optimize: optimize seata client I/O processing by adjusting thread count #7170

YongGoose · 2025-02-19T06:06:08Z

I have registered the PR changes.

Ⅰ. Describe what this PR did

Ⅱ. Does this pull request fix one issue?

fixes #7168

Ⅲ. Why don't you add test cases (unit test/integration test)?

Ⅳ. Describe how to verify it

Ⅴ. Special notes for reviews

funky-eyes

LGTM

funky-eyes · 2025-02-19T06:18:07Z

https://github.com/apache/incubator-seata/blob/2.x/script/client/conf/file.conf#L44
https://github.com/apache/incubator-seata/blob/2.x/script/client/spring/application.properties#L74
https://github.com/apache/incubator-seata/blob/2.x/script/client/spring/application.yml#L85
https://github.com/apache/incubator-seata/blob/2.x/script/config-center/config.txt#L35
Since the corresponding IO thread count for the client in the above files is 1, I believe these configurations should be changed to -1. Then, in the code, handle the value of -1 by converting it to the number of CPU cores * 2.

funky-eyes · 2025-02-19T06:18:56Z

core/src/main/java/org/apache/seata/core/rpc/netty/NettyClientConfig.java

 import static org.apache.seata.common.DefaultValues.DEFAULT_RPC_RM_REQUEST_TIMEOUT;
 import static org.apache.seata.common.DefaultValues.DEFAULT_RPC_TM_REQUEST_TIMEOUT;
 import static org.apache.seata.common.DefaultValues.DEFAULT_SELECTOR_THREAD_PREFIX;
-import static org.apache.seata.common.DefaultValues.DEFAULT_SELECTOR_THREAD_SIZE;


The value of DEFAULT_SELECTOR_THREAD_SIZE should be changed to -1

Done in f5d38a9 !

Done in f5d38a9 !

In NettyClientBootstrap, it is necessary to handle the case where selectorThreadSizeThreadSize=-1.

Done in f5d38a9 !

In NettyClientBootstrap, it is necessary to handle the case where selectorThreadSizeThreadSize=-1.

optimize: optimize seata client I/O processing by adjusting thread count #7170 (comment)

What do you think?

codecov · 2025-02-19T06:20:12Z

Codecov Report

Attention: Patch coverage is 50.00000% with 1 line in your changes missing coverage. Please review.

Project coverage is 52.18%. Comparing base (46c9b63) to head (02885a1).
Report is 3 commits behind head on 2.x.

Files with missing lines	Patch %	Lines
...apache/seata/core/rpc/netty/NettyClientConfig.java	50.00%	0 Missing and 1 partial ⚠️

Additional details and impacted files

@@            Coverage Diff            @@
##                2.x    #7170   +/-   ##
=========================================
  Coverage     52.18%   52.18%           
  Complexity     6813     6813           
=========================================
  Files          1154     1154           
  Lines         41112    41113    +1     
  Branches       4818     4819    +1     
=========================================
+ Hits          21455    21456    +1     
+ Misses        17617    17616    -1     
- Partials       2040     2041    +1

Files with missing lines	Coverage Δ
...in/java/org/apache/seata/common/DefaultValues.java	`0.00% <ø> (ø)`
...apache/seata/core/rpc/netty/NettyClientConfig.java	`52.17% <50.00%> (-0.77%)`	⬇️

... and 2 files with indirect coverage changes

YongGoose · 2025-02-19T06:32:49Z

Then, in the code, handle the value of -1 by converting it to the number of CPU cores * 2.

What do you think about changing the thread count to CPU cores * 2 when the value is negative?

funky-eyes · 2025-02-19T06:42:43Z

What do you think about changing the thread count to CPU cores * 2 when the value is negative?

If it is a negative number, directly assign the value of WorkThreadMode.Default.getValue() to it

YongGoose · 2025-02-19T06:43:05Z

core/src/main/java/org/apache/seata/core/rpc/netty/NettyClientConfig.java

+        int threadSize = CONFIG.getInt(ConfigurationKeys.CLIENT_SELECTOR_THREAD_SIZE, WorkThreadMode.Default.getValue());
+        return threadSize > 0 ? threadSize : WorkThreadMode.Default.getValue();


@funky-eyes

Handling this in NettyClientBootstrap is a good idea, but I believe it is more appropriate to process it here.
Because if we handle this in NettyClientBootstrap, we would need to process -1 in every piece of code that retrieves the thread count using this method.

You are right, handling it this way is better, good job.

YongGoose · 2025-02-19T07:41:51Z

@funky-eyes

I have resolved the failing test related to getClientSelectorThreadSize.
However, the testRemoveConfig test in ZkConfigurationTest is still failing, but only when running on Java 8.
Could you help me understand why this failure occurs specifically in Java 8?

funky-eyes · 2025-02-19T07:47:06Z

@funky-eyes

I have resolved the failing test related to getClientSelectorThreadSize. However, the testRemoveConfig test in ZkConfigurationTest is still failing, but only when running on Java 8. Could you help me understand why this failure occurs specifically in Java 8?

This is not caused by your PR; this issue is an occasional occurrence, and the exact cause is currently unknown. I have restarted the CI workflow for now.

…unt (apache#7170)

YongGoose added 2 commits February 19, 2025 15:05

optimize: Optimize seata client I/O processing by adjusting thread count

5c893b4

Add changes

0bbd7e9

funky-eyes approved these changes Feb 19, 2025

View reviewed changes

funky-eyes reviewed Feb 19, 2025

View reviewed changes

Apply comment

e7cdebc

YongGoose force-pushed the feature/7168 branch from f5d38a9 to e7cdebc Compare February 19, 2025 06:40

YongGoose commented Feb 19, 2025

View reviewed changes

YongGoose added 3 commits February 19, 2025 16:01

update test

47877b1

revert and update test

b2df221

update test

02885a1

funky-eyes merged commit aba742d into apache:2.x Feb 19, 2025
8 checks passed

funky-eyes changed the title ~~optimize: Optimize seata client I/O processing by adjusting thread count~~ optimize: optimize seata client I/O processing by adjusting thread count Feb 19, 2025

funky-eyes added this to the 2.4.0 milestone Feb 19, 2025

funky-eyes added module/core core module type: optimize labels Feb 19, 2025

slievrly pushed a commit to slievrly/fescar that referenced this pull request Oct 21, 2025

optimize: optimize seata client I/O processing by adjusting thread co…

bde7e9c

…unt (apache#7170)

YvCeung pushed a commit to YvCeung/incubator-seata that referenced this pull request Dec 25, 2025

optimize: optimize seata client I/O processing by adjusting thread co…

087f29b

…unt (apache#7170)

		int threadSize = CONFIG.getInt(ConfigurationKeys.CLIENT_SELECTOR_THREAD_SIZE, WorkThreadMode.Default.getValue());
		return threadSize > 0 ? threadSize : WorkThreadMode.Default.getValue();

optimize: optimize seata client I/O processing by adjusting thread count #7170

optimize: optimize seata client I/O processing by adjusting thread count #7170

Uh oh!

Conversation

YongGoose commented Feb 19, 2025

Ⅰ. Describe what this PR did

Ⅱ. Does this pull request fix one issue?

Ⅲ. Why don't you add test cases (unit test/integration test)?

Ⅳ. Describe how to verify it

Ⅴ. Special notes for reviews

Uh oh!

funky-eyes left a comment

Choose a reason for hiding this comment

Uh oh!

funky-eyes commented Feb 19, 2025

Uh oh!

funky-eyes Feb 19, 2025

Choose a reason for hiding this comment

Uh oh!

YongGoose Feb 19, 2025

Choose a reason for hiding this comment

Uh oh!

funky-eyes Feb 19, 2025

Choose a reason for hiding this comment

Uh oh!

YongGoose Feb 19, 2025

Choose a reason for hiding this comment

Uh oh!

codecov bot commented Feb 19, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Codecov Report

Uh oh!

YongGoose commented Feb 19, 2025

Uh oh!

funky-eyes commented Feb 19, 2025

Uh oh!

YongGoose Feb 19, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

funky-eyes Feb 19, 2025

Choose a reason for hiding this comment

Uh oh!

YongGoose commented Feb 19, 2025

Uh oh!

funky-eyes commented Feb 19, 2025

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

codecov bot commented Feb 19, 2025 •

edited

Loading

YongGoose Feb 19, 2025 •

edited

Loading