Keyboard shortcuts

Press ← or → to navigate between chapters

Press S or / to search in the book

Press ? to show this help

Press Esc to hide this help

tfmesos2

tfmesos2 runs TensorFlow tasks as an Apache Mesos framework. A Python process registers the framework, accepts matching resource offers, and starts one container per task. The containers report their port and readiness to the local API; TensorFlow can then use the generated cluster definition.

Goals

  • run parameter-server and worker tasks in parallel on Mesos
  • request CPU, memory, and optional GPU resources per job
  • use Docker containers as the task runtime
  • keep local tests reproducible without a live cluster

This documentation describes the behavior of the current Python implementation. Production values such as credentials, images, and hostnames belong in environment variables or a secret-management system.