vulragrag-starandyxxhero d4a4135f9b 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>

2026-09-15 07:35:09 +08:00

..

otel-tracing.md

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

2026-09-15 07:35:09 +08:00

towards-1.0.md

…