EDS Cache Tests

This document contains manually executed test scenarios related to PR projectcontour/contour#6906

Analyzing xDS gRPC messages

To collect gRPC debugs, you can use https://github.com/tsaarni/grpc-json-sniffer.

1. Apply the patch included in this gist patch -p1 < grpc-sniffer.patch.