[release-ocm-2.14] MGMT-23971: Add a timeout for installing-pending-user-action

internal/cluster/common.go

@@ -188,53 +188,6 @@ func getKnownMastersNodesIds(c *common.Cluster, db *gorm.DB) ([]*strfmt.UUID, er
 	return masterNodesIds, nil
 }
 
-func HostsInStatus(c *common.Cluster, statuses []string) (int, int, int) {
-	mappedMastersByRole := MapMasterHostsByStatus(c)
-	mappedArbitersByRole := MapArbiterHostsByStatus(c)
-	mappedWorkersByRole := MapWorkersHostsByStatus(c)
-	mastersInSomeInstallingStatus := 0
-	arbitersInSomeInstallingStatus := 0
-	workersInSomeInstallingStatus := 0
-
-	for _, status := range statuses {
-		mastersInSomeInstallingStatus += len(mappedMastersByRole[status])
-		arbitersInSomeInstallingStatus += len(mappedArbitersByRole[status])
-		workersInSomeInstallingStatus += len(mappedWorkersByRole[status])
-	}
-	return mastersInSomeInstallingStatus, arbitersInSomeInstallingStatus, workersInSomeInstallingStatus
-}
-
-func MapMasterHostsByStatus(c *common.Cluster) map[string][]*models.Host {
-	return mapHostsByStatus(c, models.HostRoleMaster)
-}
-
-func MapArbiterHostsByStatus(c *common.Cluster) map[string][]*models.Host {
-	return mapHostsByStatus(c, models.HostRoleArbiter)
-}
-
-func MapWorkersHostsByStatus(c *common.Cluster) map[string][]*models.Host {
-	return mapHostsByStatus(c, models.HostRoleWorker)
-}
-
-func mapHostsByStatus(c *common.Cluster, role models.HostRole) map[string][]*models.Host {
-	hostMap := make(map[string][]*models.Host)
-	for _, host := range c.Hosts {
-		if role != "" && common.GetEffectiveRole(host) != role {
-			continue
-		}
-		if _, ok := hostMap[swag.StringValue(host.Status)]; ok {
-			hostMap[swag.StringValue(host.Status)] = append(hostMap[swag.StringValue(host.Status)], host)
-		} else {
-			hostMap[swag.StringValue(host.Status)] = []*models.Host{host}
-		}
-	}
-	return hostMap
-}
-
-func MapHostsByStatus(c *common.Cluster) map[string][]*models.Host {
-	return mapHostsByStatus(c, "")
-}
-
 func UpdateMachineNetwork(db *gorm.DB, cluster *common.Cluster, machineNetwork []string) error {
 	if len(machineNetwork) > 2 {
 		return common.NewApiError(http.StatusInternalServerError,

internal/cluster/transition.go

@@ -257,7 +257,7 @@ func (th *transitionHandler) createClusterCompletionStatusInfo(ctx context.Conte
 		statusInfo = StatusInfoDegraded
 		statusInfo += ". Failed OLM operators: " + strings.Join(statuses[models.OperatorTypeOlm][models.OperatorStatusFailed], ", ")
 	} else {
-		_, _, installedWorkers := HostsInStatus(cluster, []string{models.HostStatusInstalled})
+		_, _, installedWorkers := common.HostsInStatus(cluster, []string{models.HostStatusInstalled})
 		if installedWorkers < common.NumberOfWorkers(cluster) {
 			statusInfo = StatusInfoNotAllWorkersInstalled
 		}
@@ -418,7 +418,7 @@ func (th *transitionHandler) IsFinalizing(sw stateswitch.StateSwitch, args state
 	sCluster, ok := sw.(*stateCluster)
 	installedStatus := []string{models.HostStatusInstalled}
 
-	if ok && th.enoughMastersAndWorkers(sCluster, installedStatus) {
+	if ok && common.HasEnoughMastersAndWorkers(sCluster.cluster, installedStatus) {
 		th.log.Infof("Cluster %s has at least required number of installed hosts, "+
 			"cluster is finalizing.", sCluster.cluster.ID)
 		return true, nil
@@ -431,7 +431,7 @@ func (th *transitionHandler) IsInstalling(sw stateswitch.StateSwitch, args state
 	sCluster, _ := sw.(*stateCluster)
 	installingStatuses := []string{models.HostStatusInstalling, models.HostStatusInstallingInProgress,
 		models.HostStatusInstalled, models.HostStatusInstallingPendingUserAction, models.HostStatusPreparingSuccessful}
-	return th.enoughMastersAndWorkers(sCluster, installingStatuses), nil
+	return common.HasEnoughMastersAndWorkers(sCluster.cluster, installingStatuses), nil
 }
 
 // check if we should stay in installing state
@@ -507,49 +507,6 @@ func (th *transitionHandler) PostUpdateFinalizingAMSConsoleUrl(sw stateswitch.St
 	return nil
 }
 
-// enoughMastersAndWorkers returns whether the number of master and worker nodes in the specified cluster with the given status
-// meets the required criteria. The conditions are as follows:
-//   - For SNO (Single Node OpenShift), there must be exactly one master node and zero worker nodes.
-//   - For High Availability cluster, the number of master nodes should match the user's request, and not less than the minimum. The worker node requirement depends on this request:
-//     If the user requested at least two workers, there must be at least two, indicating non-schedulable masters were intended.
-//     If the user requested fewer than two workers, any number of workers is acceptable.
-//   - For TNA Clusters the same conditions apply as for High Availability Clusters, but we also need to check that at least one arbiter node is in the correct status.
-func (th *transitionHandler) enoughMastersAndWorkers(sCluster *stateCluster, statuses []string) bool {
-	mastersInSomeInstallingStatus, arbitersInSomeInstallingStatus, workersInSomeInstallingStatus := HostsInStatus(sCluster.cluster, statuses)
-
-	if sCluster.cluster.ControlPlaneCount == 1 {
-		return mastersInSomeInstallingStatus == common.AllowedNumberOfMasterHostsInNoneHaMode &&
-			workersInSomeInstallingStatus == common.AllowedNumberOfWorkersInNoneHaMode
-	}
-
-	// hosts roles are known at this stage
-	masters, arbiters, workers, _ := common.GetHostsByEachRole(&sCluster.cluster.Cluster, false)
-	numberOfExpectedMasters := len(masters)
-	numberOfExpectedArbiters := len(arbiters)
-
-	minMasterHostsNeeded := common.MinMasterHostsNeededForInstallationInHaMode
-	if numberOfExpectedArbiters != 0 {
-		minMasterHostsNeeded = common.MinMasterHostsNeededForInstallationInHaArbiterMode
-		// validate arbiters
-		if arbitersInSomeInstallingStatus == 0 {
-			return false
-		}
-	}
-
-	// validate masters
-	if numberOfExpectedMasters < minMasterHostsNeeded ||
-		mastersInSomeInstallingStatus < numberOfExpectedMasters {
-		return false
-	}
-
-	numberOfExpectedWorkers := len(workers)
-
-	// validate workers
-	return numberOfExpectedWorkers < common.MinimumNumberOfWorkersForNonSchedulableMastersClusterInHaMode ||
-		numberOfExpectedWorkers >= common.MinimumNumberOfWorkersForNonSchedulableMastersClusterInHaMode &&
-			workersInSomeInstallingStatus >= common.MinimumNumberOfWorkersForNonSchedulableMastersClusterInHaMode
-}
-
 // check if installation reach to timeout
 func (th *transitionHandler) IsInstallationTimedOut(sw stateswitch.StateSwitch, args stateswitch.TransitionArgs) (bool, error) {
 	sCluster, ok := sw.(*stateCluster)

internal/common/common.go

@@ -781,3 +781,62 @@ func GetDefaultHighAvailabilityAndMasterCountParams(highAvailabilityMode *string
 func IsClusterTopologyHighlyAvailableArbiter(cluster *Cluster) bool {
 	return funk.NotEmpty(GetHostsByRole(cluster, models.HostRoleArbiter))
 }
+
+func HostsInStatus(c *Cluster, statuses []string) (masters, arbiters, workers int) {
+	for _, host := range c.Hosts {
+		if funk.ContainsString(statuses, swag.StringValue(host.Status)) {
+			switch GetEffectiveRole(host) {
+			case models.HostRoleMaster, models.HostRoleBootstrap:
+				masters++
+			case models.HostRoleArbiter:
+				arbiters++
+			case models.HostRoleWorker:
+				workers++
+			}
+		}
+	}
+	return
+}
+
+// HasEnoughMastersAndWorkers returns whether the number of master and worker nodes in the specified cluster with the given status
+// meets the required criteria. The conditions are as follows:
+//   - For SNO (Single Node OpenShift), there must be exactly one master node and zero worker nodes.
+//   - For High Availability cluster, the number of master nodes should match the user's request, and not less than the minimum. The worker node requirement depends on this request:
+//     If the user requested at least two workers, there must be at least two, indicating non-schedulable masters were intended.
+//     If the user requested fewer than two workers, any number of workers is acceptable.
+//   - For TNA Clusters the same conditions apply as for High Availability Clusters, but we also need to check that at least one arbiter node is in the correct status.
+func HasEnoughMastersAndWorkers(c *Cluster, statuses []string) bool {
+	mastersInStatus, arbitersInStatus, workersInStatus := HostsInStatus(c, statuses)
+
+	if c.ControlPlaneCount == 1 {
+		return mastersInStatus == AllowedNumberOfMasterHostsInNoneHaMode &&
+			workersInStatus == AllowedNumberOfWorkersInNoneHaMode
+	}
+
+	// hosts roles are known at this stage
+	masters, arbiters, workers, _ := GetHostsByEachRole(&c.Cluster, false)
+	numberOfExpectedMasters := len(masters)
+	numberOfExpectedArbiters := len(arbiters)
+
+	minMasterHostsNeeded := MinMasterHostsNeededForInstallationInHaMode
+	if numberOfExpectedArbiters != 0 {
+		minMasterHostsNeeded = MinMasterHostsNeededForInstallationInHaArbiterMode
+		// validate arbiters
+		if arbitersInStatus == 0 {
+			return false
+		}
+	}
+
+	// validate masters
+	if numberOfExpectedMasters < minMasterHostsNeeded ||
+		mastersInStatus < numberOfExpectedMasters {
+		return false
+	}
+
+	numberOfExpectedWorkers := len(workers)
+
+	// validate workers
+	return numberOfExpectedWorkers < MinimumNumberOfWorkersForNonSchedulableMastersClusterInHaMode ||
+		numberOfExpectedWorkers >= MinimumNumberOfWorkersForNonSchedulableMastersClusterInHaMode &&
+			workersInStatus >= MinimumNumberOfWorkersForNonSchedulableMastersClusterInHaMode
+}