hyperpilotio / daemon-ecs-scheduler

A ECS scheduler that places daemons on every node in ECS cluster

daemon-ecs-scheduler

A ECS scheduler that places daemons on every node in ECS cluster

Quick start

glide install

go build

daemon-ecs-scheduler --cluster weave-ecs-demo-cluster --tasks cadvisor:2

Feature

• Support the automatic registeration of tasks.

Help

daemon-ecs-scheduler --help

daemon-ecs-scheduler's Issues

Report status of current tasks and statuses

Either through API or UI

Scheduler should be long running, continously waiting for new tasks or new instances, or tasks terminated to reschedule

Create Dockerfile for automated docker builds

Use gin web framework

Add unit tests to all API calls

There is no tests in the scheduler atm. We need to at least have simple coverage to test correctness

Persist submitted task definitions for HA

RESTful API for long running event registration (register task on each host)

WHY / WHAT

• daemon-ecs-scheduler provides a RESTful interface to allow other trust clients to communicate.

GET /task-definitions/:task-definitionARN

* Query the registered task definitions in the given cluster.
:task-definitionARN can be the name of task definition or the ARN of task definition

response:

{
  ARN: string
  ShortString: string
  Cpu: int
  Memory: int
  TCPPorts: int
  UDPPorts: int
}

==========

POST /tasks/start

* start to run the given task on each host. 
After the registration of given task complete, the scheduler of daemon-ecs-scheduler
will check the status of each container instance in this cluster every 5 minutes
(depends on the configuration)

{
  cluster: string
  task-definition: string
}
the field of task-definition can be the name of task definition or the ARN.

Response
{
  statusUrl: string
}

======

GET /tasks

List the status of all the tasks in the cluster.

Response
{
running: []string
failures: []string
}

======
GET /tasks/:taskARN/status

List the status of given task. Ex: Running on which server.
Failed to start the task on which server.

Response
{
running: []string
failures: []string
}

Refer to issue #5 .

Add a logger for better logging

Filter certain nodes with labels when spawning tasks