executor: control Chunk size for TopN&Sort

[qw4990]

What problem does this PR solve?

Control the number of rows in chunks returned by LimitExec.
Following up #9354, this PR is a subtask of #9166.

What is changed and how it works?

• Modify TopN and Sort to make them support chunk size control.
• Add some unit tests.

Check List

Tests

• Unit test

[codecov-io]

Codecov Report

Merging #9364 into master will increase coverage by 0.04%.
The diff coverage is 100%.

@@            Coverage Diff             @@
##           master    #9364      +/-   ##
==========================================
+ Coverage   67.13%   67.17%   +0.04%     
==========================================
  Files         373      373              
  Lines       78136    78135       -1     
==========================================
+ Hits        52455    52487      +32     
+ Misses      20982    20954      -28     
+ Partials     4699     4694       -5

Impacted Files	Coverage Δ
executor/sort.go	84.97% <100%> (+8.16%)	⬆️
infoschema/infoschema.go	76.31% <0%> (-1.32%)	⬇️
store/tikv/lock_resolver.go	41.7% <0%> (-0.95%)	⬇️
executor/distsql.go	72.24% <0%> (+0.45%)	⬆️
executor/index_lookup_join.go	77.91% <0%> (+0.63%)	⬆️
ddl/delete_range.go	79.36% <0%> (+5.82%)	⬆️
ddl/session_pool.go	93.1% <0%> (+10.34%)	⬆️
util/systimemon/systime_mon.go	100% <0%> (+20%)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 77fb909...cc83665. Read the comment docs.

[qw4990]

/run-all-tests

[qw4990]

/run-all-tests

[zz-jason]

In SortExec.fetchRowChunks(), should we set the required rows?

[qw4990]

In SortExec.fetchRowChunks(), should we set the required rows?

Sort is a pipeline breaker and fetchRowChunks() is used to fetch all data from its child to sort.
For the value of requiredRows in this function, I think we have two choices:

1. MaxChunkSize: it indicates the child return as much data as possible once.
2. 1: it means the child should return as quickly as possible if it has any data.

Now the choice 1 is used as before.
How do you think? @zz-jason

[zz-jason]

maxChunkSize is better Yuanjia Zhang <notifications@github.com>于2019年2月20日 周三14:47写道：

…

In SortExec.fetchRowChunks(), should we set the required rows? Sort is a pipeline breaker and fetchRowChunks() is used to fetch all data from its child to sort. For the value of requiredRows in this function, I think we have two choices: 1. MaxChunkSize: it indicates the child return as much data as possible once. 2. 1: it means the child should return as quickly as possible if it has any data. How do you think? @zz-jason <https://github.com/zz-jason> — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#9364 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AFBlG3xz6mWyY4oHPklvUc44XmbA8iXMks5vPO98gaJpZM4bEVUR> .

[qw4990]

maxChunkSize is better Yuanjia Zhang notifications@github.com于2019年2月20日 周三14:47写道：
…
In SortExec.fetchRowChunks(), should we set the required rows? Sort is a pipeline breaker and fetchRowChunks() is used to fetch all data from its child to sort. For the value of requiredRows in this function, I think we have two choices: 1. MaxChunkSize: it indicates the child return as much data as possible once. 2. 1: it means the child should return as quickly as possible if it has any data. How do you think? @zz-jason https://github.com/zz-jason — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#9364 (comment)>, or mute the thread https://github.com/notifications/unsubscribe-auth/AFBlG3xz6mWyY4oHPklvUc44XmbA8iXMks5vPO98gaJpZM4bEVUR .

MaxChunkSize is the default value of RequiredRows for all Chunks.
So we don't need to set it explicitly. 😃

[zz-jason]

LGTM

[qw4990]

PTAL~ @lysu

[lysu]

LGTM

[qw4990]

/run-all-tests

[qw4990]

/run-all-tests