ik5 / distributed_test

A test i was given to write a distrubuted command and control

Scylla Test

The following repository aims to provide a Command and Control (C&C) system for managing "unlimited" number of servers to perform tasks and report back.

The idea behind the implementation:

There are three levels of services

0. Dispatcher - CnC server
1. Orchestration - Sends tasks to nodes or sub orchestration servers
2. Nodes - Actual workers that does the code.

Flow idea

The dispatcher has X Orchestration servers that it knows about, having them registered and reporting an heartbeat every second to know that they are up.

The Orchestration server has registered X amount of either nodes or additional orchestration servers. The server sends heartbeat to the Dispather if it is connected directly to it, and sends actions to nodes or other orchestration servers.

It gathers all information arrived from all nodes and other orchestration servers and pass it up to the dispatcher.

The node server is a worker that sends a heartbeat to a known orchestration server every second, and listen to an action to do.

When an action is given, it execute it, and reports back what was the result.

How to use

There is the cmd directory. It contains nodes and orchestration sub directories.

The nodes directory are the workers nodes described under this readme.

The orchestration directory contains the orchestration server described under this readme.

How I tested

Created the following file:

l: 216
t: 2

{
  "check_etc_hosts_has_8888": {
    "path": "/etc/hosts",
    "type": "file_contains",
    "check": "8.8.8.8"
  },
  "check_log_file_exists": {
    "path": "/var/log/messages.log",
    "type": "file_exists"
  }
}

Then used netcat like so:

nc 127.0.0.2:8082 < test.request