control/controlclient,health,tailcfg: refactor control health messages (#15839)

* control/controlclient,health,tailcfg: refactor control health messages

Updates tailscale/corp#27759

Signed-off-by: James Sanderson <jsanderson@tailscale.com>
Signed-off-by: Paul Scott <408401+icio@users.noreply.github.com>
Co-authored-by: Paul Scott <408401+icio@users.noreply.github.com>

control/controlclient/auto.go

@@ -12,6 +12,7 @@ import (
 	"sync/atomic"
 	"time"
 
+	"tailscale.com/health"
 	"tailscale.com/logtail/backoff"
 	"tailscale.com/net/sockstats"
 	"tailscale.com/tailcfg"
@@ -198,7 +199,11 @@ func NewNoStart(opts Options) (_ *Auto, err error) {
 	c.mapCtx, c.mapCancel = context.WithCancel(context.Background())
 	c.mapCtx = sockstats.WithSockStats(c.mapCtx, sockstats.LabelControlClientAuto, opts.Logf)
 
-	c.unregisterHealthWatch = opts.HealthTracker.RegisterWatcher(direct.ReportHealthChange)
+	c.unregisterHealthWatch = opts.HealthTracker.RegisterWatcher(func(c health.Change) {
+		if c.WarnableChanged {
+			direct.ReportWarnableChange(c.Warnable, c.UnhealthyState)
+		}
+	})
 	return c, nil
 
 }

control/controlclient/direct.go

@@ -1623,9 +1623,9 @@ func postPingResult(start time.Time, logf logger.Logf, c *http.Client, pr *tailc
 	return nil
 }
 
-// ReportHealthChange reports to the control plane a change to this node's
+// ReportWarnableChange reports to the control plane a change to this node's
 // health. w must be non-nil. us can be nil to indicate a healthy state for w.
-func (c *Direct) ReportHealthChange(w *health.Warnable, us *health.UnhealthyState) {
+func (c *Direct) ReportWarnableChange(w *health.Warnable, us *health.UnhealthyState) {
 	if w == health.NetworkStatusWarnable || w == health.IPNStateWarnable || w == health.LoginStateWarnable {
 		// We don't report these. These include things like the network is down
 		// (in which case we can't report anyway) or the user wanted things

control/controlclient/map.go

@@ -6,7 +6,10 @@ package controlclient
 import (
 	"cmp"
 	"context"
+	"crypto/sha256"
+	"encoding/hex"
 	"encoding/json"
+	"io"
 	"maps"
 	"net"
 	"reflect"
@@ -828,6 +831,16 @@ func (ms *mapSession) sortedPeers() []tailcfg.NodeView {
 func (ms *mapSession) netmap() *netmap.NetworkMap {
 	peerViews := ms.sortedPeers()
 
+	// Convert all ms.lastHealth to the new [netmap.NetworkMap.DisplayMessages].
+	var msgs map[tailcfg.DisplayMessageID]tailcfg.DisplayMessage
+	for _, h := range ms.lastHealth {
+		mak.Set(&msgs, tailcfg.DisplayMessageID("control-health-"+strhash(h)), tailcfg.DisplayMessage{
+			Title:    "Coordination server reports an issue",
+			Severity: tailcfg.SeverityMedium,
+			Text:     "The coordination server is reporting a health issue: " + h,
+		})
+	}
+
 	nm := &netmap.NetworkMap{
 		NodeKey:           ms.publicNodeKey,
 		PrivateKey:        ms.privateNodeKey,
@@ -842,7 +855,7 @@ func (ms *mapSession) netmap() *netmap.NetworkMap {
 		SSHPolicy:         ms.lastSSHPolicy,
 		CollectServices:   ms.collectServices,
 		DERPMap:           ms.lastDERPMap,
-		ControlHealth:     ms.lastHealth,
+		DisplayMessages:   msgs,
 		TKAEnabled:        ms.lastTKAInfo != nil && !ms.lastTKAInfo.Disabled,
 	}
 
@@ -868,5 +881,12 @@ func (ms *mapSession) netmap() *netmap.NetworkMap {
 	if DevKnob.ForceProxyDNS() {
 		nm.DNS.Proxied = true
 	}
+
 	return nm
 }
+
+func strhash(h string) string {
+	s := sha256.New()
+	io.WriteString(s, h)
+	return hex.EncodeToString(s.Sum(nil))
+}

control/controlclient/map_test.go

@@ -7,6 +7,7 @@ import (
 	"context"
 	"encoding/json"
 	"fmt"
+	"maps"
 	"net/netip"
 	"reflect"
 	"strings"
@@ -1148,23 +1149,36 @@ func TestNetmapHealthIntegration(t *testing.T) {
 	ht.GotStreamedMapResponse()
 
 	nm := ms.netmapForResponse(&tailcfg.MapResponse{
-		Health: []string{"Test message"},
+		Health: []string{
+			"Test message",
+			"Another message",
+		},
 	})
-	ht.SetControlHealth(nm.ControlHealth)
+	ht.SetControlHealth(nm.DisplayMessages)
 
-	state := ht.CurrentState()
-	warning, ok := state.Warnings["control-health"]
+	want := map[health.WarnableCode]health.UnhealthyState{
+		"control-health-c0719e9a8d5d838d861dc6f675c899d2b309a3a65bb9fe6b11e5afcbf9a2c0b1": {
+			WarnableCode: "control-health-c0719e9a8d5d838d861dc6f675c899d2b309a3a65bb9fe6b11e5afcbf9a2c0b1",
+			Title:        "Coordination server reports an issue",
+			Severity:     health.SeverityMedium,
+			Text:         "The coordination server is reporting a health issue: Test message",
+		},
+		"control-health-1dc7017a73a3c55c0d6a8423e3813c7ab6562d9d3064c2ec6ac7822f61b1db9c": {
+			WarnableCode: "control-health-1dc7017a73a3c55c0d6a8423e3813c7ab6562d9d3064c2ec6ac7822f61b1db9c",
+			Title:        "Coordination server reports an issue",
+			Severity:     health.SeverityMedium,
+			Text:         "The coordination server is reporting a health issue: Another message",
+		},
+	}
 
-	if !ok {
-		t.Fatal("no warning found in current state with code 'control-health'")
+	got := maps.Clone(ht.CurrentState().Warnings)
+	for k := range got {
+		if !strings.HasPrefix(string(k), "control-health") {
+			delete(got, k)
+		}
 	}
-	if got, want := warning.Title, "Coordination server reports an issue"; got != want {
-		t.Errorf("warning.Title = %q, want %q", got, want)
-	}
-	if got, want := warning.Severity, health.SeverityMedium; got != want {
-		t.Errorf("warning.Severity = %s, want %s", got, want)
-	}
-	if got, want := warning.Text, "The coordination server is reporting an health issue: Test message"; got != want {
-		t.Errorf("warning.Text = %q, want %q", got, want)
+
+	if d := cmp.Diff(want, got); d != "" {
+		t.Fatalf("CurrentStatus().Warnings[\"control-health*\"] different than expected (-want +got)\n%s", d)
 	}
 }