Class ClusterSpec

tensorflow/python/training/server_lib.py:242–491

Represents a cluster as a set of "tasks", organized into "jobs". A `tf.train.ClusterSpec` represents the set of processes that participate in a distributed TensorFlow computation. Every `tf.distribute.Server` is constructed in a particular cluster. To create a cluster with two jobs and fiv

@tf_export("train.ClusterSpec")
class ClusterSpec(object):
  """Represents a cluster as a set of "tasks", organized into "jobs".

  A `tf.train.ClusterSpec` represents the set of processes that
  participate in a distributed TensorFlow computation. Every
  `tf.distribute.Server` is constructed in a particular cluster.

  To create a cluster with two jobs and five tasks, you specify the
  mapping from job names to lists of network addresses (typically
  hostname-port pairs).

  ```python
  cluster = tf.train.ClusterSpec({"worker": ["worker0.example.com:2222",
                                             "worker1.example.com:2222",
                                             "worker2.example.com:2222"],
                                  "ps": ["ps0.example.com:2222",
                                         "ps1.example.com:2222"]})
  ```

  Each job may also be specified as a sparse mapping from task indices
  to network addresses. This enables a server to be configured without
  needing to know the identity of (for example) all other worker
  tasks:

  ```python
  cluster = tf.train.ClusterSpec({"worker": {1: "worker1.example.com:2222"},
                                  "ps": ["ps0.example.com:2222",
                                         "ps1.example.com:2222"]})
  ```
  """

  def __init__(self, cluster):
    """Creates a `ClusterSpec`.

    Args:
      cluster: A dictionary mapping one or more job names to (i) a list of
        network addresses, or (ii) a dictionary mapping integer task indices to
        network addresses; or a `tf.train.ClusterDef` protocol buffer.

    Raises:
      TypeError: If `cluster` is not a dictionary mapping strings to lists
        of strings, and not a `tf.train.ClusterDef` protobuf.
    """
    if isinstance(cluster, dict):
      self._cluster_spec = {}
      for job_name, tasks in cluster.items():
        if isinstance(tasks, (list, tuple)):
          job_tasks = {i: task for i, task in enumerate(tasks)}
        elif isinstance(tasks, dict):
          job_tasks = {i: task for i, task in tasks.items()}
        else:
          raise TypeError("The tasks for job %r must be a list or a dictionary "
                          "from integers to strings." % job_name)
        self._cluster_spec[job_name] = job_tasks
      self._make_cluster_def()
    elif isinstance(cluster, cluster_pb2.ClusterDef):
      self._cluster_def = cluster
      self._cluster_spec = {}