You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I was looking for a feature to add two Operators as dependencies to a custom operator that I'm working on. I understand that the OLM dependency feature would perfectly fit for my use case.
After reading the docs (Dependency Resolution) and (Required CRDs), I updated my ClusterServiceVersion to add the required dependencies as follows:
customresourcedefinitions:
required:
- name: infinispans.infinispan.orgversion: v1kind: InfinispandisplayName: Infinispan Clusterdescription: Represents a Infinispan cluster used internally by the Data Index Service
- name: kafkas.kafka.strimzi.ioversion: v1beta1kind: KafkadisplayName: Strimzi Kafka Clusterdescription: Represents a Kafka cluster used internally by the Data Index Serviceowned:
- version: v1alpha1kind: MyKinddescription: My Kind DescriptiondisplayName: MyKindDisplayNamename: mykind.namespace.example.org
In this case, this operator will leverage from Infinispan and Kafka operators.
I deployed the application into Quay and pushed using operator-courier. Then created an entry into the OperatorHub:
After that, I created a subscription to the operator into a new namespace. The operator wasn't installed, although the installed-custom-namespace-5767c568db-85ghx pod and the subscription were successfully created.
What features I was expecting:
During the subscription, an interface hint about the required dependencies for this custom operator taken from the ClusterServiceVersion file (required.displayName and required.description)
Those dependencies automatically installed together with my custom operator
After manually installing both dependencies, my custom operator was successfully deployed. This is nice since it's enforcing the dependencies, but from an user point of view I missed the mentioned expected features.
I suppose I did something wrong. I was hoping that you could help me figure this out. 😃
Many thanks!
The text was updated successfully, but these errors were encountered:
Am I guessing correctly that you are using openshift 4.1.z? If so, there is an issue in 4.1.z that prevents OLM from doing automatic dependency resolution.
The easiest workaround is, as you suggest, to manually select the dependency to install. This issue is fixed in 4.2.
Yes, I'm on 4.1.z, that's the reason why I'm seeing this then. Many thanks for your explanation. Makes sense, I'll set the expectations of my team regarding this issue on 4.1.z.
I was looking for a feature to add two Operators as dependencies to a custom operator that I'm working on. I understand that the OLM dependency feature would perfectly fit for my use case.
After reading the docs (Dependency Resolution) and (Required CRDs), I updated my
ClusterServiceVersion
to add the required dependencies as follows:In this case, this operator will leverage from Infinispan and Kafka operators.
I deployed the application into Quay and pushed using
operator-courier
. Then created an entry into the OperatorHub:After that, I created a subscription to the operator into a new namespace. The operator wasn't installed, although the
installed-custom-namespace-5767c568db-85ghx
pod and the subscription were successfully created.What features I was expecting:
ClusterServiceVersion
file (required.displayName
andrequired.description
)After manually installing both dependencies, my custom operator was successfully deployed. This is nice since it's enforcing the dependencies, but from an user point of view I missed the mentioned expected features.
I suppose I did something wrong. I was hoping that you could help me figure this out. 😃
Many thanks!
The text was updated successfully, but these errors were encountered: