feat: add --strict-ports option for predictable port assignment by yang1002378395-cmyk · Pull Request #21320 · sgl-project/sglang

yang1002378395-cmyk · 2026-03-24T14:44:31Z

Summary

Add --strict-ports option to disable automatic port selection
When enabled, server fails immediately if requested ports are unavailable
Ensures predictable port assignment for users who need to know exactly which ports will be used

Problem

Currently, when a requested port is occupied, settle_port automatically selects an available port. This causes issues for users who:

Don't know which port the server ended up using
Cannot reliably create clients with the same configuration
Have port conflicts across multiple server instances

Solution

Add --strict-ports option. When enabled:

Check if requested ports are available (HTTP, scheduler, master)
Fail with clear error message if any port is unavailable
Do NOT auto-select alternative ports

Usage

# Normal mode (default): auto-select ports if unavailable
python -m sglang.launch_server --model <model> --port 8080

# Strict mode: fail if port 8080 is unavailable
python -m sglang.launch_server --model <model> --port 8080 --strict-ports

Related to #21284

Test Plan

Start server with occupied port + --strict-ports → should fail with clear error
Start server with available port + --strict-ports → should succeed
Default behavior (no --strict-ports) → should auto-select ports as before

…ound This allows proper fallback to diffusers backend when native config is not available for a model. Fixes sgl-project#21311

When --strict-ports is enabled, the server will fail immediately if the requested ports are unavailable instead of automatically selecting alternative ports. This ensures predictable port assignment for users who need to know exactly which ports will be used. Related to sgl-project#21284

gemini-code-assist · 2026-03-24T14:44:51Z

Summary of Changes

Hello, I'm Gemini Code Assist¹! I'm currently reviewing this pull request and will post my feedback shortly. In the meantime, here's a summary to help you and other reviewers quickly get up to speed!

This pull request introduces a significant enhancement to the server's port management by adding a --strict-ports option. This feature addresses the problem of unpredictable port assignments that occurred when requested ports were already in use, which previously led to automatic port selection. With this change, users can now enforce exact port configurations, as the server will explicitly check for port availability and terminate with an error if any specified port is occupied, thereby ensuring reliable and consistent server deployments.

Highlights

New --strict-ports option: Added a new command-line option --strict-ports to control server port assignment behavior.
Predictable Port Assignment: When --strict-ports is enabled, the server will fail immediately if any requested port (HTTP, scheduler, master) is unavailable, ensuring predictable port assignments and preventing automatic port selection.
Improved Error Handling: Introduced clear error messages when ports are unavailable in strict mode, guiding users on how to resolve port conflicts.

🧠 New Feature in Public Preview: You can now enable Memory to help Gemini Code Assist learn from your team's feedback. This makes future code reviews more consistent and personalized to your project's style. Click here to enable Memory in your admin console.

Using Gemini Code Assist

The full guide for Gemini Code Assist can be found on our documentation page, here are some quick tips.

Invoking Gemini

You can request assistance from Gemini at any point by creating a comment using either /gemini <command> or @gemini-code-assist <command>. Below is a summary of the supported commands on the current page.

Feature	Command	Description
Code Review	`/gemini review`	Performs a code review for the current pull request in its current state.
Pull Request Summary	`/gemini summary`	Provides a summary of the current pull request in its current state.
Comment	@gemini-code-assist	Responds in comments when explicitly tagged, both in pull request comments and review comments.
Help	`/gemini help`	Displays a list of available commands.

Customization

To customize Gemini Code Assist for GitHub experience, repository maintainers can create a configuration file and/or provide a custom code review style guide (such as PEP-8 for Python) by creating and adding files to a .gemini/ folder in the base of the repository. Detailed instructions can be found here.

Limitations & Feedback

Gemini Code Assist may make mistakes. Please leave feedback on any instances where its feedback is incorrect or counter productive. You can react with 👍 and 👎 on @gemini-code-assist comments. If you're interested in giving your feedback about your experience with Gemini Code Assist for GitHub and other Google products, sign up here.

Review the Privacy Notices, Generative AI Prohibited Use Policy, Terms of Service, and learn how to configure Gemini Code Assist in GitHub here. Gemini can make mistakes, so double check it and use code with caution. ↩

gemini-code-assist

Code Review

This pull request introduces a new strict_ports option, which, when enabled, causes the server to fail if requested ports are unavailable instead of auto-selecting them. It also modifies the model registry to log a debug message and return None instead of raising a RuntimeError when model information is not found. The review suggests that changing the RuntimeError to a debug log in the registry might hide critical configuration issues and recommends reverting to an exception. Additionally, it recommends making the error messages for scheduler_port and master_port more consistent and helpful when strict_ports is enabled.

gemini-code-assist · 2026-03-24T14:50:05Z

+        logger.debug(f"No model info found for model path: {model_path}")
+        return None


Changing RuntimeError to logger.debug and returning None can hide potential configuration issues. If model info is not found for a given model_path, it's likely an error that should be surfaced to the user, rather than being silently ignored at the debug log level. This change could lead to hard-to-debug NoneType errors later in the execution flow. It's better to fail fast. Please consider reverting to raising an exception.

Suggested change

logger.debug(f"No model info found for model path: {model_path}")

return None

raise RuntimeError(f"No model info found for model path: {model_path}")

gemini-code-assist · 2026-03-24T14:50:06Z

+            if not is_port_available(self.scheduler_port):
+                raise RuntimeError(
+                    f"Scheduler port {self.scheduler_port} is unavailable and --strict-ports is enabled."
+                )
+            if self.master_port is not None and not is_port_available(self.master_port):
+                raise RuntimeError(
+                    f"Master port {self.master_port} is unavailable and --strict-ports is enabled."
+                )


The error messages for unavailable scheduler_port and master_port are not as helpful as the one for the main port. For consistency and better user experience, consider including the suggestion to 'Either use a different port or remove --strict-ports to allow auto-selection' in all error messages when strict_ports is enabled.

Suggested change

if not is_port_available(self.scheduler_port):

raise RuntimeError(

f"Scheduler port {self.scheduler_port} is unavailable and --strict-ports is enabled."

)

if self.master_port is not None and not is_port_available(self.master_port):

raise RuntimeError(

f"Master port {self.master_port} is unavailable and --strict-ports is enabled."

)

if not is_port_available(self.scheduler_port):

raise RuntimeError(

f"Scheduler port {self.scheduler_port} is unavailable and --strict-ports is enabled. "

f"Either use a different port or remove --strict-ports to allow auto-selection."

)

if self.master_port is not None and not is_port_available(self.master_port):

raise RuntimeError(

f"Master port {self.master_port} is unavailable and --strict-ports is enabled. "

f"Either use a different port or remove --strict-ports to allow auto-selection."

)

- Reverted registry.py to raise RuntimeError instead of returning None to avoid hiding critical configuration issues - Made strict_ports error messages consistent across all ports - Added helpful hints for resolving port conflicts

mickqian · 2026-03-27T06:45:48Z

/tag-and-rerun-ci

yhyang201 · 2026-03-27T06:56:50Z

/tag-and-rerun-ci

…gnment (sgl-project#21320) Co-authored-by: 阳虎 <yanghu@yanghudeMacBook-Pro.local>

yang1002378395-cmyk added 2 commits March 24, 2026 22:35

fix: return None instead of raising RuntimeError when no model info f…

d69c39b

…ound This allows proper fallback to diffusers backend when native config is not available for a model. Fixes sgl-project#21311

yang1002378395-cmyk requested review from mickqian, ping1jing2 and yhyang201 as code owners March 24, 2026 14:44

github-actions Bot added the diffusion SGLang Diffusion label Mar 24, 2026

gemini-code-assist Bot reviewed Mar 24, 2026

View reviewed changes

yang1002378395-cmyk mentioned this pull request Mar 24, 2026

[diffusion] It is may be not a good design to settle_port dynamic #21284

Closed

mickqian approved these changes Mar 27, 2026

View reviewed changes

github-actions Bot added the run-ci label Mar 27, 2026

mickqian merged commit f83b1b7 into sgl-project:main Mar 27, 2026
104 of 118 checks passed

satyamk7054 pushed a commit to satyamk7054/sglang that referenced this pull request Apr 3, 2026

[diffusion] feat: add --strict-ports option for predictable port assi…

350932f

…gnment (sgl-project#21320) Co-authored-by: 阳虎 <yanghu@yanghudeMacBook-Pro.local>

JustinTong0323 pushed a commit to JustinTong0323/sglang that referenced this pull request Apr 7, 2026

[diffusion] feat: add --strict-ports option for predictable port assi…

cb5d58c

…gnment (sgl-project#21320) Co-authored-by: 阳虎 <yanghu@yanghudeMacBook-Pro.local>

yhyang201 pushed a commit to yhyang201/sglang that referenced this pull request Apr 22, 2026

[diffusion] feat: add --strict-ports option for predictable port assi…

52c9fd4

…gnment (sgl-project#21320) Co-authored-by: 阳虎 <yanghu@yanghudeMacBook-Pro.local>

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

feat: add --strict-ports option for predictable port assignment#21320

feat: add --strict-ports option for predictable port assignment#21320
mickqian merged 3 commits intosgl-project:mainfrom
yang1002378395-cmyk:feat-strict-ports

yang1002378395-cmyk commented Mar 24, 2026

Uh oh!

gemini-code-assist Bot commented Mar 24, 2026

Uh oh!

gemini-code-assist Bot left a comment

Uh oh!

gemini-code-assist Bot Mar 24, 2026

Uh oh!

gemini-code-assist Bot Mar 24, 2026

Uh oh!

mickqian commented Mar 27, 2026

Uh oh!

yhyang201 commented Mar 27, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

		logger.debug(f"No model info found for model path: {model_path}")
		return None

	logger.debug(f"No model info found for model path: {model_path}")
	return None
	raise RuntimeError(f"No model info found for model path: {model_path}")

Conversation

yang1002378395-cmyk commented Mar 24, 2026

Summary

Problem

Solution

Usage

Test Plan

Uh oh!

gemini-code-assist Bot commented Mar 24, 2026

Summary of Changes

Highlights

Footnotes

Uh oh!

gemini-code-assist Bot left a comment

Choose a reason for hiding this comment

Code Review

Uh oh!

gemini-code-assist Bot Mar 24, 2026

Choose a reason for hiding this comment

Uh oh!

gemini-code-assist Bot Mar 24, 2026

Choose a reason for hiding this comment

Uh oh!

mickqian commented Mar 27, 2026

Uh oh!

yhyang201 commented Mar 27, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants