[xray] Fix bug when counting a task's lineage size #2600
Merged
Conversation
This file contains hidden or 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
ujvl
reviewed
Aug 8, 2018
| @@ -211,7 +211,12 @@ void LineageCache::AddReadyTask(const Task &task) { | |||
| } | |||
| } | |||
|
|
|||
| uint64_t LineageCache::CountUnsubscribedLineage(const TaskID &task_id) const { | |||
| uint64_t LineageCache::CountUnsubscribedLineage(const TaskID &task_id, | |||
| std::unordered_set<TaskID> &seen) const { | |||
There was a problem hiding this comment.
Think you can maybe default initialize here so you don't have to pass it in on the root call const std::unordered_set<TaskID> &seen = std::unordered_set<TaskID>() (but is potentially uglier).
There was a problem hiding this comment.
I think I prefer the current way.
robertnishihara
approved these changes
Aug 8, 2018
|
Test FAILed. |
|
Test PASSed. |
8 tasks
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Outdated suggestions cannot be applied.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.
What do these changes do?
In the lineage cache, we sometimes count the size of the uncommitted lineage of a task so that we can decide whether we should evict it. The method to count the size walks the task's lineage tree but does not consider whether a node has been seen before, so it ends up double-counting any nodes that appear twice in the tree.
In the test case
python -m pytest test/stress_tests.py::test_dependencies[ray_start_combination1], which submits tasks with many interleaving dependencies, this resulted in crashes of the raylet because a handler on the event loop could end up spending minutes in theCountUnsubscribedLineagemethod. Then, heartbeats were not sent quickly enough and the raylet would be marked as dead.This PR adds back the
test/stress_tests.py::test_dependenciesandtest/stress_tests.py::test_submitting_many_taskstests for raylet.