ExchangeSenderOp check writer status only if there is data to flush to MPPTunnel

[windtalker]

What problem does this PR solve?

Issue Number: ref #6233

Problem Summary:
Currently, ExchangeSenderSinkOp check the writer status in ExchangeSenderSinkOp::prepareImpl(), so even if there is currently no data to write, the ExchangeSenderSinkOp need to check the writer status, and the pipeline task maybe blocked if the writer is not ready. In this pr, ExchangeSenderSinkOp only check the writer status if it has data to flush to MPPTunnel.

What is changed and how it works?

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

• Affects user behaviors
• Contains syntax changes
• Contains variable changes
• Contains experimental features
• Changes MySQL compatibility

Release note

None

[ti-chi-bot]

[APPROVALNOTIFIER] This PR is NOT APPROVED

This pull-request has been approved by:
Once this PR has been reviewed and has the lgtm label, please ask for approval from windtalker, ensuring that each of them provides their approval before proceeding. For more information see the Code Review Process.

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

Needs approval from an approver in each of these files:

• OWNERS

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

[windtalker]

/test pull-unit-test

[windtalker]

/test pull-integration-test

[SeaRise]

Perhaps moving waitForWriter into Writer::write and Writer::flush would be better. This way, even if the tunnel is busy, write can still batch in memory.

[windtalker]

You mean if the block's size is small, it will be cached in the writer without writting to tunnel?

[SeaRise]

yes, if the total block rows is smaller than batch_send_min_limit, MPPTunnel::write would not be call.
https://github.com/pingcap/tiflash/blob/master/dbms/src/Flash/Mpp/BroadcastOrPassThroughWriter.cpp#L103-L109

[windtalker]

Good idea, I reimplement this pr by moving waitForWriter to Writer::flush. But waitForWriter is still needed in ExchangeSenderSinkOp::await

[SeaRise]

lgtm