We have an Arctic Spa at our house which we enjoy. Recently, an OpenAPI spec has been published for it, so I decided to write a Java wrapper so I could control the spa myself, and monitor temperature with Prometheus and Grafana.
The resulting code is here. The wrapper class "com.khubla.kspa.Spa" takes care of all the API details, so that status and control of the Spa is as simple as:
Spa spa = new Spa("abc123"); Status status = spa.getStatus(); System.out.println("Temperature :"+status.getTemperature());
My preferred server OS is FreeBSD. Each time I install it, I have to go through numerous steps to secure it. So, to avoid that effort, and to make sure I get it right each time, I've created an Ansible playbook to automate the entire process. You can learn about it here. Some of the tasks it completes include:
fbsd-secured is provided as-is. fbsd-secured is intended as an example of configurations which may, or may not, improve the security posture of FreeBSD systems. The author does not accept any responsibility for damages caused by the use of these configurations. Any user using these configuration should do their own research and must take full accountability for potential risk and/or damages resulting from using them
When building database ORM code, I prefer to use ebean, and I prefer to use DAO's. Since DAO's are simple to make generic, I have a Generic DAO I use for all projects. I've finally decided to open-source my DAO code, and you can find it here https://github.com/teverett/ebean-dao.
I've recently been quite interested in Graph languages, not for Graph Databases, but for representing information, in code that is essentially "things" and "relationships". For example, one perspective on Organizations is that they are simply representable as people, systems and the information that flows between them. Those three building blocks are, in theory, enough to describe anything from a small business to a government, if you have an appropriate way to describe all the information, ensure the connections make sense, and report on it. If you further layer on the idea of systems of systems, you have the ability to describe different levels of government, business units, departments and so on. Public and private flows of information allow models that accurately represent the fact that no organization is fully meshed: the systems only expose certain information.
The reporting on it is quite interesting, and led to be becoming interested in DOT as a way to describe reports resulting from queries. I didn't find a small, embeddable Java library that enabled me to read and write DOT files, so I ended up writing DOT4J.
The writing of DOT files, is trivial. The reading however, ended up requiring a lexer-parser since DOT is, essentially a programming language of its own. The grammar came from ANTLR, of course.
DOT4J is here. The Github page has the Maven co-ordinates for published binaries.
I recently had reason to get to know OpenAPI at work, so I decided to become familiar with it. I have a HomeSeer home controller at home, which exposes a JSON API, so I decided to write a Java OpenAPI server to expose the JSON API over OpenAPI.
The net result is hsOpenAPI. It exposes most of the HomeSeer API and uses hsClient, the Java wrapper I had previously built for the HomeSeer API.
After all, now that we have wireless thermostats, door locks, outlets, dimmers, alarms, garage doors, and motion sensors, it only makes sense to publish events onto a Service Bus. In my case I used Mosquitto in a FreeBSD jail as my broker.
My home controller is a HomeSeer. There are MQTT plugins available, but I preferred to use my own hsClient library which is a Java wrapper for the HomeSeer JSON API. I ended up writing hsMQTT, an MQTT bridge for HomeSeer which polls the HomeSeer API and publishes JSON messages onto an MQTT topic when device values change.
I recently got a call from a friend who has a library of thousands of PDF documents. He wanted to know there was a such thing as a personal search engine that could index his library of documents, and allow full text search inside of them.
I have a giant document library too, including PDF documents, Office documents, source code, email and so on. I decided I needed one too, and it should be based on Elastic Search.
The net result of the weekends work is "kSearch", a personal search engine. kSearch include a threaded file system indexer, automatic maintainence of the search index as the file system changes, and a web search UI.
Home automation is coming into the mainstream a with recent offerings from the big tech companies, and I'm interested in it too. I've had a number of home automation controllers, most recently an H3 Pro SEL from HomeSeer.
So, I decided to import home status information into InfluxDB and graph it. The first step was getting the information from HomeSeer. There is a JSON-over-HTTP API, but no existing Java client. So, a simple HomeSeer H3 API client in Java was written: hsclient.
The next step was getting the data into influx. The best way to do this would be to write a Telegraf plugin. However, I wanted a chance to learn a little more about Java closures and lambas, so I wrote hsinflux a threaded command-line app which connects to HomeSeer on a schedule, downloads the device status data and drops it into InfluxDB. hsinflux is here.
The net result, with Grafana, is the ability to see my house temperatures, humidity, door and window status, and thermostats. Here are some examples.
House temperatures this week on 5 minute intervals
At the time of writing, its -17C outside, and my server room temperature is showing a steady decrease, down to -6C. The increase on the far right is the result of opening the server room door to get some warm air from the house.
And since I have a TSDB, why not record the time it's taking to collect data from HomeSeer?
Since I can, I'm also querying battery charge on the devices that have batteries. It's not an interesting graph, but down the road I can use Grafana to alert me when the batteries need to be changed.
Overall, a great opportunity to become familiar with the technologies. Once the data is in Influx, building arbitrary graphs to visualize data is both simple and powerful. The graphs I've built are simple, however for IT professionals wanting insight into complex infrastructure, tools such as Influx and Grafana represent a step change in visibility and analysis.
From an operational perspective, TSDB's give DevOps engineers an opportunity to answer a question we've been trying to answer for decades:
What changed the hour before it crashed? The week before it crashed? The month before it crashed?