Zookeeper

com.loopfor.zookeeper.Zookeeper
See theZookeeper companion object
trait Zookeeper

An instance of a ZooKeeper client.

Attributes

Companion:
object
Graph
Supertypes
class Object
trait Matchable
class Any
Known subtypes

Members list

Concise view

Value members

Abstract methods

Returns a view of this client in which operations are performed ''asynchronously''.

Returns a view of this client in which operations are performed ''asynchronously''.

Attributes

Returns:

an asynchronous view of this client

def close(): Unit

Closes the client connection to the ZooKeeper cluster.

Closes the client connection to the ZooKeeper cluster.

A consequence of closing the connection is that ZooKeeper will expire the corresponding session, which further implies that all ephemeral nodes created by this client will be deleted.

Attributes

def ensure(path: String): Future[Unit]

Ensures that the value of a node, specified by the given path, is synchronized across the ZooKeeper cluster.

Ensures that the value of a node, specified by the given path, is synchronized across the ZooKeeper cluster.

'''An important note on consistency''': ZooKeeper does not guarantee, for any given point in time, that all clients will have a consistent view of the cluster. Since reads can be served by any node in the cluster, whereas writes are serialized through the leader, there exists the possibility in which two separate clients may have inconsistent views. This scenario occurs when the leader commits a change once consensus is reached, but the change has not yet propagated across the cluster. Therefore, reads occurring before the commit has propagated will be globally inconsistent. This behavior is normally acceptable, but for some use cases, writes may need to be globally visible before subsequent reads occur.

This method is particularly useful when a ''write'' occurring in one process is followed by a ''read'' in another process. For example, consider the following sequence of operations:

  • process A writes a value
  • process A sends a message to process B
  • process B reads the value

The assumption is that process B expects to see the value written by process A, but as mentioned, ZooKeeper does not make this guarantee. A call to this method before process B attempts to read the value ensures that all prior writes are consistently applied across the cluster, thus observing the write in process A.

Attributes

Returns:

a future that completes when the node is synchronized across the cluster

Returns the session associated with ZooKeeper.

Returns the session associated with ZooKeeper.

Note that the connection with ZooKeeper is established asynchronously, which means there is no implied guarantee that the session has connected nor will eventually connect at the time that an instance of this trait is created.

Attributes

Returns:

the session associated with ZooKeeper

Returns a view of this client in which operations are performed ''synchronously''.

Returns a view of this client in which operations are performed ''synchronously''.

Attributes

Returns:

a synchronous view of this client