Kubernetes Cluster: Using SkyWalking to Collect Tracing Data from Applications¶

Introduction¶

In a Kubernetes cluster, applications are deployed as images. When using SkyWalking to collect tracing data from applications, following the traditional approach—embedding the SkyWalking Agent into each application image before deployment—leads to overly large images and makes future SkyWalking version upgrades inconvenient. This article introduces a method that uses the official SkyWalking image to provide the SkyWalking Agent via initContainers for collecting tracing data.

Prerequisites¶

Install DataKit¶

Install DataKit
DataKit access version ≥ 1.4.15

Enable the Input¶

To enable the SkyWalking collector, you need to add the file /usr/local/datakit/conf.d/skywalking/skywalking.conf on the server where DataKit is running. When deploying with DaemonSet, you must first define a ConfigMap and then mount the file into the DataKit container. You will need the datakit.yaml file used during DataKit deployment.

Define skywalking.conf¶

Add the definition of skywalking.conf to the ConfigMap resource in the datakit.yaml file.

apiVersion: v1
kind: ConfigMap
metadata:
  name: datakit-conf
  namespace: datakit
data:  
    ## The following is the new content
    skywalking.conf: |- 
        [[inputs.skywalking]]
          ## skywalking grpc server listening on address
          address = "0.0.0.0:11800"

Mount skywalking.conf¶

Add the following content under volumeMounts in the datakit.yaml file.

        - mountPath: /usr/local/datakit/conf.d/skywalking/skywalking.conf
          name: datakit-conf
          subPath: skywalking.conf

Restart DataKit¶

kubectl delete -f datakit.yaml
kubectl apply -f datakit.yaml

Ingesting Java Tracing Data¶

Create a Test Application¶

The example uses datakit-springboot-demo. Download it and build it into a JAR file named service-demo-1.0-SNAPSHOT.jar.

Build the Image¶

Write a Dockerfile.

FROM openjdk:8u292

RUN /bin/cp /usr/share/zoneinfo/Asia/Shanghai /etc/localtime
RUN echo 'Asia/Shanghai' >/etc/timezone
ENV jar service-demo-1.0-SNAPSHOT.jar

ENV workdir /data/app/
RUN mkdir -p ${workdir}
COPY ${jar} ${workdir}
WORKDIR ${workdir}
ENTRYPOINT ["sh", "-ec", "exec java ${JAVA_OPTS} -jar ${jar} ${PARAMS} "]

Build the image and upload it to the image registry.

docker build -t 172.16.0.246/df-demo/service-demo:v1  .
docker push 172.16.0.246/df-demo/service-demo:v1

Deploy the Application¶

In the skywalking-demo.yaml deployment file, use initContainers to initialize the apache/skywalking-java-agent:8.7.0-alpine image. The Pod's application can then access skywalking-agent.jar. Define the startup command in JAVA_OPTS, where -Dskywalking.agent.service_name=skywalking-demo-master sets the application alias to skywalking-demo-master.

apiVersion: apps/v1
kind: Deployment
metadata:
  name: skywalking-demo-deployment
spec:
  replicas: 1
  selector:
    matchLabels:
      app: skywalking-demo-pod
  template:
    metadata:
      labels:
        app: skywalking-demo-pod
      annotations:
    spec:
      containers:
      - name: skywalking-demo-demo-container
        image: 172.16.0.246/df-demo/service-demo:v1
        env:
        - name: HOST_IP
          valueFrom:
            fieldRef:
              apiVersion: v1
              fieldPath: status.hostIP
        - name: HOST_NAME
          valueFrom:
            fieldRef:
              apiVersion: v1
              fieldPath: spec.nodeName      
        - name: JAVA_OPTS
          value: |-
                -javaagent:/skywalking/agent/skywalking-agent.jar -Dskywalking.agent.service_name=skywalking-demo-master  -Dskywalking.collector.backend_service=$(HOST_IP):11800           
        ports:
        - containerPort: 8090
          protocol: TCP
        volumeMounts:
        - mountPath: /skywalking
          name: skywalking-agent
      initContainers:
        - name: agent-container
          image: apache/skywalking-java-agent:8.7.0-alpine
          volumeMounts:
            - name: skywalking-agent   
              mountPath: /agent
          command: [ "/bin/sh" ]
          args: [ "-c", "cp -R /skywalking/agent /agent/" ]  
      restartPolicy: Always
      volumes:
        - name: skywalking-agent
          emptyDir: {}

kubectl apply -f skywalking-demo.yaml

Access the Application¶

Check the Pod's port.

Invoke the application to generate tracing data.

curl 10.244.36.98:8090/ping

Application Performance Monitoring (APM)¶

Log in to Guance and navigate to the Application Performance Monitoring (APM) module to view the skywalking-demo-master application.