Skip to content

Bisection build: 4.9.1 + writer stop_service join timeout (jointimeout1) #18896

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Closed
juanjux wants to merge 3 commits into from
Closed

Bisection build: 4.9.1 + writer stop_service join timeout (jointimeout1) #18896

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
eb1c13e
Add default timeout for join
VianneyRuhlmann Jul 7, 2026
ce42adf
catch error
VianneyRuhlmann Jul 7, 2026
3d98347
fix: correct syntax error + mark bisection build 4.9.1+jointimeout1
juanjux Jul 7, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
7 changes: 5 additions & 2 deletions ddtrace/internal/writer/writer.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -503,11 +503,14 @@ def periodic(self):

def _stop_service(
self,
timeout: Optional[float] = None,
timeout: Optional[float] = 0.1,

@chatgpt-codex-connector chatgpt-codex-connector Bot Jul 7, 2026

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

P2 Badge Preserve the blocking default for writer shutdown

When HTTPWriter.stop() is called without an explicit timeout, this new default makes shutdown wait only 100ms even though the public shutdown contract treats None as “block until flushing has successfully completed.” PeriodicThread.join(timeout) returns silently on timeout, so for slow agentless/HTTP flushes (for example direct AgentlessTraceWriter.stop() or AgentlessTraceWriter.recreate(), which calls self.stop() with no timeout) stop() can mark the service stopped and return while on_shutdown() is still draining or before the process exits, causing queued traces to be dropped or the old writer to run concurrently with the replacement.

Useful? React with 👍 / 👎.

@chatgpt-codex-connector chatgpt-codex-connector Bot Jul 7, 2026

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

P2 Badge Pass a timeout from tracer shutdown callers

The main tracer shutdown path still bypasses this default: SpanAggregator.shutdown calls self.writer.stop(timeout) even when the default timeout is None, so HTTPWriter._stop_service receives None and performs the same unbounded join for normal tracer.shutdown()/atexit shutdowns. If this build is meant to test or fix slow writer shutdown through tracer shutdown, the caller needs to supply the 0.1s value or normalize None here; otherwise only direct no-argument writer.stop() calls exercise the new timeout.

Useful? React with 👍 / 👎.

) -> None:
# FIXME: don't join() on stop(), let the caller handle this
super(HTTPWriter, self)._stop_service()
self.join(timeout=timeout)
try:
self.join(timeout=timeout)

@chatgpt-codex-connector chatgpt-codex-connector Bot Jul 7, 2026

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

P2 Badge Apply the timeout to the actual fork join path

For the fork-timeout bisection, this only bounds the join reached through Service.stop()/_stop_service; the fork hook I checked in ddtrace/internal/threads.py still stops periodic threads with thread._before_fork() and then calls thread.join() with no timeout. In processes that hang while forking with a writer periodic thread active, this build will still block on that unbounded fork join and can incorrectly rule out the writer join hypothesis.

Useful? React with 👍 / 👎.

except Exception:
log.error("Join on periodic thread failed", exc_info=True)

def on_shutdown(self):
try:
Expand Down
2 changes: 1 addition & 1 deletion pyproject.toml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -9,7 +9,7 @@ build-backend = "setuptools.build_meta"

[project]
name = "ddtrace"
version = "4.9.1"
version = "4.9.1+jointimeout1"

@chatgpt-codex-connector chatgpt-codex-connector Bot Jul 7, 2026

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

P1 Badge Keep package version compatible with release tags

If this lands on a releasable branch/tag, the local-version marker cannot pass the existing release/tag gates: .gitlab-ci.yml only treats tags matching vX.Y.Z or vX.Y.ZrcN as releases, while the package version job in .gitlab/package.yml requires CI_COMMIT_TAG to equal v${PACKAGE_VERSION}. With this value, v4.9.1 fails the package-version check and v4.9.1+jointimeout1 does not match the release regex, so release CI cannot proceed from this commit.

Useful? React with 👍 / 👎.

description = "Datadog APM client library"
readme = "README.md"
license = { text = "LICENSE.BSD3" }
Expand Down
Loading