• Overview
• Build it
• Deploy it
  • Docker
  • Kubernetes
• Use it
  • Docker
    • Preparation
    • Jaeger
    • Prometheus
    • ELK
  • Kubernetes
    • Preparation
    • Jaeger
    • Prometheus
    • ELK
• item
• order
• util
• License

Example of instrumenting a Go microservices application:

• Structured logging via zap
• Automatic endpoint monitoring exposing metrics to Prometheus
• Internal & distributed tracing via Jaeger

Deployments can be made via Docker and on Kubernetes. Additional instrumentation:

• Monitoring of demo app and internal Kubernetes components via Prometheus Operator
• Automatic container log aggregation via ELK Stack

The example application consists of two services: item and order. Both service have a seperate redis instance as their primary data store. When an order gets placed, order contacts item to check if the wished items with the requested quantities are in stock:

API endpoints of both services are instrumented via Prometheus and export the following metrics:

• in_flight_requests: a gauge of requests currently being served by the wrapped handler
• api_requests_total: a counter for requests to the wrapped handler
• request_duration_seconds: a histogram for request latencies
• response_size_bytes: a histogram for response sizes

Additionally, all requests are traced via Jaeger:

On Kubernetes, next to the application and node-specific metrics exposed via node_exporter, internal Kubernetes components are monitored as well:

• kube-apiserver
• kube-dns
• etcd
• kube-controller-manager
• kube-scheduler
• kubelet
• kube-state-metrics

Proemtheus is fully managed via the Prometheus Operator, allowing flexible deployment of new instances, rules, alertmanagers as well as convenient monitoring of Services via ServiceMonitors. A Mailhog instance is deployed to test the alerting pipeline.

For the ELK Stack, a Filebeat DaemonSet is watching all container logs on the nodes.

To build it from source you need Go 1.11+ installed.

This project uses Go Modules so you can clone the repo to anywhere:

git clone https://github.com/obitech/micro-obs.git
cd micro-obs/

Run make to test & build it:

make
# ...

Or make docker to build a Docker image:

make docker TAG=yourname/yourimage
# ...

Make sure Docker and Docker Compose are installed.

Deploy the stack:

cd deploy/docker
docker-compose up -d

Make sure Docker and Kubernets (via Minikube or microk8s, for example) are installed.

Deploy the monitoring stack first:

kubectl create -f deploy/k8s/000-monitoring --recursive

It might take a while for the Prometheus Operator CRDs to be created. If the deployment fails, try running above command again.

Check if everything is up:

$ kubectl get pods -n monitoring
NAME                                   READY     STATUS    RESTARTS   AGE
alertmanager-main-0                    2/2       Running   0          1m
elasticsearch-f8dc4db44-l7ddn          2/2       Running   0          1m
filebeat-pnx76                         2/2       Running   0          1m
grafana-5bfbfb9665-nf9c9               1/1       Running   0          1m
jaeger-deployment-ff5c4dccc-tfxns      1/1       Running   0          1m
kibana-86c4b5577b-dxtpf                1/1       Running   0          1m
kube-state-metrics-58dcbb8579-hsw78    4/4       Running   0          1m
logstash-74479c885f-5lfcw              1/1       Running   0          1m
mailhog-778bd8484d-8vqfg               1/1       Running   0          1m
node-exporter-swtd6                    2/2       Running   0          1m
prometheus-core-0                      3/3       Running   0          1m
prometheus-operator-85bb8dc95b-wqpx6   1/1       Running   0          1m

Next deploy micro-obs stack:

kubectl create -f deploy/k8s/100-micro-obs/ --recursive

Check if everything is up:

$ kubectl get pods -n micro-obs
NAME                           READY     STATUS    RESTARTS   AGE
item-54d9d7d554-2thtl          1/1       Running   0          1m
order-6549584969-k2cp8         1/1       Running   0          1m
redis-item-5bcf99c9f7-zdf2r    2/2       Running   0          1m
redis-order-68869c7986-4s7w2   2/2       Running   0          1m

First build the dummy CLI application:

make build-dummy

Create some dummy data:

./bin/dummy data all

After creating dummy data, those transactions can be found in the Jaeger Query UI at http://localhost:16686:

Send some dummy requests:

./bin/dummy requests all / /aksdasd /items /orders /delay /error -n 100 -c 5 

Start Grafana and upload the deploy/docker/dashboards/micro-obs.json dashboard:

After sending some requests, logs can be queried via Kibana running on http://localhost:5601:

Create some dummy data:

./bin/dummy data all -i http://localhost:30808 -o http://localhost:30809

After creating the dummy data, those transactions can be found in the Jaeger Query UI at http://localhost:30686:

Both the Kubernetes' internal components as well as the micro-obs application (TODO) is being monitored by Prometheus. Some pre-installed Dashboards can be found in Grafana via http://localhost:30300 (default login is admin:admin):

TODO

Request:

POST http://localhost:8080/items
[
	{
		"name": "banana",
		"desc": "a yello fruit",
		"qty": 5
	},
	{
		"name": "water",
		"desc": "bottles of water",
		"qty": 10
	},
	{
		"name": "apple",
		"desc": "delicious",
		"qty": 15
	}
]

Response:

{
    "status": 201,
    "message": "items BxYs9DiGaIMXuakIxX, GWkUo1hE3u7vTxR, JAQU27CQrTkQCNr,  created",
    "count": 3,
    "data": [
        {
            "name": "banana",
            "id": "BxYs9DiGaIMXuakIxX",
            "desc": "a yello fruit",
            "qty": 5
        },
        {
            "name": "water",
            "id": "GWkUo1hE3u7vTxR",
            "desc": "bottles of water",
            "qty": 10
        },
        {
            "name": "apple",
            "id": "JAQU27CQrTkQCNr",
            "desc": "delicious",
            "qty": 15
        }
    ]
}

Request:

POST http://localhost:8090/orders/create
{
	"items": [
		{
			"id": "BxYs9DiGaIMXuakIxX",
			"qty": 2
		},
		{
			"id": "GWkUo1hE3u7vTxR",
			"qty": 8
		}
	]
}

Response:

{
    "status": 201,
    "message": "order 1 created",
    "count": 1,
    "data": [
        {
            "id": 1,
            "items": [
                {
                    "id": "BxYs9DiGaIMXuakIxX",
                    "qty": 2
                },
                {
                    "id": "GWkUo1hE3u7vTxR",
                    "qty": 8
                }
            ]
        }
    ]
}

MIT