Interface Locker

storage.go:129–151

Locker facilitates synchronization across machines and networks. It essentially provides a distributed named-mutex service so that multiple consumers can coordinate tasks and share resources. If possible, a Locker should implement a coordinated distributed locking mechanism by generating fencing to

// operations across nodes in a cluster; not for internal, extremely
// short-lived, or high-contention locks.
type Locker interface {
	// Lock acquires the lock for name, blocking until the lock
	// can be obtained or an error is returned. Only one lock
	// for the given name can exist at a time. A call to Lock for
	// a name which already exists blocks until the named lock
	// is released or becomes stale.
	//
	// If the named lock represents an idempotent operation, callers
	// should always check to make sure the work still needs to be
	// completed after acquiring the lock. You never know if another
	// process already completed the task while you were waiting to
	// acquire it.
	//
	// Implementations should honor context cancellation.
	Lock(ctx context.Context, name string) error

	// Unlock releases named lock. This method must ONLY be called
	// after a successful call to Lock, and only after the critical
	// section is finished, even if it errored or timed out. Unlock
	// cleans up any resources allocated during Lock. Unlock should
	// only return an error if the lock was unable to be released.
	Unlock(ctx context.Context, name string) error
}

type TryLocker interface {
	// TryLock attempts to acquire the lock for name, and returns a