Publish the plugin.

When using the plugin as a part of a continuous delivery pipeline, the name of the tasks can be confusing.
For example the task deployAll could be interpretet as deploying a single unit to a Maven Repository, a docker registry and a runtime enivroment all at once.

Suggestions for new names:

==== deploy
deployToCloud
cloudDeployment

==== deployAll
deployAllToCloud
batchCloudDeployment

See this workaround to accomplish automatic refresh of the DC/OS token: https://medium.com/@richardgirges/authenticating-open-source-dc-os-with-third-party-services-125fa33a5add#.qyiniam1a

Afterwards no user interaction should be necessary.

Fix all issues Sonar detected.

Add integration tests for the following topics:

• Strategy deletion:
• MarathonReplaceStrategyIntegrationTest.testDelete()
• KubernetesReplaceStrategyIntegrationTest.testDelete()
• KubernetesResetStrategyIntegrationTest.testDelete()
• Strategy update:
• GroupResourceIntegrationTest.testUpdate()
• AppResourceIntegrationTest.testUpdate()
• MarathonUpdateStrategyIntegrationTest
• KubernetesUpdateStrategyIntegrationTest
• Deployer (use the test project to test them - should test everything):
• MarathonDeployer
• KubernetesDeployer

Add tests for commons module.

• BaseResourceConfigFactory
• ContentTreeUtil
• FileUtil
• DeployerMessageBundle
• BasePingResource
• BaseResourceFactory
• Blocker
• ClientFactory
• ResponseInterpreterUtil
• BaseDeletionStrategy
• BaseReplaceStrategy
• BaseUpdateStrategy
• BaseDeployer
• BaseResource

Create a start page and fill the wiki.

• Startpage
• Wiki
• Travis integration
• Code Climate integration
• Codedev integration

Add code examples for closures, methods and properties in the wiki pages.

Update the DC/OS test cluster, as the current marathon version (1.3.0-RC4) doesn't support app updates.
See: mesosphere/marathon#4280

Rename the existing update strategies to "RESET" and "REPLACE". Additionally create a "UPDATE" strategy (see #7).

Retry calls after a short waiting period if the server returns a 5xx or 409 error.

The complete source code has to be documented.

Hi there,

I am trying to deploy into Rancher Kubernetes and receive the following error:

de.qaware.cloud.deployer.commons.error.ResourceException: Resource supports no update (Resource: ReplicationController: /booking-service) (Environment: kub-lab)

Here's my YML:

apiVersion: v1
kind: Service
metadata:
  labels:
    name: my-booking-service
  name: my-booking-service
spec:
  ports:
    - port: 8761
  selector:
    name: booking-service
  type: LoadBalancer
---
apiVersion: v1
kind: ReplicationController
metadata:
  name: booking-service
  labels:
    name: booking-service
spec:
  containers:
    - image: <validImageName>
      name: booking-service
      ports:
        - containerPort: 8761
          name: booking-service

When trying to switch to REPLACE or RESET strategy, I receive a HTTP status code 405 because DELETE seems not to be supported.

So is this an issue with the plugin or more likely with me running kubernetes via rancher?

Overwrite dcosAuthToken(void) with dcosAuthToken(File).

As updates of Pods and ReplicationControllers are not supported, an exception should be thrown and no updates at all should be executed.

The deployTask should accept the environment as a option. For more details see the seu-as-code credentials plugin.

Build project with travis.

It's possible to use a kubernetes service account token to login.

Those methods should use a format function and not String concatenation.

If possible, all error messages should contain the id of the environment they belong to or a filename.

Remove dependencies.gradle from build.gradle and enable plugin publishing without it.

Use WireMock to mock kubernetes/marathon REST api.

• kubernetes
• marathon
• extract test credentials in setup script which adds them as env-vars
• separate reports
• feed reports in sonarqube

Create tests for the kubernetes module and reach a coverage of over 80%.

Optimize log messages for Replace- and ResetStrategy in kubernetes and marathon.

Add a task which checks whether the specified authentication options are valid or not.

Right now only the error code and not the message is logged. Printing the error message might help to detect failures.

Goal: Provide best-in-class developer support for the interface of developers to cluster orchestrators

• Connector to OpenShift [Prio 1]
• Connector to Kontena [Prio 3]
• Connector to Nomad [Prio 3]
• Integrated syntax and consistency checks for all formats [Prio 1]
• Linting of decriptors (best practices, pitfalls) [Prio 2]
• Provide a visualization (e.g. like the OpenShift visualization) [Prio 2]
• Also provide a Maven plugin [Prio 3]
• Add blocking feature, which blocks until the container started [Prio 1] (See #50)
• Remove all code style issues (Sonarqube, Codacy, Codebeat) [Prio 1]

Hi,

is there any way to make the deployAll task run and block until the deployed Pods are up and running? In my case we're trying to use this in a Jenkinsfile as part of an integration test suite and we're running in the problem, that the deployment task finishes too soon and thus the tests fail.

cheers
christian

Check authorization on start and throw an error if authorization is incorrect.

See the version eye badge to stay informed about outdated dependencies:

The ssl certificate is specified as string right now. But in future a file containing the certificate should be specified

Hi,

thanks for the plugin!
I am trying to deploy into a K8s cluster running in a Rancher setup. Basically Rancher gives you a valid kubectl config file. Is the baseURL for your gradle plugin meant to be the "server" url from the kubectl file? For when I use that one I receive a 405 from my k8s server. However deploying my .yml manually via kubectl works fine.

cheers,
Christian

The cloud package should be renamed to environment.

Implement and test the marathon module.

Add log messages for the ping tests in marathon and kubernetes.

Adding the environment id at the end of an exception message works as long as it's a single line message. Otherwise it looks strange.

Take a look at hystrix for call retries.
See: https://github.com/Netflix/Hystrix

The logging output should be printed during build.

As the tokens have different names than authentication- and api-token, they should be renamed as explained in #16.

Add @auther tag to all source files.

Move DC/OS token functionality into dcos module.

Use the seuac-credentials-plugin (https://github.com/seu-as-code/seu-as-code.plugins) to store the credentials of the plugin-test.

Optimize the plugin config and workflow.

The coverage needs to be increased to 90%.

Current coverage:

It seems, that the DC/OS authentication token, which is used to retrieve the api token, expires after 10 days.

Asked for the expiration date of such a token on stackoverflow:
https://stackoverflow.com/questions/39546227/dc-os-authentication-vs-api-token

Create a delete task which deletes the specified environment.

Test the plugin module.

• DeployerFactory
• EnvironmentConfigFactory
• EnvironmentFactory
• BaseAllEnvironmentsTask
• BaseSingleEnvironmentTask
• DeleteAllTask
• DeleteTask
• DeployAllTask
• DeployTask
• DcosAuthTokenInitializer
• DefaultTokenInitializer
• OpenIdConnectTokenInitializer
• DeployerPlugin

Create roullout update and canary release support for marathon and kubernetes.

The log messages need to be optimized.

Read the DC/OS token from the .toml config file on startup.