kolaente 0a6c0d0b37 fix(e2etests): drain pending event handlers before seeding fixtures ...

setupE2ETestEnv seeded fixtures without waiting for the previous test's
event handlers. Cancelling the test context does not stop them:
sendWebhookPayload builds its request with context.Background(), so an
in-flight webhook delivery keeps running — and keeps holding the DB
session opened by WebhookDeliveryListener.Handle — after the router shuts
down. The next test's db.LoadFixtures() then raced that session and failed
with "testfixtures: could not clean table \"webhooks\": database table is
locked: webhooks", most often in TestWebhookFailingSiblingDoesNotBlockOthers
and TestUserWebhookTasksOverdueBatchE2E.

Call events.WaitForPendingHandlers() first, the same drain the e2e testing
endpoint already does in ReplaceTableContents and TruncateAllTestingTables.

The regression test blocks a webhook target for 500ms, cancels the context
while the delivery is in flight, and asserts the next setup waits it out.
Without the drain it reproduces the CI error in ~0.1s.

2026-08-27 13:36:08 +02:00

..

integrations.go

…

main_test.go

…

setup_test.go

…

user_webhook_test.go

…

webhook_multi_test.go

…

webhook_test.go

…