Emoji.voto, a sample application for Linkerd service mesh

[[412321]]

A microservice application that allows users to vote for their favorite emoji and keeps track of the votes received on a leaderboard. May the best emoji win.

The application consists of the following 3 services:

emojivoto-web: Web frontend and REST API

• https://github.com/BuoyantIO/emojivoto/tree/main/emojivoto-web

emojivoto-emoji-svc: gRPC API for finding and listing emojis

• https://github.com/BuoyantIO/emojivoto/tree/main/emojivoto-emoji-svc

emojivoto-voting-svc: gRPC API for voting and leaderboards

• https://github.com/BuoyantIO/emojivoto/blob/main/emojivoto-voting-svc

Actual Combat

Tencent Cloud K8S Cluster Practice Service Mesh—Linkerd2 & Traefik2 Deployment of emojivoto Application

run

In Minikube

Deploy the application to Minikube using the Linkerd2 service mesh.

1. Install linkerd CLI

 curl https://run.linkerd.io/install | sh

2. Install Linkerd2

 linkerd install | kubectl apply -f -

3. Check out the dashboard!

 linkerd dashboard

4.Inject, Deploy, and Enjoy

 kubectl kustomize kustomize/deployment | \
 linkerd inject - | \
 kubectl apply -f -

5. Use apps!

 minikube -n emojivoto service web-svc

In docker-compose

It is also possible to run the application using docker-compose (without Linkerd2).

Build and run:

 make deploy- to -docker-compose

The web application will run on port 8080 of the docker host.

Via URL

Standalone deployment to an existing cluster:

 kubectl apply -k github.com/BuoyantIO/emojivoto/kustomize/deployment

Generate some traffic

The VoteBot service can bring you some traffic. It votes "randomly" on emojis like this:

• 15% of the votes went to 🍩
• When you don't vote for 🍩, it will randomly select an emoji

If you ran the application using the above instructions, VoteBot will be deployed and will start sending traffic to the voting endpoint.

If you want to run the bot manually:

 export WEB_HOST=localhost:8080 # replace   with your web location
 go run emojivoto-web/cmd/vote-bot/main.go

Release a new version

To build and push a multi-arch docker image:

1. Update the tag name in common.mk

2. Create a Buildx builder instance

 docker run --rm --privileged multiarch/qemu-user-static --reset -p yes  
 docker buildx create   --name=multiarch-builder --driver=docker-container --use  
 docker buildx inspect multiarch-builder --bootstrap  

3. Build & push multi-arch docker image to hub.docker.com

 docker login
 make multi-arch

4. Update:

• docker-compose.yml
• kustomize/deployment/emoji.yml
• kustomize/deployment/vote-bot.yml
• kustomize/deployment/voting.yml
• kustomize/deployment/web.yml

5. Distribute to the Linkerd website repo

 kubectl kustomize kustomize/deployment > ../website/run.linkerd.io/ public /emojivoto.yml
 kubectl kustomize kustomize/daemonset > ../website/run.linkerd.io/ public /emojivoto-daemonset.yml
 kubectl kustomize kustomize/statefulset > ../website/run.linkerd.io/ public /emojivoto-statefulset.yml

Prometheus metrics

By default, the voting service exposes Prometheus metrics about the current vote count on port 8801.

This can be disabled by unsetting the PROM_PORT environment variable.

Local Development

Emojivoto webapp

This application is written in React and bundled using webpack. Use the following command to run emojivoto go services and develop on the front-end.

Set up the proto file and build the application

 make build

Start voting service

 GRPC_PORT=8081 go run emojivoto-voting-svc/cmd/server.go

[In a separate terminal window] Start the emoji server

 GRPC_PORT=8082 go run emojivoto-emoji-svc/cmd/server.go

[In a separate terminal window] Bundle frontend resources

 cd emojivoto-web/webapp
 yarn install
 yarn webpack # one time asset-bundling OR  
 yarn webpack-dev-server --port 8083 # bundle/serve reloading assets  

[In a separate terminal window] Start the web server

 export WEB_PORT=8080
 export VOTINGSVC_HOST=localhost:8081
 export EMOJISVC_HOST=localhost:8082 
 
 # if you ran yarn webpack
 export INDEX_BUNDLE=emojivoto-web/webapp/dist/index_bundle.js 
 
 # if you ran yarn webpack-dev-server
 export WEBPACK_DEV_SERVER=http://localhost:8083 
 
 # start the webserver
 go run emojivoto-web/cmd/server.go

[Optional] Start a voting bot to automatically generate traffic.

 export WEB_HOST=localhost:8080
 go run emojivoto-web/cmd/vote-bot/main.go

View emojivoto

 open http://localhost:8080

Testing the Linkerd Service Configuration File

Service Profiles are a feature of Linkerd that provide per-route functionality such as telemetry, timeouts, and retries. The Emojivoto application aims to showcase service profiles with the following instructions.

Service Profiles: https://linkerd.io/2/features/service-profiles

Generate ServiceProfile definition from .proto file

The emoji and voting services are gRPC applications that have Protocol Buffers (protobuf) definition files. These .proto files can be used as input to the linkerd profile command to create a ServiceProfile definition yaml file. The Linkerd Service Profile documentation outlines the steps required to create the yaml file, and these are the commands you can use from the root of that repository:

 linkerd profile --proto proto/Emoji.proto emoji-svc -n emojivoto  
 linkerd profile --proto proto/Voting.proto voting-svc -n emojivoto  

Protocol Buffers (protobuf): https://developers.google.com/protocol-buffers

gRPC: https://grpc.io

Linkerd Service Profile documentation: https://linkerd.io/2/tasks/setting-up-service-profiles/#protobuf

Each of these commands will output YAML, which you can write to a file or pipe directly to kubectl apply. For example:

• Writing to a file:

 linkerd profile --proto proto/Emoji.proto emoji-svc -n emojivoto > emoji  
 -sp.yaml

• Apply directly:

 linkerd profile --proto proto/Voting.proto voting-svc -n emojivoto | \  
 kubectl apply -f -

Generate ServiceProfile definition for Web deployment

emojivoto's web-svc deployment is a React application hosted by a Go server. We can use linkerd profile auto creation to generate a ServiceProfile resource for web-svc using the following command:

 linkerd profile -n emojivoto web-svc --tap deploy/web --tap-duration 10s | \  
 kubectl apply -f -

Now that service configuration files are generated for all services, you can observe per-service route metrics on the Linkerd Dashboard or using the linkerd routes command.

 linkerd -n emojivoto routes deploy/web-svc --to svc/emoji-svc  

linkerd profile auto creation

• https://linkerd.io/2/tasks/setting-up-service-profiles/#auto-creation

Linkerd Dashboard

• https://linkerd.io/2/features/dashboard