[data] fix a bug that may cause async map tasks to hang #48861
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
raulchen
requested review from
scottjlee,
bveeramani,
stephanie-wang,
omatthew98,
alexeykudinkin and
srinathk10
as code owners
Signed-off-by: Hao Chen <chenh1024@gmail.com>
raulchen
force-pushed
the
fix-async-map-stuck
branch
from
dab4793 to
5af4576
Compare
bveeramani
approved these changes
Nov 22, 2024
| loop = ray.data._map_actor_context.udf_map_asyncio_loop | ||
| tasks = [loop.create_task(process_batch(x)) for x in input_iterable] | ||
| try: | ||
| loop = ray.data._map_actor_context.udf_map_asyncio_loop |
There was a problem hiding this comment.
Out of scope for this PR, but do you know why we need to use a global variable for the loop? Seems like the actor context is specific to a UDF?
There was a problem hiding this comment.
I guess it's because we already have a global var _MapActorContext to cache other stuff.
technically the loop can also be bound to the actor object.
alexeykudinkin
approved these changes
Nov 22, 2024
| @@ -389,6 +394,12 @@ async def process_all_batches(): | |||
| # from the async generator, corresponding to a | |||
There was a problem hiding this comment.
Let's simplify the conditional as well to just
while True:
# Blocking
batch = q.get()
if sentinel:
break
| @@ -352,6 +352,8 @@ def transform_fn( | |||
| # generators, and in the main event loop, yield them from | |||
| # the queue as they become available. | |||
| output_batch_queue = queue.Queue() | |||
| # Use a special object to signal the end of the queue. | |||
| end_of_queue = object() | |||
There was a problem hiding this comment.
nit: Sentinel is a pretty common term for it
jecsand838
pushed a commit
to jecsand838/ray
that referenced
this pull request
Dec 4, 2024
…48861) ## Why are these changes needed? Fix a bug that may cause async map tasks to hang. See code comments for details. This issue can be reproduced with an existing test `test_map_batches_async_generator` on slow machines. --------- Signed-off-by: Hao Chen <chenh1024@gmail.com> Signed-off-by: Connor Sanders <connor@elastiflow.com>
dentiny
pushed a commit
to dentiny/ray
that referenced
this pull request
Dec 7, 2024
…48861) ## Why are these changes needed? Fix a bug that may cause async map tasks to hang. See code comments for details. This issue can be reproduced with an existing test `test_map_batches_async_generator` on slow machines. --------- Signed-off-by: Hao Chen <chenh1024@gmail.com> Signed-off-by: hjiang <dentinyhao@gmail.com>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Why are these changes needed?
Fix a bug that may cause async map tasks to hang. See code comments for details.
This issue can be reproduced with an existing test
test_map_batches_async_generatoron slow machines.Related issue number
Checks
git commit -s) in this PR.scripts/format.shto lint the changes in this PR.method in Tune, I've added it in
doc/source/tune/api/under thecorresponding
.rstfile.