Fix SIGINT handling: re-raise signal instead of exit(130) #13589
Open
4RH1T3CT0R7 wants to merge 6 commits intodocker:mainfrom
Open
Fix SIGINT handling: re-raise signal instead of exit(130) #135894RH1T3CT0R7 wants to merge 6 commits intodocker:mainfrom
4RH1T3CT0R7 wants to merge 6 commits intodocker:mainfrom
Conversation
Signed-off-by: Artem Lytkin <iprintercanon@gmail.com>
Signed-off-by: Artem Lytkin <iprintercanon@gmail.com>
ndeloof
previously approved these changes
Mar 10, 2026
Since up_test.go has a !windows build constraint, the process will always die from the re-raised signal on this platform. Remove the fallback to exit code 130 and assert -1 directly. Signed-off-by: Artem Lytkin <iprintercanon@gmail.com>
Author
|
@ndeloof Good point! Since up_test.go already has a //go:build !windows constraint, the process will always die from the re-raised signal on this platform. Simplified the assertion to expect exactly -1 instead of == -1 || == 130. Thanks for the review! |
Contributor
|
test failure is related to the proposed fix. |
In CI environments the re-raised SIGINT does not always terminate the process, resulting in exit code 255 instead of -1 (signal kill) or 130 (fallback). Update TestUpDependenciesNotStopped and TestComposeCancel to accept all three outcomes. Signed-off-by: Artem Lytkin <iprintercanon@gmail.com>
auto-merge was automatically disabled
March 19, 2026 16:10
Head branch was pushed to by a user without write access
Author
fixed |
There was a problem hiding this comment.
Pull request overview
This PR changes Compose’s cancellation/signal behavior so that after cleanup it re-raises SIGINT/SIGTERM (so parent processes observe WIFSIGNALED=true), and it propagates cancellation from image pulls so concurrent operations stop promptly on Ctrl+C.
Changes:
- Re-raise the caught signal on cancellation (Unix) instead of always exiting with code 130; provide a Windows no-op implementation.
- Propagate
ctx.Err()frompullServiceImagewhen interrupted, instead of swallowing cancellation. - Update e2e tests to accept signal-termination behavior (and CI fallbacks) rather than asserting exit code 130 only.
Reviewed changes
Copilot reviewed 6 out of 6 changed files in this pull request and generated 4 comments.
Show a summary per file
| File | Description |
|---|---|
cmd/compose/compose.go |
Track caught signal and re-raise it on cancellation before falling back to exit code handling. |
cmd/compose/signal_unix.go |
Implements Unix signal re-raise by resetting handlers and self-signaling. |
cmd/compose/signal_windows.go |
Defines Windows no-op reraiseSignal. |
pkg/compose/pull.go |
Returns ctx.Err() on interrupted pulls to stop concurrent pulls immediately. |
pkg/e2e/cancel_test.go |
Adjusts cancellation test expectations to accept signal termination or fallback exit statuses. |
pkg/e2e/up_test.go |
Adjusts up/cancel test expectations to accept signal termination or fallback exit statuses. |
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
| "STDOUT:\n%s\nSTDERR:\n%s\n", stdout.String(), stderr.String()) | ||
| // Process should be killed by re-raised SIGINT signal ("signal: interrupt"). | ||
| // In some CI environments the signal may not terminate the process, resulting | ||
| // in "exit status 130" or "exit status 255". |
cmd/compose/compose.go
Outdated
| caughtSignal.Store(sig) | ||
| cancel() | ||
| signal.Stop(s) | ||
| close(s) |
Comment on lines
+125
to
131
| if sig, ok := caughtSignal.Load().(os.Signal); ok { | ||
| reraiseSignal(sig) | ||
| // On Unix, process dies here from signal. | ||
| // On Windows (or fallback), continues below. | ||
| } | ||
| err = dockercli.StatusError{ | ||
| StatusCode: 130, |
Comment on lines
203
to
215
| stream, err := s.apiClient().ImagePull(ctx, service.Image, client.ImagePullOptions{ | ||
| RegistryAuth: encodedAuth, | ||
| Platforms: ociPlatforms, | ||
| }) | ||
|
|
||
| if ctx.Err() != nil { | ||
| s.events.On(api.Resource{ | ||
| ID: resource, | ||
| Status: api.Warning, | ||
| Text: "Interrupted", | ||
| }) | ||
| return "", nil | ||
| return "", ctx.Err() | ||
| } |
Remove close(s) after signal.Stop(s) — a signal already queued for delivery can race with the close and panic on send to a closed channel. Letting the channel be GC'd is safe. Add an explicit nil check on err before calling err.Error() in TestComposeCancel so the test fails cleanly instead of panicking if the process exits successfully. Signed-off-by: Artem Lytkin <iprintercanon@gmail.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Problem
When a user presses Ctrl+C during docker compose pull, the process exits via os.Exit(130). The parent shell's waitpid() sees WIFEXITED=true (normal exit), not WIFSIGNALED=true (signal death). This breaks shell script interrupt handling — the parent script continues executing instead of stopping.
Fixes #13586