Communication between pods

Question

I am currently in the process to set up sentry.io but i am having problems in setting it up in openshift 3.11

I got pods running for sentry itself, postgresql, redis and memcache but according to the log messages they are not able to communicate together.

sentry.exceptions.InvalidConfiguration: Error 111 connecting to 127.0.0.1:6379. Connection refused.

Do i need to create a network like in docker or should the pods (all in the same namespace) be able to talk to each other by default? I got admin rights for the complete project so i can also work with the console and not only the web interface.

Best wishes

EDIT: Adding deployment config for sentry and its service and for the sake of simplicity the postgres config and service. I also blanked out some unnecessary information with the keyword BLANK if I went overboard please let me know and ill look it up.

Deployment config for sentry:

apiVersion: apps.openshift.io/v1
kind: DeploymentConfig
metadata:
  annotations:
    openshift.io/generated-by: OpenShiftWebConsole
  creationTimestamp: BLANK
  generation: 20
  labels:
    app: sentry
  name: sentry
  namespace: test
  resourceVersion: '506667843'
  selfLink: BLANK
  uid: BLANK
spec:
  replicas: 1
  selector:
    app: sentry
    deploymentconfig: sentry
  strategy:
    activeDeadlineSeconds: 21600
    resources: {}
    rollingParams:
      intervalSeconds: 1
      maxSurge: 25%
      maxUnavailable: 25%
      timeoutSeconds: 600
      updatePeriodSeconds: 1
    type: Rolling
  template:
    metadata:
      annotations:
        openshift.io/generated-by: OpenShiftWebConsole
      creationTimestamp: null
      labels:
        app: sentry
        deploymentconfig: sentry
    spec:
      containers:
        - env:
            - name: SENTRY_SECRET_KEY
              value: Iamsosecret
            - name: C_FORCE_ROOT
              value: '1'
            - name: SENTRY_FILESTORE_DIR
              value: /var/lib/sentry/files/data
          image: BLANK
          imagePullPolicy: Always
          name: sentry
          ports:
            - containerPort: 9000
              protocol: TCP
          resources: {}
          terminationMessagePath: /dev/termination-log
          terminationMessagePolicy: File
          volumeMounts:
            - mountPath: /var/lib/sentry/files
              name: sentry-1
      dnsPolicy: ClusterFirst
      restartPolicy: Always
      schedulerName: default-scheduler
      securityContext: {}
      terminationGracePeriodSeconds: 30
      volumes:
        - emptyDir: {}
          name: sentry-1
  test: false
  triggers:
    - type: ConfigChange
    - imageChangeParams:
        automatic: true
        containerNames:
          - sentry
        from:
          kind: ImageStreamTag
          name: 'sentry:latest'
          namespace: catcloud
        lastTriggeredImage: BLANK
      type: ImageChange
status:
  availableReplicas: 1
  conditions:
    - lastTransitionTime: BLANK
      lastUpdateTime: BLANK
      message: Deployment config has minimum availability.
      status: 'True'
      type: Available
    - lastTransitionTime: BLANK
      lastUpdateTime: BLANK
      message: replication controller "sentry-19" successfully rolled out
      reason: NewReplicationControllerAvailable
      status: 'True'
      type: Progressing
  details:
    causes:
      - type: ConfigChange
    message: config change
  latestVersion: 19
  observedGeneration: 20
  readyReplicas: 1
  replicas: 1
  unavailableReplicas: 0
  updatedReplicas: 1

Service for sentry:

apiVersion: v1
kind: Service
metadata:
  annotations:
    openshift.io/generated-by: OpenShiftWebConsole
  creationTimestamp: BLANK
  labels:
    app: sentry
  name: sentry
  namespace: test
  resourceVersion: '505555608'
  selfLink: BLANK
  uid: BLANK
spec:
  clusterIP: BLANK
  ports:
    - name: 9000-tcp
      port: 9000
      protocol: TCP
      targetPort: 9000
  selector:
    deploymentconfig: sentry
  sessionAffinity: None
  type: ClusterIP
status:
  loadBalancer: {}

Deployment config for postgresql:

apiVersion: apps.openshift.io/v1
kind: DeploymentConfig
metadata:
  annotations:
    openshift.io/generated-by: OpenShiftWebConsole
  creationTimestamp: BLANK
  generation: 10
  labels:
    app: postgres
    type: backend
  name: postgres
  namespace: test
  resourceVersion: '506664185'
  selfLink: BLANK
  uid: BLANK
spec:
  replicas: 1
  selector:
    app: postgres
    deploymentconfig: postgres
    type: backend
  strategy:
    activeDeadlineSeconds: 21600
    resources: {}
    rollingParams:
      intervalSeconds: 1
      maxSurge: 25%
      maxUnavailable: 25%
      timeoutSeconds: 600
      updatePeriodSeconds: 1
    type: Rolling
  template:
    metadata:
      annotations:
        openshift.io/generated-by: OpenShiftWebConsole
      creationTimestamp: null
      labels:
        app: postgres
        deploymentconfig: postgres
        type: backend
    spec:
      containers:
        - env:
            - name: PGDATA
              value: /var/lib/postgresql/data/sql
            - name: POSTGRES_HOST_AUTH_METHOD
              value: trust
            - name: POSTGRESQL_USER
              value: sentry
            - name: POSTGRESQL_PASSWORD
              value: sentry
            - name: POSTGRESQL_DATABASE
              value: sentry
          image: BLANK
          imagePullPolicy: Always
          name: postgres
          ports:
            - containerPort: 5432
              protocol: TCP
          resources: {}
          terminationMessagePath: /dev/termination-log
          terminationMessagePolicy: File
          volumeMounts:
            - mountPath: /var/lib/postgresql/data
              name: volume-uirge
              subPath: sql
      dnsPolicy: ClusterFirst
      restartPolicy: Always
      schedulerName: default-scheduler
      securityContext:
        runAsUser: 2000020900
      terminationGracePeriodSeconds: 30
      volumes:
        - name: volume-uirge
          persistentVolumeClaim:
            claimName: postgressql
  test: false
  triggers:
    - type: ConfigChange
    - imageChangeParams:
        automatic: true
        containerNames:
          - postgres
        from:
          kind: ImageStreamTag
          name: 'postgres:latest'
          namespace: catcloud
        lastTriggeredImage: BLANK
      type: ImageChange
status:
  availableReplicas: 1
  conditions:
    - lastTransitionTime: BLANK
      lastUpdateTime: BLANK
      message: Deployment config has minimum availability.
      status: 'True'
      type: Available
    - lastTransitionTime: BLANK
      lastUpdateTime: BLANK
      message: replication controller "postgres-9" successfully rolled out
      reason: NewReplicationControllerAvailable
      status: 'True'
      type: Progressing
  details:
    causes:
      - type: ConfigChange
    message: config change
  latestVersion: 9
  observedGeneration: 10
  readyReplicas: 1
  replicas: 1
  unavailableReplicas: 0
  updatedReplicas: 1

Service config postgresql:

apiVersion: v1
kind: Service
metadata:
  annotations:
    openshift.io/generated-by: OpenShiftWebConsole
  creationTimestamp: BLANK
  labels:
    app: postgres
    type: backend
  name: postgres
  namespace: catcloud
  resourceVersion: '506548841'
  selfLink: /api/v1/namespaces/catcloud/services/postgres
  uid: BLANK
spec:
  clusterIP: BLANK
  ports:
    - name: 5432-tcp
      port: 5432
      protocol: TCP
      targetPort: 5432
  selector:
    deploymentconfig: postgres
  sessionAffinity: None
  type: ClusterIP
status:
  loadBalancer: {}

Answer 1

Pods (even in the same namespace) are not able to talk directly to each other by default. You need to create a Service in order to allow a pod to receive connections from another pod. In general, one pod connects to another pod via the latter's service, as I illustrated below:

The connection info would look something like <servicename>:<serviceport> (e.g. elasticsearch-master:9200) rather than localhost:port.

You can read https://kubernetes.io/docs/concepts/services-networking/service/ for further info on a service.

N.B: localhost:port will only work for containers running inside the same pod to connect to each other, just like how nginx connects to gravitee-mgmt-api and gravitee-mgmt-ui in my illustration above.