[Core] Rationalize boolean environment variable handling #19550

art-dsit · 2025-06-12T12:31:00Z

Essential Elements of an Effective PR Description Checklist

The purpose of the PR, such as "Fix some issue (link existing issues this PR will resolve)".
The test plan, such as providing test command.
The test results, such as pasting the results comparison before and after, or e2e results
(Optional) The necessary documentation update, such as updating supported_models.md and examples for a new model.

Purpose

The handling of boolean env vars is confusing at the moment. For example

VLLM_DEBUG_LOG_API_SERVER_RESPONSE must be set to "true" (or "TRUE", "True", "tRuE" etc) - "1" won't work

VLLM_USE_V1 must be set to "1" - "True" won't work

I've added parse_bool_env and changed all the handlers to use it.

Test Plan

Test Result

(Optional) Documentation Update

github-actions · 2025-06-12T12:31:12Z

👋 Hi! Thank you for contributing to the vLLM project.

💬 Join our developer Slack at https://slack.vllm.ai to discuss your PR in #pr-reviews, coordinate on features in #feat- channels, or join special interest groups in #sig- channels.

Just a reminder: PRs would not trigger full CI run by default. Instead, it would only run fastcheck CI which starts running only a small and essential subset of CI tests to quickly catch errors. You can run other CI tests on top of those by going to your fastcheck build on Buildkite UI (linked in the PR checks section) and unblock them. If you do not have permission to unblock, ping simon-mo or khluu to add you in our Buildkite org.

Once the PR is approved and ready to go, your PR reviewer(s) can run CI to test the changes comprehensively before merging.

To run CI, PR reviewers can either: Add ready label to the PR or enable auto-merge.

🚀

gemini-code-assist

Summary of Changes

Hello @art-dsit, 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 addresses the current inconsistency in handling boolean environment variables by introducing a dedicated helper function. This function provides a single, clear way to parse boolean values from the environment, improving code readability and maintainability by replacing varied parsing methods with a consistent approach across the codebase.

Highlights

Consistent Boolean Parsing: Introduced a new helper function parse_bool_env in vllm/envs.py to standardize how boolean environment variables are interpreted. This function accepts case-insensitive "true" or "1" as True and uses a specified default otherwise.
Refactor Environment Variable Handling: Updated numerous existing environment variable parsing lambdas within the _ENV_VARS dictionary to utilize the new parse_bool_env helper, replacing inconsistent ad-hoc parsing logic (e.g., bool(int(...)), .lower() == "true").

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 in your pull request via creating an issue comment (i.e. comment on the pull request page) using either /gemini <command> or @gemini-code-assist <command>. Below is a summary of the supported commands.

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 issue 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 is currently in preview and 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 to provide feedback.

You can also get AI-powered code generation, chat, as well as code reviews directly in the IDE at no cost with the Gemini Code Assist IDE Extension.

Review the Privacy Notices, Generative AI Prohibited Use Policy, Terms of Service, and learn how to configureGemini 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 helper function parse_bool_env to standardize the parsing of boolean environment variables. This is a good step towards improving code consistency and maintainability. The changes update numerous environment variable handlers to use this new function. Note that this standardization changes the exact set of string values interpreted as boolean for some variables, which is the intended rationalization but might affect users relying on previous, less consistent behavior.

gemini-code-assist · 2025-06-12T12:32:27Z

vllm/envs.py

    "VLLM_TEST_DYNAMO_FULLGRAPH_CAPTURE":
-    lambda: bool(
-        os.environ.get("VLLM_TEST_DYNAMO_FULLGRAPH_CAPTURE", "1") != "0"),
+    lambda: parse_bool_env("VLLM_TEST_DYNAMO_FULLGRAPH_CAPTURE", True),

    # Feature flag to enable/disable Inductor standalone compile.
    # In torch <= 2.7 we ignore this flag; in torch >= 2.8 this is
    # enabled by default.
    "VLLM_USE_STANDALONE_COMPILE":
-    lambda: os.environ.get("VLLM_USE_STANDALONE_COMPILE", "1") == "1",
+    lambda: parse_bool_env("VLLM_USE_STANDALONE_COMPILE", True),


The original logic for VLLM_USE_STANDALONE_COMPILE treated only the exact string "1" as True (when unset, it defaulted to "1" which was then compared to "1"). The new parse_bool_env function with a default of True will treat unset, "1", and "true" as True, and "0", "false", and other strings as False. This is a change in behavior for string values other than "1". While standardizing to "1"/"true" is good, this change in behavior for non-"1" strings should be noted, as users might have relied on the old behavior.

gemini-code-assist · 2025-06-12T12:32:27Z

vllm/envs.py

    # If set, the OpenAI API server will stay alive even after the underlying
    # AsyncLLMEngine errors and stops serving requests
    "VLLM_KEEP_ALIVE_ON_ENGINE_DEATH":
-    lambda: bool(os.getenv("VLLM_KEEP_ALIVE_ON_ENGINE_DEATH", 0)),
+    lambda: parse_bool_env("VLLM_KEEP_ALIVE_ON_ENGINE_DEATH"),


The original logic for VLLM_KEEP_ALIVE_ON_ENGINE_DEATH used bool(os.getenv("...")) with a default of 0. This would treat any non-empty string (including "0", "false", "true", etc.) as True, and only None or 0 as False. The new parse_bool_env function with a default of False will only treat "1" and "true" as True, and everything else (including other non-empty strings) as False. This is a significant change in how various string values are interpreted.

gemini-code-assist · 2025-06-12T12:32:27Z

vllm/envs.py

    # If this env var is set to 1, vLLM will skip the peer-to-peer check,
    # and trust the driver's peer-to-peer capability report.
    "VLLM_SKIP_P2P_CHECK":
-    lambda: os.getenv("VLLM_SKIP_P2P_CHECK", "0") == "1",
+    lambda: parse_bool_env("VLLM_SKIP_P2P_CHECK"),


The original logic for VLLM_SKIP_P2P_CHECK treated only the exact string "1" as True. The new parse_bool_env function will treat "1" and "true" as True. This is a change in behavior for the string value "true".

gemini-code-assist · 2025-06-12T12:32:28Z

vllm/envs.py

    # Gaudi3. This is only applicable to HPU contiguous cache. If set to true,
    # contiguous cache fetch will be used.
    "VLLM_USE_HPU_CONTIGUOUS_CACHE_FETCH":
-    lambda: os.environ.get("VLLM_CONTIGUOUS_PA", "true").lower() in
-    ("1", "true"),
+    lambda: parse_bool_env("VLLM_CONTIGUOUS_PA", True),


The original logic for VLLM_USE_HPU_CONTIGUOUS_CACHE_FETCH treated "1" and "true" (case-insensitive) as True, and other strings like "0" or "false" as True (due to the default "true" and the in check). The new parse_bool_env function with a default of True will treat unset, "1", and "true" as True, and "0", "false", and other strings as False. This changes the behavior for "0" and "false".

gemini-code-assist · 2025-06-12T12:32:28Z

vllm/envs.py

+def parse_bool_env(name: str, default: bool = False) -> bool:
+    """Parse a boolean environment variable consistently.
+
+    Accepts "1" or "true" (case-insensitive) as True,
+    and anything else (including blank/unset) as the supplied default value.
+    """
+    value = os.environ.get(name)
+    if value is None:
+        return default
+    return value.lower() in ("1", "true")


This new helper function is a great addition for standardizing how boolean environment variables are parsed. It significantly improves consistency and maintainability.

gemini-code-assist · 2025-06-12T12:32:28Z

vllm/envs.py

    # If true, will load models from ModelScope instead of Hugging Face Hub.
    # note that the value is true or false, not numbers
    "VLLM_USE_MODELSCOPE":
-    lambda: os.environ.get("VLLM_USE_MODELSCOPE", "False").lower() == "true",
+    lambda: parse_bool_env("VLLM_USE_MODELSCOPE"),


The comment for VLLM_USE_MODELSCOPE mentions that the value is "true" or "false", not numbers. However, the new parse_bool_env function also accepts "1" as True. Consider updating the comment to reflect this.

Suggested change

# If true, will load models from ModelScope instead of Hugging Face Hub.

# note that the value is true or false, not numbers

"VLLM_USE_MODELSCOPE":

lambda: os.environ.get("VLLM_USE_MODELSCOPE", "False").lower() == "true",

lambda: parse_bool_env("VLLM_USE_MODELSCOPE"),

# If true, will load models from ModelScope instead of Hugging Face Hub.

# Accepts "1" or "true" (case-insensitive).

"VLLM_USE_MODELSCOPE":

lambda: parse_bool_env("VLLM_USE_MODELSCOPE"),

gemini-code-assist · 2025-06-12T12:32:28Z

vllm/envs.py


    # Whether to log responses from API Server for debugging
    "VLLM_DEBUG_LOG_API_SERVER_RESPONSE":
-    lambda: os.environ.get("VLLM_DEBUG_LOG_API_SERVER_RESPONSE", "False"
-                           ).lower() == "true",
+    lambda: parse_bool_env("VLLM_DEBUG_LOG_API_SERVER_RESPONSE"),


Similar to VLLM_USE_MODELSCOPE, the comment for VLLM_DEBUG_LOG_API_SERVER_RESPONSE implies only "true" is accepted, but "1" is now also accepted. Consider updating the comment.

Suggested change

# Whether to log responses from API Server for debugging

"VLLM_DEBUG_LOG_API_SERVER_RESPONSE":

lambda: os.environ.get("VLLM_DEBUG_LOG_API_SERVER_RESPONSE", "False"

).lower() == "true",

lambda: parse_bool_env("VLLM_DEBUG_LOG_API_SERVER_RESPONSE"),

# Whether to log responses from API Server for debugging

# Accepts "1" or "true" (case-insensitive).

"VLLM_DEBUG_LOG_API_SERVER_RESPONSE":

lambda: parse_bool_env("VLLM_DEBUG_LOG_API_SERVER_RESPONSE"),

Signed-off-by: Art.OCathain <[email protected]>

mergify · 2025-06-19T04:55:39Z

This pull request has merge conflicts that must be resolved before it can be
merged. Please rebase the PR, @art-dsit.

https://docs.github.com/en/pull-requests/collaborating-with-pull-requests/working-with-forks/syncing-a-fork

gemini-code-assist bot reviewed Jun 12, 2025

View reviewed changes

art-dsit force-pushed the bool_env_vars branch 3 times, most recently from 3870229 to 9f64589 Compare June 12, 2025 12:34

Rationalize boolean environment variable handling

4db83ac

Signed-off-by: Art.OCathain <[email protected]>

art-dsit force-pushed the bool_env_vars branch from 9f64589 to 4db83ac Compare June 12, 2025 12:46

art-dsit changed the title ~~Rationalize boolean environment variable handling~~ [core] Rationalize boolean environment variable handling Jun 12, 2025

art-dsit changed the title ~~[core] Rationalize boolean environment variable handling~~ [Core] Rationalize boolean environment variable handling Jun 12, 2025

DarkLight1337 requested a review from youkaichao June 13, 2025 05:30

mergify bot added the needs-rebase label Jun 19, 2025

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

Uh oh!

[Core] Rationalize boolean environment variable handling #19550

[Core] Rationalize boolean environment variable handling #19550

Uh oh!

art-dsit commented Jun 12, 2025 •

edited

Loading

Uh oh!

github-actions bot commented Jun 12, 2025

Uh oh!

gemini-code-assist bot left a comment

Uh oh!

gemini-code-assist bot left a comment

Uh oh!

gemini-code-assist bot Jun 12, 2025

Uh oh!

gemini-code-assist bot Jun 12, 2025

Uh oh!

gemini-code-assist bot Jun 12, 2025

Uh oh!

gemini-code-assist bot Jun 12, 2025

Uh oh!

gemini-code-assist bot Jun 12, 2025

Uh oh!

gemini-code-assist bot Jun 12, 2025

Uh oh!

gemini-code-assist bot Jun 12, 2025

Uh oh!

mergify bot commented Jun 19, 2025

Uh oh!

Uh oh!

Uh oh!

[Core] Rationalize boolean environment variable handling #19550

Are you sure you want to change the base?

[Core] Rationalize boolean environment variable handling #19550

Uh oh!

Conversation

art-dsit commented Jun 12, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Essential Elements of an Effective PR Description Checklist

Purpose

Test Plan

Test Result

(Optional) Documentation Update

Uh oh!

github-actions bot commented Jun 12, 2025

Uh oh!

gemini-code-assist bot left a comment

Choose a reason for hiding this comment

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 Jun 12, 2025

Choose a reason for hiding this comment

Uh oh!

gemini-code-assist bot Jun 12, 2025

Choose a reason for hiding this comment

Uh oh!

gemini-code-assist bot Jun 12, 2025

Choose a reason for hiding this comment

Uh oh!

gemini-code-assist bot Jun 12, 2025

Choose a reason for hiding this comment

Uh oh!

gemini-code-assist bot Jun 12, 2025

Choose a reason for hiding this comment

Uh oh!

gemini-code-assist bot Jun 12, 2025

Choose a reason for hiding this comment

Uh oh!

gemini-code-assist bot Jun 12, 2025

Choose a reason for hiding this comment

Uh oh!

mergify bot commented Jun 19, 2025

Uh oh!

Uh oh!

art-dsit commented Jun 12, 2025 •

edited

Loading