mirror of
https://github.com/tailscale/tailscale.git
synced 2024-11-25 19:15:34 +00:00
5336362e64
- Wrap each prober function into a probe class that allows associating metric labels and custom metrics with a given probe; - Make sure all existing probe classes set a `class` metric label; - Move bandwidth probe size from being a metric label to a separate gauge metric; this will make it possible to use it to calculate average used bandwidth using a PromQL query; - Also export transfer time for the bandwidth prober (more accurate than the total probe time, since it excludes connection establishment time). Updates tailscale/corp#17912 Signed-off-by: Anton Tolchanov <anton@tailscale.com>
403 lines
10 KiB
Go
403 lines
10 KiB
Go
// Copyright (c) Tailscale Inc & AUTHORS
|
|
// SPDX-License-Identifier: BSD-3-Clause
|
|
|
|
package prober
|
|
|
|
import (
|
|
"context"
|
|
"errors"
|
|
"fmt"
|
|
"strings"
|
|
"sync"
|
|
"sync/atomic"
|
|
"testing"
|
|
"time"
|
|
|
|
"github.com/prometheus/client_golang/prometheus/testutil"
|
|
"tailscale.com/tstest"
|
|
)
|
|
|
|
const (
|
|
probeInterval = 8 * time.Second // So expvars that are integer numbers of seconds change
|
|
halfProbeInterval = probeInterval / 2
|
|
quarterProbeInterval = probeInterval / 4
|
|
convergenceTimeout = time.Second
|
|
convergenceSleep = time.Millisecond
|
|
aFewMillis = 20 * time.Millisecond
|
|
)
|
|
|
|
var epoch = time.Unix(0, 0)
|
|
|
|
func TestProberTiming(t *testing.T) {
|
|
clk := newFakeTime()
|
|
p := newForTest(clk.Now, clk.NewTicker)
|
|
|
|
invoked := make(chan struct{}, 1)
|
|
|
|
notCalled := func() {
|
|
t.Helper()
|
|
select {
|
|
case <-invoked:
|
|
t.Fatal("probe was invoked earlier than expected")
|
|
default:
|
|
}
|
|
}
|
|
called := func() {
|
|
t.Helper()
|
|
select {
|
|
case <-invoked:
|
|
case <-time.After(2 * time.Second):
|
|
t.Fatal("probe wasn't invoked as expected")
|
|
}
|
|
}
|
|
|
|
p.Run("test-probe", probeInterval, nil, FuncProbe(func(context.Context) error {
|
|
invoked <- struct{}{}
|
|
return nil
|
|
}))
|
|
|
|
waitActiveProbes(t, p, clk, 1)
|
|
|
|
called()
|
|
notCalled()
|
|
clk.Advance(probeInterval + halfProbeInterval)
|
|
called()
|
|
notCalled()
|
|
clk.Advance(quarterProbeInterval)
|
|
notCalled()
|
|
clk.Advance(probeInterval)
|
|
called()
|
|
notCalled()
|
|
}
|
|
|
|
func TestProberTimingSpread(t *testing.T) {
|
|
clk := newFakeTime()
|
|
p := newForTest(clk.Now, clk.NewTicker).WithSpread(true)
|
|
|
|
invoked := make(chan struct{}, 1)
|
|
|
|
notCalled := func() {
|
|
t.Helper()
|
|
select {
|
|
case <-invoked:
|
|
t.Fatal("probe was invoked earlier than expected")
|
|
default:
|
|
}
|
|
}
|
|
called := func() {
|
|
t.Helper()
|
|
select {
|
|
case <-invoked:
|
|
case <-time.After(2 * time.Second):
|
|
t.Fatal("probe wasn't invoked as expected")
|
|
}
|
|
}
|
|
|
|
probe := p.Run("test-spread-probe", probeInterval, nil, FuncProbe(func(context.Context) error {
|
|
invoked <- struct{}{}
|
|
return nil
|
|
}))
|
|
|
|
waitActiveProbes(t, p, clk, 1)
|
|
|
|
notCalled()
|
|
// Name of the probe (test-spread-probe) has been chosen to ensure that
|
|
// the initial delay is smaller than half of the probe interval.
|
|
clk.Advance(halfProbeInterval)
|
|
called()
|
|
notCalled()
|
|
|
|
// We need to wait until the main (non-initial) ticker in Probe.loop is
|
|
// waiting, or we could race and advance the test clock between when
|
|
// the initial delay ticker completes and before the ticker for the
|
|
// main loop is created. In this race, we'd first advance the test
|
|
// clock, then the ticker would be registered, and the test would fail
|
|
// because that ticker would never be fired.
|
|
err := tstest.WaitFor(convergenceTimeout, func() error {
|
|
clk.Lock()
|
|
defer clk.Unlock()
|
|
for _, tick := range clk.tickers {
|
|
tick.Lock()
|
|
stopped, interval := tick.stopped, tick.interval
|
|
tick.Unlock()
|
|
|
|
if stopped {
|
|
continue
|
|
}
|
|
// Test for the main loop, not the initialDelay
|
|
if interval == probe.interval {
|
|
return nil
|
|
}
|
|
}
|
|
|
|
return fmt.Errorf("no ticker with interval %d found", probe.interval)
|
|
})
|
|
if err != nil {
|
|
t.Fatal(err)
|
|
}
|
|
|
|
clk.Advance(quarterProbeInterval)
|
|
notCalled()
|
|
clk.Advance(probeInterval)
|
|
called()
|
|
notCalled()
|
|
}
|
|
|
|
func TestProberRun(t *testing.T) {
|
|
clk := newFakeTime()
|
|
p := newForTest(clk.Now, clk.NewTicker)
|
|
|
|
var (
|
|
mu sync.Mutex
|
|
cnt int
|
|
)
|
|
|
|
const startingProbes = 100
|
|
var probes []*Probe
|
|
|
|
for i := 0; i < startingProbes; i++ {
|
|
probes = append(probes, p.Run(fmt.Sprintf("probe%d", i), probeInterval, nil, FuncProbe(func(context.Context) error {
|
|
mu.Lock()
|
|
defer mu.Unlock()
|
|
cnt++
|
|
return nil
|
|
})))
|
|
}
|
|
|
|
checkCnt := func(want int) {
|
|
t.Helper()
|
|
err := tstest.WaitFor(convergenceTimeout, func() error {
|
|
mu.Lock()
|
|
defer mu.Unlock()
|
|
if cnt == want {
|
|
cnt = 0
|
|
return nil
|
|
}
|
|
return fmt.Errorf("wrong number of probe counter increments, got %d want %d", cnt, want)
|
|
})
|
|
if err != nil {
|
|
t.Fatal(err)
|
|
}
|
|
}
|
|
|
|
waitActiveProbes(t, p, clk, startingProbes)
|
|
checkCnt(startingProbes)
|
|
clk.Advance(probeInterval + halfProbeInterval)
|
|
checkCnt(startingProbes)
|
|
if c, err := testutil.GatherAndCount(p.metrics, "prober_result"); c != startingProbes || err != nil {
|
|
t.Fatalf("expected %d prober_result metrics; got %d (error %s)", startingProbes, c, err)
|
|
}
|
|
|
|
keep := startingProbes / 2
|
|
|
|
for i := keep; i < startingProbes; i++ {
|
|
probes[i].Close()
|
|
}
|
|
waitActiveProbes(t, p, clk, keep)
|
|
|
|
clk.Advance(probeInterval)
|
|
checkCnt(keep)
|
|
if c, err := testutil.GatherAndCount(p.metrics, "prober_result"); c != keep || err != nil {
|
|
t.Fatalf("expected %d prober_result metrics; got %d (error %s)", keep, c, err)
|
|
}
|
|
}
|
|
|
|
func TestPrometheus(t *testing.T) {
|
|
clk := newFakeTime()
|
|
p := newForTest(clk.Now, clk.NewTicker).WithMetricNamespace("probe")
|
|
|
|
var succeed atomic.Bool
|
|
p.Run("testprobe", probeInterval, map[string]string{"label": "value"}, FuncProbe(func(context.Context) error {
|
|
clk.Advance(aFewMillis)
|
|
if succeed.Load() {
|
|
return nil
|
|
}
|
|
return errors.New("failing, as instructed by test")
|
|
}))
|
|
|
|
waitActiveProbes(t, p, clk, 1)
|
|
|
|
err := tstest.WaitFor(convergenceTimeout, func() error {
|
|
want := fmt.Sprintf(`
|
|
# HELP probe_interval_secs Probe interval in seconds
|
|
# TYPE probe_interval_secs gauge
|
|
probe_interval_secs{class="",label="value",name="testprobe"} %f
|
|
# HELP probe_start_secs Latest probe start time (seconds since epoch)
|
|
# TYPE probe_start_secs gauge
|
|
probe_start_secs{class="",label="value",name="testprobe"} %d
|
|
# HELP probe_end_secs Latest probe end time (seconds since epoch)
|
|
# TYPE probe_end_secs gauge
|
|
probe_end_secs{class="",label="value",name="testprobe"} %d
|
|
# HELP probe_result Latest probe result (1 = success, 0 = failure)
|
|
# TYPE probe_result gauge
|
|
probe_result{class="",label="value",name="testprobe"} 0
|
|
`, probeInterval.Seconds(), epoch.Unix(), epoch.Add(aFewMillis).Unix())
|
|
return testutil.GatherAndCompare(p.metrics, strings.NewReader(want),
|
|
"probe_interval_secs", "probe_start_secs", "probe_end_secs", "probe_result")
|
|
})
|
|
if err != nil {
|
|
t.Fatal(err)
|
|
}
|
|
|
|
succeed.Store(true)
|
|
clk.Advance(probeInterval + halfProbeInterval)
|
|
|
|
err = tstest.WaitFor(convergenceTimeout, func() error {
|
|
start := epoch.Add(probeInterval + halfProbeInterval)
|
|
end := start.Add(aFewMillis)
|
|
want := fmt.Sprintf(`
|
|
# HELP probe_interval_secs Probe interval in seconds
|
|
# TYPE probe_interval_secs gauge
|
|
probe_interval_secs{class="",label="value",name="testprobe"} %f
|
|
# HELP probe_start_secs Latest probe start time (seconds since epoch)
|
|
# TYPE probe_start_secs gauge
|
|
probe_start_secs{class="",label="value",name="testprobe"} %d
|
|
# HELP probe_end_secs Latest probe end time (seconds since epoch)
|
|
# TYPE probe_end_secs gauge
|
|
probe_end_secs{class="",label="value",name="testprobe"} %d
|
|
# HELP probe_latency_millis Latest probe latency (ms)
|
|
# TYPE probe_latency_millis gauge
|
|
probe_latency_millis{class="",label="value",name="testprobe"} %d
|
|
# HELP probe_result Latest probe result (1 = success, 0 = failure)
|
|
# TYPE probe_result gauge
|
|
probe_result{class="",label="value",name="testprobe"} 1
|
|
`, probeInterval.Seconds(), start.Unix(), end.Unix(), aFewMillis.Milliseconds())
|
|
return testutil.GatherAndCompare(p.metrics, strings.NewReader(want),
|
|
"probe_interval_secs", "probe_start_secs", "probe_end_secs", "probe_latency_millis", "probe_result")
|
|
})
|
|
if err != nil {
|
|
t.Fatal(err)
|
|
}
|
|
}
|
|
|
|
func TestOnceMode(t *testing.T) {
|
|
clk := newFakeTime()
|
|
p := newForTest(clk.Now, clk.NewTicker).WithOnce(true)
|
|
|
|
p.Run("probe1", probeInterval, nil, FuncProbe(func(context.Context) error { return nil }))
|
|
p.Run("probe2", probeInterval, nil, FuncProbe(func(context.Context) error { return fmt.Errorf("error2") }))
|
|
p.Run("probe3", probeInterval, nil, FuncProbe(func(context.Context) error {
|
|
p.Run("probe4", probeInterval, nil, FuncProbe(func(context.Context) error {
|
|
return fmt.Errorf("error4")
|
|
}))
|
|
return nil
|
|
}))
|
|
|
|
p.Wait()
|
|
wantCount := 4
|
|
for _, metric := range []string{"prober_result", "prober_end_secs"} {
|
|
if c, err := testutil.GatherAndCount(p.metrics, metric); c != wantCount || err != nil {
|
|
t.Fatalf("expected %d %s metrics; got %d (error %s)", wantCount, metric, c, err)
|
|
}
|
|
}
|
|
}
|
|
|
|
type fakeTicker struct {
|
|
ch chan time.Time
|
|
interval time.Duration
|
|
|
|
sync.Mutex
|
|
next time.Time
|
|
stopped bool
|
|
}
|
|
|
|
func (t *fakeTicker) Chan() <-chan time.Time {
|
|
return t.ch
|
|
}
|
|
|
|
func (t *fakeTicker) Stop() {
|
|
t.Lock()
|
|
defer t.Unlock()
|
|
t.stopped = true
|
|
}
|
|
|
|
func (t *fakeTicker) fire(now time.Time) {
|
|
t.Lock()
|
|
defer t.Unlock()
|
|
// Slight deviation from the stdlib ticker: time.Ticker will
|
|
// adjust t.next to make up for missed ticks, whereas we tick on a
|
|
// fixed interval regardless of receiver behavior. In our case
|
|
// this is fine, since we're using the ticker as a wakeup
|
|
// mechanism and not a precise timekeeping system.
|
|
select {
|
|
case t.ch <- now:
|
|
default:
|
|
}
|
|
for now.After(t.next) {
|
|
t.next = t.next.Add(t.interval)
|
|
}
|
|
}
|
|
|
|
type fakeTime struct {
|
|
sync.Mutex
|
|
*sync.Cond
|
|
curTime time.Time
|
|
tickers []*fakeTicker
|
|
}
|
|
|
|
func newFakeTime() *fakeTime {
|
|
ret := &fakeTime{
|
|
curTime: epoch,
|
|
}
|
|
ret.Cond = &sync.Cond{L: &ret.Mutex}
|
|
return ret
|
|
}
|
|
|
|
func (t *fakeTime) Now() time.Time {
|
|
t.Lock()
|
|
defer t.Unlock()
|
|
ret := t.curTime
|
|
return ret
|
|
}
|
|
|
|
func (t *fakeTime) NewTicker(d time.Duration) ticker {
|
|
t.Lock()
|
|
defer t.Unlock()
|
|
ret := &fakeTicker{
|
|
ch: make(chan time.Time, 1),
|
|
interval: d,
|
|
next: t.curTime.Add(d),
|
|
}
|
|
t.tickers = append(t.tickers, ret)
|
|
t.Cond.Broadcast()
|
|
return ret
|
|
}
|
|
|
|
func (t *fakeTime) Advance(d time.Duration) {
|
|
t.Lock()
|
|
defer t.Unlock()
|
|
t.curTime = t.curTime.Add(d)
|
|
for _, tick := range t.tickers {
|
|
if t.curTime.After(tick.next) {
|
|
tick.fire(t.curTime)
|
|
}
|
|
}
|
|
}
|
|
|
|
func (t *fakeTime) activeTickers() (count int) {
|
|
t.Lock()
|
|
defer t.Unlock()
|
|
for _, tick := range t.tickers {
|
|
if !tick.stopped {
|
|
count += 1
|
|
}
|
|
}
|
|
return
|
|
}
|
|
|
|
func waitActiveProbes(t *testing.T, p *Prober, clk *fakeTime, want int) {
|
|
t.Helper()
|
|
err := tstest.WaitFor(convergenceTimeout, func() error {
|
|
if got := p.activeProbes(); got != want {
|
|
return fmt.Errorf("installed probe count is %d, want %d", got, want)
|
|
}
|
|
if got := clk.activeTickers(); got != want {
|
|
return fmt.Errorf("active ticker count is %d, want %d", got, want)
|
|
}
|
|
return nil
|
|
})
|
|
if err != nil {
|
|
t.Fatal(err)
|
|
}
|
|
}
|