Home Monitoring Upgrade

I’ve been monitoring stats from my meter, weather and hotwater tank for over two years now (see http://blog.v-s-f.co.uk/2015/04/home-monitoring-home-made-reborn/) and the application now needs an upgrade.

I now want to log more data from the weather station (temperature and humidity). This should be as simple as adding two new columns to the HSQLDB, changing the application to write in to the two new fields and adding two new fields to the service definition, but it’s not quite that straight forward…

The old app uses an out dated version of Mule on Tomcat in Docker and it’s far too heavy weight for what it needs to be. Therefore it’s time to give it a revamp.

It’s also occured to me recently that instead of storing the data in five separate tables (one for generation, upload info, hotwater, meter and weather data), why not store it in one table. This saves a significant amount of space as there are four less records per minute and it makes adding new columns for additional data sources relatively quick. The HSQLDB that I’ve been using for a while now is over 400M!

So the first task, which is possibly the biggest, is to migrate the data from the five tables in HSQLDB to a single table and then stop using HSQLDB and migrate to MySQL. Why MySQL – it’s actually quite a performant database, it’s free and easy to get running.

Elasticsearch, Logstash and Kibana (E.L.K.) on Docker – Part 3 Kibana

This is the final part of setting up Elasticsearch, Logstash and Kibana using the official Docker Hub images. If you haven’t already read Part 1 Logstash or Part 2 Elasticsearch, it might be good to read them first.

So far I have a running Logstash sending messages to Elasticsearch both of which are running in separate Docker containers and now I’m going to add Kibana. Kibana adds the graphical UI that enables you to visualise, create dashboards and search for messages etc.

Kibana talks to Elasticsearch to query the data, so for our Docker run statement, we need to link the kibana instance to the elasticsearch-node. For consistency, the Kibana instance is given the name kibana-node and it’s given a parameter to talk to the elasticsearch-node via a fully qualified search url.

docker run --link elasticsearch-node:elasticsearch-node --name kibana-node -p 5601:5601 -d -e ELASTICSEARCH_URL=http://elasticsearch-node:9200 kibana

When you first log into Kibana, it will ask for an index to be created. This bit did catch me out as you can’t create an index until Elasticsearch has received a few events. Once it has, it’s as simple as giving it a name and selecting the right date field, then clicking create.

Clicking on the Discover menu will then allow you to see the rececent events and start creating new queries for turning into dashboard widgets.


I haven’t yet had enough time to create any fancy dashboards to give an example as it’s a little different to the tool I’m more familiar with (Splunk).

The last point to mention is that all of the examples so far have not saved data other than to the Docker image. Therefore if the Docker image is removed, so is all your historic data! To persist the data, I’d suggest having a look at Docker volumes – but since I haven’t tried it yet, can’t guarantee it is the right answer!

That concludes my mini-series on the subject of Elasticsearch, Logstash and Kibana (E.L.K.) on Docker 🙂

Elasticsearch, Logstash and Kibana (E.L.K.) on Docker – Part 2 Elasticsearch

It’s worth reading Part 1 Logstash first.

So, today I had a chance to try out Elasticsearch on docker and it was semi easy to get it to work… the trickiest part was linking the Logstash and Kibana instances with the Elasticsearch instance. The trick is to name everything!

So here’s the command to run Elasticsearch, note I’ve given it a name of elasticsearch-node.

docker run -d --name elasticsearch-node elasticsearch

In order to then link Logstash to the Elasticsearch node, we need to change the command used to run Logstash from this…

docker run -p 13456:9999 -it --rm -v "$PWD":/config-dir logstash -f /config-dir/logstash.conf

To this…

docker run -p 13456:9999 -d -it -v "$PWD":/config-dir --link elasticsearch-node:elasticsearch-node --name logstash-node logstash -f /config-dir/logstash.conf

The differences being that the Logstash image now has a name of logstash-node when it’s run and it links to the Elasticsearch node via the name and identical alias.

The Logstash config file has also been changed to reference the Elasticsearch node as shown below.

input {
  tcp {
    port => 9999
    codec => line

filter {
  kv {
    source => "message"
    recursive => "true"

output {
  stdout {codec => rubydebug}
  elasticsearch {
    hosts => ["elasticsearch-node"]

Coming up next… Kibana on Docker!