Skip to content

fix(ci): silence sidecar flush-timeout log noise in client_side_stats test#3949

Draft
Leiyks wants to merge 2 commits into
masterfrom
leiyks/fix-ci-crashtracker-threads
Draft

fix(ci): silence sidecar flush-timeout log noise in client_side_stats test#3949
Leiyks wants to merge 2 commits into
masterfrom
leiyks/fix-ci-crashtracker-threads

Conversation

@Leiyks
Copy link
Copy Markdown
Contributor

@Leiyks Leiyks commented Jun 3, 2026

Under concurrent CI load, tests/ext/request-replayer/client_side_stats.phpt intermittently failed on a stray [ddtrace] [error] Failed synchronously flushing traces: Kind(TimedOut) line from the best-effort 5s synchronous flush — the stats themselves still arrive via the polled async sidecar path. Add DD_TRACE_LOG_LEVEL=off, matching its sibling tests client_side_stats_peer_tags / client_side_stats_trace_filters. Verified 60/60 under load in docker.

Note: the crashtracker thread-format fix originally here is now redundant — landed on master via #3953 — so this branch was rebased to contain only the client_side_stats fix.

@datadog-official
Copy link
Copy Markdown

datadog-official Bot commented Jun 3, 2026

Pipelines  Tests

Fix all issues with BitsAI

⚠️ Warnings

🚦 6 Pipeline jobs failed

DataDog/apm-reliability/dd-trace-php | ASAN test_c with multiple observers: [8.5]   View in Datadog   GitLab

DataDog/apm-reliability/dd-trace-php | benchmarks-tracer   View in Datadog   GitLab

DataDog/apm-reliability/dd-trace-php | min install tests   View in Datadog   GitLab

View all 6 failed jobs.

ℹ️ Info

No other issues found (see more)

🧪 All tests passed
❄️ No new flaky tests detected

🔄 Datadog auto-retried 1 job - 1 passed on retry View in Datadog

🎯 Code Coverage (details)
Patch Coverage: 100.00%
Overall Coverage: 54.15% (+0.03%)

Useful? React with 👍 / 👎

This comment will be updated automatically if new data arrives.
🔗 Commit SHA: ba019ca | Docs | Datadog PR Page | Give us feedback!

… test [test_extension_ci]

Under concurrent run-tests load the best-effort dd_trace_synchronous_flush()
can exceed its 5s window, emitting '[ddtrace] [error] Failed synchronously
flushing traces: Kind(TimedOut)' on stderr. The stats still arrive via the
async sidecar path that the test polls with waitForStats(), so the expected
output is produced, but the stray error line breaks the output diff.

Set DD_TRACE_LOG_LEVEL=off to match the sibling tests
(client_side_stats_peer_tags, client_side_stats_trace_filters), which already
suppress this benign log noise.
@Leiyks Leiyks force-pushed the leiyks/fix-ci-crashtracker-threads branch from d340f37 to 6da4162 Compare June 5, 2026 12:49
@Leiyks Leiyks changed the title fix(ci): read flattened error.threads list in crashtracker collect-all-threads test fix(ci): silence sidecar flush-timeout log noise in client_side_stats test Jun 5, 2026
…flush timing [test_extension_ci]

Under concurrent run-tests -j4 load two background-sender/sidecar tests
flaked because best-effort flushing to the shared request-replayer can be
delayed beyond their wait windows:

- agent_headers_container_id_fargate.phpt threw 'wait for replay timeout':
  its DD_TRACE_AGENT_FLUSH_INTERVAL=333 gives a ~16s replay-poll window
  (maxIteration * flushInterval), half that of the reliably-passing sibling
  agent_headers_container_id.phpt (666). Bump to 666 to match the sibling and
  double the window the BGS delivery can use under load.

- shm_data_internal_fns.phpt got 'Failed synchronously flushing traces:
  Kind(TimedOut)' stderr noise and a missing agent sampling config
  (bool(false)/float(-1)) because the best-effort synchronous_flush timed out,
  delaying the trace that makes the replayer return rate_by_service. Set
  DD_TRACE_LOG_LEVEL=off (same convention as client_side_stats) to silence the
  benign log line, and re-trigger the flush periodically while polling 100x100ms
  so the sampling config reliably propagates into SHM.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant