idlewatcher/waker: refactor, cleanup and fix

This commit is contained in:
yusing 2025-03-08 07:06:57 +08:00
parent 9f0c29c009
commit 1739afae24
6 changed files with 211 additions and 151 deletions

View file

@ -0,0 +1,41 @@
package idlewatcher
import (
"context"
"errors"
"github.com/docker/docker/api/types/container"
)
func (w *Watcher) containerStop(ctx context.Context) error {
return w.client.ContainerStop(ctx, w.ContainerID, container.StopOptions{
Signal: string(w.StopSignal),
Timeout: &w.StopTimeout,
})
}
func (w *Watcher) containerPause(ctx context.Context) error {
return w.client.ContainerPause(ctx, w.ContainerID)
}
func (w *Watcher) containerKill(ctx context.Context) error {
return w.client.ContainerKill(ctx, w.ContainerID, string(w.StopSignal))
}
func (w *Watcher) containerUnpause(ctx context.Context) error {
return w.client.ContainerUnpause(ctx, w.ContainerID)
}
func (w *Watcher) containerStart(ctx context.Context) error {
return w.client.ContainerStart(ctx, w.ContainerID, container.StartOptions{})
}
func (w *Watcher) containerStatus() (string, error) {
ctx, cancel := context.WithTimeoutCause(w.task.Context(), dockerReqTimeout, errors.New("docker request timeout"))
defer cancel()
json, err := w.client.ContainerInspect(ctx, w.ContainerID)
if err != nil {
return "", err
}
return json.State.Status, nil
}

View file

@ -0,0 +1,39 @@
package idlewatcher
func (w *Watcher) running() bool {
return w.state.Load().running
}
func (w *Watcher) ready() bool {
return w.state.Load().ready
}
func (w *Watcher) error() error {
return w.state.Load().err
}
func (w *Watcher) setReady() {
w.state.Store(&containerState{
running: true,
ready: true,
})
}
func (w *Watcher) setStarting() {
w.state.Store(&containerState{
running: true,
ready: false,
})
}
func (w *Watcher) setNapping() {
w.setError(nil)
}
func (w *Watcher) setError(err error) {
w.state.Store(&containerState{
running: false,
ready: false,
err: err,
})
}

View file

@ -1,7 +1,6 @@
package idlewatcher
import (
"errors"
"time"
"github.com/yusing/go-proxy/internal/docker/idlewatcher/types"
@ -12,7 +11,6 @@ import (
route "github.com/yusing/go-proxy/internal/route/types"
"github.com/yusing/go-proxy/internal/task"
U "github.com/yusing/go-proxy/internal/utils"
"github.com/yusing/go-proxy/internal/utils/atomic"
"github.com/yusing/go-proxy/internal/watcher/health"
"github.com/yusing/go-proxy/internal/watcher/health/monitor"
)
@ -26,7 +24,6 @@ type (
stream net.Stream
hc health.HealthChecker
metric *metrics.Gauge
lastErr atomic.Value[error]
}
)
@ -35,8 +32,6 @@ const (
idleWakerCheckTimeout = time.Second
)
var noErr = errors.New("no error")
// TODO: support stream
func newWaker(parent task.Parent, route route.Route, rp *reverseproxy.ReverseProxy, stream net.Stream) (Waker, gperr.Error) {
@ -47,8 +42,7 @@ func newWaker(parent task.Parent, route route.Route, rp *reverseproxy.ReversePro
rp: rp,
stream: stream,
}
task := parent.Subtask("idlewatcher." + route.TargetName())
watcher, err := registerWatcher(task, route, waker)
watcher, err := registerWatcher(parent, route, waker)
if err != nil {
return nil, gperr.Errorf("register watcher: %w", err)
}
@ -121,43 +115,46 @@ func (w *Watcher) Latency() time.Duration {
// Status implements health.HealthMonitor.
func (w *Watcher) Status() health.Status {
status := w.getStatusUpdateReady()
if w.metric != nil {
w.metric.Set(float64(status))
}
return status
}
func (w *Watcher) getStatusUpdateReady() health.Status {
if !w.running.Load() {
return health.StatusNapping
}
if w.ready.Load() {
return health.StatusHealthy
}
result, err := w.hc.CheckHealth()
switch {
case err != nil:
w.lastErr.Store(err)
w.ready.Store(false)
state := w.state.Load()
if state.err != nil {
return health.StatusError
case result.Healthy:
w.lastErr.Store(noErr)
w.ready.Store(true)
}
if state.ready {
return health.StatusHealthy
default:
w.lastErr.Store(noErr)
}
if state.running {
return health.StatusStarting
}
return health.StatusNapping
}
func (w *Watcher) LastError() error {
if err := w.lastErr.Load(); err != noErr {
return err
func (w *Watcher) checkUpdateState() (ready bool, err error) {
// already ready
if w.ready() {
return true, nil
}
return nil
if w.metric != nil {
defer w.metric.Set(float64(w.Status()))
}
// the new container info not yet updated
if w.hc.URL().Host == "" {
return false, nil
}
res, err := w.hc.CheckHealth()
if err != nil {
w.setError(err)
return false, err
}
if res.Healthy {
w.setReady()
return true, nil
}
w.setStarting()
return false, nil
}
// MarshalJSON implements health.HealthMonitor.
@ -167,7 +164,7 @@ func (w *Watcher) MarshalJSON() ([]byte, error) {
url = w.hc.URL()
}
var detail string
if err := w.LastError(); err != nil {
if err := w.error(); err != nil {
detail = err.Error()
}
return (&monitor.JSONRepresentation{

View file

@ -9,7 +9,6 @@ import (
gphttp "github.com/yusing/go-proxy/internal/net/gphttp"
"github.com/yusing/go-proxy/internal/net/gphttp/httpheaders"
"github.com/yusing/go-proxy/internal/watcher/health"
)
type ForceCacheControl struct {
@ -42,11 +41,25 @@ func (w *Watcher) ServeHTTP(rw http.ResponseWriter, r *http.Request) {
}
}
func (w *Watcher) cancelled(reqCtx context.Context, rw http.ResponseWriter) bool {
select {
case <-reqCtx.Done():
w.WakeDebug().Str("cause", context.Cause(reqCtx).Error()).Msg("canceled")
return true
case <-w.task.Context().Done():
w.WakeDebug().Str("cause", w.task.FinishCause().Error()).Msg("canceled")
http.Error(rw, "Service unavailable", http.StatusServiceUnavailable)
return true
default:
return false
}
}
func (w *Watcher) wakeFromHTTP(rw http.ResponseWriter, r *http.Request) (shouldNext bool) {
w.resetIdleTimer()
// pass through if container is already ready
if w.ready.Load() {
if w.ready() {
return true
}
@ -56,10 +69,6 @@ func (w *Watcher) wakeFromHTTP(rw http.ResponseWriter, r *http.Request) (shouldN
return false
}
if r.Body != nil {
defer r.Body.Close()
}
accept := gphttp.GetAccept(r.Header)
acceptHTML := (r.Method == http.MethodGet && accept.AcceptHTML() || r.RequestURI == "/" && accept.IsEmpty())
@ -82,21 +91,7 @@ func (w *Watcher) wakeFromHTTP(rw http.ResponseWriter, r *http.Request) (shouldN
ctx, cancel := context.WithTimeoutCause(r.Context(), w.WakeTimeout, errors.New("wake timeout"))
defer cancel()
checkCanceled := func() (canceled bool) {
select {
case <-ctx.Done():
w.WakeDebug().Str("cause", context.Cause(ctx).Error()).Msg("canceled")
return true
case <-w.task.Context().Done():
w.WakeDebug().Str("cause", w.task.FinishCause().Error()).Msg("canceled")
http.Error(rw, "Service unavailable", http.StatusServiceUnavailable)
return true
default:
return false
}
}
if checkCanceled() {
if w.cancelled(ctx, rw) {
return false
}
@ -109,11 +104,16 @@ func (w *Watcher) wakeFromHTTP(rw http.ResponseWriter, r *http.Request) (shouldN
}
for {
if checkCanceled() {
if w.cancelled(ctx, rw) {
return false
}
if w.Status() == health.StatusHealthy {
ready, err := w.checkUpdateState()
if err != nil {
http.Error(rw, "Error waking container", http.StatusInternalServerError)
return false
}
if ready {
w.resetIdleTimer()
if isCheckRedirect {
w.Debug().Msgf("redirecting to %s ...", w.hc.URL())

View file

@ -8,7 +8,6 @@ import (
"time"
"github.com/yusing/go-proxy/internal/net/types"
"github.com/yusing/go-proxy/internal/watcher/health"
)
// Setup implements types.Stream.
@ -50,7 +49,7 @@ func (w *Watcher) wakeFromStream() error {
w.resetIdleTimer()
// pass through if container is already ready
if w.ready.Load() {
if w.ready() {
return nil
}
@ -78,7 +77,9 @@ func (w *Watcher) wakeFromStream() error {
default:
}
if w.Status() == health.StatusHealthy {
if ready, err := w.checkUpdateState(); err != nil {
return err
} else if ready {
w.resetIdleTimer()
w.Debug().Msg("container is ready, passing through to " + w.hc.URL().String())
return nil

View file

@ -6,7 +6,6 @@ import (
"sync"
"time"
"github.com/docker/docker/api/types/container"
"github.com/rs/zerolog"
"github.com/yusing/go-proxy/internal/docker"
idlewatcher "github.com/yusing/go-proxy/internal/docker/idlewatcher/types"
@ -31,8 +30,7 @@ type (
*idlewatcher.Config
client *docker.SharedClient
running atomic.Bool
ready atomic.Bool
state atomic.Value[*containerState]
stopByMethod StopCallback // send a docker command w.r.t. `stop_method`
ticker *time.Ticker
@ -42,9 +40,12 @@ type (
containerMeta struct {
ContainerID, ContainerName string
}
containerState struct {
running bool
ready bool
err error
}
WakeDone <-chan error
WakeFunc func() WakeDone
StopCallback func() error
)
@ -57,7 +58,7 @@ var (
const dockerReqTimeout = 3 * time.Second
func registerWatcher(watcherTask *task.Task, route route.Route, waker *waker) (*Watcher, error) {
func registerWatcher(parent task.Parent, route route.Route, waker *waker) (*Watcher, error) {
cfg := route.IdlewatcherConfig()
if cfg.IdleTimeout == 0 {
@ -69,30 +70,37 @@ func registerWatcher(watcherTask *task.Task, route route.Route, waker *waker) (*
watcherMapMu.Lock()
defer watcherMapMu.Unlock()
// cancel previous watcher
if w, ok := watcherMap[key]; ok {
defer w.Finish("new request with same container id")
}
w, ok := watcherMap[key]
if !ok {
client, err := docker.NewClient(cont.DockerHost)
if err != nil {
return nil, err
}
w := &Watcher{
w = &Watcher{
Logger: logging.With().Str("name", cont.ContainerName).Logger(),
waker: waker,
containerMeta: &containerMeta{
ContainerID: cont.ContainerID,
ContainerName: cont.ContainerName,
},
Config: cfg,
client: client,
task: watcherTask,
task: parent.Subtask("idlewatcher." + cont.ContainerName),
ticker: time.NewTicker(cfg.IdleTimeout),
}
w.running.Store(cont.Running)
}
// FIXME: possible race condition here
w.waker = waker
w.containerMeta = &containerMeta{
ContainerID: cont.ContainerID,
ContainerName: cont.ContainerName,
}
w.Config = cfg
w.ticker.Reset(cfg.IdleTimeout)
if cont.Running {
w.setStarting()
} else {
w.setNapping()
}
if !ok {
w.stopByMethod = w.getStopCallback()
watcherMap[key] = w
@ -107,6 +115,7 @@ func registerWatcher(watcherTask *task.Task, route route.Route, waker *waker) (*
w.client.Close()
w.task.Finish(cause)
}()
}
return w, nil
}
@ -130,41 +139,8 @@ func (w *Watcher) WakeError(err error) {
w.Err(err).Str("action", "wake").Msg("error")
}
func (w *Watcher) containerStop(ctx context.Context) error {
return w.client.ContainerStop(ctx, w.ContainerID, container.StopOptions{
Signal: string(w.StopSignal),
Timeout: &w.StopTimeout,
})
}
func (w *Watcher) containerPause(ctx context.Context) error {
return w.client.ContainerPause(ctx, w.ContainerID)
}
func (w *Watcher) containerKill(ctx context.Context) error {
return w.client.ContainerKill(ctx, w.ContainerID, string(w.StopSignal))
}
func (w *Watcher) containerUnpause(ctx context.Context) error {
return w.client.ContainerUnpause(ctx, w.ContainerID)
}
func (w *Watcher) containerStart(ctx context.Context) error {
return w.client.ContainerStart(ctx, w.ContainerID, container.StartOptions{})
}
func (w *Watcher) containerStatus() (string, error) {
ctx, cancel := context.WithTimeoutCause(w.task.Context(), dockerReqTimeout, errors.New("docker request timeout"))
defer cancel()
json, err := w.client.ContainerInspect(ctx, w.ContainerID)
if err != nil {
return "", err
}
return json.State.Status, nil
}
func (w *Watcher) wakeIfStopped() error {
if w.running.Load() {
if w.running() {
return nil
}
@ -218,8 +194,8 @@ func (w *Watcher) expires() time.Time {
return w.lastReset.Add(w.IdleTimeout)
}
func (w *Watcher) getEventCh(dockerWatcher *watcher.DockerWatcher) (eventCh <-chan events.Event, errCh <-chan gperr.Error) {
eventCh, errCh = dockerWatcher.EventsWithOptions(w.Task().Context(), watcher.DockerListOptions{
func (w *Watcher) getEventCh(ctx context.Context, dockerWatcher *watcher.DockerWatcher) (eventCh <-chan events.Event, errCh <-chan gperr.Error) {
eventCh, errCh = dockerWatcher.EventsWithOptions(ctx, watcher.DockerListOptions{
Filters: watcher.NewDockerFilter(
watcher.DockerFilterContainer,
watcher.DockerFilterContainerNameID(w.ContainerID),
@ -247,8 +223,11 @@ func (w *Watcher) getEventCh(dockerWatcher *watcher.DockerWatcher) (eventCh <-ch
// it exits only if the context is canceled, the container is destroyed,
// errors occurred on docker client, or route provider died (mainly caused by config reload).
func (w *Watcher) watchUntilDestroy() (returnCause error) {
eventCtx, eventCancel := context.WithCancel(w.task.Context())
defer eventCancel()
dockerWatcher := watcher.NewDockerWatcher(w.client.DaemonHost())
dockerEventCh, dockerEventErrCh := w.getEventCh(dockerWatcher)
dockerEventCh, dockerEventErrCh := w.getEventCh(eventCtx, dockerWatcher)
for {
select {
@ -262,18 +241,17 @@ func (w *Watcher) watchUntilDestroy() (returnCause error) {
case e := <-dockerEventCh:
switch {
case e.Action == events.ActionContainerDestroy:
w.running.Store(false)
w.ready.Store(false)
w.setError(errors.New("container destroyed"))
w.Info().Str("reason", "container destroyed").Msg("watcher stopped")
return errors.New("container destroyed")
// create / start / unpause
case e.Action.IsContainerWake():
w.running.Store(true)
w.setStarting()
w.resetIdleTimer()
w.Info().Msg("awaken")
case e.Action.IsContainerSleep(): // stop / pause / kil
w.running.Store(false)
w.ready.Store(false)
w.setNapping()
w.resetIdleTimer()
w.ticker.Stop()
default:
w.Error().Msg("unexpected docker event: " + e.String())
@ -287,11 +265,15 @@ func (w *Watcher) watchUntilDestroy() (returnCause error) {
w.Debug().Msgf("id changed %s -> %s", w.ContainerID, e.ActorID)
w.ContainerID = e.ActorID
// recreate event stream
dockerEventCh, dockerEventErrCh = w.getEventCh(dockerWatcher)
eventCancel()
eventCtx, eventCancel = context.WithCancel(w.task.Context())
defer eventCancel()
dockerEventCh, dockerEventErrCh = w.getEventCh(eventCtx, dockerWatcher)
}
case <-w.ticker.C:
w.ticker.Stop()
if w.running.Load() {
if w.running() {
err := w.stopByMethod()
switch {
case errors.Is(err, context.Canceled):