go-orchestrator

Repository for working through Build An Orchestrator in Go

Details

• Go Version: 1.18

Running Locally

1. Start up application with task start
2. Add tasks by firing REST Call

curl -v --request POST \                                                            
  --header 'Content-Type: application/json' \
  --data '{
    "ID": "266592cd-960d-4091-981c-8c25c44b1018",
    "State": 2,
    "Task": {
        "State": 1,
        "ID": "266592cd-960d-4091-981c-8c25c44b1018",
        "Name": "task-1",
        "Image": "strm/helloworld-http"
    }
}

3. List all tasks

curl http://localhost:8989/tasks |jq .

4. Delete task

curl -v --request DELETE "localhost:8989/tasks/75e260da-e9f7-4601-bca2-d52461df12cc"

Task File

Task is a task runner / build tool that aims to be simply common commands. The commands can be found in the Taskfile.yml file.

Task File Installation

Installation instructions for Task can be found here.

List Task Commands

To list all available commands for this project:

$ task --list

Definitions

Orchestrator

• Automates deploying, scaling & managing containers.

Task

• Smallest unit of work.
  • Should specify it's needs for memory, CPU & disk.
  • What to do in case of failure (restart_policy).
  • Name of container image used to run the task.

Job (not included in this implementation)

• Group of tasks to perform a set of functions.
• Should specify details at a higher level than tasks.
  • Each task that makes up the job.
  • Which data centers the job should run in.
  • The type of job.

Scheduler

• Decides which machine can best host the tasks defined.
• Could be a process within the machine or possibly a different service.
• Decisions can vary in model complexity from round-robin to score calculation based on variables and deciding node by best score
  • Determine set of candidate machines to run task
  • Score machines from best to worst
  • Pick machine with best score

Manager

• Brain of orchestrator & main entry point
• Jobs go to manager, manager uses scheduler to determine where to run job.
• Can gather metrics, to use for scheduling.
  • Accept requests from users to start/stop tasks
  • Schedule tasks onto worker machines
  • Keep track of tasks, states & the machine they run on

Worker

• Runs the tasks assigned to it, attempts task restarts on failure, makes metrics about task and machine health for master
  • Runs asks as containers
  • Accepts tasks to run from manager
  • provides relevant statistics to the manager to schedule tasks
  • Keeps track of tasks & their state

Cluster

• Logical grouping of all above components.

CLI

• Main user interface
  • Start/Stop Tasks
  • Get status of tasks
  • See state of machines
  • Start manager
  • Start worker