Check whether or not we know the version once we start receiving mess…

…ages from services

Fixes a case where daemon is up but service is not for the initial "Get Version" messages

internal/metrics/crawler.go

@@ -30,6 +30,7 @@ type CrawlerServiceMetrics struct {
 	metrics *Metrics
 
 	// General Service Metrics
+	gotVersionResponse bool
 	version *prometheus.GaugeVec
 
 	// Current network
@@ -128,6 +129,7 @@ func (s *CrawlerServiceMetrics) SetupPollingMetrics(ctx context.Context) {}
 // Disconnected clears/unregisters metrics when the connection drops
 func (s *CrawlerServiceMetrics) Disconnected() {
 	s.version.Reset()
+	s.gotVersionResponse = false
 	s.totalNodes5Days.Unregister()
 	s.reliableNodes.Unregister()
 	s.ipv4Nodes5Days.Unregister()
@@ -143,9 +145,18 @@ func (s *CrawlerServiceMetrics) Reconnected() {
 
 // ReceiveResponse handles crawler responses that are returned over the websocket
 func (s *CrawlerServiceMetrics) ReceiveResponse(resp *types.WebsocketResponse) {
+	// Sometimes, when we reconnect, or start exporter before chia is running
+	// the daemon is up before the service, and the initial request for the version
+	// doesn't make it to the service
+	// daemon doesn't queue these messages for later, they just get dropped
+	if !s.gotVersionResponse {
+		utils.LogErr(s.metrics.client.FullNodeService.GetVersion(&rpc.GetVersionOptions{}))
+	}
+
 	switch resp.Command {
 	case "get_version":
 		versionHelper(resp, s.version)
+		s.gotVersionResponse = true
 	case "get_peer_counts":
 		fallthrough
 	case "loaded_initial_peers":

internal/metrics/farmer.go

@@ -24,6 +24,7 @@ type FarmerServiceMetrics struct {
 	metrics *Metrics
 
 	// General Service Metrics
+	gotVersionResponse bool
 	version *prometheus.GaugeVec
 
 	// Connection Metrics
@@ -108,6 +109,7 @@ func (s *FarmerServiceMetrics) SetupPollingMetrics(ctx context.Context) {}
 // Disconnected clears/unregisters metrics when the connection drops
 func (s *FarmerServiceMetrics) Disconnected() {
 	s.version.Reset()
+	s.gotVersionResponse = false
 	s.connectionCount.Reset()
 	s.plotFilesize.Reset()
 	s.plotCount.Reset()
@@ -123,9 +125,18 @@ func (s *FarmerServiceMetrics) Reconnected() {
 
 // ReceiveResponse handles crawler responses that are returned over the websocket
 func (s *FarmerServiceMetrics) ReceiveResponse(resp *types.WebsocketResponse) {
+	// Sometimes, when we reconnect, or start exporter before chia is running
+	// the daemon is up before the service, and the initial request for the version
+	// doesn't make it to the service
+	// daemon doesn't queue these messages for later, they just get dropped
+	if !s.gotVersionResponse {
+		utils.LogErr(s.metrics.client.FullNodeService.GetVersion(&rpc.GetVersionOptions{}))
+	}
+
 	switch resp.Command {
 	case "get_version":
 		versionHelper(resp, s.version)
+		s.gotVersionResponse = true
 	case "get_connections":
 		s.GetConnections(resp)
 	case "new_farming_info":

internal/metrics/fullnode.go

@@ -230,9 +230,9 @@ func (s *FullNodeServiceMetrics) Reconnected() {
 
 // ReceiveResponse handles full node related responses that are returned over the websocket
 func (s *FullNodeServiceMetrics) ReceiveResponse(resp *types.WebsocketResponse) {
-	// Sometimes, when we reconnect, or start exporter before the full node is up
-	// the daemon is up before the full node, and the initial request for the version
-	// doesn't make it to the service, since it wasn't up yet
+	// Sometimes, when we reconnect, or start exporter before chia is running
+	// the daemon is up before the service, and the initial request for the version
+	// doesn't make it to the service
 	// daemon doesn't queue these messages for later, they just get dropped
 	if !s.gotVersionResponse {
 		utils.LogErr(s.metrics.client.FullNodeService.GetVersion(&rpc.GetVersionOptions{}))

internal/metrics/harvester.go

@@ -25,6 +25,7 @@ type HarvesterServiceMetrics struct {
 	metrics *Metrics
 
 	// General Service Metrics
+	gotVersionResponse bool
 	version *prometheus.GaugeVec
 
 	// Connection Metrics
@@ -124,6 +125,7 @@ func (s *HarvesterServiceMetrics) httpGetPlots() {
 // Disconnected clears/unregisters metrics when the connection drops
 func (s *HarvesterServiceMetrics) Disconnected() {
 	s.version.Reset()
+	s.gotVersionResponse = false
 	s.connectionCount.Reset()
 	s.totalPlots.Unregister()
 	s.plotFilesize.Reset()
@@ -140,9 +142,18 @@ func (s *HarvesterServiceMetrics) Reconnected() {
 
 // ReceiveResponse handles crawler responses that are returned over the websocket
 func (s *HarvesterServiceMetrics) ReceiveResponse(resp *types.WebsocketResponse) {
+	// Sometimes, when we reconnect, or start exporter before chia is running
+	// the daemon is up before the service, and the initial request for the version
+	// doesn't make it to the service
+	// daemon doesn't queue these messages for later, they just get dropped
+	if !s.gotVersionResponse {
+		utils.LogErr(s.metrics.client.FullNodeService.GetVersion(&rpc.GetVersionOptions{}))
+	}
+
 	switch resp.Command {
 	case "get_version":
 		versionHelper(resp, s.version)
+		s.gotVersionResponse = true
 	case "get_connections":
 		s.GetConnections(resp)
 	case "farming_info":

internal/metrics/timelord.go

@@ -23,6 +23,7 @@ type TimelordServiceMetrics struct {
 	metrics *Metrics
 
 	// General Service Metrics
+	gotVersionResponse bool
 	version *prometheus.GaugeVec
 
 	// Timelord Metrics
@@ -62,6 +63,7 @@ func (s *TimelordServiceMetrics) SetupPollingMetrics(ctx context.Context) {}
 // Disconnected clears/unregisters metrics when the connection drops
 func (s *TimelordServiceMetrics) Disconnected() {
 	s.version.Reset()
+	s.gotVersionResponse = false
 	s.fastestTimelord.Unregister()
 	s.slowTimelord.Unregister()
 	s.estimatedIPS.Unregister()
@@ -75,10 +77,19 @@ func (s *TimelordServiceMetrics) Reconnected() {
 
 // ReceiveResponse handles crawler responses that are returned over the websocket
 func (s *TimelordServiceMetrics) ReceiveResponse(resp *types.WebsocketResponse) {
+	// Sometimes, when we reconnect, or start exporter before chia is running
+	// the daemon is up before the service, and the initial request for the version
+	// doesn't make it to the service
+	// daemon doesn't queue these messages for later, they just get dropped
+	if !s.gotVersionResponse {
+		utils.LogErr(s.metrics.client.FullNodeService.GetVersion(&rpc.GetVersionOptions{}))
+	}
+
 	//("finished_pot_challenge", "new_compact_proof", "skipping_peak", "new_peak")
 	switch resp.Command {
 	case "get_version":
 		versionHelper(resp, s.version)
+		s.gotVersionResponse = true
 	case "finished_pot":
 		s.FinishedPoT(resp)
 	case "new_compact_proof":

internal/metrics/wallet.go

@@ -25,6 +25,7 @@ type WalletServiceMetrics struct {
 	metrics *Metrics
 
 	// General Service Metrics
+	gotVersionResponse bool
 	version *prometheus.GaugeVec
 
 	// Connection Metrics
@@ -93,6 +94,7 @@ func (s *WalletServiceMetrics) SetupPollingMetrics(ctx context.Context) {
 // Disconnected clears/unregisters metrics when the connection drops
 func (s *WalletServiceMetrics) Disconnected() {
 	s.version.Reset()
+	s.gotVersionResponse = false
 	s.connectionCount.Reset()
 	s.walletSynced.Unregister()
 	s.confirmedBalance.Reset()
@@ -109,9 +111,18 @@ func (s *WalletServiceMetrics) Reconnected() {
 
 // ReceiveResponse handles wallet responses that are returned over the websocket
 func (s *WalletServiceMetrics) ReceiveResponse(resp *types.WebsocketResponse) {
+	// Sometimes, when we reconnect, or start exporter before chia is running
+	// the daemon is up before the service, and the initial request for the version
+	// doesn't make it to the service
+	// daemon doesn't queue these messages for later, they just get dropped
+	if !s.gotVersionResponse {
+		utils.LogErr(s.metrics.client.FullNodeService.GetVersion(&rpc.GetVersionOptions{}))
+	}
+
 	switch resp.Command {
 	case "get_version":
 		versionHelper(resp, s.version)
+		s.gotVersionResponse = true
 	case "get_connections":
 		s.GetConnections(resp)
 	case "coin_added":