Add scheduling strategies in CA#9205
Conversation
To be more exact, it can: - Insert a number. Complexity: Amortized O(1). - Check if all numbers upto a given number are inserted. Complexity: O(1). - Efficiently wait until all numbers inserted up to a given number.
|
[APPROVALNOTIFIER] This PR is NOT APPROVED This pull-request has been approved by: walidghallab The full list of commands accepted by this bot can be found here. DetailsNeeds approval from an approver in each of these files:Approvers can indicate their approval by writing |
|
Hi @walidghallab. Thanks for your PR. I'm waiting for a kubernetes member to verify that this patch is reasonable to test. If it is, they should reply with Once the patch is verified, the new status will be reflected by the I understand the commands that are listed here. DetailsInstructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes-sigs/prow repository. |
6155035 to
73eefa2
Compare
|
/cc x13n |
|
/uncc elmiko |
|
/uncc aleksandra-malinowska |
73eefa2 to
62531e3
Compare
|
/ok-to-test |
To be more exact: - An iterator to control the order of nodes to go through. - An option to prefer the smallest number of iterations in case of multiple matches, even if parallelism is enabled.
62531e3 to
ad16378
Compare
|
PR needs rebase. DetailsInstructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes-sigs/prow repository. |
What type of PR is this?
/kind feature
What this PR does / why we need it:
Add scheduling strategies in CA.
To be more exact:
- An iterator to control the order of nodes to go through.
- An option to prefer the smallest number of iterations in case of
multiple matches, even if parallelism is enabled.
Why we need it? Because currently RunFiltersUntilPassingNode has only one scheduling strategy, which is to always start from the last node it tried. While this work in many cases, there are some cases where we don't want this behavior.
Adding this option is a way to:
The only alternative that exist right now is to run CheckPredicates N times, but this is very slow because:
In the current benchmarks, setting PreferredSmallestSteps to true has overhead of ~12%.
Other changes shouldn't have any noticeable effect.
Which issue(s) this PR fixes:
Fixes #
Special notes for your reviewer:
Does this PR introduce a user-facing change?
Additional documentation e.g., KEPs (Kubernetes Enhancement Proposals), usage docs, etc.: