This repo contains a environment for demonstrating OpenTelemetry tracing support in Promscale.
This repo is used to deliver workshops on OpenTelemetry Traces and SQL. If you are here to explore the workshop, head to this page.
A password generator service is instrumented with OpenTelemetry tracing. This is an absurd service and should not be taken as a shining example of architecture nor coding. It exists as a playground example to generate traces. The lower service generates random lowercase letters. The upper service service generates random uppercase letters. The digit service generates random digits, and the special service generates random special characters. There is a generator service which makes calls to the other services to compose a random password. Finally, there is a load script which continuously calls the generator service in order to simulate user load.
The lower service is a Ruby app using Sinatra framework while the other services use Python with Flask.
There are two versions of the services. Those under the instrumented directory have explicit instrumentation for tracing, whereas the versions under the uninstrumented directory solely use auto-instrumentation.
All of the microservices forward their traces to an instance of the OpenTelemetry Collector. The collector sends the traces on to an instance of the Promscale Collector which stores them in a TimescaleDB database. An instance of the Jaeger UI is pointed to the Promscale instance, and an instance of Grafana is pointed at both Jaeger and the TimescaleDB database. In this way, you can use SQL to query the traces directly in the database, and visualize tracing data in Jaeger and Grafana dashboards.
When the system runs in docker it is configured via the docker compose file, and is operated with docker-compose. Run the following command from the root of the repo to (re)start the system.
docker-compose up --remove-orphans --build --detach
Once running, the following links will let you explore the various components of the system:
When you are ready to shutdown the system, use the following command.
docker-compose down
When you want to run the system in k8s you can use the following commands (if you have certmanager installed skip that step):
kubectl apply -f https://github.com/cert-manager/cert-manager/releases/download/v1.8.0/cert-manager.yaml
kubectl apply -f https://github.com/open-telemetry/opentelemetry-operator/releases/latest/download/opentelemetry-operator.yaml
kubectl apply -k ./yaml
You should wait for the OpenTelemetry Operator pod to come up before you run the second command.
You should see data flowing into Grafana automatically.
To access individual services as per the Docker instructions below you will need to run
sh ./yaml/port-forward.sh
You can use any SQL client that supports PostgreSQL to connect to the TimescaleDB database. The database is not password protected.
Use the psql client to connect to the Timescaledb instance via:
psql -h localhost -p 5999 -d otel_demo -U postgresNote: the default Grafana login credentials are user "admin" and password "admin". Once you log in for the first time you will be asked to update your password.