compose/pkg
Nicolas De Loof 023544aae7
Some checks are pending
ci / validate (lint) (push) Waiting to run
ci / validate (relay-lint) (push) Waiting to run
ci / validate (relay-test) (push) Waiting to run
ci / validate (validate-docs) (push) Waiting to run
ci / validate (validate-go-mod) (push) Waiting to run
ci / validate (validate-headers) (push) Waiting to run
ci / validate (validate-mocks) (push) Waiting to run
ci / binary (push) Waiting to run
ci / binary-finalize (push) Blocked by required conditions
ci / bin-image-test (push) Waiting to run
ci / relay-image-test (push) Waiting to run
ci / test (push) Waiting to run
ci / e2e (plugin, oldstable, graphdriver) (push) Waiting to run
ci / e2e (standalone, oldstable, graphdriver) (push) Waiting to run
ci / e2e (plugin, stable, containerd) (push) Waiting to run
ci / e2e (plugin, stable, graphdriver) (push) Waiting to run
ci / e2e (standalone, stable, containerd) (push) Waiting to run
ci / e2e (standalone, stable, graphdriver) (push) Waiting to run
ci / coverage (push) Blocked by required conditions
ci / release (push) Blocked by required conditions
merge / bin-image-prepare (push) Waiting to run
merge / bin-image (push) Blocked by required conditions
merge / relay-image (push) Waiting to run
merge / module-image (push) Waiting to run
Scorecards supply-chain security / Scorecards analysis (push) Waiting to run
zizmor / zizmor (push) Waiting to run
fix(up): catch a late starter on a graceful teardown too, not just on-exit
glours' review on this PR (pullrequestreview-5277692679): stopLateStarter
only ever got armed inside stopOnFirstExit's own closure, so it caught a
service racing the on-exit cascade's sweep but not one racing
gracefulTeardown's (Ctrl+C/SIGTERM) identical one-shot stopApplication
sweep -- same root cause, same file, just the other trigger. Worse for
that path: Ctrl+C can land at t~=0, before monitor.Start even starts and
while s.start() (its own uncancelable context) is still climbing the
dependency graph, so most services may still be created, not running.
The only recovery today is a second manual Ctrl+C.

u.isTerminated already flags "termination is underway" and was already
set by gracefulTeardown (just not consulted for late-starter catching,
and set after stopApplication rather than before it). Extract the
late-starter watch out of stopOnFirstExit into its own listener,
stopLateStarters, armed on u.isTerminated regardless of which path set
it, and register it unconditionally (Ctrl+C works regardless of the
on-exit policy, unlike stopOnFirstExit's own listener). stopOnFirstExit
now sets isTerminated itself before its sweep, symmetric with
gracefulTeardown (also reordered to set it before stopApplication, not
after).

isLateStarter is extracted as a pure predicate, following the existing
shouldFollowStartEvent precedent, and unit-tested (TestIsLateStarter):
the previous fix (ca867482a) shipped only against a flaky test hitting
the on-exit case by chance, with no dedicated regression test at all.

Signed-off-by: Nicolas De Loof <nicolas.deloof@gmail.com>
2026-09-22 14:45:23 +02:00
..
api build(deps): adopt compose-go jobs and the container-spec layering 2026-09-21 16:35:20 +02:00
bridge build(deps): adopt compose-go jobs and the container-spec layering 2026-09-21 16:35:20 +02:00
compose fix(up): catch a late starter on a graceful teardown too, not just on-exit 2026-09-22 14:45:23 +02:00
dryrun fix(dry-run): commit no longer creates a real image; the interception set becomes declarative 2026-09-16 13:44:21 +02:00
e2e test(watch): budget the watcher's cold start beyond the default poll timeout 2026-09-22 14:45:23 +02:00
mocks PRIVATE_PORT optional for docker compose port 2026-09-11 11:35:18 +02:00
remote golangci-lint: enable perfsprint linter 2026-09-10 10:25:34 +02:00
utils fix: make e2e tests pass reliably locally with Docker Desktop 2026-05-18 10:55:41 +02:00
watch golangci-lint: enable perfsprint linter 2026-09-10 10:25:34 +02:00