Scheduling main loop blocked and timeout due to un-released PreBind lock in Volcano

[guoqinwill]

Description

Volcano performs asynchronous Pod binding. After integrating with DRA, the PreBind phase introduces a lock that may not be released if an external service call hangs (default timeout: 600s), causing the main scheduling workflow to block when loading predicate plugins until the timeout releases the lock.

Steps to reproduce the issue

1. Create a PVC, but specify an unavailable availability zone.
2. Create a Pod that binds to this PVC.

Describe the results you received and expected

The Pod remains in the Pending state because PV binding fails. However, Volcano’s session cycle time remains normal, and subsequent Pod scheduling latency is also normal.

What version of Volcano are you using?

1.12

Any other relevant information

No response