Skip to content

[Core][Label scheduling 3/n] Implement node label scheduling strategy#37339

Merged
jjyao merged 1 commit intoray-project:masterfrom
larrylian:labels_scheduling_3
Jul 25, 2023
Merged

[Core][Label scheduling 3/n] Implement node label scheduling strategy#37339
jjyao merged 1 commit intoray-project:masterfrom
larrylian:labels_scheduling_3

Conversation

@larrylian
Copy link
Copy Markdown
Contributor

@larrylian larrylian commented Jul 12, 2023 •
edited
Loading

Why are these changes needed?

Implement node label scheduling strategy

Related issue number

#34894
(P2)Implement the node affinity with labels interface in Python and transparently transmit it to the CoreWorker.
(P2)Implement the node affinity with labels scheduling policy.

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 :(

@larrylian larrylian force-pushed the labels_scheduling_3 branch from 819e95a to e8fa814 Compare July 13, 2023 02:24
@larrylian larrylian changed the title [Core][Label scheduling 2/n] Add node label scheduling strategy [Core][Label scheduling 3/n] Implement node label scheduling strategy Jul 18, 2023
@larrylian larrylian force-pushed the labels_scheduling_3 branch from e8fa814 to 4088418 Compare July 18, 2023 10:11
@larrylian larrylian requested a review from jjyao July 18, 2023 10:13
jjyao
jjyao reviewed Jul 19, 2023
View reviewed changes
src/ray/raylet/scheduling/policy/node_label_scheduling_policy.cc Outdated
Copy link
Copy Markdown
Contributor

@jjyao jjyao Jul 19, 2023

Choose a reason for hiding this comment

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

I don't think we need to score soft match: you either match the entire expressions or not. There is no partial match. We only need to score after we support OR with weights.

Copy link
Copy Markdown
Contributor Author

@larrylian larrylian Jul 20, 2023

Choose a reason for hiding this comment

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

It's okay. We can add it if necessary in the future.

@larrylian larrylian force-pushed the labels_scheduling_3 branch from 4088418 to 0ddeee6 Compare July 21, 2023 09:57
@larrylian larrylian requested a review from jjyao July 21, 2023 10:02
jjyao
jjyao reviewed Jul 21, 2023
View reviewed changes
Copy link
Copy Markdown
Contributor

@jjyao jjyao left a comment

Choose a reason for hiding this comment

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

Will continue.

jjyao
jjyao reviewed Jul 21, 2023
View reviewed changes
src/ray/common/task/task_spec.h Outdated
Copy link
Copy Markdown
Contributor

@jjyao jjyao Jul 21, 2023

Choose a reason for hiding this comment

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

SerializeAsString as well?

Copy link
Copy Markdown
Contributor

@jjyao jjyao Jul 24, 2023

Choose a reason for hiding this comment

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

The hash function should guarantee that if two label scheduling strategies are the same, they must have the same hash value. Can we actually examine each field of NodeLabelSchedulingStrategy like what we do for other scheduling strategies?

Copy link
Copy Markdown
Contributor

@jjyao jjyao Jul 25, 2023

Choose a reason for hiding this comment

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

Let's at least change to SerializeAsString for now.

Copy link
Copy Markdown
Contributor Author

@larrylian larrylian Jul 25, 2023

Choose a reason for hiding this comment

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

Because SerializeAsString is unsafe for hash. I think just use DebugString for now.

@larrylian larrylian force-pushed the labels_scheduling_3 branch from 0ddeee6 to 25ce51f Compare July 24, 2023 03:19
@larrylian larrylian requested a review from jjyao July 24, 2023 03:20
jjyao
jjyao reviewed Jul 24, 2023
View reviewed changes
src/ray/common/task/task_spec.h Outdated
Copy link
Copy Markdown
Contributor

@jjyao jjyao Jul 24, 2023

Choose a reason for hiding this comment

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

The hash function should guarantee that if two label scheduling strategies are the same, they must have the same hash value. Can we actually examine each field of NodeLabelSchedulingStrategy like what we do for other scheduling strategies?

@larrylian larrylian force-pushed the labels_scheduling_3 branch from 25ce51f to c8c71fa Compare July 24, 2023 07:19
@larrylian
Copy link
Copy Markdown
Contributor Author

larrylian commented Jul 24, 2023

The hash function should guarantee that if two label scheduling strategies are the same, they must have the same hash value. Can we actually examine each field of NodeLabelSchedulingStrategy like what we do for other scheduling strategies?

That's fine. I'd prefer to create a new PR for this modification. The struct has quite a few member variables, so the code will be substantial. I'll also write some test cases.

jjyao
jjyao approved these changes Jul 25, 2023
View reviewed changes
src/ray/common/task/task_spec.h Outdated
Copy link
Copy Markdown
Contributor

@jjyao jjyao Jul 25, 2023

Choose a reason for hiding this comment

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

Let's at least change to SerializeAsString for now.

@larrylian
[Core][Label scheduling 3/n] Implement node label scheduling strategy
e367a4b 
Signed-off-by: 稚鱼 <554538252@qq.com>
Signed-off-by: LarryLian <554538252@qq.com>
@larrylian larrylian force-pushed the labels_scheduling_3 branch from c8c71fa to e367a4b Compare July 25, 2023 09:50
@jjyao jjyao merged commit 9680b82 into ray-project:master Jul 25, 2023
NripeshN pushed a commit to NripeshN/ray that referenced this pull request Aug 15, 2023
@larrylian @NripeshN
[Core][Label scheduling 3/n] Implement node label scheduling strategy (…
3dbce5b 
…ray-project#37339)

Signed-off-by: LarryLian <554538252@qq.com>
Signed-off-by: NripeshN <nn2012@hw.ac.uk>
arvind-chandra pushed a commit to lmco/ray that referenced this pull request Aug 31, 2023
@larrylian @arvind-chandra
[Core][Label scheduling 3/n] Implement node label scheduling strategy (…
81eed48 
…ray-project#37339)

Signed-off-by: LarryLian <554538252@qq.com>
Signed-off-by: e428265 <arvind.chandramouli@lmco.com>
@larrylian larrylian mentioned this pull request Sep 4, 2023
Open
28 tasks
vymao pushed a commit to vymao/ray that referenced this pull request Oct 11, 2023
@larrylian
[Core][Label scheduling 3/n] Implement node label scheduling strategy (…
86ba2c9 
…ray-project#37339)

Signed-off-by: LarryLian <554538252@qq.com>
Signed-off-by: Victor <vctr.y.m@example.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@jjyao jjyao jjyao approved these changes

@wuisawesome wuisawesome Awaiting requested review from wuisawesome

@ericl ericl Awaiting requested review from ericl

@AmeerHajAli AmeerHajAli Awaiting requested review from AmeerHajAli

@robertnishihara robertnishihara Awaiting requested review from robertnishihara

@pcmoritz pcmoritz Awaiting requested review from pcmoritz

@raulchen raulchen Awaiting requested review from raulchen

Assignees

@jjyao jjyao

Labels

None yet

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

2 participants

@larrylian @jjyao