remote execution framework
Project description
# Odium: Remote Execution / Distributed Application Framework
## Requirements:
* Python 3 (>= 3.5) and git are installed.
* Remote execution: Arbitrary execution of code on remote systems under the connected user account.
* Versioned state: Each version of the configuration state has a unique id (hash of the state declaration), so that remotely executed commands can ensure the correct state for that command, thus ensuring that the calculation is done with the expected code / library / application / version.
* Any instance is both a hub and a node. It is a node to its source hub, it is a hub to any nodes that have connected to it.
* All communication encrypted between hub and node. Each publishes its public key for this purpose.
## Design:
* JSON for all messages.
* ZeroMQ for all network messaging.
* TCP port: default 18202, but configurable.
(2018-02-02 = date of "Past Life" Agents of S.H.I.E.L.D. episode, first reference to odium)
* Node and hub communicate using PUB/SUB (broadcast/query => nodes), REQ/REP (query => hub)
Also PUSH/PULL (job queue => nodes), and PULL/PUSH (result collection => hub).
* hub PUB / node SUB @ 18202
* node REQ / hub REP @ 18203
* hub PUSH / node PULL @ 18220
* node PUSH / hub PULL @ 18221
* Instance uses relational database to store configuration state, job queue, log, ...
(sqlite in the simple case)
* All communication includes the current configuration state id for the sending instance.
* Configuration state id = Application's current git commit hash.
* Configuration state includes: pip & system packages installed, state / environment data
* Commands (broadcast and queued) are executed in the application environment using multiprocess & subprocess. Results, logs, and errors are collected and returned via REQ => hub or PUSH => hub
* Asynchronous and multiprocessing:
* node listens on a main thread for instructions, then pushes instructions to other threads for processing. It can thus handle an unlimited number of instructions before the first have finished.
* hub listens on a main thread for requests, then pushes request processing to other threads.
* hub & node both send messages on main thread.
## Requirements:
* Python 3 (>= 3.5) and git are installed.
* Remote execution: Arbitrary execution of code on remote systems under the connected user account.
* Versioned state: Each version of the configuration state has a unique id (hash of the state declaration), so that remotely executed commands can ensure the correct state for that command, thus ensuring that the calculation is done with the expected code / library / application / version.
* Any instance is both a hub and a node. It is a node to its source hub, it is a hub to any nodes that have connected to it.
* All communication encrypted between hub and node. Each publishes its public key for this purpose.
## Design:
* JSON for all messages.
* ZeroMQ for all network messaging.
* TCP port: default 18202, but configurable.
(2018-02-02 = date of "Past Life" Agents of S.H.I.E.L.D. episode, first reference to odium)
* Node and hub communicate using PUB/SUB (broadcast/query => nodes), REQ/REP (query => hub)
Also PUSH/PULL (job queue => nodes), and PULL/PUSH (result collection => hub).
* hub PUB / node SUB @ 18202
* node REQ / hub REP @ 18203
* hub PUSH / node PULL @ 18220
* node PUSH / hub PULL @ 18221
* Instance uses relational database to store configuration state, job queue, log, ...
(sqlite in the simple case)
* All communication includes the current configuration state id for the sending instance.
* Configuration state id = Application's current git commit hash.
* Configuration state includes: pip & system packages installed, state / environment data
* Commands (broadcast and queued) are executed in the application environment using multiprocess & subprocess. Results, logs, and errors are collected and returned via REQ => hub or PUSH => hub
* Asynchronous and multiprocessing:
* node listens on a main thread for instructions, then pushes instructions to other threads for processing. It can thus handle an unlimited number of instructions before the first have finished.
* hub listens on a main thread for requests, then pushes request processing to other threads.
* hub & node both send messages on main thread.
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
odium-0.0.1.tar.gz
(2.3 kB
view details)
Built Distribution
File details
Details for the file odium-0.0.1.tar.gz
.
File metadata
- Download URL: odium-0.0.1.tar.gz
- Upload date:
- Size: 2.3 kB
- Tags: Source
- Uploaded using Trusted Publishing? No
File hashes
Algorithm | Hash digest | |
---|---|---|
SHA256 | 27410fc28ea236a15ea15a1fef5b3d6704b88eaa48a9a640500a9019d9bf9d7b |
|
MD5 | 8e8191b0951309b3dad7540db6fc7d1b |
|
BLAKE2b-256 | b126ec55534bbcf18782d0f850e946bb24fdcf7467190a9be35bdc0a34cf8923 |
File details
Details for the file odium-0.0.1-py2.py3-none-any.whl
.
File metadata
- Download URL: odium-0.0.1-py2.py3-none-any.whl
- Upload date:
- Size: 2.3 kB
- Tags: Python 2, Python 3
- Uploaded using Trusted Publishing? No
File hashes
Algorithm | Hash digest | |
---|---|---|
SHA256 | ca6b61ae269a22af07eb04fbf434d1c7b814881dc5e2f90e5d51fffc279d927f |
|
MD5 | c73be5cd5c947e9d04f01acb7a017aa4 |
|
BLAKE2b-256 | d8cdaacdd7a59ca0fc303ee5dfc7711ec8a8b3cdeb0f15491d7fa11b6069f6d7 |