Refresh port status every minute unless there are events to process

Signed-off-by: Anthony Floeder <[email protected]>

internal/switchtec/cmd/fabric/perf.go

@@ -51,7 +51,7 @@ func (cmd *BandwidthCmd) Run() error {
 		return err
 	}
 
-	// Resetting the bandwidth counter takes about 1s
+	// Resetting the bandwidht counter takes about 1s
 	time.Sleep(time.Second)
 
 	// Record the bandwidth counters

internal/switchtec/pkg/switchtec/events.go

@@ -728,7 +728,7 @@ func (dev *Device) GetGfmsEvents() ([]GfmsEvent, error) {
 	type response struct {
 		ResponseNumber       uint16 // Number of GFMS Event Entries in this response
 		RemainingNumberFlags uint16 // [0:14] Number of GFMS Event Entries remaining in GFMS Event Queue
-		// [15]   Flag to indicatre whether the event entry buffer queue has
+		// [15]   Flag to indicate whether the event entry buffer queue has
 		//        been overwritten as a result of not being read in time.
 		Data [maxDataLength - 4]byte
 	}

pkg/manager-event/manager.go

@@ -61,8 +61,8 @@ import (
 //					"EventType": "Alert",
 //					"EventId": "ABCD123",
 //					"Severity": "Warning",
-//					"Message": "Fabric switch port is down",
-//					"MessageId": "Alert.1.0.FabricSwitchPortDown",
+//					"Message": "Fabirc switch port is down",
+//					"MessageId": "Alert.1.0.FabircSwitchPortDown",
 //					"MessageArgs": [
 //						"Rabbit", "0", "Port 0"
 //					],
@@ -73,16 +73,16 @@ import (
 //			]
 //		}
 //
-// There should always be a Message Registry that is paired with a Message ID. In this case
+// There should always be a Message Registry that is paried with a Message ID. In this case
 // there would be a Message Registry with the following
 //
 //		{
 //			"Id": "Alert.1.0.0",
 //			"RegistryPrefix": "Alert",
 //			"RegistryVersion": "1.0.0",
 //			"Messages": {
-//				"FabricSwitchPortDown": {
-//					"Description": "A Port is down on Fabric %1 Switch %2 Port %3",
+//				"FabircSwitchPortDown": {
+//					"Description": "A Port is down on Fabirc %1 Switch %2 Port %3",
 //					"Message": "Port connectivity was lost on Fabric %1 Switch %2 Port %3",
 //					"Severity": "Critical",
 //					"NumberOfArgs": 3,
@@ -97,14 +97,14 @@ import (
 //
 // When a client receives an event (by creating a subscription with the event service), they
 // first use the MessageId to find the corresponding registry, version, and message.
-// In the above example, the MessageId is "Alert.1.0.FabricSwitchPortDown".  This is broken
+// In the above example, the MessageId is "Alert.1.0.FabircSwitchPortDown".  This is broken
 // down by the client into three parts
 // 		- Registry Prefix: 			"Alert"
 //		- Major / Minor Version: 	"1.0"
-//		- Message Identifier:		"FabricSwitchPortDown"
+//		- Message Identifier:		"FabircSwitchPortDown"
 //
 // The client must then use that information to scan the available registries and locate the
-// correct Message Registry. The Event Service contains a link of the registries in use, so
+// correct Message Registry. The Event Service contains a link of the registires in use, so
 // the client can scan those registries and do a lookup.
 //
 // There are a bunch of registries already created by DMTF: https://redfish.dmtf.org/registries/

pkg/manager-fabric/config_default.yaml

@@ -15,7 +15,6 @@ switches:
       - name: Rabbit
         type: ManagementPort
         port: 24
-        slot: -1
         width: 16
       - name: Compute 0
         type: UpstreamPort

pkg/manager-fabric/manager.go

@@ -386,17 +386,17 @@ func (s *Switch) identify() error {
 			return err
 		}
 
-		paxID, err := dev.Identify()
+		paxId, err := dev.Identify()
 		if err != nil {
 			log.Error(err, "Identify error")
 			return err
 		}
 
-		log.V(2).Info("Identified switch device", "pax", paxID)
-		if id := strconv.Itoa(int(paxID)); id == s.id {
+		log.V(2).Info("Identified switch device", "pax", paxId)
+		if id := strconv.Itoa(int(paxId)); id == s.id {
 			s.dev = dev
 			s.path = path
-			s.paxId = paxID
+			s.paxId = paxId
 
 			s.model = s.getModel()
 			s.manufacturer = s.getManufacturer()
@@ -920,6 +920,8 @@ func Initialize(log ec.Logger, ctrl SwitchtecControllerInterface) error {
 			log.Error(err, "Failed to identify switch")
 		}
 
+		log.Info("Switch identified", "pax", s.paxId)
+
 		for portIdx, portConf := range switchConf.Ports {
 			portType := portConf.getPortType()
 			id := strconv.Itoa(portIdx)

pkg/manager-fabric/monitor.go

@@ -27,7 +27,7 @@ import (
 	"github.com/NearNodeFlash/nnf-ec/internal/switchtec/pkg/switchtec"
 )
 
-// The Fabric Monitor is responsible for ensuring that the fabric and related sub-resource
+// The Fabric Monitor is responsible for ensuring that the fabriic and related sub-resource
 // are updated with the latest information from the switch. This runs as a background
 // thread, and periodically queries the fabric.
 func NewFabricMonitor(f *Fabric) *monitor {
@@ -38,7 +38,7 @@ type monitor struct {
 	fabric *Fabric
 }
 
-// Run will run the Fabric Monitor forever
+// Run Fabric Monitor forever
 func (m *monitor) Run() {
 
 	for {
@@ -49,10 +49,18 @@ func (m *monitor) Run() {
 			s := &m.fabric.switches[idx]
 
 			// The normal path is when the switch is operating without issue and we can
-			// poll the switch for any events, and process those events
+			// poll the switch for any events then process those events
 			if s.isReady() {
 
 				if events, err := s.dev.GetEvents(); err == nil {
+
+					// In the steady state there will be no events.
+					// Refresh the port status to ensure we're up to date.
+					if len(events) == 0 {
+						s.refreshPortStatus()
+						continue
+					}
+
 					for _, event := range events {
 						physPortId, isDown := m.getEventInfo(event)
 
@@ -96,7 +104,7 @@ func (*monitor) checkSwitchStatus(s *Switch) {
 
 const invalidPhysicalPortId = math.MaxUint8
 
-func (m *monitor) getEventInfo(e switchtec.GfmsEvent) (uint8, bool) {
+func (m *monitor) getEventInfo(e switchtec.GfmsEvent) (uint8, bool /* is down event? */) {
 
 	switch e.Id {
 	case switchtec.FabricLinkUp_GfmsEvent, switchtec.FabricLinkDown_GfmsEvent:

pkg/manager-nnf/manager.go

@@ -396,7 +396,7 @@ func (s *StorageService) cleanupVolumes() {
 }
 
 // Initialize is responsible for initializing the NNF Storage Service; the
-// Storage Service must complete initialization without error prior any
+// Storage Service must complete initialization without error prior to any
 // access to the Storage Service. Failure to initialize will cause the
 // storage service to misbehave.
 func (s *StorageService) Initialize(log ec.Logger, ctrl NnfControllerInterface) error {
@@ -529,7 +529,7 @@ func (s *StorageService) EventHandler(e event.Event) error {
 	}
 
 	// Check if the fabric is ready; that is all devices are enumerated and discovery
-	// is complete. We
+	// is complete.
 	if e.Is(msgreg.FabricReadyNnf("")) {
 		log.V(1).Info("Fabric ready")