7. How Things Work

This section explains how some of the functions inside PRRTE perform their job. If you have questions about, for example, how session directory trees work, then you will find information on that subject here.

• 7.1. Session Directories
  • 7.1.1. Client Session Directories
  • 7.1.2. Tool and Server Session Directories
• 7.2. Per-App-Context Mapping
  • 7.2.1. When to Use It
  • 7.2.2. Command-Line Syntax
  • 7.2.3. Job-Level-Only Directives
  • 7.2.4. Per-App NOLOCAL
  • 7.2.5. PMIx Spawn Path
  • 7.2.6. Inheritance and Fallback
  • 7.2.7. How the Dispatch Works
  • 7.2.8. Attribute Storage
  • 7.2.9. Framework Version
• 7.3. Scheduler Integration
  • 7.3.1. Overview
  • 7.3.2. Objects and Definitions
  • 7.3.3. Flow-of-Control
• 7.4. Job Launch State Machine
  • 7.4.1. Architecture
  • 7.4.2. Job State Definitions
  • 7.4.3. The Daemon Launch Sequence
  • 7.4.4. Process State Machine
  • 7.4.5. Termination and Error States
  • 7.4.6. Key Source Files
  • 7.4.7. Debugging
  • 7.4.8. DVM Extension and the Daemon-Launch Race
• 7.5. The Runtime Messaging Layer (RML)
  • 7.5.1. The OOB TCP Transport
  • 7.5.2. Reliable Messaging (RELM)
  • 7.5.3. Historical note: one directory, once three frameworks
  • 7.5.4. Architecture and key state
  • 7.5.5. Sending a message
  • 7.5.6. Receiving, matching, and relaying
  • 7.5.7. Routing: the radix tree
  • 7.5.8. Fault tolerance and reliable messaging
  • 7.5.9. Elastic growth and launcher-less bootstrap
  • 7.5.10. Where to look