CI Test infrastructure

.gitignore

@@ -12,3 +12,6 @@ super-linter.log
 .ansible
 /ocp-primary.yaml
 /ocp-secondary.yaml
+tests/ci/.results/
+tests/ci/.pytest_cache/
+tests/ci/__pycache__/

Changes.md

@@ -3,8 +3,8 @@
 v1.3 - September 2026
 
 * Drop the ODF-MCO `drpartner-s4` variant and rename `rhdr-s4` to `drpartner-s4`.
-* `drpartner-minimal` uses the same preview RHDR catalog/IDMS/subscription and
-  defaults to `byoc: true`. The `odf` variant is unchanged.
+* `drpartner-minimal` uses the same preview RHDR catalog/IDMS/subscription.
+* Patterns are now defaulted to `byoc: true`.
 
 v1.3 - August 2026
 

Makefile

@@ -10,3 +10,4 @@ install-byoc: ramen-prereq pattern-install ## Installs the pattern onto a cluste
 ramen-prereq: ## Check if values.byoc false do nothing, else run the precheck against clusters accessed from values-secrets
 	echo "Running precheck for ramendr"
 	cd ansible && ansible-playbook -i hosts $(EXTRA_ARGS) $(EXTRA_VARS) playbooks/validate_byoc.yml
+

README.md

@@ -28,7 +28,7 @@ Spoke BOMs nest under the install variant as
 
 | Variant | Select | Purpose |
 |---------|--------|---------|
-| `odf` (default) | `main.variant: odf` | Baseline: full ODF Regional DR + Virtualization |
+| `odf` (default) | `main.variant: odf` | Baseline: full ODF Regional DR + Virtualization; `byoc: true` |
 | `drpartner-s4` | `main.variant: drpartner-s4` | Partner CSI + hub S4: OADP, OCP-V, preview RHDR; infrastructure DRClusters + `2m-novm` DRPolicy (no `2m-vm`/DRPC/VMs); Submariner disabled; `byoc: true` |
 | `drpartner-minimal` | `main.variant: drpartner-minimal` | Partner CSI without S4, Submariner, or DRCluster sync/validation: OADP, OCP-V, preview RHDR; Hive/BYOC only (`byoc: true`) |
 

overrides/values-cluster-names.yaml

@@ -34,4 +34,4 @@ drpc:
 # ./pattern.sh make install-byoc.
 # That target will run a validation play to ensure the cluster
 # setup will work with ramen patterns
-byoc: false
+byoc: true

overrides/values-odf-chart.yaml

@@ -7,3 +7,24 @@ storageSystem:
 
 objectStorage:
   enable: true
+
+odf:
+  # Workaround for a Rook/ODF bug: the default zone topology spread constraint on OSD
+  # prepare pods requires device-class=ssd, a label only added after OSD initialization,
+  # not during prepare. The constraint matches zero pods and is a no-op, allowing all
+  # 3 prepare pods to land on the same node. This sets a hard per-hostname constraint
+  # that actually matches rook-ceph-osd-prepare pods, ensuring one prepare pod per node.
+  # Requires openshift-data-foundations chart >= 0.3.2 (exposes odf.preparePlacement).
+  # When using OPP chart <= 0.3.1, the odf-prepare-patch ArgoCD app applies this patch
+  # as a PostSync hook instead.
+  preparePlacement:
+    topologySpreadConstraints:
+      - maxSkew: 1
+        topologyKey: kubernetes.io/hostname
+        whenUnsatisfiable: DoNotSchedule
+        labelSelector:
+          matchExpressions:
+            - key: app
+              operator: In
+              values:
+                - rook-ceph-osd-prepare

pattern-metadata.yaml

@@ -20,12 +20,12 @@ requirements:
       platform:
         aws:
           replicas: 3
-          type: m5.2xlarge
+          type: m5.4xlarge
     controlPlane:
       platform:
         aws:
           replicas: 3
-          type: m5.xlarge
+          type: m5.4xlarge
   spoke: # Managed DR clusters (primary/secondary)
     compute:
       platform:

tests/ci/README.md

@@ -0,0 +1,32 @@
+# CI Tests for ramendr-starter-kit
+
+Pytest is invoked per-clustergroup from the repository root via the `interop-test` Tekton task:
+
+```console
+pytest -lv test_<TARGET_CLUSTERGROUP>.py --junit-xml .results/test_<TARGET_CLUSTERGROUP>.xml
+```
+
+## Required environment variables
+
+| Variable | Description |
+|---|---|
+| `VP_HUBCONFIG` | Path to kubeconfig for the hub cluster |
+| `VP_SPOKECONFIG` | Path to kubeconfig for `ocp-primary` (DR spoke 1) |
+| `VP_SPOKECONFIG_SECONDARY` | Path to kubeconfig for `ocp-secondary` (DR spoke 2) |
+| `TARGET_CLUSTERGROUP` | Clustergroup to test (defaults to `hub`) |
+
+## Running tests locally
+
+From the repository root:
+
+```bash
+VP_HUBCONFIG=~/.kube/hub-config \
+VP_SPOKECONFIG=~/.kube/spoke-primary-config \
+VP_SPOKECONFIG_SECONDARY=~/.kube/spoke-secondary-config \
+TARGET_CLUSTERGROUP=odf \
+./pattern.sh make run-ci-tests
+```
+
+## Test coverage
+
+- `test_odf.py` — hub-side operator subscriptions, pod health, ACM ManagedCluster status, ArgoCD reachability and application health for all three clusters

tests/ci/__init__.py

@@ -0,0 +1 @@
+__version__ = "0.1.0"

tests/ci/conftest.py

@@ -0,0 +1 @@
+from validatedpatterns_tests.interop.conftest_openshift import *  # noqa

tests/ci/requirements.txt

@@ -0,0 +1,15 @@
+ansible-core==2.18.*
+ansible-runner
+awxkit
+kubernetes
+openshift
+boto3>=1.21
+botocore>=1.24
+awscli>=1.22
+azure-cli>=2.34
+gcloud
+humanize
+pytz
+pytest
+junitparser
+vp-qe-test-common @ git+https://github.com/validatedpatterns/vp-qe-test-common.git@v1