Remote Backend State Locking and Ghost Lock Resolution

Terraform utilizes state locking in remote backends, such as S3 with DynamoDB, to prevent concurrent modifications that could lead to state corruption. When a process terminates unexpectedly or a network partition occurs between the client and the lock provider, a "ghost lock" may persist in the backend despite no active operation.

The terraform force-unlock command allows for the manual release of these locks using a specific Lock ID. However, there is a critical risk of permanent state corruption if the lock is released while a hidden process is still actively writing to the state file.

• What is the recommended method for verifying that a process is truly defunct before executing a force-unlock?
• Are there backend-specific timeout configurations to automatically expire locks without manual intervention?