serve_flmdata.js is a Javascript script running on node.js to receive and persist Fluksometer readings in a database and visualizing them in a panel and chart web service - so it needs node.js installed... It connects to the FLM's MQTT broker on the discovered IP address(es) using the multicast DNS service discovery - so there is no further configuration to change beside the required setup steps described in the following.
To use the combined Fluksometer persistence and visualization script you have to perform some preparation steps.
The following steps are based on a fresh installation of Raspbian/Debian on a Raspberry Pi. I chose the Jessie lite variant; it should also work with an equivalent, Debian-based PC installation or on a Macintosh. In brief, for the Jessie lite variant you have to additionally install following packages:
sudo apt-get install libavahi-compat-libdnssd-dev git sqlite3 node
Note that for node(.js) you have to get the link to the corresponding Adafruit package first; see for more details in the following or jump to the actual script install directly.
As the combined script stores Fluksometer data, a proper database must be installed; in my case it is SQlite. An alternative use of MySQL is also possible.
sudo apt-get install sqlite3
This installs the database and all dependent packages. Everything else is done from the serving script.
For MySQL install it by
sudo apt-get install mysql-server
and prepare the required database settings by executing the script
./createdb.sh
As written, either use SQlite or MySQL.
To run the persistence and visualization script install node.js. For this you may take the sources from http://github.com/nodejs/node and compile it on your Raspberry Pi. Use following command sequence - make sure to select/checkout the branch to your convenience
cd ~
git clone http://github.com/nodejs/node
cd node
git checkout <the node.js release you want to install>
./configure
nohup make &
This will take some while; to check what the make process is doing, use
tail nohup.out
After finished make install node.js on your computer.
sudo make install
Note again, that the make process takes quite some while on a RasPi, so be patient; you may start the make process also directly (thus, without nohup and &); I chose to run it in the background with "nohup make &" allowing a logoff during processing. By the tail command you may have a look what’s currently going on. You'll recognize a finished make either if the nohup.out does not increase any further or the tail provides evidence that the make ended after all its compiling steps - if there where errors, hey, Google is your friend...
After the sudo make install check if node was installed properly
node -v
npm -v
This shows the respective versions of node and the node package manager. Or it shows that your make failed (grmph) - therefore:
As an easy alternative you may obtain a precompiled installation from Adafruit Industries; follow the description.
Now the preparation is finished and you may continue with the actual Fluksometer service. Install it via git
cd ~
git clone http://github.com/gebhardm/flmdisplay
To use mdns with npm on a Raspberry Pi you need to have installed also the avahi compatibility library.
sudo apt-get install libavahi-compat-libdnssd-dev
Now install the required mqtt, mdns, sqlite3, and socket.io modules by executing
cd flmdisplay
npm install
utilizing the provided package.json file. Note: For the MySQL variant install also mysql.
Be aware that these modules evolve; for example, there was an issue with incompatible changes on socket.io v0.9 vs. v1.0 - so, even though I tested the stuff, it may not work with a next version of the used modules... Again, as this is free stuff, help yourself finding out what's up - there are zillions of possibilities what might have happened. (Have you used git checkout? Have you called ./configure?)
As mdns uses a compatibility layer, be aware that it throws warnings on use; follow the links within the warnings to get an understanding what has happened (and then ignore them).
To start the combined persistence and visualization service, run the provided start script, that pushes the exeuction into the background
./flmdata.sh
As an alternative you may also run it using node serve_flmdata.js. This applies also for the MySQL alternative; this needs to be started by node serve_flmdata_mysql.js.
After start, the script checks if the corresponding database exists; if not, it creates the corresponding files and tables. Now the script discovers the existing Fluksometer MQTT broker(s) and subscribes to the respective sensor topics.
If there is no message that your FLM was detected (the message shows the IP address and the corresponding port) something went wrong with the mDNS installation; check this - again, Google is your friend...
On detection of one or more Fluksometers, the script starts to insert received values (i.e. mqtt messages on topic /sensor/+/gauge) into the database for later retrieval, e.g. by the chart.
With the script also a web server is started that you can reach on the (RasPi’s) IP address from any computer in the LAN; access the visualization on port 1080
http://"RasPi IP address":1080
Note that the service also advertises its existence; check with a Bonjour browser; so alternatively you may access the service also (at least on a Mac) via
http://raspberrypi.local:1080
This assumes that you kept the default hostname; it may be different if you changed it, for example during raspi-config.
That's it. Now you should have an up and running local Fluksometer data visualization with gauge, panel, graph, and chart.
##A "better" solution From the Fluksometer's v2.4.x firmware release there are some very interesting features that lead to installing persistence and visualization directly on the FLM. For this purpose development on this repository will rather not be evolving, but its successor flmlocal that provides all you need to make a full-fledged visualization without the need for an extra web server.
##Licenses
All code provided under the respective licences; if not denoted otherwise the MIT license is the one to care about...
(c) 2014-2016, Markus Gebhard, Karlsruhe