Fix: get_stats blocks main thread on storage failure

[UranusSeven]

Hi from novita.ai team 👋

Motivation

When using HiCache with 3fs backend, the main scheduler thread could become blocked when collecting storage metrics via get_stats(). This issue occurs due to lock contention between multiple threads:

Backup thread: Holds the lock during long-running filesystem I/O operations in _batch_set() (decorated with @synchronized())
Main thread: Periodically calls get_stats() (also decorated with @synchronized()) to collect storage metrics for monitoring

When storage is slow or down, the backup thread holds the lock for extended periods, causing the main thread to block when trying to acquire the same lock.

Modifications

Changed get_stats() to use non-blocking lock acquisition: uses self.lock.acquire(blocking=False) instead of the @synchronized() decorator.

Accuracy Tests

Benchmarking and Profiling

Checklist

• Format your code according to the Format code with pre-commit.
• Add unit tests according to the Run and add unit tests.
• Update documentation according to Write documentations.
• Provide accuracy and speed benchmark results according to Test the accuracy and Benchmark the speed.
• Follow the SGLang code style guidance.
• Work with maintainers to merge your PR. See the PR Merge Process

[gemini-code-assist]

Warning

You have reached your daily quota limit. Please wait up to 24 hours and I will start processing your requests again!

[ShangmingCai]

Looks reasonable to me. cc: @xiezhq-hermann what do you think

[hzh0425]

This issue can be resolved by this PR. Thank you! @UranusSeven

#13407

[ShangmingCai]

Feel free to reopen this PR if #13407 is not sufficient.