grpc: add caller info to client and wrap innerClient

[okJiang]

What problem does this PR solve?

Issue Number: Ref #8593

What is changed and how does it work?

Check List

Tests

• Unit test
• Integration test

Release note

None.

[ti-chi-bot]

Skipping CI for Draft Pull Request.
If you want CI signal for your change, please convert it to an actual PR.
You can still manually trigger a test run with /test all

[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 assign zhouqiang-cl for approval. 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

[okJiang]

/retest

[okJiang]

/retest

[codecov]

Codecov Report

Attention: Patch coverage is 85.07463% with 50 lines in your changes missing coverage. Please review.

Project coverage is 75.73%. Comparing base (2637331) to head (550c343).
Report is 2 commits behind head on master.

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #8704      +/-   ##
==========================================
- Coverage   75.80%   75.73%   -0.07%     
==========================================
  Files         461      461              
  Lines       72315    72072     -243     
==========================================
- Hits        54817    54584     -233     
- Misses      14000    14002       +2     
+ Partials     3498     3486      -12     

Flag	Coverage Δ
unittests	75.73% <85.07%> (-0.07%)	⬇️

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

[okJiang]

/retest

[okJiang]

/cc @JmPotato

[JmPotato]

A nested struct without introducing the field inner may reduce the code changes. But I'm not sure whether the pointer here still works, maybe we can have a try.

[okJiang]

But I'm not sure whether the pointer here still works, maybe we can have a try.

I tried it before and indeed encountered some issues, so I wrapped the inner client.

[okJiang]

And I think wrapping innerclient also helps to reduce the amount of code in client.go, as the code lines in client.go are a bit excessive.

[nolouch]

Can we provide an automatic method to ingest the CallerID and Component? Like:

package processutil

import "os"

// GetCallerID returns the name of the currently running process
func GetCallerID() string {
	return os.Args[0]
}


func GetTheComponent() string {
	// Get the program counter for the calling function
	pc, _, _, ok := runtime.Caller(1)
	if !ok {
		return "unknown"
	}

	// Retrieve the full function name, including the package path
	fullFuncName := runtime.FuncForPC(pc).Name()

	// Find the last slash, which separates the package path from the function name
	lastSlash := strings.LastIndex(fullFuncName, "/")
	if lastSlash == -1 {
		// No slash, use the last period instead, which separates the package and function
		lastSlash = strings.LastIndex(fullFuncName, ".")
	}
	// Extract the package name
	return fullFuncName[:lastSlash]
}

[okJiang]

Very good suggestion👍, I will consider if there are any flaws

[JmPotato]

Perhaps the HTTP client could also adopt this method.

[okJiang]

I have added these two functions to this pr. In the default situation, CreateClient will set the callerID and callerComponent by these two functions. I removed callerID input from CreateClient and keep the callerComponent to allow caller set component.

And I expose the upperLayer to allow user to set it. For example, client-go can set it to catch the real caller in TiDB.

How do you think about these changes?

[nolouch]

rest lgtm

[nolouch]

why not move to innerClient, and make callerComponent as ClientOption?

[okJiang]

What you said is exactly what I thought at the beginning. You can see it here

The ClientOption is not actually universal; if we want to apply it across all interfaces, we need to further organize the various options. Therefore, I adopted @JmPotato 's suggestion. In my view, I can accept both implementation methods.

[nolouch]

I think @JmPotato's suggestion also doesn't require changing the function signature. However, if you intend to require the client to set the component name, I'm fine with changing it.

[okJiang]

Yes, I intend to require the client to set it.