mirror of
https://github.com/tailscale/tailscale.git
synced 2025-05-09 17:16:50 +00:00
control/controlclient,health,tailcfg: refactor control health messages
Updates tailscale/corp#27759 Signed-off-by: James Sanderson <jsanderson@tailscale.com>
This commit is contained in:
parent
165b99278b
commit
2f9907d4ea
@ -133,6 +133,7 @@ tailscale.com/cmd/derper dependencies: (generated by github.com/tailscale/depawa
|
||||
tailscale.com/tsweb from tailscale.com/cmd/derper+
|
||||
tailscale.com/tsweb/promvarz from tailscale.com/cmd/derper
|
||||
tailscale.com/tsweb/varz from tailscale.com/tsweb+
|
||||
tailscale.com/types/bools from tailscale.com/tailcfg
|
||||
tailscale.com/types/dnstype from tailscale.com/tailcfg+
|
||||
tailscale.com/types/empty from tailscale.com/ipn
|
||||
tailscale.com/types/ipproto from tailscale.com/tailcfg+
|
||||
|
@ -894,7 +894,7 @@ tailscale.com/cmd/k8s-operator dependencies: (generated by github.com/tailscale/
|
||||
tailscale.com/tsweb from tailscale.com/util/eventbus
|
||||
tailscale.com/tsweb/varz from tailscale.com/util/usermetric+
|
||||
tailscale.com/types/appctype from tailscale.com/ipn/ipnlocal
|
||||
tailscale.com/types/bools from tailscale.com/tsnet
|
||||
tailscale.com/types/bools from tailscale.com/tsnet+
|
||||
tailscale.com/types/dnstype from tailscale.com/ipn/ipnlocal+
|
||||
tailscale.com/types/empty from tailscale.com/ipn+
|
||||
tailscale.com/types/ipproto from tailscale.com/net/flowtrack+
|
||||
|
@ -61,6 +61,7 @@ tailscale.com/cmd/stund dependencies: (generated by github.com/tailscale/depawar
|
||||
tailscale.com/tsweb from tailscale.com/cmd/stund+
|
||||
tailscale.com/tsweb/promvarz from tailscale.com/cmd/stund
|
||||
tailscale.com/tsweb/varz from tailscale.com/tsweb+
|
||||
tailscale.com/types/bools from tailscale.com/tailcfg
|
||||
tailscale.com/types/dnstype from tailscale.com/tailcfg
|
||||
tailscale.com/types/ipproto from tailscale.com/tailcfg
|
||||
tailscale.com/types/key from tailscale.com/tailcfg
|
||||
|
@ -139,6 +139,7 @@ tailscale.com/cmd/tailscale dependencies: (generated by github.com/tailscale/dep
|
||||
tailscale.com/tstime/rate from tailscale.com/cmd/tailscale/cli+
|
||||
tailscale.com/tsweb from tailscale.com/util/eventbus
|
||||
tailscale.com/tsweb/varz from tailscale.com/util/usermetric+
|
||||
tailscale.com/types/bools from tailscale.com/tailcfg
|
||||
tailscale.com/types/dnstype from tailscale.com/tailcfg+
|
||||
tailscale.com/types/empty from tailscale.com/ipn
|
||||
tailscale.com/types/ipproto from tailscale.com/ipn+
|
||||
|
@ -373,6 +373,7 @@ tailscale.com/cmd/tailscaled dependencies: (generated by github.com/tailscale/de
|
||||
tailscale.com/tsweb from tailscale.com/util/eventbus
|
||||
tailscale.com/tsweb/varz from tailscale.com/cmd/tailscaled+
|
||||
tailscale.com/types/appctype from tailscale.com/ipn/ipnlocal
|
||||
tailscale.com/types/bools from tailscale.com/tailcfg
|
||||
tailscale.com/types/dnstype from tailscale.com/ipn/ipnlocal+
|
||||
tailscale.com/types/empty from tailscale.com/ipn+
|
||||
tailscale.com/types/flagtype from tailscale.com/cmd/tailscaled
|
||||
|
@ -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
|
||||
|
||||
}
|
||||
|
@ -1625,7 +1625,7 @@ func postPingResult(start time.Time, logf logger.Logf, c *http.Client, pr *tailc
|
||||
|
||||
// ReportHealthChange 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
|
||||
|
@ -7,6 +7,7 @@ import (
|
||||
"cmp"
|
||||
"context"
|
||||
"encoding/json"
|
||||
"fmt"
|
||||
"maps"
|
||||
"net"
|
||||
"reflect"
|
||||
@ -828,6 +829,23 @@ func (ms *mapSession) sortedPeers() []tailcfg.NodeView {
|
||||
func (ms *mapSession) netmap() *netmap.NetworkMap {
|
||||
peerViews := ms.sortedPeers()
|
||||
|
||||
var displayMessages map[tailcfg.DisplayMessageID]tailcfg.DisplayMessage
|
||||
|
||||
if len(ms.lastHealth) > 0 {
|
||||
// As they all resolve to the same ID, we can only pass on one message
|
||||
// from ControlHealth. In practice we generally send 0 or 1, but
|
||||
// if there are multiple, the last one wins.
|
||||
h := ms.lastHealth[len(ms.lastHealth)-1]
|
||||
displayMessages = map[tailcfg.DisplayMessageID]tailcfg.DisplayMessage{
|
||||
"control-health": {
|
||||
Title: "Coordination server reports an issue",
|
||||
Severity: tailcfg.SeverityMedium,
|
||||
Text: fmt.Sprintf("The coordination server is reporting a health issue: %s", h),
|
||||
},
|
||||
}
|
||||
|
||||
}
|
||||
|
||||
nm := &netmap.NetworkMap{
|
||||
NodeKey: ms.publicNodeKey,
|
||||
PrivateKey: ms.privateNodeKey,
|
||||
@ -842,7 +860,7 @@ func (ms *mapSession) netmap() *netmap.NetworkMap {
|
||||
SSHPolicy: ms.lastSSHPolicy,
|
||||
CollectServices: ms.collectServices,
|
||||
DERPMap: ms.lastDERPMap,
|
||||
ControlHealth: ms.lastHealth,
|
||||
DisplayMessages: displayMessages,
|
||||
TKAEnabled: ms.lastTKAInfo != nil && !ms.lastTKAInfo.Disabled,
|
||||
}
|
||||
|
||||
|
@ -1150,7 +1150,7 @@ func TestNetmapHealthIntegration(t *testing.T) {
|
||||
nm := ms.netmapForResponse(&tailcfg.MapResponse{
|
||||
Health: []string{"Test message"},
|
||||
})
|
||||
ht.SetControlHealth(nm.ControlHealth)
|
||||
ht.SetControlHealth(nm.DisplayMessages)
|
||||
|
||||
state := ht.CurrentState()
|
||||
warning, ok := state.Warnings["control-health"]
|
||||
@ -1164,7 +1164,7 @@ func TestNetmapHealthIntegration(t *testing.T) {
|
||||
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 {
|
||||
if got, want := warning.Text, "The coordination server is reporting a health issue: Test message"; got != want {
|
||||
t.Errorf("warning.Text = %q, want %q", got, want)
|
||||
}
|
||||
}
|
||||
|
147
health/health.go
147
health/health.go
@ -88,34 +88,35 @@ type Tracker struct {
|
||||
// sysErr maps subsystems to their current error (or nil if the subsystem is healthy)
|
||||
// Deprecated: using Warnables should be preferred
|
||||
sysErr map[Subsystem]error
|
||||
watchers set.HandleSet[func(*Warnable, *UnhealthyState)] // opt func to run if error state changes
|
||||
watchers set.HandleSet[func(Change)] // opt func to run if error state changes
|
||||
timer tstime.TimerController
|
||||
|
||||
latestVersion *tailcfg.ClientVersion // or nil
|
||||
checkForUpdates bool
|
||||
applyUpdates opt.Bool
|
||||
|
||||
inMapPoll bool
|
||||
inMapPollSince time.Time
|
||||
lastMapPollEndedAt time.Time
|
||||
lastStreamedMapResponse time.Time
|
||||
lastNoiseDial time.Time
|
||||
derpHomeRegion int
|
||||
derpHomeless bool
|
||||
derpRegionConnected map[int]bool
|
||||
derpRegionHealthProblem map[int]string
|
||||
derpRegionLastFrame map[int]time.Time
|
||||
derpMap *tailcfg.DERPMap // last DERP map from control, could be nil if never received one
|
||||
lastMapRequestHeard time.Time // time we got a 200 from control for a MapRequest
|
||||
ipnState string
|
||||
ipnWantRunning bool
|
||||
ipnWantRunningLastTrue time.Time // when ipnWantRunning last changed false -> true
|
||||
anyInterfaceUp opt.Bool // empty means unknown (assume true)
|
||||
controlHealth []string
|
||||
lastLoginErr error
|
||||
localLogConfigErr error
|
||||
tlsConnectionErrors map[string]error // map[ServerName]error
|
||||
metricHealthMessage *metrics.MultiLabelMap[metricHealthMessageLabel]
|
||||
inMapPoll bool
|
||||
inMapPollSince time.Time
|
||||
lastMapPollEndedAt time.Time
|
||||
lastStreamedMapResponse time.Time
|
||||
lastNoiseDial time.Time
|
||||
derpHomeRegion int
|
||||
derpHomeless bool
|
||||
derpRegionConnected map[int]bool
|
||||
derpRegionHealthProblem map[int]string
|
||||
derpRegionLastFrame map[int]time.Time
|
||||
derpMap *tailcfg.DERPMap // last DERP map from control, could be nil if never received one
|
||||
lastMapRequestHeard time.Time // time we got a 200 from control for a MapRequest
|
||||
ipnState string
|
||||
ipnWantRunning bool
|
||||
ipnWantRunningLastTrue time.Time // when ipnWantRunning last changed false -> true
|
||||
anyInterfaceUp opt.Bool // empty means unknown (assume true)
|
||||
lastNotifiedControlMessages map[tailcfg.DisplayMessageID]tailcfg.DisplayMessage // latest control messages processed, kept for change detection
|
||||
controlMessages map[tailcfg.DisplayMessageID]tailcfg.DisplayMessage // latest control messages received
|
||||
lastLoginErr error
|
||||
localLogConfigErr error
|
||||
tlsConnectionErrors map[string]error // map[ServerName]error
|
||||
metricHealthMessage *metrics.MultiLabelMap[metricHealthMessageLabel]
|
||||
}
|
||||
|
||||
func (t *Tracker) now() time.Time {
|
||||
@ -397,12 +398,18 @@ func (t *Tracker) setUnhealthyLocked(w *Warnable, args Args) {
|
||||
prevWs := t.warnableVal[w]
|
||||
mak.Set(&t.warnableVal, w, ws)
|
||||
if !ws.Equal(prevWs) {
|
||||
|
||||
change := Change{
|
||||
WarnableChanged: true,
|
||||
Warnable: w,
|
||||
UnhealthyState: w.unhealthyState(ws),
|
||||
}
|
||||
for _, cb := range t.watchers {
|
||||
// If the Warnable has been unhealthy for more than its TimeToVisible, the callback should be
|
||||
// executed immediately. Otherwise, the callback should be enqueued to run once the Warnable
|
||||
// becomes visible.
|
||||
if w.IsVisible(ws, t.now) {
|
||||
cb(w, w.unhealthyState(ws))
|
||||
cb(change)
|
||||
continue
|
||||
}
|
||||
|
||||
@ -415,7 +422,7 @@ func (t *Tracker) setUnhealthyLocked(w *Warnable, args Args) {
|
||||
// Check if the Warnable is still unhealthy, as it could have become healthy between the time
|
||||
// the timer was set for and the time it was executed.
|
||||
if t.warnableVal[w] != nil {
|
||||
cb(w, w.unhealthyState(ws))
|
||||
cb(change)
|
||||
delete(t.pendingVisibleTimers, w)
|
||||
}
|
||||
})
|
||||
@ -448,8 +455,23 @@ func (t *Tracker) setHealthyLocked(w *Warnable) {
|
||||
delete(t.pendingVisibleTimers, w)
|
||||
}
|
||||
|
||||
change := Change{
|
||||
WarnableChanged: true,
|
||||
Warnable: w,
|
||||
}
|
||||
for _, cb := range t.watchers {
|
||||
cb(w, nil)
|
||||
cb(change)
|
||||
}
|
||||
}
|
||||
|
||||
// notifyWatchersControlChangedLocked calls each watcher to signal that control
|
||||
// health messages have changed (and should be fetched via CurrentState).
|
||||
func (t *Tracker) notifyWatchersControlChangedLocked() {
|
||||
change := Change{
|
||||
ControlHealthChanged: true,
|
||||
}
|
||||
for _, cb := range t.watchers {
|
||||
cb(change)
|
||||
}
|
||||
}
|
||||
|
||||
@ -476,23 +498,52 @@ func (t *Tracker) AppendWarnableDebugFlags(base []string) []string {
|
||||
return ret
|
||||
}
|
||||
|
||||
// RegisterWatcher adds a function that will be called whenever the health state of any Warnable changes.
|
||||
// If a Warnable becomes unhealthy or its unhealthy state is updated, the callback will be called with its
|
||||
// current Representation.
|
||||
// If a Warnable becomes healthy, the callback will be called with ws set to nil.
|
||||
// The provided callback function will be executed in its own goroutine. The returned function can be used
|
||||
// to unregister the callback.
|
||||
func (t *Tracker) RegisterWatcher(cb func(w *Warnable, r *UnhealthyState)) (unregister func()) {
|
||||
return t.registerSyncWatcher(func(w *Warnable, r *UnhealthyState) {
|
||||
go cb(w, r)
|
||||
// Change is used to communicate a change to health. This could either be due to
|
||||
// a Warnable changing from health to unhealthy (or vice-versa), or because the
|
||||
// health messages received from the control-plane have changed.
|
||||
type Change struct {
|
||||
// Health messages from the POV of the control-plane server changed. If set, WarnableChanged, Warnable, and UnhealthyState are not set.
|
||||
ControlHealthChanged bool
|
||||
|
||||
// A client Warnable changed state. Mutually exclusive with ControlHealthChanged. If set, Warnable is set and ControlHealthChanged is not set.
|
||||
WarnableChanged bool
|
||||
|
||||
// Warnable that changed state. If it is now unhealthy, UnhealthyState will
|
||||
// be set.
|
||||
Warnable *Warnable
|
||||
|
||||
// UnhealthyState is set if the Warnable is unhealthy, and unset if the
|
||||
// Warnable is now healthy.
|
||||
UnhealthyState *UnhealthyState
|
||||
}
|
||||
|
||||
// RegisterWatcher adds a function that will be called whenever the health state
|
||||
// of any Warnable changes or the health messages from the control-plane change.
|
||||
//
|
||||
// If a Warnable becomes unhealthy or its unhealthy state is updated, the
|
||||
// callback will be called with WarnableChanged set to true and the Warnable and
|
||||
// its UnhealthyState.
|
||||
//
|
||||
// If a Warnable becomes healthy, the callback will be called with
|
||||
// WarnableChanged set to true, the Warnable set, and UnhealthyState set to nil
|
||||
//
|
||||
// If the health messages from the control-plane change, the callback will be
|
||||
// called with ControlHealthChanged set to true. Clients can fetch the set of
|
||||
// control-plane health messages by calling [Tracker.CurrentState].
|
||||
//
|
||||
// The provided callback function will be executed in its own goroutine. The
|
||||
// returned function can be used to unregister the callback.
|
||||
func (t *Tracker) RegisterWatcher(cb func(Change)) (unregister func()) {
|
||||
return t.registerSyncWatcher(func(c Change) {
|
||||
go cb(c)
|
||||
})
|
||||
}
|
||||
|
||||
// registerSyncWatcher adds a function that will be called whenever the health
|
||||
// state of any Warnable changes. The provided callback function will be
|
||||
// executed synchronously. Call RegisterWatcher to register any callbacks that
|
||||
// won't return from execution immediately.
|
||||
func (t *Tracker) registerSyncWatcher(cb func(w *Warnable, r *UnhealthyState)) (unregister func()) {
|
||||
// state changes. The provided callback function will be executed synchronously.
|
||||
// Call RegisterWatcher to register any callbacks that won't return from
|
||||
// execution immediately.
|
||||
func (t *Tracker) registerSyncWatcher(cb func(c Change)) (unregister func()) {
|
||||
if t.nil() {
|
||||
return func() {}
|
||||
}
|
||||
@ -500,7 +551,7 @@ func (t *Tracker) registerSyncWatcher(cb func(w *Warnable, r *UnhealthyState)) (
|
||||
t.mu.Lock()
|
||||
defer t.mu.Unlock()
|
||||
if t.watchers == nil {
|
||||
t.watchers = set.HandleSet[func(*Warnable, *UnhealthyState)]{}
|
||||
t.watchers = set.HandleSet[func(Change)]{}
|
||||
}
|
||||
handle := t.watchers.Add(cb)
|
||||
if t.timer == nil {
|
||||
@ -647,13 +698,15 @@ func (t *Tracker) updateLegacyErrorWarnableLocked(key Subsystem, err error) {
|
||||
}
|
||||
}
|
||||
|
||||
func (t *Tracker) SetControlHealth(problems []string) {
|
||||
func (t *Tracker) SetControlHealth(problems map[tailcfg.DisplayMessageID]tailcfg.DisplayMessage) {
|
||||
if t.nil() {
|
||||
return
|
||||
}
|
||||
t.mu.Lock()
|
||||
defer t.mu.Unlock()
|
||||
t.controlHealth = problems
|
||||
|
||||
t.controlMessages = problems
|
||||
|
||||
t.selfCheckLocked()
|
||||
}
|
||||
|
||||
@ -1159,14 +1212,10 @@ func (t *Tracker) updateBuiltinWarnablesLocked() {
|
||||
t.setHealthyLocked(derpRegionErrorWarnable)
|
||||
}
|
||||
|
||||
if len(t.controlHealth) > 0 {
|
||||
for _, s := range t.controlHealth {
|
||||
t.setUnhealthyLocked(controlHealthWarnable, Args{
|
||||
ArgError: s,
|
||||
})
|
||||
}
|
||||
} else {
|
||||
t.setHealthyLocked(controlHealthWarnable)
|
||||
// Check if control health messages have changed
|
||||
if !maps.EqualFunc(t.lastNotifiedControlMessages, t.controlMessages, tailcfg.DisplayMessage.Equal) {
|
||||
t.lastNotifiedControlMessages = t.controlMessages
|
||||
t.notifyWatchersControlChangedLocked()
|
||||
}
|
||||
|
||||
if err := envknob.ApplyDiskConfigError(); err != nil {
|
||||
|
@ -25,6 +25,7 @@ func TestAppendWarnableDebugFlags(t *testing.T) {
|
||||
w := Register(&Warnable{
|
||||
Code: WarnableCode(fmt.Sprintf("warnable-code-%d", i)),
|
||||
MapDebugFlag: fmt.Sprint(i),
|
||||
Text: StaticMessage(""),
|
||||
})
|
||||
defer unregister(w)
|
||||
if i%2 == 0 {
|
||||
@ -114,7 +115,9 @@ func TestWatcher(t *testing.T) {
|
||||
becameUnhealthy := make(chan struct{})
|
||||
becameHealthy := make(chan struct{})
|
||||
|
||||
watcherFunc := func(w *Warnable, us *UnhealthyState) {
|
||||
watcherFunc := func(c Change) {
|
||||
w := c.Warnable
|
||||
us := c.UnhealthyState
|
||||
if w != testWarnable {
|
||||
t.Fatalf("watcherFunc was called, but with an unexpected Warnable: %v, want: %v", w, testWarnable)
|
||||
}
|
||||
@ -184,7 +187,9 @@ func TestSetUnhealthyWithTimeToVisible(t *testing.T) {
|
||||
becameUnhealthy := make(chan struct{})
|
||||
becameHealthy := make(chan struct{})
|
||||
|
||||
watchFunc := func(w *Warnable, us *UnhealthyState) {
|
||||
watchFunc := func(c Change) {
|
||||
w := c.Warnable
|
||||
us := c.UnhealthyState
|
||||
if w != mw {
|
||||
t.Fatalf("watcherFunc was called, but with an unexpected Warnable: %v, want: %v", w, w)
|
||||
}
|
||||
@ -457,21 +462,36 @@ func TestControlHealth(t *testing.T) {
|
||||
ht.SetIPNState("NeedsLogin", true)
|
||||
ht.GotStreamedMapResponse()
|
||||
|
||||
ht.SetControlHealth([]string{"Test message"})
|
||||
ht.SetControlHealth(map[tailcfg.DisplayMessageID]tailcfg.DisplayMessage{
|
||||
"control-health-test": {},
|
||||
})
|
||||
state := ht.CurrentState()
|
||||
warning, ok := state.Warnings["control-health"]
|
||||
warning, ok := state.Warnings["control-health-test"]
|
||||
|
||||
if !ok {
|
||||
t.Fatal("no warning found in current state with code 'control-health'")
|
||||
}
|
||||
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.WarnableCode, "control-health-test"; string(got) != want {
|
||||
t.Errorf("warning.WarnableCode = %q, want %q", got, want)
|
||||
}
|
||||
if got, want := warning.Severity, 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)
|
||||
}
|
||||
|
||||
func TestControlHealthNotifiesOnSet(t *testing.T) {
|
||||
ht := Tracker{}
|
||||
ht.SetIPNState("NeedsLogin", true)
|
||||
ht.GotStreamedMapResponse()
|
||||
|
||||
gotNotified := false
|
||||
ht.registerSyncWatcher(func(_ Change) {
|
||||
gotNotified = true
|
||||
})
|
||||
|
||||
ht.SetControlHealth(map[tailcfg.DisplayMessageID]tailcfg.DisplayMessage{
|
||||
"test": {},
|
||||
})
|
||||
|
||||
if !gotNotified {
|
||||
t.Errorf("watcher did not get called, want it to be called")
|
||||
}
|
||||
}
|
||||
|
||||
@ -480,12 +500,45 @@ func TestControlHealthNotifiesOnChange(t *testing.T) {
|
||||
ht.SetIPNState("NeedsLogin", true)
|
||||
ht.GotStreamedMapResponse()
|
||||
|
||||
ht.SetControlHealth(map[tailcfg.DisplayMessageID]tailcfg.DisplayMessage{
|
||||
"test-1": {},
|
||||
})
|
||||
|
||||
gotNotified := false
|
||||
ht.registerSyncWatcher(func(_ *Warnable, _ *UnhealthyState) {
|
||||
ht.registerSyncWatcher(func(_ Change) {
|
||||
gotNotified = true
|
||||
})
|
||||
|
||||
ht.SetControlHealth([]string{"Test message"})
|
||||
ht.SetControlHealth(map[tailcfg.DisplayMessageID]tailcfg.DisplayMessage{
|
||||
"test-2": {},
|
||||
})
|
||||
|
||||
if !gotNotified {
|
||||
t.Errorf("watcher did not get called, want it to be called")
|
||||
}
|
||||
}
|
||||
|
||||
func TestControlHealthNotifiesOnDetailsChange(t *testing.T) {
|
||||
ht := Tracker{}
|
||||
ht.SetIPNState("NeedsLogin", true)
|
||||
ht.GotStreamedMapResponse()
|
||||
|
||||
ht.SetControlHealth(map[tailcfg.DisplayMessageID]tailcfg.DisplayMessage{
|
||||
"test-1": {
|
||||
Title: "Title",
|
||||
},
|
||||
})
|
||||
|
||||
gotNotified := false
|
||||
ht.registerSyncWatcher(func(_ Change) {
|
||||
gotNotified = true
|
||||
})
|
||||
|
||||
ht.SetControlHealth(map[tailcfg.DisplayMessageID]tailcfg.DisplayMessage{
|
||||
"test-1": {
|
||||
Title: "Updated title",
|
||||
},
|
||||
})
|
||||
|
||||
if !gotNotified {
|
||||
t.Errorf("watcher did not get called, want it to be called")
|
||||
@ -498,16 +551,20 @@ func TestControlHealthNoNotifyOnUnchanged(t *testing.T) {
|
||||
ht.GotStreamedMapResponse()
|
||||
|
||||
// Set up an existing control health issue
|
||||
ht.SetControlHealth([]string{"Test message"})
|
||||
ht.SetControlHealth(map[tailcfg.DisplayMessageID]tailcfg.DisplayMessage{
|
||||
"test": {},
|
||||
})
|
||||
|
||||
// Now register our watcher
|
||||
gotNotified := false
|
||||
ht.registerSyncWatcher(func(_ *Warnable, _ *UnhealthyState) {
|
||||
ht.registerSyncWatcher(func(_ Change) {
|
||||
gotNotified = true
|
||||
})
|
||||
|
||||
// Send the same control health message again - should not notify
|
||||
ht.SetControlHealth([]string{"Test message"})
|
||||
ht.SetControlHealth(map[tailcfg.DisplayMessageID]tailcfg.DisplayMessage{
|
||||
"test": {},
|
||||
})
|
||||
|
||||
if gotNotified {
|
||||
t.Errorf("watcher got called, want it to not be called")
|
||||
@ -519,11 +576,13 @@ func TestControlHealthIgnoredOutsideMapPoll(t *testing.T) {
|
||||
ht.SetIPNState("NeedsLogin", true)
|
||||
|
||||
gotNotified := false
|
||||
ht.registerSyncWatcher(func(_ *Warnable, _ *UnhealthyState) {
|
||||
ht.registerSyncWatcher(func(_ Change) {
|
||||
gotNotified = true
|
||||
})
|
||||
|
||||
ht.SetControlHealth([]string{"Test message"})
|
||||
ht.SetControlHealth(map[tailcfg.DisplayMessageID]tailcfg.DisplayMessage{
|
||||
"control-health": {},
|
||||
})
|
||||
|
||||
state := ht.CurrentState()
|
||||
_, ok := state.Warnings["control-health"]
|
||||
|
@ -5,6 +5,8 @@ package health
|
||||
|
||||
import (
|
||||
"time"
|
||||
|
||||
"tailscale.com/tailcfg"
|
||||
)
|
||||
|
||||
// State contains the health status of the backend, and is
|
||||
@ -21,7 +23,8 @@ type State struct {
|
||||
}
|
||||
|
||||
// UnhealthyState contains information to be shown to the user to inform them
|
||||
// that a Warnable is currently unhealthy.
|
||||
// that a Warnable is currently unhealthy or DisplayMessage is being sent from
|
||||
// the control-plane.
|
||||
type UnhealthyState struct {
|
||||
WarnableCode WarnableCode
|
||||
Severity Severity
|
||||
@ -98,11 +101,39 @@ func (t *Tracker) CurrentState() *State {
|
||||
wm[w.Code] = *w.unhealthyState(ws)
|
||||
}
|
||||
|
||||
for id, message := range t.lastNotifiedControlMessages {
|
||||
s := UnhealthyStateFromDisplayMessage(id, message)
|
||||
wm[s.WarnableCode] = s
|
||||
}
|
||||
|
||||
return &State{
|
||||
Warnings: wm,
|
||||
}
|
||||
}
|
||||
|
||||
func UnhealthyStateFromDisplayMessage(id tailcfg.DisplayMessageID, message tailcfg.DisplayMessage) UnhealthyState {
|
||||
severity := SeverityMedium
|
||||
|
||||
switch message.Severity {
|
||||
case tailcfg.SeverityHigh:
|
||||
severity = SeverityHigh
|
||||
case tailcfg.SeverityMedium:
|
||||
severity = SeverityMedium
|
||||
case tailcfg.SeverityLow:
|
||||
severity = SeverityLow
|
||||
}
|
||||
|
||||
state := UnhealthyState{
|
||||
WarnableCode: WarnableCode(id),
|
||||
Severity: severity,
|
||||
Title: message.Title,
|
||||
Text: message.Text,
|
||||
ImpactsConnectivity: message.ImpactsConnectivity,
|
||||
}
|
||||
|
||||
return state
|
||||
}
|
||||
|
||||
// isEffectivelyHealthyLocked reports whether w is effectively healthy.
|
||||
// That means it's either actually healthy or it has a dependency that
|
||||
// that's unhealthy, so we should treat w as healthy to not spam users
|
||||
|
@ -238,16 +238,6 @@ var applyDiskConfigWarnable = Register(&Warnable{
|
||||
},
|
||||
})
|
||||
|
||||
// controlHealthWarnable is a Warnable that warns the user that the coordination server is reporting an health issue.
|
||||
var controlHealthWarnable = Register(&Warnable{
|
||||
Code: "control-health",
|
||||
Title: "Coordination server reports an issue",
|
||||
Severity: SeverityMedium,
|
||||
Text: func(args Args) string {
|
||||
return fmt.Sprintf("The coordination server is reporting an health issue: %v", args[ArgError])
|
||||
},
|
||||
})
|
||||
|
||||
// warmingUpWarnableDuration is the duration for which the warmingUpWarnable is reported by the backend after the user
|
||||
// has changed ipnWantRunning to true from false.
|
||||
const warmingUpWarnableDuration = 5 * time.Second
|
||||
|
@ -931,11 +931,15 @@ func (b *LocalBackend) linkChange(delta *netmon.ChangeDelta) {
|
||||
}
|
||||
}
|
||||
|
||||
func (b *LocalBackend) onHealthChange(w *health.Warnable, us *health.UnhealthyState) {
|
||||
if us == nil {
|
||||
b.logf("health(warnable=%s): ok", w.Code)
|
||||
} else {
|
||||
b.logf("health(warnable=%s): error: %s", w.Code, us.Text)
|
||||
func (b *LocalBackend) onHealthChange(change health.Change) {
|
||||
if change.WarnableChanged {
|
||||
w := change.Warnable
|
||||
us := change.UnhealthyState
|
||||
if us == nil {
|
||||
b.logf("health(warnable=%s): ok", w.Code)
|
||||
} else {
|
||||
b.logf("health(warnable=%s): error: %s", w.Code, us.Text)
|
||||
}
|
||||
}
|
||||
|
||||
// Whenever health changes, send the current health state to the frontend.
|
||||
@ -6132,7 +6136,7 @@ func (b *LocalBackend) setNetMapLocked(nm *netmap.NetworkMap) {
|
||||
b.pauseOrResumeControlClientLocked()
|
||||
|
||||
if nm != nil {
|
||||
b.health.SetControlHealth(nm.ControlHealth)
|
||||
b.health.SetControlHealth(nm.DisplayMessages)
|
||||
} else {
|
||||
b.health.SetControlHealth(nil)
|
||||
}
|
||||
|
@ -20,6 +20,7 @@ import (
|
||||
"strings"
|
||||
"time"
|
||||
|
||||
"tailscale.com/types/bools"
|
||||
"tailscale.com/types/dnstype"
|
||||
"tailscale.com/types/key"
|
||||
"tailscale.com/types/opt"
|
||||
@ -2028,7 +2029,7 @@ type MapResponse struct {
|
||||
// plane's perspective. A nil value means no change from the previous
|
||||
// MapResponse. A non-nil 0-length slice restores the health to good (no
|
||||
// known problems). A non-zero length slice are the list of problems that
|
||||
// the control place sees.
|
||||
// the control plane sees.
|
||||
//
|
||||
// Note that this package's type, due its use of a slice and omitempty, is
|
||||
// unable to marshal a zero-length non-nil slice. The control server needs
|
||||
@ -2078,6 +2079,65 @@ type MapResponse struct {
|
||||
DefaultAutoUpdate opt.Bool `json:",omitempty"`
|
||||
}
|
||||
|
||||
// DisplayMessage represents a health state of the node from the control plane's
|
||||
// perspective. It is deliberately similar to health.Warnable as both get
|
||||
// converted into health.UnhealthyState to be sent to the GUI.
|
||||
type DisplayMessage struct {
|
||||
// Title is a string that the GUI uses as title for this message. The title
|
||||
// should be short and fit in a single line.
|
||||
Title string
|
||||
|
||||
// Text is an extended string that the GUI will display to the user.
|
||||
Text string
|
||||
|
||||
// Severity is the severity of the DisplayMessage, which the GUI can use to
|
||||
// determine how to display it. Maps to health.Severity.
|
||||
Severity DisplayMessageSeverity
|
||||
|
||||
// ImpactsConnectivity is whether the health problem will impact the user's
|
||||
// ability to connect to the Internet or other nodes on the tailnet, which
|
||||
// the GUI can use to determine how to display it.
|
||||
ImpactsConnectivity bool `json:",omitempty"`
|
||||
}
|
||||
|
||||
// DisplayMessageID is a string that uniquely identifies the kind of health
|
||||
// issue (e.g. "session-expired").
|
||||
type DisplayMessageID string
|
||||
|
||||
// Equal returns true iff all fields are equal.
|
||||
func (m DisplayMessage) Equal(o DisplayMessage) bool {
|
||||
if c := cmp.Compare(m.Title, o.Title); c != 0 {
|
||||
return false
|
||||
}
|
||||
if c := cmp.Compare(m.Text, o.Text); c != 0 {
|
||||
return false
|
||||
}
|
||||
if c := cmp.Compare(m.Severity, o.Severity); c != 0 {
|
||||
return false
|
||||
}
|
||||
if c := bools.Compare(m.ImpactsConnectivity, o.ImpactsConnectivity); c != 0 {
|
||||
return false
|
||||
}
|
||||
return true
|
||||
}
|
||||
|
||||
// DisplayMessageSeverity represents how serious a [DisplayMessage] is. Analogous
|
||||
// to health.Severity.
|
||||
type DisplayMessageSeverity string
|
||||
|
||||
const (
|
||||
// SeverityHigh is the highest severity level, used for critical errors that need immediate attention.
|
||||
// On platforms where the client GUI can deliver notifications, a SeverityHigh message will trigger
|
||||
// a modal notification.
|
||||
SeverityHigh DisplayMessageSeverity = "high"
|
||||
// SeverityMedium is used for errors that are important but not critical. This won't trigger a modal
|
||||
// notification, however it will be displayed in a more visible way than a SeverityLow message.
|
||||
SeverityMedium DisplayMessageSeverity = "medium"
|
||||
// SeverityLow is used for less important notices that don't need immediate attention. The user will
|
||||
// have to go to a Settings window, or another "hidden" GUI location to see these messages.
|
||||
SeverityLow DisplayMessageSeverity = "low"
|
||||
)
|
||||
|
||||
// ClientVersion is information about the latest client version that's available
|
||||
// for the client (and whether they're already running it).
|
||||
//
|
||||
|
@ -878,3 +878,79 @@ func TestCheckTag(t *testing.T) {
|
||||
})
|
||||
}
|
||||
}
|
||||
|
||||
func TestDisplayMessageEqual(t *testing.T) {
|
||||
base := DisplayMessage{
|
||||
Title: "title",
|
||||
Text: "text",
|
||||
Severity: SeverityHigh,
|
||||
ImpactsConnectivity: false,
|
||||
}
|
||||
|
||||
type test struct {
|
||||
name string
|
||||
value DisplayMessage
|
||||
wantEqual bool
|
||||
}
|
||||
|
||||
for _, test := range []test{
|
||||
{
|
||||
name: "same",
|
||||
value: DisplayMessage{
|
||||
Title: "title",
|
||||
Text: "text",
|
||||
Severity: SeverityHigh,
|
||||
ImpactsConnectivity: false,
|
||||
},
|
||||
wantEqual: true,
|
||||
},
|
||||
{
|
||||
name: "different-title",
|
||||
value: DisplayMessage{
|
||||
Title: "different title",
|
||||
Text: "text",
|
||||
Severity: SeverityHigh,
|
||||
ImpactsConnectivity: false,
|
||||
},
|
||||
wantEqual: false,
|
||||
},
|
||||
{
|
||||
name: "different-text",
|
||||
value: DisplayMessage{
|
||||
Title: "title",
|
||||
Text: "different text",
|
||||
Severity: SeverityHigh,
|
||||
ImpactsConnectivity: false,
|
||||
},
|
||||
wantEqual: false,
|
||||
},
|
||||
{
|
||||
name: "different-severity",
|
||||
value: DisplayMessage{
|
||||
Title: "title",
|
||||
Text: "text",
|
||||
Severity: SeverityMedium,
|
||||
ImpactsConnectivity: false,
|
||||
},
|
||||
wantEqual: false,
|
||||
},
|
||||
{
|
||||
name: "different-impactsConnectivity",
|
||||
value: DisplayMessage{
|
||||
Title: "title",
|
||||
Text: "text",
|
||||
Severity: SeverityHigh,
|
||||
ImpactsConnectivity: true,
|
||||
},
|
||||
wantEqual: false,
|
||||
},
|
||||
} {
|
||||
t.Run(test.name, func(t *testing.T) {
|
||||
got := base.Equal(test.value)
|
||||
|
||||
if got != test.wantEqual {
|
||||
t.Errorf("Equal: got %t, want %t", got, test.wantEqual)
|
||||
}
|
||||
})
|
||||
}
|
||||
}
|
||||
|
@ -323,7 +323,7 @@ tailscale.com/tsnet dependencies: (generated by github.com/tailscale/depaware)
|
||||
tailscale.com/tsweb from tailscale.com/util/eventbus
|
||||
tailscale.com/tsweb/varz from tailscale.com/tsweb+
|
||||
tailscale.com/types/appctype from tailscale.com/ipn/ipnlocal
|
||||
tailscale.com/types/bools from tailscale.com/tsnet
|
||||
tailscale.com/types/bools from tailscale.com/tsnet+
|
||||
tailscale.com/types/dnstype from tailscale.com/client/local+
|
||||
tailscale.com/types/empty from tailscale.com/ipn+
|
||||
tailscale.com/types/ipproto from tailscale.com/ipn+
|
||||
|
@ -54,12 +54,12 @@ type NetworkMap struct {
|
||||
// between updates and should not be modified.
|
||||
DERPMap *tailcfg.DERPMap
|
||||
|
||||
// ControlHealth are the list of health check problems for this
|
||||
// DisplayMessages are the list of health check problems for this
|
||||
// node from the perspective of the control plane.
|
||||
// If empty, there are no known problems from the control plane's
|
||||
// point of view, but the node might know about its own health
|
||||
// check problems.
|
||||
ControlHealth []string
|
||||
DisplayMessages map[tailcfg.DisplayMessageID]tailcfg.DisplayMessage
|
||||
|
||||
// TKAEnabled indicates whether the tailnet key authority should be
|
||||
// enabled, from the perspective of the control plane.
|
||||
|
Loading…
x
Reference in New Issue
Block a user