WIP Add dataplane service dependency graph and fallback ordering

rabi · rabi · commit d2bce4ee15b1 · 2026-05-13T18:32:02.000+05:30
Introduce dependsOn on services and fallbackToListOrder (*bool,
default true) on deployments. Services are topologically sorted into
execution levels. Explicit dependsOn stays authoritative, while services
without dependsOn can temporarily fall back to list predecessor order.

Change-Id: I3bdeb6e51dbba7dda14f2c7a60a496cc3df85260
Signed-off-by: rabi &lt;ramishra@redhat.com&gt;
diff --git a/api/bases/dataplane.openstack.org_openstackdataplanedeployments.yaml b/api/bases/dataplane.openstack.org_openstackdataplanedeployments.yaml
@@ -96,6 +96,12 @@ spec:
                 description: Time before the deployment is requeued in seconds
                 minimum: 1
                 type: integer
+              fallbackToListOrder:
+                default: true
+                description: |-
+                  FallbackToListOrder preserves service list order for services
+                  without explicit dependsOn. Defaults to true.
+                type: boolean
               nodeSets:
                 description: NodeSets is the list of NodeSets deployed
                 items:
diff --git a/api/bases/dataplane.openstack.org_openstackdataplaneservices.yaml b/api/bases/dataplane.openstack.org_openstackdataplaneservices.yaml
@@ -116,6 +116,14 @@ spec:
                       x-kubernetes-map-type: atomic
                   type: object
                 type: array
+              dependsOn:
+                description: |-
+                  DependsOn is a list of EDPMServiceType or CR name values that must
+                  complete before this service runs. If empty, the service depends on
+                  the preceding entry in the NodeSet service list.
+                items:
+                  type: string
+                type: array
               deployOnAllNodeSets:
                 description: |-
                   DeployOnAllNodeSets - should the service be deploy across all nodesets
diff --git a/api/dataplane/v1beta1/openstackdataplanedeployment_types.go b/api/dataplane/v1beta1/openstackdataplanedeployment_types.go
@@ -78,6 +78,13 @@ type OpenStackDataPlaneDeploymentSpec struct {
 	// variables to inject into the Ansible Execution Environment pod.
 	// If not specified, defaults to "openstack-aee-default-env".
 	AnsibleEEEnvConfigMapName string `json:"ansibleEEEnvConfigMapName,omitempty"`
+
+	// FallbackToListOrder preserves service list order for services
+	// without explicit dependsOn. Defaults to true.
+	// +kubebuilder:validation:Optional
+	// +kubebuilder:default=true
+	// +operator-sdk:csv:customresourcedefinitions:type=spec,xDescriptors={"urn:alm:descriptor:com.tectonic.ui:booleanSwitch"}
+	FallbackToListOrder *bool `json:"fallbackToListOrder"`
 }
 
 // AnsibleExecutionSummary captures the final ansible-runner execution result
diff --git a/api/dataplane/v1beta1/openstackdataplaneservice_types.go b/api/dataplane/v1beta1/openstackdataplaneservice_types.go
@@ -117,6 +117,12 @@ type OpenStackDataPlaneServiceSpec struct {
 	// to manage the service. If not set, will default to the
 	// OpenStackDataPlaneService name.
 	EDPMServiceType string `json:"edpmServiceType,omitempty" yaml:"edpmServiceType,omitempty"`
+
+	// DependsOn is a list of EDPMServiceType or CR name values that must
+	// complete before this service runs. If empty, the service depends on
+	// the preceding entry in the NodeSet service list.
+	// +kubebuilder:validation:Optional
+	DependsOn []string `json:"dependsOn,omitempty" yaml:"dependsOn,omitempty"`
 }
 
 // OpenStackDataPlaneServiceStatus defines the observed state of OpenStackDataPlaneService
diff --git a/api/dataplane/v1beta1/zz_generated.deepcopy.go b/api/dataplane/v1beta1/zz_generated.deepcopy.go
diff --git a/bindata/crds/crds.yaml b/bindata/crds/crds.yaml
@@ -19035,6 +19035,12 @@ spec:
                 description: Time before the deployment is requeued in seconds
                 minimum: 1
                 type: integer
+              fallbackToListOrder:
+                default: true
+                description: |-
+                  FallbackToListOrder preserves service list order for services
+                  without explicit dependsOn. Defaults to true.
+                type: boolean
               nodeSets:
                 description: NodeSets is the list of NodeSets deployed
                 items:
@@ -21399,6 +21405,14 @@ spec:
                       x-kubernetes-map-type: atomic
                   type: object
                 type: array
+              dependsOn:
+                description: |-
+                  DependsOn is a list of EDPMServiceType or CR name values that must
+                  complete before this service runs. If empty, the service depends on
+                  the preceding entry in the NodeSet service list.
+                items:
+                  type: string
+                type: array
               deployOnAllNodeSets:
                 description: |-
                   DeployOnAllNodeSets - should the service be deploy across all nodesets
diff --git a/config/crd/bases/dataplane.openstack.org_openstackdataplanedeployments.yaml b/config/crd/bases/dataplane.openstack.org_openstackdataplanedeployments.yaml
@@ -96,6 +96,12 @@ spec:
                 description: Time before the deployment is requeued in seconds
                 minimum: 1
                 type: integer
+              fallbackToListOrder:
+                default: true
+                description: |-
+                  FallbackToListOrder preserves service list order for services
+                  without explicit dependsOn. Defaults to true.
+                type: boolean
               nodeSets:
                 description: NodeSets is the list of NodeSets deployed
                 items:
diff --git a/config/crd/bases/dataplane.openstack.org_openstackdataplaneservices.yaml b/config/crd/bases/dataplane.openstack.org_openstackdataplaneservices.yaml
@@ -116,6 +116,14 @@ spec:
                       x-kubernetes-map-type: atomic
                   type: object
                 type: array
+              dependsOn:
+                description: |-
+                  DependsOn is a list of EDPMServiceType or CR name values that must
+                  complete before this service runs. If empty, the service depends on
+                  the preceding entry in the NodeSet service list.
+                items:
+                  type: string
+                type: array
               deployOnAllNodeSets:
                 description: |-
                   DeployOnAllNodeSets - should the service be deploy across all nodesets
diff --git a/config/services/dataplane_v1beta1_openstackdataplaneservice_bootstrap.yaml b/config/services/dataplane_v1beta1_openstackdataplaneservice_bootstrap.yaml
@@ -6,3 +6,6 @@ spec:
   playbook: osp.edpm.bootstrap
   edpmServiceType: bootstrap
   caCerts: combined-ca-bundle
+  dependsOn:
+    - redhat
+    - download-cache
diff --git a/config/services/dataplane_v1beta1_openstackdataplaneservice_configure_network.yaml b/config/services/dataplane_v1beta1_openstackdataplaneservice_configure_network.yaml
@@ -5,3 +5,5 @@ metadata:
 spec:
   playbook: osp.edpm.configure_network
   edpmServiceType: configure-network
+  dependsOn:
+    - bootstrap
diff --git a/config/services/dataplane_v1beta1_openstackdataplaneservice_configure_os.yaml b/config/services/dataplane_v1beta1_openstackdataplaneservice_configure_os.yaml
@@ -5,3 +5,5 @@ metadata:
 spec:
   playbook: osp.edpm.configure_os
   edpmServiceType: configure-os
+  dependsOn:
+    - install-os
diff --git a/config/services/dataplane_v1beta1_openstackdataplaneservice_download_cache.yaml b/config/services/dataplane_v1beta1_openstackdataplaneservice_download_cache.yaml
@@ -5,3 +5,5 @@ metadata:
 spec:
   playbook: osp.edpm.download_cache
   edpmServiceType: download-cache
+  dependsOn:
+    - redhat
diff --git a/config/services/dataplane_v1beta1_openstackdataplaneservice_frr.yaml b/config/services/dataplane_v1beta1_openstackdataplaneservice_frr.yaml
@@ -4,6 +4,8 @@ metadata:
   name: frr
 spec:
   playbook: osp.edpm.frr
+  dependsOn:
+    - run-os
   containerImageFields:
   - EdpmFrrImage
   edpmServiceType: frr
diff --git a/config/services/dataplane_v1beta1_openstackdataplaneservice_install_certs.yaml b/config/services/dataplane_v1beta1_openstackdataplaneservice_install_certs.yaml
@@ -6,3 +6,5 @@ spec:
   playbook: osp.edpm.install_certs
   addCertMounts: True
   edpmServiceType: install-certs
+  dependsOn:
+    - bootstrap
diff --git a/config/services/dataplane_v1beta1_openstackdataplaneservice_install_os.yaml b/config/services/dataplane_v1beta1_openstackdataplaneservice_install_os.yaml
@@ -5,3 +5,5 @@ metadata:
 spec:
   playbook: osp.edpm.install_os
   edpmServiceType: install-os
+  dependsOn:
+    - bootstrap
diff --git a/config/services/dataplane_v1beta1_openstackdataplaneservice_libvirt.yaml b/config/services/dataplane_v1beta1_openstackdataplaneservice_libvirt.yaml
@@ -4,6 +4,9 @@ metadata:
   name: libvirt
 spec:
   playbook: osp.edpm.libvirt
+  dependsOn:
+    - run-os
+    - install-certs
   dataSources:
     # NOTE: this Secret needs to be created before deploying the data plane.
     # It should contain the libvirt sasl auth password using the key LibvirtPassword
diff --git a/config/services/dataplane_v1beta1_openstackdataplaneservice_logging.yaml b/config/services/dataplane_v1beta1_openstackdataplaneservice_logging.yaml
@@ -3,6 +3,8 @@ kind: OpenStackDataPlaneService
 metadata:
   name: logging
 spec:
+  dependsOn:
+    - run-os
   dataSources:
     - secretRef:
         name: logging-compute-config-data
diff --git a/config/services/dataplane_v1beta1_openstackdataplaneservice_neutron_dhcp.yaml b/config/services/dataplane_v1beta1_openstackdataplaneservice_neutron_dhcp.yaml
@@ -4,6 +4,8 @@ metadata:
   name: neutron-dhcp
 spec:
   playbook: osp.edpm.neutron_dhcp
+  dependsOn:
+    - run-os
   dataSources:
     - secretRef:
         name: neutron-dhcp-agent-neutron-config
diff --git a/config/services/dataplane_v1beta1_openstackdataplaneservice_neutron_metadata.yaml b/config/services/dataplane_v1beta1_openstackdataplaneservice_neutron_metadata.yaml
@@ -4,6 +4,8 @@ metadata:
   name: neutron-metadata
 spec:
   playbook: osp.edpm.neutron_metadata
+  dependsOn:
+    - ovn
   dataSources:
     - secretRef:
         name: neutron-ovn-metadata-agent-neutron-config
diff --git a/config/services/dataplane_v1beta1_openstackdataplaneservice_neutron_ovn.yaml b/config/services/dataplane_v1beta1_openstackdataplaneservice_neutron_ovn.yaml
@@ -4,6 +4,8 @@ metadata:
   name: neutron-ovn
 spec:
   playbook: osp.edpm.neutron_ovn
+  dependsOn:
+    - ovn
   dataSources:
     - secretRef:
         name: neutron-ovn-agent-neutron-config
diff --git a/config/services/dataplane_v1beta1_openstackdataplaneservice_neutron_sriov.yaml b/config/services/dataplane_v1beta1_openstackdataplaneservice_neutron_sriov.yaml
@@ -4,6 +4,8 @@ metadata:
   name: neutron-sriov
 spec:
   playbook: osp.edpm.neutron_sriov
+  dependsOn:
+    - run-os
   dataSources:
     - secretRef:
         name: neutron-sriov-agent-neutron-config
diff --git a/config/services/dataplane_v1beta1_openstackdataplaneservice_nova.yaml b/config/services/dataplane_v1beta1_openstackdataplaneservice_nova.yaml
@@ -3,6 +3,13 @@ kind: OpenStackDataPlaneService
 metadata:
   name: nova
 spec:
+  dependsOn:
+    - libvirt
+    - ssh-known-hosts
+    - neutron-ovn
+    - neutron-sriov
+    - neutron-dhcp
+    - neutron-metadata
   dataSources:
     - secretRef:
         name: nova-cell1-compute-config
diff --git a/config/services/dataplane_v1beta1_openstackdataplaneservice_os_reboot.yaml b/config/services/dataplane_v1beta1_openstackdataplaneservice_os_reboot.yaml
@@ -5,3 +5,5 @@ metadata:
 spec:
   playbook: osp.edpm.reboot
   edpmServiceType: reboot-os
+  dependsOn:
+    - run-os
diff --git a/config/services/dataplane_v1beta1_openstackdataplaneservice_ovn.yaml b/config/services/dataplane_v1beta1_openstackdataplaneservice_ovn.yaml
@@ -4,6 +4,10 @@ metadata:
   name: ovn
 spec:
   playbook: osp.edpm.ovn
+  dependsOn:
+    - run-os
+    - configure-network
+    - install-certs
   dataSources:
     - configMapRef:
         name: ovncontroller-config
diff --git a/config/services/dataplane_v1beta1_openstackdataplaneservice_ovn_bgp_agent.yaml b/config/services/dataplane_v1beta1_openstackdataplaneservice_ovn_bgp_agent.yaml
@@ -4,6 +4,8 @@ metadata:
   name: ovn-bgp-agent
 spec:
   playbook: osp.edpm.ovn_bgp_agent
+  dependsOn:
+    - ovn
   dataSources:
     - secretRef:
         name: neutron-ovn-agent-neutron-config
diff --git a/config/services/dataplane_v1beta1_openstackdataplaneservice_run_os.yaml b/config/services/dataplane_v1beta1_openstackdataplaneservice_run_os.yaml
@@ -8,3 +8,5 @@ spec:
   - EdpmLogrotateCrondImage
   - EdpmIscsidImage
   edpmServiceType: run-os
+  dependsOn:
+    - configure-os
diff --git a/config/services/dataplane_v1beta1_openstackdataplaneservice_ssh_known_hosts.yaml b/config/services/dataplane_v1beta1_openstackdataplaneservice_ssh_known_hosts.yaml
@@ -6,3 +6,6 @@ spec:
   playbook: osp.edpm.ssh_known_hosts
   deployOnAllNodeSets: true
   edpmServiceType: ssh-known-hosts
+  dependsOn:
+    - install-os
+    - configure-network
diff --git a/config/services/dataplane_v1beta1_openstackdataplaneservice_swift.yaml b/config/services/dataplane_v1beta1_openstackdataplaneservice_swift.yaml
@@ -4,6 +4,8 @@ metadata:
   name: swift
 spec:
   playbook: osp.edpm.swift
+  dependsOn:
+    - run-os
   dataSources:
     - secretRef:
         name: swift-conf
diff --git a/config/services/dataplane_v1beta1_openstackdataplaneservice_telemetry.yaml b/config/services/dataplane_v1beta1_openstackdataplaneservice_telemetry.yaml
@@ -3,6 +3,8 @@ kind: OpenStackDataPlaneService
 metadata:
   name: telemetry
 spec:
+  dependsOn:
+    - nova
   dataSources:
     - secretRef:
         name: ceilometer-compute-config-data
diff --git a/config/services/dataplane_v1beta1_openstackdataplaneservice_telemetry_power_monitoring.yaml b/config/services/dataplane_v1beta1_openstackdataplaneservice_telemetry_power_monitoring.yaml
@@ -3,6 +3,8 @@ kind: OpenStackDataPlaneService
 metadata:
   name: telemetry-power-monitoring
 spec:
+  dependsOn:
+    - nova
   dataSources:
     - secretRef:
         name: ceilometer-ipmi-config-data
diff --git a/config/services/dataplane_v1beta1_openstackdataplaneservice_validate_network.yaml b/config/services/dataplane_v1beta1_openstackdataplaneservice_validate_network.yaml
@@ -5,3 +5,5 @@ metadata:
 spec:
   playbook: osp.edpm.validate_network
   edpmServiceType: validate-network
+  dependsOn:
+    - configure-network
diff --git a/internal/controller/dataplane/openstackdataplanedeployment_controller.go b/internal/controller/dataplane/openstackdataplanedeployment_controller.go
@@ -19,10 +19,12 @@ package dataplane
 
 import (
 	"context"
+	"errors"
 	"fmt"
 	"time"
 
 	"github.com/go-playground/validator/v10"
+	"github.com/iancoleman/strcase"
 	batchv1 "k8s.io/api/batch/v1"
 	corev1 "k8s.io/api/core/v1"
 	k8s_errors "k8s.io/apimachinery/pkg/api/errors"
@@ -250,12 +252,47 @@ func (r *OpenStackDataPlaneDeploymentReconciler) Reconcile(ctx context.Context,
 	shouldRequeue := false
 	haveError := false
 	deploymentErrMsg := ""
-	var nodesetServiceMap map[string][]string
+	var nodesetServiceMap map[string][][]string
 	backoffLimitReached := false
 
+	fallbackToListOrder := true
+	if instance.Spec.FallbackToListOrder != nil {
+		fallbackToListOrder = *instance.Spec.FallbackToListOrder
+	}
+
 	if nodesetServiceMap, err = deployment.DedupeServices(ctx, helper, nodeSets.Items,
-		instance.Spec.ServicesOverride); err != nil {
+		instance.Spec.ServicesOverride, fallbackToListOrder); err != nil {
 		util.LogErrorForObject(helper, err, "OpenStackDeployment error for deployment", instance)
+		var missingServiceErr *deployment.MissingServiceError
+		if errors.As(err, &missingServiceErr) {
+			nsConditions := instance.Status.NodeSetConditions[missingServiceErr.NodeSet]
+			nodeSetServiceErrorMessage := fmt.Sprintf(
+				dataplanev1.NodeSetServiceDeploymentErrorMessage,
+				missingServiceErr.Service) + " error %s"
+			nsConditions.MarkFalse(
+				dataplanev1.NodeSetDeploymentReadyCondition,
+				condition.ErrorReason,
+				condition.SeverityError,
+				nodeSetServiceErrorMessage,
+				err.Error())
+			serviceCondition := condition.Type(fmt.Sprintf(
+				"Service%sDeploymentReady",
+				strcase.ToCamel(missingServiceErr.Service)))
+			nsConditions.MarkFalse(
+				serviceCondition,
+				condition.ErrorReason,
+				condition.SeverityError,
+				nodeSetServiceErrorMessage,
+				err.Error())
+			instance.Status.NodeSetConditions[missingServiceErr.NodeSet] = nsConditions
+			instance.Status.Conditions.MarkFalse(
+				condition.DeploymentReadyCondition,
+				condition.ErrorReason,
+				condition.SeverityWarning,
+				condition.DeploymentReadyErrorMessage,
+				fmt.Sprintf("nodeSet: %s error: %s", missingServiceErr.NodeSet, err.Error()))
+			return ctrl.Result{}, err
+		}
 		instance.Status.Conditions.MarkFalse(
 			condition.DeploymentReadyCondition,
 			condition.ErrorReason,
@@ -309,11 +346,9 @@ func (r *OpenStackDataPlaneDeploymentReconciler) Reconcile(ctx context.Context,
 			Version:                     version,
 		}
 
-		// When ServicesOverride is set on the OpenStackDataPlaneDeployment,
-		// deploy those services for each OpenStackDataPlaneNodeSet. Otherwise,
-		// deploy with the OpenStackDataPlaneNodeSet's Services.
 		var deployResult *ctrl.Result
-		deployResult, err = deployer.Deploy(nodesetServiceMap[nodeSet.Name])
+		deployResult, err = deployer.Deploy(
+			nodesetServiceMap[nodeSet.Name])
 
 		nsConditions := instance.Status.NodeSetConditions[nodeSet.Name]
 		nsConditions.Set(nsConditions.Mirror(dataplanev1.NodeSetDeploymentReadyCondition))
diff --git a/internal/dataplane/depgraph.go b/internal/dataplane/depgraph.go
diff --git a/internal/dataplane/depgraph_test.go b/internal/dataplane/depgraph_test.go
diff --git a/internal/dataplane/deployment.go b/internal/dataplane/deployment.go
diff --git a/internal/dataplane/service.go b/internal/dataplane/service.go
diff --git a/test/functional/dataplane/openstackdataplanedeployment_controller_test.go b/test/functional/dataplane/openstackdataplanedeployment_controller_test.go
diff --git a/test/kuttl/tests/dataplane-deploy-global-service-test/01-assert.yaml b/test/kuttl/tests/dataplane-deploy-global-service-test/01-assert.yaml
diff --git a/test/kuttl/tests/dataplane-deploy-no-nodes-test/01-assert.yaml b/test/kuttl/tests/dataplane-deploy-no-nodes-test/01-assert.yaml
