Watch Conditions, TriggerBindings, TriggerTemples #492
Conversation
| @@ -288,7 +307,7 @@ func reconcileActiveVersions(stackResource *kabanerov1alpha2.Stack, c client.Cli | |||
| renderingContext["CollectionId"] = cID | |||
| renderingContext["StackId"] = cID | |||
|
|
|||
| ownerIsController := false | |||
| ownerIsController := true | |||
There was a problem hiding this comment.
So, this falls apart when you have two stacks that use the same pipeline. The stack controller tries to add the second stack to the owners array, but the request fails because only one owner can have ownerIsController set. I thought it would be more consistent to have them all set to false here, rather than have the first Stack set it to true and the rest set it to false. There is really nothing special about the first stack.
| - kind: Condition | ||
| name: '' | ||
| version: v1alpha1 | ||
| - kind: TriggerBinding |
There was a problem hiding this comment.
The trigger bindings and trigger templates don't follow the object ownership model (they are in a different namespace) so there will never be any of these objects in the openshift console. It's probably best to leave them off of the resources view (unless there's another reason why they are here, in which case please let me know what that is).
This should fix issue #483 and work towards fixing issue #491
For created assets (tasks/pipelines/triggers) set ownerReference controller=true.
In turn they are properly watched and recreated if modified/deleted.
Add Watch for Condititions
Add Watch for TriggerBindings, TriggerTemplates...
It turns out the TriggerBindings and TriggerTemplates can't really be watched unless Kabanero is a cluster operator. The objects are created in
tekton-pipelinesnamespace, but the operator env WATCH_NAMESPACE is scoped only tokabanero. They don't get an ownerRef.Should probably be picked for 0.6