Execution Servers - aces/cbrain GitHub Wiki

Introduction

An execution server or Bourreau is a Rails application. In CBRAIN you can configure as many execution servers as you want, each one corresponding to a computing site where tasks are performed.

Table of contents

  1. Execution server setup
  2. How to create an execution server
  3. Reports

1. Execution server setup

Consult the Bourreau Setup documentation, for information on how to set up an execution server.

2. How to create an execution server

  • Go to the "Servers" tab.
  • Click "Create New Server".
  • Fill in the form:
    • Name: The name of the execution server. Note: The name must also be changed accordingly in the config file Bourreau/config/initializers/config_bourreau.rb for the server to restart properly later on.
    • System 'From' reply address: This is optional. If this is set, then messages sent automatically by the system contain this return adress.
    • Description: The first line is a short description which is shown in the Servers table. After that any special note for the users can be added.
    • Owner: The owner of the execution server.
    • Project: Access to an execution server can be limited to members of a specific project. See the Projects section for more information.
    • Status: The execution server can be "online" or "offline" in CBRAIN. If the execution server is created with "offline" status, then it is not accessible to users.
    • Timeout for is alive check (seconds): Time after which an execution server is considered not alive.
    • Time Zone: The time zone of the execution server.
    • SSH Remote Control Configuration:
      • Hostname: The UNIX hostname where the execution server is installed.
      • Username: The UNIX username on the host.
      • Port Number: This is optional and is usually 22 for SSH. If your SSHD server listens on a different port, then specify it here.
      • Rails Server Directory: The full path where the Bourreau Rails application code is installed, for instance, /home/user/cbrain/Bourreau.
      • Second-level effective host: Sometimes you will have to enter a second level of connection.
    • Tunnelling Configuration:
      • Database Server Remote Tunnel Port: The choice of port number is arbitrary and can be any number between 1024 and 65535. The Bourreau application uses this port on the remote host to connect back to the MySQL server used by BrainPortal. The tunnel is set up automatically, so it is only necessary to make sure this port number is not in use by any other application on the host where the Bourreau runs.
      • ActiveResource Remote Tunnel Port: Again, the choice of port number is arbitrary, but this time the port is open on the BrainPortal side and allows the BrainPortal to send commands the Bourreau side.
    • Cache Management Configuration:
      • Path to Data Provider caches: Each Bourreau needs its own directory to cache data. Create a new empty directory on the Bourreau's host and enter its full path here. As usual, make sure this directory is not shared with anything else and not even used as a cache by any other CBRAIN Rails application. If the Bourreau is on the frontend of a supercomputer, then this directory should be on a filesystem visible from all the compute nodes of that supercomputer.
      • Patterns for filenames to ignore: Enter any particular pattern of filenames to ignore; typically the '.DS_Store' and the '._*' file are ignored.
      • Cache Expiration Timeout
    • Tool Version Configuration: A tool config can only be created for an existing execution server. So create the execution server first and then create the associated tool config afterwards. See the section about Tools for more information.
    • Cluster Management System Configuration:
      • Type of cluster: The Bourreau schedules tasks on a supercomputer cluster. Enter here the type of cluster you have access to on the machine where the Bourreau is installed. Typically, supercomputers have cluster management systems with names like SGE (Sun Grid Engine), Torque or MOAB. UNIX can also be selected, in which case no cluster management system is used and the Bourreau simply launches the tasks as standard UNIX processes.
      • Path to shared work directory: Just like for the cache directory, this is configured with the full path to an empty directory on the Bourreau side. And again, it should not be shared with any other resource. This directory is the location where subdirectories are created for each task launched on this Bourreau. If the Bourreau is on the frontend of a supercomputer, then this directory should be on a filesystem visible from all the compute nodes of that supercomputer.
      • Default queue name: Name od the queue
      • Extra 'qsub' options: Extra option for qsub
    • Bourreau Workers Configuration:
      • Number of Workers: Configure a small number of worker subprocesses that are launched on the Bourreau side to handle the tasks running there. In the original platform there are usually two to four workers for each execution server.
      • Check interval: The interval used by the worker to check for a new task.
      • Log destination: Default is good for production can be changed in development.
      • Log verbosity: Default is good for production can be changed in development.
    • Task Limits: The task limits can only be defined once the execution server is created. Useful when you want to limit the number of active task for a specific execution server.

Note: Original author of this document is Natacha Beck

⚠️ **GitHub.com Fallback** ⚠️