Delete deployment and associated pods?

[de1m]

Hi,

const Api = require('kubernetes-client');

const deploy = new Api.Extensions({
    url: 'https://SERVER:6443',
    version: 'v1beta1',  // Defaults to 'v1beta1'
    insecureSkipTlsVerify: true,
    auth: {
        user: 'root',
        pass: 'pass'
    }
});

function print(err, result) {
    console.log(JSON.stringify(err || result, null, 2));
}

// deploy.ns('fortesting').deploy('echoserver').get(print);
deploy.ns('fortesting').deploy({name: 'echoserver', qs: { PropagationPolicy: false }}).delete(print);

Output:

{
  "code": 404
}

I can delete the deploy with 'deploy.ns('fortesting').deploy('echoserver').delete(print);` but this command cannot delete the pods, which I've created through deploy.

I've found the issue #62 , the option orphanDependents is depricated ([https://kubernetes.io/docs/api-reference/v1.6/#deployment-v1beta1-extensions](doc api)) it's now PropagationPolicy.

[silasbw]

Sorry for the delay .. did you resolve this? Did you try:

deploy.ns('fortesting').deploy('echoserver').delete({ qs: { PropagationPolicy: false }}, print);

or

deploy.ns('fortesting').deploy.delete({ name: 'echoserver' qs: { PropagationPolicy: false }}, print);

[gempesaw]

Somewhat confusingly, the Concepts docs say you can send a body in to the DELETE endpoint as well: https://kubernetes.io/docs/concepts/workloads/controllers/garbage-collection/#setting-the-cascading-deletion-policy

$ kubectl proxy --port=8080 &
$ curl -X DELETE localhost:8080/apis/extensions/v1beta1/namespaces/default/replicasets/my-repset \
-d '{"kind":"DeleteOptions","apiVersion":"v1","propagationPolicy":"Foreground"}' \
-H "Content-Type: application/json"

which IIRC you can do with this library something like

// warning: untested code
deploy.ns('foresting').deploy('echoserver').delete({ body: { kind: 'DeleteOptions', apiVersion: 'v1', propagationPolicy: 'Foreground' } })

although I'm not 100% sure on that, since we don't do that method anymore. With our version of Kubernetes (1.8), the propagationPolicy: 'Foreground' was deleting the deployment and the pods, but it was orphaning replica sets, which recreated the pods on their own.

Our solution was to trace kubectl delete deployment echoserver -v9 to see how they did it, and then we tried to replicate that logic in our code:

1. get the deployment (to make sure it exists, i guess)
2. pause the deployment and set replicas to 0. (you can use the body from 1 to PUT the whole thing along with the pause and replicas: 0. or we ended up doing a PATCH with body { spec: { replicas: 0, revisionHistoryLimit: 0, paused: true } }.
3. get all the matching replica sets for the paused deployment
4. for each replica set, PUT or PATCH set its desired replicas to 0 to delete the pods via spec.replicas
5. loop on a GET of the replica sets until the status.replicas is 0
6. delete the now-empty replica sets
7. delete the now-empty paused deployment

[RobertDiebels]

Has someone tried this out yet? I've made an attempt though it was unsuccessful. Also, the fact that kubectl does it like described above is kind of disturbing. I'd assume a delete request would be sufficient.