Conversation
Signed-off-by: Ken Sipe <kensipe@gmail.com>
| status: | ||
| phase: Running | ||
|
|
||
| # confirms that kudo is running |
There was a problem hiding this comment.
Why is it necessary to check this here?
There was a problem hiding this comment.
fail fast and portability
frankly if kudo is running this is a quick sub second check... what are the concerns?
There was a problem hiding this comment.
it's not a pattern we should promote: this is something that is covered beforeAll tests (by using kudo init --wait) option. I guess you could see it as very defensive programming but frankly, this shouldn't be necessary
| kind: TestStep | ||
| commands: | ||
| - command: kubectl delete storageclass standard | ||
| - command: kubectl apply -f https://raw.githubusercontent.com/rancher/local-path-provisioner/master/deploy/local-path-storage.yaml |
There was a problem hiding this comment.
This is applying the same StorageClass that has just been deleted, see https://github.com/rancher/local-path-provisioner/blob/master/deploy/local-path-storage.yaml#L82. Why is this necessary to do here? This also means that the error condition above will always fire.
There was a problem hiding this comment.
@kensipe I see this as an infra step. If we run these tests along with other tests present in this repository we might be changing the outcome of the tests. What about just using the storage class that is already present and which is being deleted?
There was a problem hiding this comment.
yeah... these tests are great for setting up and running / verifying the flink demo... perhaps we shouldn't hook them into the infra... is there any suggestion?
| apiVersion: kuttl.dev/v1beta1 | ||
| kind: TestStep | ||
| commands: | ||
| - command: kubectl kudo install zookeeper --operator-version=0.3.0 --skip-instance |
There was a problem hiding this comment.
Let's use local packages here, i.e. relative paths instead of packages from the community repository. This will allow us to see if (local) changes to any of these operators break this test.
Here and for kafka and flink below.
There was a problem hiding this comment.
perhaps these tests are different... and should be relocated or not hooked into the infra... this is under the "demo" folder and these teams are similar to our use of kuttl for MWT.. they setup a cluster with the flink demo.
On the other hand... I could see value in having the demo checked on PRs to confirm it still works... hmm..
zen-dog
left a comment
There was a problem hiding this comment.
This PR is basically outdated: newest (but unmerged) flink-demo requires only this command to install:
$ kubectl kudo install repository/flink/docs/demo/financial-fraud/demo-operator --instance flink-demo
and none of the --skip-instance prerequisites. The only reason the above PR is not merged is that we needed to bump KUDO to 0.15 first
| apiVersion: kuttl.dev/v1beta1 | ||
| kind: TestStep | ||
| commands: | ||
| - command: kubectl kudo install zookeeper --operator-version=0.3.0 --skip-instance |
There was a problem hiding this comment.
None of these steps are necessary once #279 is merged
|
Let's update this to use a recent version of |
|
One thing might be a problem though: We already install O/OVs for a lot of operators as part of the test setup: https://github.com/kudobuilder/operators/blob/master/kuttl-test.yaml. We should run this test in a separate namespace to ensure that new O/OVs are installed for the |
|
@nfnt unless kuttl is configured to run in a specific namespace... a new namespace is create for each test |
Signed-off-by: Ken Sipe kensipe@gmail.com