Previously we passed along this piece of state via the kwargs for
a task, but this runs the risk of the task accidentally receiving
the extra kwarg unless we've covered all the code paths that could
invoke it directly e.g. retries don't invoke __call__.
This switches to using Celery "headers" to pass the extra state. It
turns out that a Celery has two "header" concepts, which leads to
some confusion and even a bug with the framework [1]:
- In older (pre v4.4) versions of Celery, the "headers" specified
by apply_async() would become _the_ headers in the message that
gets passed around workers, etc. These would be available later on
via "self.request.headers".
- Since Celery protocol v2, the meaning of "headers" in the message
changed to become (basically) _all_ metadata about the task [2],
with the "headers" option in apply_async() being merged [3] into
the big dict of metadata.
This makes using headers a bit confusing unfortunately, since the
data structure we put in is subtly different to what comes out in
the request context. Nonetheless, it still works. I've added some
comments to try and clarify it.
Note that one of the original tests is no longer necessary, since we
don't need to worry about argument passing styles with headers.
[1]: https://github.com/celery/celery/issues/4875
[2]: 663e4d3a0b (diff-07a65448b2db3252a9711766beec23372715cd7597c3e309bf53859eabc0107fR343)
[3]: 681a922220/celery/app/amqp.py (L495)
This applies the same change we made in other apps [1][2]. Adding
the override here is special, though, because it means the others
will now get triggered, since this app is the start of the chain
of tasks for a request. We will also retain existing request_id
tracing for tasks within this app, since "apply_async" calls the
"send_task" method internally, which is the one we're overriding.
[1]: 6f3c118a1e
[2]: 2e08b7aa95
Celery's apply_async function accepts 'kwargs' as (get ready to be
confused) either a positional argument, or a keyword argument:
Positional: apply_async(['args'], {'kw': 'args'})
Keyword: apply_async(args=['args'], kwargs={'kw': 'args'})
We rely on the positional form in at least one place [1]. This fixes
the overload of apply_async to cope with both forms, and continue to
pass through any other (confusion time again) keyword args to super(),
such as queue="queue".
Note that we've also decided to stop accepting other positional args,
since this is unnecessarily confusing, and we don't currently rely on
it in our code. This stops it creeping in in future.
[1]: fde927e00e/app/job/rest.py (L186)
Tasks will fail if we leave the kwarg in, so I think it's quite
important that we test this works. We don't cover this in any other
test because we call the task functions directly, so the request_id
kwarg doesn't get injected beforehand.
This requires upgrading freezegun, as time.monotonic wasn't frozen
by v1.0. Note that we need to explicitly specify the base class for
the task in the test, the reason for which is quite subtle:
- Normally, by using the 'notify_api' fixture, the base class is set
to NotifyTask automatically by running app.create_app [1].
- However, when run alongside other tests, the imports of files with
other celery tasks cause the base class to be instantiated and cached
as the default Celery one. This means none of our tests actually use
our custom superclass when testing tasks.
Because we can't run 'apply_async' directly (since this would require
an actual Celery broker), we need to manually push/pop the request
Context that's normally done as part of sending a task.
Note also that we use a UUID as the name for a task, since these are
global. We want to avoid the task polluting other tests in future,
as well as make it clear the task is being reused.
[1]: dea5828d0e/app/__init__.py (L113)
