CODEWITHSUNDEEP
openshift
dagrun
dagrun

Reputation: 651

Openshift Init:CrashLoopBackOff

When trying to deploy my project to openshift with the yml-file underneath I get an error of Init:CrashLoopBackOff. Does anyone see what is wrong or have any tips on how to investigate this further than just looking at the Events of the pod. The events say "Container image "url/the-app:snapshot" already present on machine"

apiVersion: apps/v1
kind: Deployment
metadata:
  labels:
    app: the-app
  name: the-app
spec:
  replicas: 1
  selector:
    matchLabels:
      app: the-app
  strategy:
    type: Recreate
    redeployOnConfigChange: true
  template:
    metadata:
      labels:
        app: the-app
    spec:
      containers:
        - image: {{ image_registry }}/the-app:{{ the-app-version }}
          imagePullPolicy: Always
          name: the-app
          ports:
            - containerPort: 10202
              protocol: TCP
          livenessProbe:
            httpGet:
              path: /actuator/health
              port: 10202
              scheme: HTTP
            initialDelaySeconds: 20
            timeoutSeconds: 10
          readinessProbe:
            httpGet:
              path: /actuator/health
              port: 10202
              scheme: HTTP
            initialDelaySeconds: 20
            timeoutSeconds: 10
          env:
            - name: SPRING_PROFILES_ACTIVE
              value: "prod"
            - name: ENVIRONMENT_SERVER_PORT
              value: "10202"
          volumeMounts:
            - name: my-config
              mountPath: /my-config
            - name: input-data
              mountPath: /input-data
      initContainers:
        - name: init-test-data
          image: {{ image_registry }}/the-app:{{ the-app-version }}
          command: ["/bin/sh","-c"]
          args: ['apt-get install --yes git clone https://MYGITREPO.git']
          volumeMounts:
            - mountPath: /tmp
              name: input-data
      volumes:
        - name: my-config
          configMap:
            name: my-configmap
        - name: input-data
          emptyDir: {}
  identity:
    enabled: true
  logging:
    index: "{{ splunk_index }}"

---
apiVersion: v1
kind: Service
metadata:
  labels:
    app: the-app
  name: the-app
spec:
  ports:
    - name: 10202-tcp
      port: 10202
      protocol: TCP
      targetPort: 10202
  selector:
    app: the-app
status:
  loadBalancer: {}

Upvotes: 0

Views: 2372

Answers (1)

larsks
larsks

Reputation: 311387

The state Init:CrashLoopBackOff means that one of your initContainers is crashing. You would normally diagnose this by looking at the logs for the container (kubectl logs the-app -c init-test-data), but in this case there's an obvious problem.

You've set the args in your init-test-data container to...

apt-get install --yes git clone https://MYGITREPO.git

...but that's not a valid command. It looks like you have accidentally mashed two commands together. Something like this might work:

command:
  - /bin/sh
  - -c
args:
  - |
    apt-get install --yes git
    git clone https://MYGITREPO.git

Although there's really no particular reason to use both command and args; this will also work just fine:

command:
  - /bin/sh
  - -c
  - |
    apt-get install --yes git
    git clone https://MYGITREPO.git

(But you probably want to explicitly cd to an appropriate directory first).

NB: In YAML, this:

command:
  - /bin/sh
  - -c

Is exactly identical to:

command: ["/bin/sh", "-c"]

But I prefer the previous form.

Upvotes: 1

Related Questions