[fix] [issue 1067] Fix the excessive dataChan capacity

[Gleiphir2769]

Fixes #1067

Motivation

wait to edit...

Modifications

Describe the modifications you've done.

Verifying this change

• Make sure that the change passes the CI checks.

[gunli]

Hi @Gleiphir2769, as discussed in #1055 , we will move some logic from internalSend() back to internalAsyncSend(), reserving resource for the requset will be done before putting it into dataChan, if there is enough semaphore and pending queue spaces, continue, or it will be block or failed, as the JAVA client did, and this won't be a problem any more.

[Gleiphir2769]

we will move some logic from internalSend() back to internalAsyncSend(), reserving resource for the requset will be done before putting it into dataChan

If we have made some progress in this work, this PR can be closed. But I think we should fix the existing bugs first and this fix has no side effects on this work. What do you think? @gunli

[gunli]

If we have made some progress in this work, this PR can be closed. But I think we should fix the existing bugs first and this fix has no side effects on this work. What do you think? @gunli

@Gleiphir2769 I will do the refactoring work this week.

[Gleiphir2769]

I will do the refactoring work this week.

Good @gunli ! I will close this PR after your refactor work.

[gunli]

@Gleiphir2769 I have pushed the PR #1071 , please help to review it, thanks!

[tisonkun]

Closing...

You may review #1121 and I still think that we can have a much better solution than even after #1121 merged.