mc2-symmetry 0.4.0

Symmetry - A cluster management system for micro-clusters

symmetry

Cluster management tool for the MC2.

Build

To create a virtual environment and install the requirements, run (by default it uses python3.7, you can change that in the Makefile variable VENV_BIN)

make venv

Build the Docker image for local use (may need sudo)

make docker-build

Run Symmetry with Docker

Populate etc/nodes.ini with the information from your cluster. The application can then be started with

make start

And point your browser to http://localhost:5000/ui/dashboard

Stop with

make stop

Both commands use docker-compose on docker/symmetry/docker-compose.yml, and may need sudo

Run Symmetry components manually

Prerequisites

Make sure the virtualenv is activated

source .venv/bin/activate

Symmetry publishes and reads telemetry data to and from redis, so you need a running redis instance first:

redis-server

Start the telemetry deamon

Start the telemetry daemon in the activated virtualenv via

python -m symmetry.cli.telemd

telemd also takes various arguments to control runtime aspects (activate the -h flag). For exampe, to disable power monitoring and probe nodes every 0.5 seconds, run

python -m symmetry.cli.telemd --power-disable --agent-interval=.5

You can verify that the telemetry daemon sends data by suscribing to the topic via the redis CLI:

redis-cli psubscribe "telemetry:*"

Start the cluster deamon

Start the cluster daemon (which controls the cluster state)

python -m symmetry.cli.clusterd

Websocket server

To start the websocket server for the dashboard:

python -m symmetry.cli.websocket

The websocket is running on port 8082 and automatically connects to the local redis-server on localhost:6379. Connect with ws://ip:8082/topic to receive all published messages for the given topic.

Run the dashboard

Serve the app with gunicorn

gunicorn -w 4 --preload -b 0.0.0.0:5000 \
    -c symmetry/webapp/gunicorn.conf.py \
    symmetry.webapp.wsgi:api

Point your browser to http://localhost:5000/ui/dashboard.

Using Symmetry

Creating services

To create a new service in the cluster a POST request must be sent to the api/services/{service_id} endpoint, with service_id being the new service.

The POST request has one optional (tar-exported docker image) and one mandatory (service description) body element.

The service description is structured this way:

%YAML 1.2
---
id: #must be same id as in endpoint
maintainer: #the uploader of the service
name: #"Pretty" version of the name
version: #any version number
description: #simple description
image_tag: #if docker image uploaded then this tag must be provided by the image. otherwise it is pulled 
port: #the port which is exposed by the service

After correct execution the answer should look somewhat like this:

{
    "name": "Test Service",
    "version": "1.0.0",
    "desc": "This is a simple test_service",
    "maintainer": "silvio",
    "10.0.0.3": "1"
}

Other Client Scripts

Recording telemetry data

You can use the telemetry client cli.telemc to consume telemetry data in various ways.

# prints to system out
python -m symmetry.cli.telemc

# writes to a csv file
python -m symmetry.cli.telemc --file /tmp/telemetry-results.csv

Arduino Power Meter

To sample wattage from the arduino power monitor every 0.5 seconds call:

python -m symmetry.cli.power --interval 0.5

This will write the data into the Redis instance like the telemetry daemon.