influxdb

InfluxDB

This directory contains a docker-compose environment that demonstrates a data pipeline composed of slow_cooker, linkerd, Telegraf, InfluxDB, and Grafana.

Startup

The docker-compose.yml file that's included in this directory is configured to run the demo. Start everything with:

$ docker-compose build && docker-compose up -d

That command will start Grafana on port 3000 on your Docker host, which you can use to view the demo in action. Set the DOCKER_IP environment variable to your Docker IP (e.g. DOCKER_IP=localhost for Docker For Mac, or DOCKER_IP=$(docker-machine ip) for Docker Toolkit), and then open the dashboard with:

$ open http://$DOCKER_IP:3000 # on OS X

It will look like this:

The Docker environment runs the following containers, listed in order of front-to-back with respect to the data pipeline:

• grafana: A Grafana instance, serving its UI on port 3000, surfacing metrics from influxdb. The Docker environment also boots a grafana_bootstrap container, to configure Grafana at runtime to connect to InfluxDB.

• influxdb: An InfluxDB instance, configuring to receive metrics from Telegraf and serve metrics to Grafana.

• telegraf: A Telegraf instance, configuring to collect metrics from each linkerd, and send them to influxdb.

• apps: Four app instances, grouped two each as app1 and app2. The app service is defined in app.go. The app service responds to HTTP requests with the string "pong". Each app server is configured with varying levels or latency and success rate.

• linkerd: Two linkerd instances, both configured to route requests to the app services. The linkerd instances vary in the distribution of traffic between app1 and app2, defined in their dtabs in linkerd1.yml and linkerd2.yml.

• slow_cooker: Two slow_cooker instances, slow_cooker1 and slow_cooker2, generate load on linkerd1 and linkerd2, respectively. The slow_cooker instances vary in the amount of traffic sent to their respective linkerd's.

Troubleshooting

If you have any issues getting the demo up and running, pop into linkerd's Slack and we'll help you get it sorted out.

Thanks! 👋