How to ovewrite an existing pod in kubernetes cluster

[yosser.mahfoudh]

I want to deploy a pod name core-x but that pod is already running in my cluster what I want is to overwrite the old one with the new one.

Knowing that my image version has the same tag?

I need your help please

[yosser.mahfoudh]

- pipe: atlassian/kubectl-run:3.2.0

variables:

KUBE_CONFIG: $KUBE_CONFIG

KUBECTL_COMMAND: 'delete -f masster/build/kubernetes/core-srv.yaml
 | apply'

RESOURCE_PATH: 'core-srv.yaml'

sleep: 10s 

@Yevhen  like this?

[Yevhen]

@yosser.mahfoudh you can't create a new pod with the same name and you can't change name of the existing pod too since name and namespace fields are immutable. Some of the fields can be changed though.

[yosser.mahfoudh]

So if I create a new pod named core-Y. how the core-X stop runing in my kubernetes cluster?

[Yevhen]

@yosser.mahfoudh you will have to explicitly delete it.

[yosser.mahfoudh]

could you pleqse give me an example.

[Yevhen]

That would be "kubectl delete pod $podName -n $namespace" if you're using kubectl.

[yosser.mahfoudh]

This is what I have

step:

name: deploy

trigger: manual

script:

- echo VERSION_BACK=$(grep '<version>' pom.xml | head -1 | tail -1 | sed 's/[<>version/ \t]//g') > env.sh

- source env.sh

- cd /opt/atlassian/pipelines/agent/build/ && ls

- sed -i "s|{{image}}|imagename/core-srv:${VERSION_BACK}|g" core-srv.yaml

- pipe: atlassian/kubectl-run:3.2.0

variables:

KUBE_CONFIG: $KUBE_CONFIG

KUBECTL_COMMAND: 'apply'

RESOURCE_PATH: 'core-srv.yaml'

so should I replace apply by delete ?

[Yevhen]

Yes. Perhaps, to delete the pod you don't need to run any sed against yaml. It'll be deleted as long as a pod with such a name exists in the namespace (its spec does not really matter)

[yosser.mahfoudh]

But it will know which pod I want to delete ?

[Yevhen]

It's will delete whatever is in metadata.name in your yaml.

[yosser.mahfoudh]

I tried what you said but it deletes it and never restart creating the pod from the begin.

[Yevhen]

@yosser.mahfoudh yes, this is what delete command is expected to do. In the next step you can apply again (I'd recommend adding a 10-15 second sleep to give apiserver time to delete the pod).

[yosser.mahfoudh]

@Yevhen 

do you mean like this 

 pipe: atlassian/kubectl-run:3.2.0

variables:

KUBE_CONFIG: $KUBE_CONFIG

KUBECTL_COMMAND: 'delete -f masster/vuild/core-srv.yaml' | apply 

RESOURCE_PATH: 'core-srv.yaml'

[Yevhen]

@yosser.mahfoudh I was more thinking about a new step. So your steps in the pipeline would differ by one command only.

Also, I'd recommend looking at an official K8s documentation and play with kubectl and your pod yaml in the command line to better understand kubectl and cluster behavior.