Skip to content
This repository has been archived by the owner on Sep 23, 2020. It is now read-only.

nimbusproject/processregistry

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

13 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

⚠️ The Nimbus infrastructure project is no longer under development. ⚠️

For more information, please read the news announcement. If you are interested in providing IaaS capabilities to the scientific community, see CHI-in-a-Box, a packaging of the Chameleon testbed, which has been in development since 2014.


Process Registry

A basic prototype Process Registry.

Installation

Use the setup.py to install. The settings.py file is configured to use sqlite by default. If you want to use mysql, refer to the standard instructions for databases (https://docs.djangoproject.com/en/dev/ref/settings/#databases) on the Django documentation site.

Usage

To query, the process registry, you can do GET/POST/PUT/DELETE against

/api/process_definition/ 

to do standard Create/Update/Read/Delete operations.

For example to get a list of records:

$ curl -u $USERNAME:$PASSWORD http://127.0.0.1:8000/api/process_definition/
{
  "meta": {
    "limit": 20,
    "next": null,
    "offset": 0,
    "previous": null,
    "total_count": 2
  },
  "objects": [
    {
      "definition": {
        "exec": "/path/to/executable"
      },
      "id": 1,
      "resource_uri": "/api/process_definition/1/"
    },
    {
      "definition": {
        "exec": "/path/to/executable"
      },
      "id": 2,
      "resource_uri": "/api/process_definition/2/"
    }
  ]
}

To create a process definition:

$ curl -u $USERNAME:$PASSWORD --dump-header - -H "Content-Type: application/json" -X POST --data '{"definition": {"exec": "/path/to/executable"}}' http://localhost:8000/api/process_definition/
HTTP/1.0 201 CREATED
Date: Wed, 28 Aug 2013 18:19:43 GMT
Server: WSGIServer/0.1 Python/2.7.2
Vary: Accept
Content-Type: text/html; charset=utf-8
Location: http://localhost:8000/api/process_definition/3/

etc.

Process Definition Format

The basic Process Definition format is a dictionary like the following:

{
    "exec": "/path/to/executable",
    "application": "referencetodtindtrs",
    "process_type": "typeofprocess",
    "timeout": 60,
    "cwd": "/path/to/cwd",
    "environment": {"ENVAR": "val},
    "argv": ["arg1", "arg2"],
    "input_adapter": {
        "input_type": "intype",
        "input_params": {
            "chunk_size": "10KB",
            "tcp_port": 8090
        }
    },
    "output_adapter" {
        "output_type": "outtype",
        "output_params": {
             "file": "path/to/file"
        }
    }

}

With the following parameters:

exec: Path to the executable that should be run

application: Reference to the application that this runs on. This is a DT in the DTRS, which includes the VM image it will run on, and the chef recipes / configuration that needs to be installed process_type: Type of process. Choose from:

  • single: a process that takes an input file, and produces an output (think gcc)
  • service: a process that runs for a long time

cwd: Current working directory of executable

environment: Environment variables that should be set for a process

argv: List of arguments that should be passed to the process

input_adapter: Defines a way of moving data from the messaging system to the process. A dictionary that can take the following parameters:

  • input_type: The type of input that the process expects, could be a single file that is read once, input to a port, appending to a file, writing to a unix pipe, etc.
  • input_parameters: Depends on the input type, but a dictionary of parameters to configure the input adapter

output_adapter: Defines a way of moving data produced by the process to the messaging system. A dictionary that can take the following parameters:

  • output_type: The type of output that the process produces, could be a single file, output read from a port, a file that is appended to, a unix pipe that can be read, etc.
  • output_parameters: Depends on the output type, but a dictionary of parameters to configure the output adapter

Input and Output Adapters

Since the ways in which processes get input and produce output is widely varied, we need an extensible and flexible way of defining how to adapt messages from the messaging framework to be fed to a process, and a way of reading the output produced from that process and put on the messaging framework. Some examples:

Input

file: A simple file that is fed to a process. The message is written to a file, and then fed somehow to the process, either by standard input or as an argument to the process.

For example, a file that is fed by STDIN:

{
    "exec": "someproc",
    "input_adapter": {
         "input_type": "file",
         "input_parameters": {
             "file_type": "stream",
             "stream_name": "STDIN"
         }
    }
}

This produces the equivalent of:

$ someproc < /path/to/file

Another example, a file that is fed by trailing argument:

{
    "exec": "someproc",
    "args": ["--quiet", "--input", "${input_file}"],
    "input_adapter": {
        "input_type": "file",
        "input_parameters": {
            "file_path": "${input_file}"
        }
    }
}

This produces the equivalent of:

$ someproc --quiet --input /path/to/file

Output

file: A simple file that is produced by a process. The file is read from the process somehow, either from stdout or an a.out style file, and then is put on the message bus:

For example, a file that is read from STDOUT:

{
    "exec": "someproc",
    "output_adapter": {
        "output_type": "file",
        "output_parameters": {
            "file_type": "stream",
            "stream_name": "STDOUT"
        }
    }
}

This produces the equivalent of:

$ someproc > /path/to/file

Another example, a file that is produced by trailing argument:

{
    "exec": "someproc",
    "args": ["--quiet", "${output_file}"],
    "output_adapter": {
        "output_type": "file",
        "output_parameters": {
            "file_path": "${output_file}"
        }
    }
}

This produces the equivalent of:

$ someproc --quiet --output /path/to/file

Practical Example: tr

tr is a simple text stream transformation process. We will use it to apply a simple transformation process to a text stream where we remove every 'x' character. Here is an example process configuration for it:

{
    "exec": "/usr/bin/tr",
    "argv": ["-d", "x"],
    "input_adapter": {
        "input_type": "file",
        "input_parameters": {
            "file_type": "stream",
            "stream_name": "STDIN"
        }
    },
    "output_adapter": {
        "output_type": "file",
        "output_parameters": {
            "file_type": "stream",
            "stream_name": "STDOUT"
        }
    }
}

Practical Example: ffmpeg

ffmpeg is a tool to convert and stream audio and video. In this example, we will encode an audio sample into the MP3 format.

{
    "exec": "/usr/local/bin/ffmpeg",
    "argv": ["-i", "${input_file}", "-q:a", "2", "${output_file}"],
    "input_adapter": {
        "input_type": "file",
        "input_parameters": {
            "file_path": "${input_file}"
        }
    },
    "output_adapter": {
        "output_type": "file",
        "output_parameters": {
            "file_path": "${output_file}"
        }
    }
}

About

No description, website, or topics provided.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages