feat(taskbroker-client): Add pass_headers option to task registration by untitaker · Pull Request #623 · getsentry/taskbroker

untitaker · 2026-05-07T13:49:29Z

Summary

Adds pass_headers parameter to @namespace.register() decorator
When pass_headers=True, the task function receives activation headers as a headers keyword argument (dict[str, str])
Provides a simpler alternative to implementing a full ContextHook for tasks that just need direct access to header values

We need this kind of functionality when using raw mode together with the profiles consumer. Mainly for killswitches.

ref STREAM-882

Usage

from taskbroker_client.types import TaskHeaders

@namespace.register(name="my.task", pass_headers=True)
def my_task(org_id: int, headers: TaskHeaders) -> None:
    tenant_id = headers.get("x-tenant-id", "")
    trace_id = headers.get("sentry-trace", "")

Test plan

Added unit tests for pass_headers attribute on Task
Added tests for ALWAYS_EAGER mode (empty headers dict passed)
Added worker integration test verifying headers are passed during execution
All 158 tests pass

🤖 Generated with Claude Code

@register

Allow tasks to opt into receiving activation headers by setting pass_headers=True in the @register() decorator. When enabled, the task function receives headers as a keyword argument (dict[str, str]). This provides a simpler alternative to implementing a full ContextHook for tasks that just need direct access to header values. Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

Add `process_profile_from_kafka` task that accepts raw Kafka message bytes for use with taskbroker's passthrough mode. This allows taskbroker to read directly from the ingest-profiles Kafka topic. Changes: - New task `process_profile_from_kafka` in `ingest.profiling.passthrough` namespace - Refactored `_process_profile_message` to share logic between consumer and task - Removed base64 encoding - tasks now accept bytes directly - Use Kafka headers for killswitch/sampled when available, fall back to message body Depends on getsentry/taskbroker#623 for exposing Kafka headers to tasks. STREAM-882

markstory

I think the general design is reasonable. Another potential solution would be to pass the entire activation into the task, and let userland code pick the headers out. I prefer passing just the headers over the the entire activation though.

markstory · 2026-05-07T15:41:30Z

                compression_type=compression_type,
                report_timeout_errors=report_timeout_errors,
                silenced_exceptions=silenced_exceptions,
+                pass_headers=pass_headers,


Since we're going to inject parameters into the task, should we validate that the task doesn't already have a headers parameter with an incompatible type?

Another potential solution would be to pass the entire activation into the task,

as in, you have a pass_activation option?

Since we're going to inject parameters into the task, should we validate that the task doesn't already have a headers parameter with an incompatible type?

this feels magical but i can do it

I think we can enforce that the headers kwarg is there explicitly. def mytask(**kwargs): is then just invalid.

I think we can enforce that the headers kwarg is there explicitly. def mytask(**kwargs): is then just invalid.

I was more thinking of the scenario where a task has an incompatible headers parameter like:

def some_task(headers: set[str]):

as in, you have a pass_activation option?

Yeah, but as I think about it more I like this idea less, and pass_headers is the better solution.

i think the current version covers this. it requires the header arg to be there, to have the right type, and to be explicit

…=True Use inspect module to check at task registration time that: - The function has a 'headers' parameter when pass_headers=True - The 'headers' parameter is not positional-only Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

evanh · 2026-05-08T14:17:10Z

    activation = multi_task.create_activation([], {})
    assert activation.headers["x-test-context"] == "dispatched"
    assert activation.headers["x-another"] == "also-here"
+


Can you add a test for the case when the user passes headers as one of the kwargs to their function?

ok. testing for a specific error doesn't make a lot of sense if there's no explicit errorhandling for it, so I added more checks to the code too. i think it's a bit bloated now, but probably more user-friendly.

markstory

Looks good to me. The test addition that Evan mentioned would be a good one to have.

cursor

Cursor Bugbot has reviewed your changes and found 1 potential issue.

^{❌ Bugbot Autofix is OFF. To automatically fix reported issues with cloud agents, enable autofix in the Cursor dashboard.}

^{Reviewed by Cursor Bugbot for commit 6e002f9. Configure here.}

cursor · 2026-05-08T17:01:04Z

+        if origin not in expected_types:
+            raise TypeError(
+                f"{context}: {param_name!r} parameter has type {param.annotation!r}. "
+                f"Expected one of: {', '.join(t.__name__ for t in expected_types)}."


Any.__name__ crashes error formatting on Python 3.10

Low Severity

The error message in assert_typed_kwarg uses t.__name__ to format all types in expected_types, but typing.Any does not have a __name__ attribute on Python 3.10 (where it's a _SpecialForm instance, not a class). This causes an AttributeError to be raised instead of the intended helpful TypeError when a user provides an incompatible type annotation for headers. The expected_types tuple at the call site includes Any, so the generator expression t.__name__ for t in expected_types will crash when it reaches Any.

Additional Locations (1)

clients/python/src/taskbroker_client/task.py#L134-L135

^{Reviewed by Cursor Bugbot for commit 6e002f9. Configure here.}

I don't care. Pick better types.

untitaker requested a review from a team as a code owner May 7, 2026 13:49

untitaker force-pushed the feat/pass-headers-to-task branch from 77075a8 to 15615c2 Compare May 7, 2026 13:52

getsentry deleted a comment from linear-code Bot May 7, 2026

untitaker mentioned this pull request May 7, 2026

feat(profiling): Add task for taskbroker passthrough mode getsentry/sentry#115065

Draft

2 tasks

cursor Bot reviewed May 7, 2026

View reviewed changes

Comment thread clients/python/src/taskbroker_client/worker/workerchild.py

markstory reviewed May 7, 2026

View reviewed changes

untitaker and others added 2 commits May 7, 2026 18:16

also check type hints, and move it to a separate function

4408594

cursor Bot reviewed May 7, 2026

View reviewed changes

Comment thread clients/python/src/taskbroker_client/task.py

fix

a99382e

sentry Bot reviewed May 7, 2026

View reviewed changes

Comment thread clients/python/src/taskbroker_client/task.py

cursor Bot reviewed May 7, 2026

View reviewed changes

Comment thread clients/python/src/taskbroker_client/task.py

untitaker requested review from evanh and markstory May 7, 2026 18:23

evanh approved these changes May 8, 2026

View reviewed changes

markstory approved these changes May 8, 2026

View reviewed changes

add testcase

6e002f9

cursor Bot reviewed May 8, 2026

View reviewed changes

untitaker merged commit f94fcdc into main May 8, 2026
23 checks passed

untitaker deleted the feat/pass-headers-to-task branch May 8, 2026 17:07

Uh oh!

Conversation

untitaker commented May 7, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary

Usage

Test plan

Uh oh!

Uh oh!

markstory left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

markstory May 7, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

markstory left a comment

Choose a reason for hiding this comment

Uh oh!

cursor Bot left a comment

Choose a reason for hiding this comment

Uh oh!

cursor Bot May 8, 2026

Choose a reason for hiding this comment

Any.__name__ crashes error formatting on Python 3.10

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

untitaker commented May 7, 2026 •

edited

Loading

markstory May 7, 2026 •

edited

Loading

`Any.name` crashes error formatting on Python 3.10