Skip to content

fix(vllm): accept --disable-log-requests as no-op for vLLM 0.17+ compat #783

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
lilyz-ai wants to merge 1 commit into
base: main
Choose a base branch
from
Open

fix(vllm): accept --disable-log-requests as no-op for vLLM 0.17+ compat #783

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
11 changes: 11 additions & 0 deletions model-engine/model_engine_server/inference/vllm/vllm_server.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -23,6 +23,17 @@

def parse_args(parser: FlexibleArgumentParser):
parser = make_arg_parser(parser)
# Backward compatibility: older model-engine versions pass --disable-log-requests
# which was removed from vLLM's arg parser in v0.17+. Accept it as a no-op.
if not any(
"--disable-log-requests" in getattr(a, "option_strings", []) for a in parser._actions
):
parser.add_argument(
"--disable-log-requests",
action="store_true",
default=False,
help="(deprecated, no-op) Kept for backward compatibility with older model-engine versions.",
)
Comment on lines +26 to +36
Copy link

@greptile-apps greptile-apps bot Mar 19, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

P1 Silent no-op may break sensitive_log_mode compliance guarantees

The comment says "older model-engine versions pass --disable-log-requests", but the current model-engine code (llm_model_endpoint_use_cases.py:996-997) still hardcodes this flag whenever hmi_config.sensitive_log_mode is True:

if hmi_config.sensitive_log_mode:
    vllm_args.disable_log_requests = True

By silently accepting --disable-log-requests as a no-op in vLLM 0.17+, request logging will not be disabled for endpoints that have sensitive_log_mode set. If sensitive_log_mode is used for privacy or compliance reasons, this silent behavioral regression could expose request-level logs that were previously suppressed.

It would be worth either:

  1. Confirming that vLLM 0.17+ disables request logging by default (or via a different mechanism) and documenting that here, or
  2. Identifying the replacement mechanism in vLLM 0.17+ (e.g., a new flag or VLLM_CONFIGURE_LOGGING env var) and wiring sensitive_log_mode through that path alongside this no-op shim.
Prompt To Fix With AI 
This is a comment left during a code review.
Path: model-engine/model_engine_server/inference/vllm/vllm_server.py
Line: 26-36

Comment:
**Silent no-op may break `sensitive_log_mode` compliance guarantees**

The comment says "older model-engine versions pass `--disable-log-requests`", but the *current* model-engine code (`llm_model_endpoint_use_cases.py:996-997`) still hardcodes this flag whenever `hmi_config.sensitive_log_mode` is `True`:

```python
if hmi_config.sensitive_log_mode:
    vllm_args.disable_log_requests = True
```

By silently accepting `--disable-log-requests` as a no-op in vLLM 0.17+, request logging will **not** be disabled for endpoints that have `sensitive_log_mode` set. If `sensitive_log_mode` is used for privacy or compliance reasons, this silent behavioral regression could expose request-level logs that were previously suppressed.

It would be worth either:
1. Confirming that vLLM 0.17+ disables request logging by default (or via a different mechanism) and documenting that here, or
2. Identifying the replacement mechanism in vLLM 0.17+ (e.g., a new flag or `VLLM_CONFIGURE_LOGGING` env var) and wiring `sensitive_log_mode` through that path alongside this no-op shim.

How can I resolve this? If you propose a fix, please make it concise.

return parser.parse_args()


Expand Down