fix: tell a stopped run apart from one the service killed

[ogabrielluiz]

Stacked on #14403 and merges after it. The diff below is against that branch, not release-1.12.0, and green CI here only proves this change against its parent.

What

asyncio delivers two very different events as the same CancelledError, and the flow span reported both the same way:

• a user pressing stop
• a server-imposed execution ceiling (asyncio.wait_for around the v2 build driver, a2a, or the agentic assistant), a worker shutdown, a parent task being torn down

Both came out as status="cancelled" with OTel span status UNSET. So every timeout stayed out of error-rate alerting, while the client was served a terminal error and the job row was written FAILED. The span was the only place in the system claiming nothing had gone wrong.

They are separate outcomes now:

Ending	status	Span status
User pressed stop	cancelled	UNSET
Ceiling, shutdown, teardown	aborted	ERROR, error.type=CancelledError

A withdrawn request is genuinely not a service fault, so the stop button keeps its UNSET. Everything else now agrees with what the client and the job row already say.

One constant instead of five strings

The discriminator already existed: the producers stamp a marker on the exception args and the job service reads it to choose CANCELLED over FAILED. It was a bare "LANGFLOW_USER_CANCELLED" repeated at five sites, which is why the span could not have trusted it. It is one constant now, imported by both the producers and the span.

It lives in lfx.constants, not next to the graph exceptions where it more naturally belongs. Importing lfx.graph.exceptions from a langflow service module runs lfx/graph/__init__.py, which imports Graph from a half-initialized base.py, and from langflow.main import create_app then fails outright. lfx.constants imports nothing but pathlib. I hit that while writing this and would rather leave the reason here than have someone move it back.

Verification

New test drives a real asyncio.wait_for timeout rather than hand-raising CancelledError, so it exercises the way the ceiling actually cancels a run. The user-stop test raises with the marker the producers stamp.

Collapsing the split back to one value turns the timeout test red with 'cancelled' == 'aborted'.

Suites: 12 flow-span tests, 1763 in services/, 145 v2 background/cancellation tests, plus an explicit from langflow.main import create_app check for the import cycle above.

Known gaps

aborted still covers a few distinguishable things (ceiling, shutdown, teardown). They all mean "the service did not deliver this run", which is the distinction that matters for alerting, so I did not split further.

[coderabbitai]

Important

Review skipped

Auto reviews are disabled on base/target branches other than the default branch.

🗂️ Base branches to auto review (1)

• release-.*

Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro Plus

Run ID: c4b234c9-f19a-4adb-8914-9d1797b0a7d0

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

• 🔍 Trigger review

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}

[github-actions]

✅ Test Coverage Advisor

No source changes detected without accompanying tests. Thanks for keeping coverage up! 🎉

Advisory check only — never blocks merge.

[github-actions]

✅ Migration Validation Passed

All migrations follow the Expand-Contract pattern correctly.

[github-actions]

Frontend Unit Test Coverage Report

Coverage Summary

Lines	Statements	Branches	Functions
	49.57% (71705/144626)	70.35% (10042/14274)	46.77% (1644/3515)

Unit Test Results

Tests	Skipped	Failures	Errors	Time
5529	0 💤	0 ❌	0 🔥	20m 14s ⏱️

[codecov]

Codecov Report

❌ Patch coverage is 64.70588% with 6 lines in your changes missing coverage. Please review.
✅ Project coverage is 62.03%. Comparing base (1d6c5ee) to head (7b9466f).
⚠️ Report is 2 commits behind head on feat/flow-span-protocol.

Files with missing lines	Patch %	Lines
src/lfx/src/lfx/graph/graph/base.py	28.57%	5 Missing ⚠️
src/backend/base/langflow/services/task/service.py	50.00%	1 Missing ⚠️

Additional details and impacted files

@@                     Coverage Diff                     @@
##           feat/flow-span-protocol   #14407      +/-   ##
===========================================================
+ Coverage                    61.50%   62.03%   +0.53%     
===========================================================
  Files                         2406     2416      +10     
  Lines                       240085   242243    +2158     
  Branches                     36189    36190       +1     
===========================================================
+ Hits                        147654   150286    +2632     
+ Misses                       90496    90022     -474     
  Partials                      1935     1935              

Flag	Coverage Δ
backend	70.35% <88.88%> (+2.66%)	⬆️
lfx	61.35% <37.50%> (-0.01%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

Files with missing lines	Coverage Δ
...e/langflow/services/background_execution/runner.py	98.49% <100.00%> (+0.51%)	⬆️
...ackend/base/langflow/services/job_queue/service.py	85.09% <100.00%> (+2.93%)	⬆️
src/backend/base/langflow/services/jobs/service.py	87.40% <100.00%> (-1.02%)	⬇️
src/lfx/src/lfx/constants.py	100.00% <100.00%> (ø)
src/backend/base/langflow/services/task/service.py	62.90% <50.00%> (+0.60%)	⬆️
src/lfx/src/lfx/graph/graph/base.py	64.11% <28.57%> (-0.15%)	⬇️

... and 142 files with indirect coverage changes

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.