Description of problem: Running RHACM in the IBM Cloud Have successfully imported 1 OpenShift cluster, 1 native K8S cluster and 1 minikube cluster. So we are managing 4 clusters, including the local one. Trying to distribute a small mariadb application to all 4 clusters. It works fine on the OCP clusters, but will not spin the pods up on the non-OCP clusters. Here is what I get on an OCP cluster: $ oc get all -n mariadb NAME READY STATUS RESTARTS AGE pod/mariadb-1-deploy 0/1 Completed 0 46m pod/mariadb-1-vrmqc 1/1 Running 0 46m pod/mariadb-1-z2cks 1/1 Running 0 46m NAME DESIRED CURRENT READY AGE replicationcontroller/mariadb-1 2 2 2 46m NAME TYPE CLUSTER-IP EXTERNAL-IP PORT(S) AGE service/mariadb ClusterIP 172.21.237.26 <none> 3306/TCP 46m NAME REVISION DESIRED CURRENT TRIGGERED BY deploymentconfig.apps.openshift.io/mariadb 1 2 2 config And here is the result on K8s and minikube: $ kubectl get all -n mariadb NAME TYPE CLUSTER-IP EXTERNAL-IP PORT(S) AGE service/mariadb ClusterIP 10.107.147.194 <none> 3306/TCP 12m The OCP clusters are downloading the mariadb pods from quay.io/bitnami/mariadb:10.3 and the non-OCP clusters are not able to do that. I'm able to pull the mariadb image manually on the non-OCP clusters using podman pull: $ podman images REPOSITORY TAG IMAGE ID CREATED SIZE quay.io/bitnami/mariadb 10.3 79113d63dc86 32 hours ago 324 MB Even when I pull the image manually, the application won't start on the non-OCP clusters. I had to modify the MultiClusterHub resource to get the non-OCP clusters to be managed - is there a similar setting for the applications? Version-Release number of selected component (if applicable): 2.2
G2Bsync 823365288 comment xiangjingli Tue, 20 Apr 2021 15:25:16 UTC G2Bsync The mariadb app has a openshift specific resource - `deploymentconfig.apps.openshift.io/mariadb` . This is not supported in native k8s cluster. that caused the app deployment success on OCP but failure on native ks8 cluster. This is not a ACM app lifecycle issue. To workaround this, need to wrap the application using the generic `deployment` KIND instead of OCP specific `deploymentconfig` KIND resource
Thanks! - didn't notice that. I feel dumb now :) Just to confirm - there is no issue deploying applications from RHACM to a native K8S cluster as long as it is coded correctly. Correct?
G2Bsync 824127299 comment xiangjingli Wed, 21 Apr 2021 14:55:12 UTC G2Bsync Yes, please make sure all the resource KINDs involved in the application should have their relative CRDs pre-deployed in the managed k8s cluster. If no, you can create another ACM subscription (subscriptions.apps.open-cluster-management.io ) to deploy the CRDs ahead of the real application resources deployment.
Thanks - will do.
Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory (Moderate: Red Hat Advanced Cluster Management 2.2.3 security and bug fix update), and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. https://access.redhat.com/errata/RHSA-2021:1499