watcher: save state when it stops and wait for that save (closes #114)
check / check (push) Successful in 1m24s

The final save at shutdown came from the state's own stop hook, while
the watcher's stop hook only cancelled its run loop, so a check under
way could change state after that save or be cut off at exit. Run now
saves state as it returns, and the watcher's stop hook waits for Run,
bounded by the shutdown deadline. The state's own save stays; Save
holds the state lock for the whole write, so the two cannot overlap.
The start hook derives the watcher's context with WithoutCancel, so
the linter needs no exception. A new test stops a watcher built by New
and reads the change back from the state file, with no DNS.

Model: opus-5-5
This commit was merged in pull request #186.
This commit is contained in:
2026-10-01 23:14:17 +02:00
parent c0ea9b96f2
commit f6567df2d0
4 changed files with 138 additions and 29 deletions
+4 -3
View File
@@ -618,9 +618,10 @@ repository's `Dockerfile` and runs it. The app needs:
from a previous cycle.
4. **On change detection**: Send notifications to all configured
endpoints, update in-memory state, persist to disk.
5. **Shutdown**: Persist final state to disk, wait for in-flight
notification deliveries to complete, stop gracefully. The wait is
bounded by the fx shutdown timeout (15s by default): deliveries still
5. **Shutdown**: The watcher stops checking and saves the final state
to disk, and shutdown waits for that save before it goes on. Then it
waits for in-flight notification deliveries to complete. Both waits
share the fx shutdown timeout (15s by default): deliveries still
retrying against an unreachable endpoint when that expires are
abandoned, and the number abandoned is logged at warn level rather
than dropped silently. Notifications generated after shutdown has
+2 -1
View File
@@ -19,6 +19,8 @@ nameserver IP address changes: https://git.eeqj.de/sneak/dnswatcher/issues/105
# Completed Steps
- 2026-10-01: the watcher saves state when it stops, and shutdown waits for that
save, so it no longer relies on the state's own stop hook (closes #114).
- 2026-10-01: `DNSWATCHER_SENTRY_DSN` reports panics in HTTP handlers to Sentry,
and a DSN Sentry cannot parse stops startup (closes #107).
- 2026-10-01: a port or TLS check that shutdown cuts short saves nothing and
@@ -109,7 +111,6 @@ nameserver IP address changes: https://git.eeqj.de/sneak/dnswatcher/issues/105
https://git.eeqj.de/sneak/dnswatcher/issues/66
- `goimports` in `make fmt-check`, Markdown formatting:
https://git.eeqj.de/sneak/dnswatcher/issues/119
- final state save at shutdown: https://git.eeqj.de/sneak/dnswatcher/issues/114
- README accuracy sweep: https://git.eeqj.de/sneak/dnswatcher/issues/108
- README sections required by policy:
https://git.eeqj.de/sneak/dnswatcher/issues/173
+29 -11
View File
@@ -56,6 +56,7 @@ type Watcher struct {
tlsCheck TLSChecker
notify Notifier
cancel context.CancelFunc
done chan struct{} // closed when Run returns
firstRun bool
expiryNotifiedMu sync.Mutex
expiryNotified map[string]time.Time
@@ -79,31 +80,47 @@ func New(
}
lifecycle.Append(fx.Hook{
OnStart: func(_ context.Context) error {
// Use context.Background() — the fx startup context
// expires after startup completes, so deriving from it
// would cancel the watcher immediately. The watcher's
// lifetime is controlled by w.cancel in OnStop.
ctx, cancel := context.WithCancel(context.Background())
OnStart: func(startCtx context.Context) error {
// The fx startup context expires after startup
// completes, so the watcher's context drops its
// cancellation. The watcher's lifetime is controlled
// by w.cancel in OnStop.
ctx, cancel := context.WithCancel(
context.WithoutCancel(startCtx),
)
w.cancel = cancel
w.done = make(chan struct{})
go w.Run(ctx) //nolint:contextcheck // intentionally not derived from startCtx
go func() {
defer close(w.done)
w.Run(ctx)
}()
return nil
},
OnStop: func(_ context.Context) error {
if w.cancel != nil {
OnStop: func(ctx context.Context) error {
w.cancel()
}
// Run saves state as it returns. Waiting for it here
// means the save is done before shutdown goes on.
select {
case <-w.done:
return nil
case <-ctx.Done():
return fmt.Errorf(
"waiting for the watcher to stop: %w",
ctx.Err(),
)
}
},
})
return w, nil
}
// Run starts the monitoring loop with periodic scheduling.
// Run starts the monitoring loop with periodic scheduling. When ctx
// is cancelled, it saves state and returns.
func (w *Watcher) Run(ctx context.Context) {
w.log.Info(
"watcher starting",
@@ -125,6 +142,7 @@ func (w *Watcher) Run(ctx context.Context) {
for {
select {
case <-ctx.Done():
w.saveState()
w.log.Info("watcher stopped")
return
+101 -12
View File
@@ -4,6 +4,7 @@ import (
"context"
"fmt"
"log/slog"
"os"
"slices"
"sync"
"testing"
@@ -135,6 +136,7 @@ type testDeps struct {
notifier *mockNotifier
state *state.State
config *config.Config
log *logger.Logger
}
func newTestWatcher(
@@ -143,6 +145,23 @@ func newTestWatcher(
) (*watcher.Watcher, *testDeps) {
t.Helper()
deps := newTestDeps(t, cfg)
w := watcher.NewForTest(
deps.config,
deps.state,
resolver.NewFromLogger(slog.Default()),
deps.portChecker,
deps.tlsChecker,
deps.notifier,
)
return w, deps
}
func newTestDeps(t *testing.T, cfg *config.Config) *testDeps {
t.Helper()
deps := &testDeps{
portChecker: &mockPortChecker{},
tlsChecker: &mockTLSChecker{
@@ -157,30 +176,21 @@ func newTestWatcher(
t.Fatalf("globals.New: %v", err)
}
log, err := logger.New(nil, logger.Params{Globals: g})
deps.log, err = logger.New(nil, logger.Params{Globals: g})
if err != nil {
t.Fatalf("logger.New: %v", err)
}
// The watcher saves state after every check, into cfg.DataDir.
deps.state, err = state.New(fxtest.NewLifecycle(t), state.Params{
Logger: log,
Logger: deps.log,
Config: cfg,
})
if err != nil {
t.Fatalf("state.New: %v", err)
}
w := watcher.NewForTest(
deps.config,
deps.state,
resolver.NewFromLogger(slog.Default()),
deps.portChecker,
deps.tlsChecker,
deps.notifier,
)
return w, deps
return deps
}
func defaultTestConfig(t *testing.T) *config.Config {
@@ -539,6 +549,85 @@ func TestGracefulShutdown(t *testing.T) {
}
}
// TestStopSavesState stops a watcher built by New the way fx stops it,
// and checks that a change made to the state after the last check is in
// the state file afterwards. The state's own stop hook never runs here,
// so only the watcher can have saved it. Nothing is configured to
// check, so no DNS is involved.
func TestStopSavesState(t *testing.T) {
t.Parallel()
cfg := defaultTestConfig(t)
deps := newTestDeps(t, cfg)
lc := fxtest.NewLifecycle(t)
_, err := watcher.New(lc, watcher.Params{
Logger: deps.log,
Config: cfg,
State: deps.state,
Resolver: resolver.NewFromLogger(slog.Default()),
PortCheck: deps.portChecker,
TLSCheck: deps.tlsChecker,
Notify: deps.notifier,
})
if err != nil {
t.Fatalf("watcher.New: %v", err)
}
lc.RequireStart()
// The first check saves state once. Wait for that save before
// changing the state, so the change can reach the file only
// through the save made at stop.
deadline := time.Now().Add(5 * time.Second)
for {
_, err = os.Stat(cfg.StatePath())
if err == nil {
break
}
if time.Now().After(deadline) {
t.Fatalf("the first check saved no state: %v", err)
}
time.Sleep(10 * time.Millisecond)
}
deps.state.SetDomainState(testDomain, &state.DomainState{
Nameservers: []string{oldNS1},
})
ctx, cancel := context.WithTimeout(t.Context(), 5*time.Second)
defer cancel()
err = lc.Stop(ctx)
if err != nil {
t.Fatalf("stopping the watcher: %v", err)
}
saved, err := state.New(fxtest.NewLifecycle(t), state.Params{
Logger: deps.log,
Config: cfg,
})
if err != nil {
t.Fatalf("state.New: %v", err)
}
err = saved.Load()
if err != nil {
t.Fatalf("loading the state file: %v", err)
}
ds, ok := saved.GetDomainState(testDomain)
if !ok || !slices.Equal(ds.Nameservers, []string{oldNS1}) {
t.Errorf(
"state file after stop has %+v for %s, want nameservers %v",
ds, testDomain, []string{oldNS1},
)
}
}
func TestDNSRunsBeforePortAndTLSChecks(t *testing.T) {
t.Parallel()