Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

[Data] Fix parallelism deriving heuristic to ensure parallelism stays w/in min/max bounds #47695

Open
wants to merge 23 commits into
base: master
Choose a base branch
from

Conversation

alexeykudinkin
Copy link
Contributor

Why are these changes needed?

Currently, min/max parallelism isn't actually being enforced correctly -- for large enough clusters we will be scaling out too aggressively purely based on the # of available CPUs disregarding the target block-sizes.

This change

  • Fixes parallelism detection heuristic to appropriately respect min/target block-sizes
  • Makes block-sizes configs' defaults env-var-configurable
  • Adjusts default min-block-size from 1Mb to 16Mb

Related issue number

Checks

  • I've signed off every commit(by using the -s flag, i.e., git commit -s) in this PR.
  • I've run scripts/format.sh to lint the changes in this PR.
  • I've included any doc changes needed for https://docs.ray.io/en/master/.
    • I've added any new APIs to the API Reference. For example, if I added a
      method in Tune, I've added it in doc/source/tune/api/ under the
      corresponding .rst file.
  • I've made sure the tests are passing. Note that there might be a few flaky tests, see the recent failures at https://flakey-tests.ray.io/
  • Testing Strategy
    • Unit tests
    • Release tests
    • This PR is not tested :(
@alexeykudinkin alexeykudinkin added the go add ONLY when ready to merge, run all tests label Sep 17, 2024
@alexeykudinkin alexeykudinkin force-pushed the ak/prlsm-drv-fix branch 2 times, most recently from 206c0a6 to b99aea7 Compare October 1, 2024 07:26
Signed-off-by: Alexey Kudinkin <ak@anyscale.com>
Signed-off-by: Alexey Kudinkin <ak@anyscale.com>
… for parallelism based on target block-sizes

Signed-off-by: Alexey Kudinkin <ak@anyscale.com>
Signed-off-by: Alexey Kudinkin <ak@anyscale.com>
Signed-off-by: Alexey Kudinkin <ak@anyscale.com>
Signed-off-by: Alexey Kudinkin <ak@anyscale.com>
Signed-off-by: Alexey Kudinkin <ak@anyscale.com>
Signed-off-by: Alexey Kudinkin <ak@anyscale.com>
Signed-off-by: Alexey Kudinkin <ak@anyscale.com>
Signed-off-by: Alexey Kudinkin <ak@anyscale.com>
Signed-off-by: Alexey Kudinkin <ak@anyscale.com>
Signed-off-by: Alexey Kudinkin <ak@anyscale.com>
Signed-off-by: Alexey Kudinkin <ak@anyscale.com>
Signed-off-by: Alexey Kudinkin <ak@anyscale.com>
Updating fixtures

Signed-off-by: Alexey Kudinkin <ak@anyscale.com>
Signed-off-by: Alexey Kudinkin <ak@anyscale.com>
Signed-off-by: Alexey Kudinkin <ak@anyscale.com>
Fixed incorrect usage of `_autodetect_parallelism`;

Signed-off-by: Alexey Kudinkin <ak@anyscale.com>
Tidying up

Signed-off-by: Alexey Kudinkin <ak@anyscale.com>
Signed-off-by: Alexey Kudinkin <ak@anyscale.com>
Signed-off-by: Alexey Kudinkin <ak@anyscale.com>
…et_min_block_size`

Signed-off-by: Alexey Kudinkin <ak@anyscale.com>
Signed-off-by: Alexey Kudinkin <ak@anyscale.com>

assert (
min_safe_parallelism <= max_reasonable_parallelism
), f"Parallelism boundaries have to overlap: {estimation_context}"
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

nit, use an if statement to avoid generating the error message string when not needed.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
go add ONLY when ready to merge, run all tests
2 participants