We discuss what is behind the clouddata.dustcloud.org server, and how exactly the server is built.

Requirements

Goal

clouddata.dustcloud.org allows a customer to publish sensor data to the Internet less than 5 min after having opened a SmartMesh IP starter kit

This translates into the following requirements:

clouddata.dustcloud.org must be able to receive sensor data, and store data as timelines for 12 hours.
clouddata.dustcloud.org must offer a web interface to visualize the data.

Architecture

The solution consists of the PublishToWeb SmartMesh SDK program and the clouddata.dustcloud.org server.

`PublishToWeb` Smartmesh SDK application

This application is part of the default SmartMesh SDK, and runs on the user's computer. It:

connects to the SmartMesh IP Manager
subscribes to data notifications
forwards all the OAP temperature data readings produced by the SmartMesh IP motes to clouddata.dustcloud.org.

clouddata.dusctloud.org server

The clouddata.dusctloud.org server is built on Ubuntu Server. It has the following internal architecture:

the iptables firewall filters out all external traffic not going to TCP port 80
the NGINX load balancer redirects:
- HTTP requests to http://clouddata.dustcloud.org/api/v1 to the clouddata_server script running on TCP port 8080
- HTTP requests to http://clouddata.dustcloud.org/ to the grafana frontend running on TCP port 3000
the clouddata_server script is part of the SmartMesh SDK. It receives data from the PublishToWeb.py application and inserts it into the database.
the influxdb database efficiently stores time series data. In this example, the database is configured to automatically drop data after 12 hours.
the grafana data visualization tools offer a clean dashboard for visualizing the data

Installation Steps

Vagrant setup files for the step-by-step guide below are maintained at https://github.com/dustcloud/clouddata-server.

Ubuntu

At the time of writing, we have installed Ubuntu Server 16.04 LTS.

If you are installing the server on a virtual machine using VirtualBox for testing purposes:

configure two network adapters:
- Adapter 1 as "NAT" (this allows your VM to access the web)
- Adaptor 2 as "Host-only Adapter" (this allows your host OS to access the web server running your VM)
after booting your VM, edit /etc/network/interfaces to add the last lines corresponding to the second network interface
```
 
```
restart the networking services:
```
 
```

Once installed, log into the server.

Update

The commands above update your entire OS. This can take 10-15min.

SSH

Install the OpenSSH server:

Add your public key to the server:

Edit /etc/ssh/sshd_config so it contains the following (un-commented) lines:

Restart the SSH server:

influxdb

install database

Install and start InfluxDB automatically:

You should see the InfluxDB admin panel at http://192.168.56.101:8083/.

By default, the HTTP API of InfluxDB runs on TCP port 8086.

create database

Use the influx client to create a database "grafana" and associate a retention policy so datapoints are automatically deleted after 12 hours:

Recent versions of the influxdb package include the command-line client. If your distribution has an older package, you may need to install the client separately.

grafana

install

The following commands install grafana, start it, and ensure it also starts on the next boot:

You should now be able to see the Grafana front end running at http://192.168.56.101:3000/.

We are not using authentication between Grafana and InfluxDB, but you still have to provide a username/password in the "InfluxDB Details" section below.

change admin username and password

Connect to the Grafana front end with username "admin" and password "admin". Though the profile page, change both the username and password associated with the Grafana administrator.

connect to database

Through the admin interface, add InfluxDB as a data source. From the drop down menu in the upper left, choose Data Sources and click on Add data source.

create dashboard

Through the admin interface, create a new dashboard named Temperature. From the drop down menu in the upper left, select Dashboards → New. The first row is already created, indicated by the vertical line on the left edge. Hovering over that indicator will reveal a menu for editing. Select Add Panel → Graph to create a graph of the temperature data. Under Metrics, select Toggle edit mode under the drop down menu (on the right next to the eye icon) to change the input mode so you can paste in the following query:

To verify everything works, on the Admin interface of InfluxDB (http://192.168.56.101:8083/), select the "grafana" database, click "Write Data" and enter:

enable anonymous access

Change the /etc/grafana/grafana.ini file so it contains the following lines:

set as default dashboard

log into Grafana as admin
star the dashboard you create
in the dropdown menu, edit the setting for the default organization, and set the dashboard as default home dashboard

`clouddata_server` script

install Python

install `clouddata_server` script

You can download the cloudata_server.py script as part of the SmartMesh SDK. You will find it in the app/PublishToWeb/ directory.

Install the Python dependencies of the SmartMesh SDK:

run at startup

Install supervisor:

Create a new file for supervisor to manage the clouddata_server process:

Include the following contents, where _USER_ is replaced with the current username:

Start the supervisor daemon and configure the daemon to start when the system is booted:

NGINX

installation

nginx starts automatically. Pointing your browser at http://192.168.56.101/ shows you the default interface:

configuration

We will use nginx in proxy mode. The role is that it redirects per the table below:

url	point to
`http://<ip_address>/api/v1/oap`	`http://<ip_address>:8080/oap`
`http://<ip_address>/`	`http://<ip_address>:3000/`

Create file /etc/nginx/sites-available/clouddata_server with the following contents:

Install the script:

Restart the server:

iptables

It is important you step through the iptables instructions in the order they are written!

configure

Type the following commands in this order to configure iptables (the firewall):

There lines have the following actions (respectively):

[line 1] all ports are open for communication internal to the server (needed for the communication between grafana, influxdb, and clouddata_server)
[line 2] accept incoming SSH traffic (needed for you to configure the server later on)
[line 3] accept incoming HTTP traffic (needed for a client to push sensor data and see the grafana web interface)
[lines 4,5,6,7] allow outgoing DNS queries
[lines 8,9] allow outgoing HTTP traffic (needed for example to use apt-get)
[lines 10,11] allow incoming ping requests (needed so you can ping your server)
[lines 12,13] allow outgoing ping requests (needed so from the server you can ping other hosts to verify connectivity)
[line 14] drop all other incoming traffic

verify

Type the following command to verify all rules are active:

run at boot

By default, your iptables rules will disappear at reboot. The iptables-persistent packages allows them to be applied at each reboot.

Install the iptables-persistent package:

At installation, make sure to click "yes" to save the current IPv4 and IPv6 rules

Testing

Once you have the server running, you can test it by using the following scripts:

PublishRandom.py publishes a random temperature coming from a (fake) random MAC address
PublishToWeb.py publishes a temperature published through OAP by a real node

Make sure to reboot reboot the server and verify everything starts up as expected. You should be able to ping the server, but not access the grafana website on port 3000, or the influxdb admin interface on port 8083.

Building clouddata.dustcloud.org