I am currently working on a template for OpenShift and my ImageChange trigger gets deleted when I initally instantiate the application. My Template contains the following objects
I guess the route is irrelevant but this is what it looks like so far (for better overview I will post the objects seperated, but they are all items in my Template)
ImageStream
- kind: ImageStream
apiVersion: v1
metadata:
labels:
app: my-app
name: my-app
namespace: ${IMAGE_NAMESPACE}
BuildConfig
- kind: BuildConfig
apiVersion: v1
metadata:
labels:
app: my-app
deploymentconfig: my-app
name: my-app
namespace: ${IMAGE_NAMESPACE}
selfLink: /oapi/v1/namespaces/${IMAGE_NAMESPACE}/buildconfigs/my-app
spec:
runPolicy: Serial
source:
git:
ref: pre-prod
uri: 'ssh://git@git.myreopo.net:port/project/my-app.git'
sourceSecret:
name: git-secret
type: Git
strategy:
type: Source
sourceStrategy:
env:
- name: HTTP_PROXY
value: 'http://user:password@proxy.com:8080'
- name: HTTPS_PROXY
value: 'http://user:password@proxy.com:8080'
- name: NO_PROXY
value: .something.net
from:
kind: ImageStreamTag
name: 'nodejs:8'
namespace: openshift
output:
to:
kind: ImageStreamTag
name: 'my-app:latest'
namespace: ${IMAGE_NAMESPACE}
Service
- kind: Service
apiVersion: v1
metadata:
name: my-app
labels:
app: my-app
spec:
selector:
deploymentconfig: my-app
ports:
- name: 8080-tcp
port: 8080
protocol: TCP
targetPort: 8080
sessionAffinity: None
type: ClusterIP
DeploymentConfig
Now what is already weird in the DeploymentConfig is that under spec.template.spec.containers[0].image I have to specify the full path to the repository to make it work, otherwise I get an error pulling the image. (even though documentation says my-app:latest would be correct)
- kind: DeploymentConfig
apiVersion: v1
metadata:
labels:
app: my-app
deploymentconfig: my-app
name: my-app
namespace: ${IMAGE_NAMESPACE}
selfLink: /oapi/v1/namespaces/${IMAGE_NAMESPACE}/deploymentconfigs/my-app
spec:
selector:
app: my-app
deploymentconfig: my-app
strategy:
type: Rolling
rollingParams:
intervalSeconds: 1
maxSurge: 25%
maxUnavailability: 25%
timeoutSeconds: 600
updatePeriodSeconds: 1
replicas: 1
template:
metadata:
labels:
app: my-app
deploymentconfig: my-app
spec:
containers:
- name: my-app-container
image: "${REPOSITORY_IP}:${REPOSITORY_PORT}/${IMAGE_NAMESPACE}/my-app:latest"
imagePullPolicy: Always
ports:
- containerPort: 8080
protocol: TCP
- containerPort: 8081
protocol: TCP
env:
- name: MONGODB_USERNAME
valueFrom:
secretKeyRef:
name: my-app-database
key: database-user
- name: MONGODB_PASSWORD
valueFrom:
secretKeyRef:
name: my-app-database
key: database-password
- name: MONGODB_DATABASE
value: "myapp"
- name: ROUTE_PATH
value: /my-app
- name: MONGODB_AUTHDB
value: "myapp"
- name: MONGODB_PORT
value: "27017"
- name: HTTP_PORT
value: "8080"
- name: HTTPS_PORT
value: "8082"
restartPolicy: Always
dnsPolicy: ClusterFirst
triggers:
- type: ImageChange
imageChangeParams:
automatic: true
from:
kind: ImageStreamTag
name: 'my-app:latest'
namespace: ${IMAGE_NAMESPACE}
containerNames:
- my-app-container
- type: ConfigChange
I deploy the application using
oc process -f ./openshift/template.yaml ..Parameters... | oc apply -f -
But the outcome is the same when I use oc new-app.
The weird thing is. The application gets deployed and is running fine, but image changes will have no effect. So I exported DeploymentConfig and found that it was missing the ImageChangeTrigger leaving the trigger part being
triggers:
- type: ConfigChange
At first I thought this was due to the fact that maybe the build was not ready when I tried to apply the DeploymentConfig so I created a build first and waited for it to finish. Afterwards I deployed the rest of the application (Service, Route, DeploymentConfig). The outcome was the same however. If I use the Webgui and change the DeploymentConfig there from
to this, fill out namespace, app and tag (latest) and hit apply everything works as it should. I just can't figure out why the trigger is beeing ignored initially. Would be great if someone has an idea where I'm wrong
Versions I am using are
oc: v3.9.0
kubernetes: v1.6.1
openshift v3.6.173.0.140
OK the answer was pretty simple. Turned out it was just an indentation error in the yaml file for the DeploymentConfig. Instead of
dnsPolicy: ClusterFirst
restartPolicy: Always
terminationGracePeriodSeconds: 30
triggers:
- type: ImageChange
imageChangeParams:
automatic: true
containerNames:
- alpac-studio-container
from:
kind: ImageStreamTag
name: alpac-studio:latest
- type: ConfigChange
It has to be
dnsPolicy: ClusterFirst
restartPolicy: Always
terminationGracePeriodSeconds: 30
triggers:
- type: ImageChange
imageChangeParams:
automatic: true
containerNames:
- alpac-studio-container
from:
kind: ImageStreamTag
name: alpac-studio:latest
- type: ConfigChange
So the triggers have to be on the same level as e.g. template and strategy