Try scheduling as much as available #4528
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
Previously, it would just schedule about ~1500 tasks unless the regions were totally full. Now we will schedule up to 15K tasks. Also, we will take into account batch's queueing (there could be other reasons for this besides CPU quota, though there shouldn't) and tasks that were already scheduler but not sent to batch or preprocessed so we don't overload the queue.
vitorguidi
reviewed
Dec 19, 2024
| # TODO(metzman): This doesn't distinguish between fuzz and non-fuzz | ||
| # tasks (nor preemptible and non-preemptible CPUs). Fix this. | ||
| waiting_tasks = sum( | ||
| batch.count_queued_or_scheduled_tasks(project, region) |
There was a problem hiding this comment.
One option to simulate these behaviors is https://simpy.readthedocs.io/
https://brooker.co.za/blog/2022/04/11/simulation.html
It is hard to imagine what these policies imply, by just the description
vitorguidi
reviewed
Dec 19, 2024
| logs.info(f'Soon committed CPUs: {soon_commited_cpus}') | ||
| available_cpus = sum( | ||
| get_available_cpus_for_region(project, region) for region in regions) | ||
| available_cpus = max(available_cpus - soon_commited_cpus, 0) |
There was a problem hiding this comment.
Now that we take the queue size into account, we can go back to running this very frequently, right?
There was a problem hiding this comment.
Hmmm...I had the opposite thought, that because we can schedule so many more at once, there's no need to run it so often. I think there can be a slight delay between publishing and reaching the queue, so probably above 5 minutes makes most sense.
vitorguidi
approved these changes
Dec 19, 2024
vitorguidi
added a commit
that referenced
this pull request
Dec 28, 2024
This reverts commit d222215.
Merged
vitorguidi
added a commit
that referenced
this pull request
Dec 28, 2024
The preprocess count for fuzz task went to zero after #4564 got deployed, reverting. #4528 is also being reverted because it introduced the following error into the fuzz task scheduler, which caused fuzz tasks to stop being scheduled: ``` Traceback (most recent call last): File "/mnt/scratch0/clusterfuzz/src/python/bot/startup/run_cron.py", line 68, in <module> sys.exit(main()) ^^^^^^ File "/mnt/scratch0/clusterfuzz/src/python/bot/startup/run_cron.py", line 64, in main return 0 if task_module.main() else 1 ^^^^^^^^^^^^^^^^^^ File "/mnt/scratch0/clusterfuzz/src/clusterfuzz/_internal/cron/schedule_fuzz.py", line 304, in main return schedule_fuzz_tasks() ^^^^^^^^^^^^^^^^^^^^^ File "/mnt/scratch0/clusterfuzz/src/clusterfuzz/_internal/cron/schedule_fuzz.py", line 284, in schedule_fuzz_tasks available_cpus = get_available_cpus(project, regions) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/mnt/scratch0/clusterfuzz/src/clusterfuzz/_internal/cron/schedule_fuzz.py", line 247, in get_available_cpus result = pool.starmap_async( # pylint: disable=no-member ^^^^^^^^^^^^^^^^^^ AttributeError: 'ProcessPoolExecutor' object has no attribute 'starmap_async' ```
jonathanmetzman
added a commit
that referenced
this pull request
Dec 30, 2024
jonathanmetzman
added a commit
that referenced
this pull request
Jan 8, 2025
Instead of trying to schedule a small amount of fuzz tasks every 2 minutes and hoping this leads to fuzzing at full capacity, just schedule almost the full amount at once.
jonathanmetzman
pushed a commit
that referenced
this pull request
Jan 8, 2025
The preprocess count for fuzz task went to zero after #4564 got deployed, reverting. #4528 is also being reverted because it introduced the following error into the fuzz task scheduler, which caused fuzz tasks to stop being scheduled: ``` Traceback (most recent call last): File "/mnt/scratch0/clusterfuzz/src/python/bot/startup/run_cron.py", line 68, in <module> sys.exit(main()) ^^^^^^ File "/mnt/scratch0/clusterfuzz/src/python/bot/startup/run_cron.py", line 64, in main return 0 if task_module.main() else 1 ^^^^^^^^^^^^^^^^^^ File "/mnt/scratch0/clusterfuzz/src/clusterfuzz/_internal/cron/schedule_fuzz.py", line 304, in main return schedule_fuzz_tasks() ^^^^^^^^^^^^^^^^^^^^^ File "/mnt/scratch0/clusterfuzz/src/clusterfuzz/_internal/cron/schedule_fuzz.py", line 284, in schedule_fuzz_tasks available_cpus = get_available_cpus(project, regions) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/mnt/scratch0/clusterfuzz/src/clusterfuzz/_internal/cron/schedule_fuzz.py", line 247, in get_available_cpus result = pool.starmap_async( # pylint: disable=no-member ^^^^^^^^^^^^^^^^^^ AttributeError: 'ProcessPoolExecutor' object has no attribute 'starmap_async' ```
jonathanmetzman
added a commit
that referenced
this pull request
Jan 8, 2025
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.
Instead of trying to schedule a small amount of fuzz tasks every 2 minutes and hoping this leads
to fuzzing at full capacity, just schedule almost the full amount at once.