store/copr: balance region for batch cop task (#24521)

[ti-srebot]

cherry-pick #24521 to release-5.0
You can switch your code base to this Pull Request by using git-extras:

# In tidb repo:
git pr https://github.com/pingcap/tidb/pull/24724

After apply modifications, you can push your change to this PR via:

git push git@github.com:ti-srebot/tidb.git pr/24724:release-5.0-66c8cd96b0ea

---

What problem does this PR solve?

Issue Number: close pingcap/tiflash#1854

Problem Summary:

For mpp queries, if a table has more than 1 replicas, the table scan task of this table does not balance the scan region between the TiFlash nodes, so it is possible that although the regions are evenly distributed on several TiFlash nodes, the
scanned regions on each TiFlash nodes for a specific mpp query is very uneven

What is changed and how it works?

Proposal: xxx

What's Changed:

How it Works:
In buildBatchCopTasks, when all the regions to be read is collected, it will call balanceBatchCopTask to balance region for batch cop task.

Related changes

• PR to update pingcap/docs/pingcap/docs-cn:
• Need to cherry-pick to the release branch

Check List

Tests

• Manual test (add detailed scripts or steps below)
  By add some extra log in TiDB, the balance result looks like this:

[2021/05/10 17:48:03.318 +08:00] [INFO] [batch_coprocessor.go:293] ["before task balance store 172.16.5.59:7640 : 527 regions,  store 172.16.5.85:7640 : 359 regions,  store 172.16.5.81:7640 : 243 regions, "]
[2021/05/10 17:48:03.320 +08:00] [INFO] [batch_coprocessor.go:299] ["after task balance store 172.16.5.81:7640 : 376 regions,  store 172.16.5.59:7640 : 377 regions,  store 172.16.5.85:7640 : 376 regions, "]
[2021/05/10 17:48:03.320 +08:00] [INFO] [batch_coprocessor.go:293] ["before task balance store 172.16.5.81:7640 : 1 regions, "]
[2021/05/10 17:48:03.320 +08:00] [INFO] [batch_coprocessor.go:299] ["after task balance store 172.16.5.81:7640 : 1 regions, "]
[2021/05/10 17:48:03.321 +08:00] [INFO] [batch_coprocessor.go:293] ["before task balance store 172.16.5.85:7640 : 80 regions,  store 172.16.5.81:7640 : 227 regions,  store 172.16.5.59:7640 : 66 regions, "]
[2021/05/10 17:48:03.322 +08:00] [INFO] [batch_coprocessor.go:299] ["after task balance store 172.16.5.81:7640 : 135 regions,  store 172.16.5.59:7640 : 103 regions,  store 172.16.5.85:7640 : 135 regions, "]

Side effects

• Performance regression
  • Consumes more CPU

Release note

• balance region for batch cop task

[ti-chi-bot]

@ti-srebot: This cherry pick PR is for a release branch and has not yet been approved by release team.
Adding the do-not-merge/cherry-pick-not-approved label.

To merge this cherry pick, it must first be approved (/lgtm + /merge) by the collaborators.

AFTER it has been approved by collaborators, please ping the release team in a comment to request a cherry pick review.

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.

[ti-srebot]

/run-all-tests

[ti-srebot]

@windtalker you're already a collaborator in bot's repo.

[windtalker]

/run-all-tests

[windtalker]

/run-unit-test

[fzhedu]

/LGTM

[ti-chi-bot]

[REVIEW NOTIFICATION]

This pull request has been approved by:

• fzhedu
• hanfei1991

To complete the pull request process, please ask the reviewers in the list to review by filling /cc @reviewer in the comment.
After your PR has acquired the required number of LGTMs, you can assign this pull request to the committer in the list by filling /assign @committer in the comment to help you merge this pull request.

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

Reviewer can indicate their review by submitting an approval review.
Reviewer can cancel approval by submitting a request changes review.

[wjhuang2016]

The level should be warn or error

[windtalker]

/merge

[ti-chi-bot]

This pull request has been accepted and is ready to merge.

Commit hash: 9ebdbbb

[windtalker]

/run-common-test