fix: cancel kubedog-tracked helm subprocesses on SIGINT/SIGTERM (#2791)

* fix: cancel kubedog-tracked helm subprocesses on SIGINT/SIGTERM

Kubedog tracking rooted helm and tracker contexts at Background (via
traceOnlyContext), so App.Cancel never reached those subprocesses and
Ctrl+C blocked in CleanWaitGroup until helm exited on its own.

Thread the app cancel context into HelmState (SetCancelContext) and use
it for the three kubedog tracking call sites, keeping the per-release
WithCancel safety valve. Hooks stay on the non-canceling trace bridge.

Fixes #2770

Signed-off-by: Jason Wang <vulragrag@gmail.com>

* docs/test: align cancel-context docs with #2791 and pin the kubedog wiring

- traceOnlyContext comment no longer lists kubedog tracking among the
  detached paths; only hooks (#2771) remain, with a pointer to #2791.
- docs/proposals/otel-tracing.md sections 4.3/4.4/7/9 now mark the
  kubedog cancellation gap as fixed by #2791 (SetCancelContext), so the
  design doc stops misdescribing main.
- TestBufferHelmOutputRootsAtCancelContext pins the actual #2770 wiring:
  the context bufferHelmOutput hands to ContextSwapper.WithContext must
  become Done when the app cancel context is canceled, and stay
  non-cancelable when unset.

Signed-off-by: yxxhero <aiopsclub@163.com>

---------

Signed-off-by: Jason Wang <vulragrag@gmail.com>
Signed-off-by: yxxhero <aiopsclub@163.com>
Co-authored-by: yxxhero <aiopsclub@163.com>
This commit is contained in:
vulragrag-star
2026-09-15 07:35:09 +08:00
committed by GitHub
co-authored by yxxhero
parent fff84390b1
commit d4a4135f9b
5 changed files with 113 additions and 16 deletions
+14 -9
View File
@@ -233,7 +233,7 @@ All claims below were checked against the code; line numbers are anchors for rev
|---|---|---|
| cmd → app | `app.New` roots at `context.Background()`; `ctx, Cancel = WithCancel(ctx)` (`pkg/app/app.go`, `New`) | span must be injected here (§4.2) |
| app → all helm execs | `getHelm()` constructs the `ShellRunner` with `Ctx: a.ctx` (`pkg/app/app.go:1008`); the resulting `execer` is **cached per (helm binary, kube-context)** in `a.helms` and shared by all releases and workers (`pkg/app/app.go:982–1021`) | once `App.ctx` is span-rooted, every non-kubedog helm call nests automatically, with **zero changes** to `getHelm`. The shared-instance cache is also why per-release contexts must ride per-call parameters, never mutation of the shared execer |
| kubedog path (sync with tracking) | `startBackgroundKubedogTracking(gocontext.Background(), …)` (`pkg/state/state.go:1294`) → `bufferHelmOutput` derives `releaseCtx := context.WithCancel(ctx)` and swaps it in via `execer.WithContext(releaseCtx)` (`pkg/state/helmx.go:363–365`) | helm execs on this path run on a **Background-rooted** context; runner-level spans would become **orphan traces**. Bridged in §4.4 |
| kubedog path (sync with tracking) | originally `startBackgroundKubedogTracking(gocontext.Background(), …)` (`pkg/state/state.go`), with `bufferHelmOutput` deriving `releaseCtx := context.WithCancel(ctx)` and swapping it in via `execer.WithContext(releaseCtx)` (`pkg/state/helmx.go`). **Since #2791** the three call sites pass `st.releaseCancelContext()` — the app cancel context injected via `HelmState.SetCancelContext` — so SIGINT/SIGTERM reaches these helm subprocesses too (#2770) | trace context on this path is the same command span as everywhere else: the §4.4 bridge kept spans attached while cancellation was detached, and #2791 subsequently re-attached cancellation |
| hooks | both `event.Bus` constructions (`triggerGlobalReleaseEvent`, `triggerReleaseEvent`, `pkg/state/state.go:3666, 3703`) duplicate the same literal and pass **no** `Runner`, so the default kicks in: `ShellRunner{Dir: bus.BasePath, Logger: bus.Logger, Ctx: goContext.TODO()}` with an inline comment acknowledging it should be `app.Ctx` (`pkg/event/bus.go:61–71`) | hook execs are detached; spans would be orphans. Bridged in §4.4 |
| non-kubedog release workers | release loops (`SyncReleases` etc., `pkg/state/state.go:1212 ff.`) call the shared `helmexec.Interface` with a `HelmContext` (`pkg/helmexec/context.go`) that carries **no go-context** | per-release spans need the §4.4 mechanism |
| subprocess funnel | exactly three `ShellRunner` construction sites exist (verified exhaustive): `pkg/app/app.go:129` (`Init`, `Ctx: a.ctx`), `pkg/app/app.go:1006` (`getHelm`, `Ctx: a.ctx`), and the hooks default (`pkg/event/bus.go:62`, `Ctx: TODO` — §4.4 bridge). Every external process helmfile itself starts goes through `Execute`/`ExecuteStdIn` (`pkg/helmexec/runner.go`); helm commands additionally funnel through `execer.exec()` (`pkg/helmexec/exec.go:1207`). Exception: kustomize executes inside the chartify library, outside this funnel (§12) | one instrumentation point covers everything except chartify-internal execs; spans nest wherever the runner's `Ctx` carries a span |
@@ -241,7 +241,8 @@ All claims below were checked against the code; line numbers are anchors for rev
Two pre-existing gaps surfaced by this analysis — kubedog tracking not being cancellable via
`App.ctx`, and hooks likewise — are **out of scope** for this proposal beyond trace-context
bridging (§4.4), because fixing their *cancellation* semantics would be a behavior change.
They should be reported as separate issues.
They should be reported as separate issues. (Update: the kubedog gap was fixed by
#2791 via `HelmState.SetCancelContext`; the hooks gap remains open as #2771.)
### 4.4 Context plan
@@ -258,12 +259,15 @@ They should be reported as separate issues.
2. Runner-level spans in `ShellRunner.Execute`/`ExecuteStdIn` nest for all non-kubedog,
non-hook execs automatically.
3. **Orphan-bridge for kubedog and hooks, with identical cancellation semantics:**
- `pkg/state/state.go:1294`: pass `context.WithoutCancel(telemetry.CommandContext())`
instead of `gocontext.Background()`. `WithoutCancel` preserves values (the span)
- kubedog: originally pass `context.WithoutCancel(telemetry.CommandContext())`
instead of `gocontext.Background()`. `WithoutCancel` preserved values (the span)
while dropping cancellation — and `Background` never carried cancellation anyway, so
SIGINT/timeout behavior is **bit-for-bit unchanged**; only trace context is added.
(Phase 1 uses the root span from `telemetry.CommandContext()`, which requires no new
plumbing in `pkg/state`; phase 2 re-parents under the per-release `st.traceCtx`.)
SIGINT/timeout behavior stayed **bit-for-bit unchanged**; only trace context was
added. (Phase 1 used the root span from `telemetry.CommandContext()`, which required
no new plumbing in `pkg/state`; phase 2 re-parented under the per-release
`st.traceCtx`.) Superseded by #2791: the three kubedog call sites now pass
`st.releaseCancelContext()` — same command-span trace context, plus app
cancellation, fixing #2770.
- `pkg/event/bus.go`: add an optional `Ctx context.Context` field to `Bus`; the default
runner construction uses `bus.Ctx` when set, `TODO` when nil (so behavior is unchanged
for any nil-Ctx caller). The two construction sites in `pkg/state/state.go:3666, 3703`
@@ -414,7 +418,8 @@ Traces leave the machine they run on. Ground rules, checked against what exists
churn; cancellation semantics identical.
3. Kubedog and hook bridging uses `context.WithoutCancel`, which drops cancellation and
keeps values — the swapped-out parents (`Background`/`TODO`) never propagated
cancellation either, so SIGINT/timeout behavior is unchanged.
cancellation either, so SIGINT/timeout behavior is unchanged. (Superseded for kubedog
by #2791, which re-roots tracking under the app cancel context; hooks stay bridged.)
4. Telemetry setup or export failures never fail or slow the run (warning log only, export
off the critical path).
5. All pre-existing behavior, including the known cancellation gaps of §4.3 and the exact
@@ -455,7 +460,7 @@ pkg/helmexec/redact.go // shared args redaction, legacy+strict profile
pkg/helmexec/exit_error.go // calls shared helper with legacy profile (output byte-identical)
pkg/helmexec/context.go // HelmContext.Ctx field (phase 2)
pkg/helmexec/exec.go // execCtx funnel beside exec/execStdIn (phase 2)
pkg/state/state.go // WithoutCancel bridges (kubedog call site + both event.Bus constructions); release spans (phase 2)
pkg/state/state.go // WithoutCancel bridges (both event.Bus constructions; the kubedog call sites moved to st.releaseCancelContext() in #2791); release spans (phase 2)
pkg/state/helmx.go // (no change — bridge happens at its caller)
pkg/event/bus.go // optional Ctx field consumed by the default runner
docs/experimental-features.md // feature entry → promoted out when stable
+2
View File
@@ -975,6 +975,8 @@ func (a *App) loadDesiredStateFromYamlWithBaseDir(file string, baseDir string, o
// Per-release spans (pkg/state) parent under the load span.
st.SetTraceContext(loadCtx)
// Kubedog tracking / buffered helm subprocesses cancel with the app.
st.SetCancelContext(a.ctx)
st.SetKubeconfig(a.Kubeconfig)
+24 -4
View File
@@ -15,10 +15,12 @@ import (
// traceOnlyContext returns a context carrying trace context — the command
// span, or the given parent (typically a per-release span) — while never
// propagating cancellation. The historically detached paths (kubedog
// tracking, hook execution) must keep their cancellation semantics, so only
// trace context is bridged (see docs/proposals/otel-tracing.md §4.4). With
// tracing disabled it is indistinguishable from Background.
// propagating cancellation. Hook execution must keep its detached
// cancellation semantics (#2771), so only trace context is bridged (see
// docs/proposals/otel-tracing.md §4.4). Kubedog tracking formerly rooted
// here too; it now derives from the app cancel context so SIGINT reaches
// the buffered helm subprocesses (#2770, #2791). With tracing disabled it
// is indistinguishable from Background.
func traceOnlyContext(parent ...gocontext.Context) gocontext.Context {
ctx := telemetry.CommandContext()
if len(parent) > 0 && parent[0] != nil {
@@ -35,6 +37,14 @@ func (st *HelmState) SetTraceContext(ctx gocontext.Context) {
st.traceCtx = ctx
}
// SetCancelContext sets the context canceled with the app on SIGINT/SIGTERM.
// Kubedog tracking (and the helm subprocesses it buffers) derive from this so
// process-wide cancellation reaches them. A nil context keeps the historical
// Background-rooted behavior.
func (st *HelmState) SetCancelContext(ctx gocontext.Context) {
st.cancelCtx = ctx
}
func (st *HelmState) releaseSpanParent() gocontext.Context {
if st.traceCtx != nil {
return st.traceCtx
@@ -42,6 +52,16 @@ func (st *HelmState) releaseSpanParent() gocontext.Context {
return gocontext.Background()
}
// releaseCancelContext returns the context kubedog tracking should root under.
// When unset it falls back to Background, matching the pre-#2770 detached
// semantics used by tests that construct HelmState literals without an App.
func (st *HelmState) releaseCancelContext() gocontext.Context {
if st.cancelCtx != nil {
return st.cancelCtx
}
return gocontext.Background()
}
// startReleaseSpan starts one helmfile.release.<verb> span for a release,
// parented from the state's trace context. The returned context is meant to
// be stamped into the per-release helmexec.HelmContext so the release's helm
+65
View File
@@ -229,6 +229,71 @@ func TestSetTraceContext(t *testing.T) {
assert.Equal(t, ctx, st.releaseSpanParent())
}
// TestSetCancelContext pins #2770: kubedog tracking roots under the app
// cancel context so SIGINT reaches buffered helm subprocesses. Unset keeps
// the historical Background fallback (detached / non-app-cancelable).
func TestSetCancelContext(t *testing.T) {
st := &HelmState{}
assert.Equal(t, gocontext.Background(), st.releaseCancelContext(), "unset cancel context must fall back to Background")
ctx, cancel := gocontext.WithCancel(gocontext.Background())
defer cancel()
st.SetCancelContext(ctx)
got := st.releaseCancelContext()
assert.Equal(t, ctx, got)
cancel()
select {
case <-got.Done():
default:
t.Fatal("releaseCancelContext must become Done when the app cancel context is canceled")
}
}
// ctxRecordingHelm captures the context handed to it via ContextSwapper so
// tests can assert which context roots the buffered helm subprocesses. The
// embedded nil Interface panics on any other method call, which
// bufferHelmOutput never makes.
type ctxRecordingHelm struct {
helmexec.Interface
ctx gocontext.Context
}
func (h *ctxRecordingHelm) WithLogger(_ *zap.SugaredLogger) helmexec.Interface { return h }
func (h *ctxRecordingHelm) WithContext(ctx gocontext.Context) helmexec.Interface {
h.ctx = ctx
return h
}
// TestBufferHelmOutputRootsAtCancelContext pins the actual #2770 wiring at
// the unit level: the release-scoped context that bufferHelmOutput swaps in
// via WithContext must derive from the app cancel context (so the runner's
// ctx.Done() SIGINT path fires), and stay non-cancelable when no cancel
// context is set.
func TestBufferHelmOutputRootsAtCancelContext(t *testing.T) {
appCtx, cancel := gocontext.WithCancel(gocontext.Background())
defer cancel()
tracked := &HelmState{logger: zap.NewNop().Sugar()}
tracked.SetCancelContext(appCtx)
trackedHelm := &ctxRecordingHelm{}
tracked.bufferHelmOutput(trackedHelm, tracked.releaseCancelContext(), "demo")
require.NotNil(t, trackedHelm.ctx, "bufferHelmOutput must swap in a release-scoped context via WithContext")
cancel()
select {
case <-trackedHelm.ctx.Done():
default:
t.Fatal("the context handed to WithContext must become Done when the app cancel context is canceled")
}
detached := &HelmState{logger: zap.NewNop().Sugar()}
detachedHelm := &ctxRecordingHelm{}
detached.bufferHelmOutput(detachedHelm, detached.releaseCancelContext(), "demo")
assert.NoError(t, detachedHelm.ctx.Err(), "unset cancel context must keep the release context detached (Background-rooted)")
}
func TestSkipUndesired(t *testing.T) {
assert.False(t, skipUndesired(&ReleaseSpec{}), "installed unset means desired")
+8 -3
View File
@@ -148,6 +148,11 @@ type HelmState struct {
// traceCtx parents per-release spans; set via SetTraceContext (see span.go).
traceCtx gocontext.Context
// cancelCtx is canceled with the app on SIGINT/SIGTERM. Kubedog tracking
// and its buffered helm subprocesses derive from it so Ctrl+C reaches them
// (see issue #2770). Set via SetCancelContext; nil falls back to Background.
cancelCtx gocontext.Context
ReleaseSetSpec `yaml:",inline"`
logger *zap.SugaredLogger
@@ -1312,10 +1317,10 @@ func (st *HelmState) SyncReleases(affectedReleases *AffectedReleases, helm helme
} else if release.UpdateStrategy == UpdateStrategyReinstallIfForbidden {
relErr = st.performSyncOrReinstallOfRelease(affectedReleases, helm, context, release, chart, m, flags...)
if relErr == nil {
relErr = st.trackReleaseIfEnabled(traceOnlyContext(), release, helm, opts)
relErr = st.trackReleaseIfEnabled(st.releaseCancelContext(), release, helm, opts)
}
} else {
trackHandle, trackStarted := st.startBackgroundKubedogTracking(traceOnlyContext(), release, helm, opts)
trackHandle, trackStarted := st.startBackgroundKubedogTracking(st.releaseCancelContext(), release, helm, opts)
// trackHandle.Helm is a logger-scoped helm clone that
// captures output to an in-memory buffer while tracking is
// active. When tracking isn't running it's the original
@@ -1368,7 +1373,7 @@ func (st *HelmState) SyncReleases(affectedReleases *AffectedReleases, helm helme
if trackStarted {
trackErr = trackHandle.Wait()
} else {
trackErr = st.trackReleaseIfEnabled(traceOnlyContext(), release, helm, opts)
trackErr = st.trackReleaseIfEnabled(st.releaseCancelContext(), release, helm, opts)
}
if trackErr != nil {
m.Lock()