Fix excessive logging noise by reducing the log level of the MinimaxM2ToolParser import success message

[minatoaquaMK2]

Purpose

Fix excessive logging noise in online serving by reducing the log level of the MinimaxM2ToolParser import success message from INFO to DEBUG.

Problem:
When running vLLM online serving with the Minimax M2 model, the tool parser import success message was logged at INFO level on every request, causing significant log pollution. This made it difficult to monitor important system metrics and request logs.

Example of the issue:

(APIServer pid=7) INFO 10-27 23:02:11 [loggers.py:208] Engine 000: Avg prompt throughput: 0.0 tokens/s, Avg generation throughput: 30.6 tokens/s, Running: 1 reqs, Waiting: 0 reqs, GPU KV cache usage: 5.0%, Prefix cache hit rate: 57.4%
(APIServer pid=7) INFO 10-27 23:02:12 [minimax_m2_tool_parser.py:97] vLLM Successfully import tool parser MinimaxM2ToolParser !
(APIServer pid=7) INFO:     192.168.201.117:55940 - "POST /v1/chat/completions HTTP/1.1" 200 OK
(APIServer pid=7) INFO 10-27 23:02:12 [minimax_m2_tool_parser.py:97] vLLM Successfully import tool parser MinimaxM2ToolParser !
(APIServer pid=7) INFO 10-27 23:02:14 [minimax_m2_tool_parser.py:97] vLLM Successfully import tool parser MinimaxM2ToolParser !
(APIServer pid=7) INFO:     192.168.201.117:55940 - "POST /v1/chat/completions HTTP/1.1" 200 OK
(APIServer pid=7) INFO 10-27 23:02:15 [minimax_m2_tool_parser.py:97] vLLM Successfully import tool parser MinimaxM2ToolParser !

Solution:
Changed the log level from logger.info() to logger.debug() for the tool parser import success message. This message is only useful for debugging purposes and should not clutter the INFO-level logs during normal operation.

Test Plan

1. Start vLLM online serving with Minimax M2 model
2. Send multiple chat completion requests
3. Verify that the tool parser import message no longer appears in INFO-level logs
4. Verify that the message still appears when running with DEBUG log level enabled

Test Result

Before the fix:

• Tool parser import message appeared in logs on every request at INFO level
• Logs were cluttered with repetitive messages, making it hard to track important metrics

After the fix:

• Tool parser import message no longer appears in default INFO-level logs
• Logs are clean and only show relevant operational information (throughput metrics, request logs, etc.)
• Message is still available when debugging with DEBUG log level

---

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.
• (Optional) Release notes update. If your change is user facing, please update the release notes draft in the Google Doc.

[gemini-code-assist]

Code Review

This pull request correctly addresses excessive logging by changing a log level from INFO to DEBUG. This is a good improvement. However, this change also reveals a more significant performance issue where the MinimaxM2ToolParser is re-initialized on every request, leading to repeated expensive operations like regex compilation. I've added a comment with a suggestion to refactor the class for better performance by separating one-time initializations from per-request state.

[minatoaquaMK2]

Hi @xuebi

Since you are the author of minimax_m2_tool_parser.py, I was hoping you might have a moment to review this PR?

Thanks in advance!

[bbrowning]

The change is reasonable. That log line could probably go away entirely, but will defer to others if there's some reason this is here at all. Good job on pushing back on the Gemini suggestion about instantiating a parser on every request, as we do that by design and your analysis was accurate there.

[youkaichao]

LGTM