Home Monitoring Upgrade (Part 2) – HSQLDB to MySQL

As mentioned in my previous post (see http://blog.v-s-f.co.uk/2017/04/home-monitoring-upgrade/), the first task I have is to migrate from HSQLDB to MySQL.

Because the system logs data every minute while I have power in the house and I want to minimise downtime when I actually have a fully working upgrade, I’ve experimented with a copy of the live HSQL database.

Once I’d copied it from my server to my laptop I then attempted to view the data in notepad++ – yeah, not a particularly smart move! NP++ cannot handle files over about 100M. It also turns out (having more’d the .data file) that the data is not in a readable state. I had a performance issue with a select query a long time back and changed the table to cached.

So, luckily at work a colleague had introduced us all to a great database tool called SQL Workbench. It can work with most databases and unlike SQuirreL, it doesn’t crash when looking at the work DB2 database.

Using SQL Workbench, I’ve then created a script file which creates the new consolidated table and loads data from the old tables in to the new + drops the old tables. The end result is a 122,398KB HSQLDB script file which is human readable.

Next step was to get MySQL running on my server. Instead of installing it directly though, there’s a Docker image available.

My first few attempts at inserting the data from the HSQLDB file in to the MySQL database were less than impressive! One of the attempts had the server running flat out (100% cpu) for over an hour when I finally decided that it was probably not going to complete the import this year and nuked it!

So having learnt a few lessons about not using single row inserts(!), but batching them in to multi-row inserts of 100,000 and a few MySQL deafult parameter increases (although I’m not sure if these are necessary as the batch inserts seemed to make most difference), I was finally able to import the data. It still took a few minutes from running the MySQL container to it being available – but that’s significantly better than running for hours importing the data!!

Now that I have the necessary scripts and knowledge to migrate the data, the next part is re-writing the application that receives the Arduino data, uploads to PVOutput and serves the hot water display Arduino.