Reproduce flaky: shutdown 'executes only once'

[p-datadog]

What does this PR do?

Reproducer for flaky test spec/datadog/tracing/integration_spec.rb:592 — the shutdown "executes only once" integration test. Forces the suspected race condition to validate the hypothesis in CI. Expected result: the test fails deterministically.

Do not merge — this PR exists for CI validation only.

Motivation:

Flaky test reported in PR #5581. The test intermittently fails with traces_flushed: 0 under CI load.

Hypothesis: The test fails when the HTTP round-trip to the agent takes longer than DEFAULT_SHUTDOWN_TIMEOUT (1 second). The worker thread is mid-HTTP-call when shutdown! calls join(1), the join times out, and traces_flushed is still 0 when the assertion runs.

Reproducer technique: Stubs tracer.writer.transport.send_traces with and_wrap_original to add a 2-second sleep before the real HTTP call. This exceeds the 1-second shutdown timeout, forcing the join to time out deterministically.

Change log entry

None.

How to test the change?

CI should show the "executes only once" / "flushed trace" test failing deterministically with traces_flushed: 0. If it doesn't, the hypothesis is wrong.

[datadog-datadog-prod-us1]

⚠️ Tests

✨ Fix all issues with BitsAI or with Cursor

⚠️ Other Violations

🧪 2 Tests failed

Tracer integration tests shutdown executes only once behaves like flushed trace is expected to include {:traces_flushed => 1} from rspec     (Fix with Cursor)

expected {:traces_flushed => 0, :transport => #<Datadog::Tracing::Transport::Statistics::Counts:0x00007ffbccdcee78 @success=0, @client_error=0, @server_error=0, @internal_error=0, @consecutive_errors=0>} to include {:traces_flushed => 1}
Diff:
@@ -1 +1,2 @@
-:traces_flushed => 1,
+:traces_flushed => 0,
+:transport => #<Datadog::Tracing::Transport::Statistics::Counts:0x00007ffbccdcee78 @success=0, @client_error=0, @server_error=0, @internal_error=0, @consecutive_errors=0>,

Failure/Error: expect(stats).to include(traces_flushed: 1)

  expected {:traces_flushed => 0, :transport => #<Datadog::Tracing::Transport::Statistics::Counts:0x00007ffbccdcee78 @success=0, @client_error=0, @server_error=0, @internal_error=0, @consecutive_errors=0>} to include {:traces_flushed => 1}
...

Tracer integration tests shutdown executes only once behaves like flushed trace is expected to include {traces_flushed: 1} from rspec     (Fix with Cursor)

expected {traces_flushed: 0, transport: #<Datadog::Tracing::Transport::Statistics::Counts:0x00007ff59f2cdf70 @success=0, @client_error=0, @server_error=0, @internal_error=0, @consecutive_errors=0>} to include {traces_flushed: 1}
Diff:
@@ -1 +1,2 @@
-:traces_flushed => 1,
+:traces_flushed => 0,
+:transport => #<Datadog::Tracing::Transport::Statistics::Counts:0x00007ff59f2cdf70 @success=0, @client_error=0, @server_error=0, @internal_error=0, @consecutive_errors=0>,

Failure/Error: expect(stats).to include(traces_flushed: 1)

  expected {traces_flushed: 0, transport: #<Datadog::Tracing::Transport::Statistics::Counts:0x00007ff59f2cdf70 @success=0, @client_error=0, @server_error=0, @internal_error=0, @consecutive_errors=0>} to include {traces_flushed: 1}
...

ℹ️ Info

No other issues found (see more)

❄️ No new flaky tests detected

_{This comment will be updated automatically if new data arrives.
🔗 Commit SHA: c69ca6e | Docs | Datadog PR Page | Was this helpful? React with 👍/👎 or give us feedback!}