disttask: add operator abstraction #46279

tangenta · 2023-08-21T08:06:28Z

What problem does this PR solve?

Issue Number: ref #46258

Problem Summary:

What is changed and how it works?

This PR introduces the async data operator interface, which provide an abstraction for the planner of dist task framework.

// Operator is the basic operation unit in the task execution.
type Operator interface {
	Open() error
	Close() error
	Display() string
}

By composing the async operators, we can process the data concurrently:

words := `Bob hiT a ball, the hIt BALL flew far after it was hit.`
var mostCommonWord string
splitter := makeSplitter(words)
lower := makeLower()
counter := makeCounter()
collector := makeCollector(&mostCommonWord)

Compose[string](splitter, lower)
Compose[string](lower, counter)
Compose[strCnt](counter, collector)

pipeline := NewAsyncPipeline(splitter, lower, counter, collector)
pipeline.Execute()

Check List

Tests

Unit test
Integration test
Manual test (add detailed scripts or steps below)
No code

Side effects

Performance regression: Consumes more CPU
Performance regression: Consumes more Memory
Breaking backward compatibility

Documentation

Release note

Please refer to Release Notes Language Style Guide to write a quality release note.

None

tiprow · 2023-08-21T08:06:51Z

Hi @tangenta. Thanks for your PR.

PRs from untrusted users cannot be marked as trusted with /ok-to-test in this repo meaning untrusted PR authors can never trigger tests themselves. Collaborators can still trigger tests on the PR using /test all.

I understand the commands that are listed here.

Instructions 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/test-infra repository.

codecov · 2023-08-21T08:15:41Z

Codecov Report

Merging #46279 (3726322) into master (533998e) will decrease coverage by 0.6618%.
Report is 12 commits behind head on master.
The diff coverage is 95.4545%.

Additional details and impacted files

@@               Coverage Diff                @@
##             master     #46279        +/-   ##
================================================
- Coverage   73.3815%   72.7198%   -0.6618%     
================================================
  Files          1285       1309        +24     
  Lines        394832     399639      +4807     
================================================
+ Hits         289734     290617       +883     
- Misses        86650      90608      +3958     
+ Partials      18448      18414        -34

Flag	Coverage Δ
integration	`25.6501% <0.0000%> (?)`
unit	`73.3979% <95.4545%> (+0.0163%)`	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

Components	Coverage Δ
dumpling	`54.0444% <ø> (ø)`
parser	`85.0845% <ø> (+0.0178%)`	⬆️
br	`47.9067% <ø> (-4.3111%)`	⬇️

ywqzzy

LGTM

ywqzzy · 2023-08-22T07:07:42Z

disttask/operator/wrapper.go

+	return "simpleSink"
+}
+
+type simpleOperator[T, R any] struct {


For add index, will we use IngestOperator which wraps AsyncOperator instead of simpleOperator?

Yes, you can overwrite the Display() method.

ti-chi-bot · 2023-08-22T07:19:46Z

@ywqzzy: adding LGTM is restricted to approvers and reviewers in OWNERS files.

In response to this:

LGTM

Instructions 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/test-infra repository.

D3Hunter

3/10， will review later

disttask/operator/operator.go

disttask/operator/pipeline.go

disttask/operator/operator.go

disttask/operator/wrapper.go

D3Hunter · 2023-08-22T10:04:00Z

disttask/operator/operator.go

+}
+
+// NewAsyncOperator create an AsyncOperator.
+func NewAsyncOperator[T, R any](name string, workerNum int, transform func(T) R) *AsyncOperator[T, R] {


previously i thought we need impl each operator separately, now it seems we have a common operator and we need to impl a different transform

and it's not that flexible as each input T we generate a output R, but in some case we only generate one output from all input data.(right now, we're using a special sink as a collector)

and it's not that flexible as each input T we generate a output R, but in some case we only generate one output from all input data.(right now, we're using a special sink as a collector)

Do you mean that one operator should have multiple type of input？

Let me change it to two constructors:

// NewAsyncOperatorWithTransform create an AsyncOperator with a transform function. func NewAsyncOperatorWithTransform[T, R any](name string, workerNum int, transform func(T) R) *AsyncOperator[T, R] { pool := workerpool.NewWorkerPool(name, util.DistTask, workerNum, newAsyncWorkerCtor(transform)) return NewAsyncOperator(pool) } // NewAsyncOperator create an AsyncOperator. func NewAsyncOperator[T, R any](pool *workerpool.WorkerPool[T, R]) *AsyncOperator[T, R] { return &AsyncOperator[T, R]{ pool: pool, } }

We can pass a custom worker pool for different businesses.

If this still cannot meet the requirement, we can define our own async operator by implementing Operator + WithSource/WithSink interface.

Do you mean that one operator should have multiple type of input？

now we have one output row for each input row(suppose it's row-based), we may want a operator that aggregate multiple or all input rows into a single output row.

Benjamin2037 · 2023-08-22T11:42:24Z

Seems no err handling code, next pr?

tangenta · 2023-08-22T12:25:00Z

@Benjamin2037 Yes, I will implement the error handling in next PR.

D3Hunter

lgtm

GMHDBJD

LGTM

ti-chi-bot · 2023-08-22T15:01:31Z

[APPROVALNOTIFIER] This PR is APPROVED

This pull-request has been approved by: D3Hunter, GMHDBJD, ywqzzy

The full list of commands accepted by this bot can be found here.

The pull request process is described here

Needs approval from an approver in each of these files:

~~OWNERS~~ [D3Hunter,GMHDBJD]

Approvers can indicate their approval by writing /approve in a comment
Approvers can cancel approval by writing /approve cancel in a comment

ti-chi-bot · 2023-08-22T15:01:36Z

[LGTM Timeline notifier]

Timeline:

2023-08-22 13:43:39.764532506 +0000 UTC m=+1243384.313548494: ☑️ agreed by D3Hunter.
2023-08-22 15:01:34.93883819 +0000 UTC m=+1248059.487854161: ☑️ agreed by GMHDBJD.

disttask: add operator abstraction

1e803b2

ti-chi-bot bot added do-not-merge/needs-linked-issue release-note-none Denotes a PR that doesn't merit a release note. size/XL Denotes a PR that changes 500-999 lines, ignoring generated files. labels Aug 21, 2023

add BUILD.bazel

bd38f1a

ti-chi-bot bot removed the do-not-merge/needs-linked-issue label Aug 21, 2023

tangenta added 6 commits August 21, 2023 17:25

add comments

0c349e0

fix linter

c6f8f27

fix operator close issue

8659b1a

fix linter

f6e727d

refine code and improve code coverage

630a132

remove redundant log

c1f6aad

ywqzzy approved these changes Aug 22, 2023

View reviewed changes

ywqzzy mentioned this pull request Aug 21, 2023

Enhance distributed task execution framework #46258

Closed

76 tasks

D3Hunter reviewed Aug 22, 2023

View reviewed changes

address comment

3726322

D3Hunter approved these changes Aug 22, 2023

View reviewed changes

ti-chi-bot bot added approved needs-1-more-lgtm Indicates a PR needs 1 more LGTM. labels Aug 22, 2023

GMHDBJD approved these changes Aug 22, 2023

View reviewed changes

ti-chi-bot bot added lgtm and removed needs-1-more-lgtm Indicates a PR needs 1 more LGTM. labels Aug 22, 2023

ti-chi-bot bot merged commit 1ea6499 into pingcap:master Aug 22, 2023
15 of 18 checks passed

tangenta mentioned this pull request Aug 25, 2023

ddl, disttask: implement add index operators #46414

Merged

12 tasks

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

disttask: add operator abstraction #46279

disttask: add operator abstraction #46279

tangenta commented Aug 21, 2023 •

edited by Benjamin2037

Loading

tiprow bot commented Aug 21, 2023

codecov bot commented Aug 21, 2023 •

edited

Loading

ywqzzy left a comment

ywqzzy Aug 22, 2023

tangenta Aug 22, 2023

ti-chi-bot bot commented Aug 22, 2023

D3Hunter left a comment

D3Hunter Aug 22, 2023

D3Hunter Aug 22, 2023

ywqzzy Aug 22, 2023

tangenta Aug 22, 2023 •

edited

Loading

D3Hunter Aug 22, 2023

Benjamin2037 commented Aug 22, 2023

tangenta commented Aug 22, 2023

D3Hunter left a comment

GMHDBJD left a comment

ti-chi-bot bot commented Aug 22, 2023

ti-chi-bot bot commented Aug 22, 2023

disttask: add operator abstraction #46279

disttask: add operator abstraction #46279

Conversation

tangenta commented Aug 21, 2023 • edited by Benjamin2037 Loading

What problem does this PR solve?

What is changed and how it works?

Check List

Release note

tiprow bot commented Aug 21, 2023

codecov bot commented Aug 21, 2023 • edited Loading

Codecov Report

ywqzzy left a comment

Choose a reason for hiding this comment

ywqzzy Aug 22, 2023

Choose a reason for hiding this comment

tangenta Aug 22, 2023

Choose a reason for hiding this comment

ti-chi-bot bot commented Aug 22, 2023

D3Hunter left a comment

Choose a reason for hiding this comment

D3Hunter Aug 22, 2023

Choose a reason for hiding this comment

D3Hunter Aug 22, 2023

Choose a reason for hiding this comment

ywqzzy Aug 22, 2023

Choose a reason for hiding this comment

tangenta Aug 22, 2023 • edited Loading

Choose a reason for hiding this comment

D3Hunter Aug 22, 2023

Choose a reason for hiding this comment

Benjamin2037 commented Aug 22, 2023

tangenta commented Aug 22, 2023

D3Hunter left a comment

Choose a reason for hiding this comment

GMHDBJD left a comment

Choose a reason for hiding this comment

ti-chi-bot bot commented Aug 22, 2023

ti-chi-bot bot commented Aug 22, 2023

[LGTM Timeline notifier]

tangenta commented Aug 21, 2023 •

edited by Benjamin2037

Loading

codecov bot commented Aug 21, 2023 •

edited

Loading

tangenta Aug 22, 2023 •

edited

Loading