Right, should be back in action now.
The immediate cause of the issue was a full root volume causing the database process to eventually die. The cause of the full root volume was an automated job that runs every two minutes to scrape the Alerts data from SW2. This job utilises an external library to parse the HTML via a regular expression matcher. The expression used by this external library is not semantically correct for the more modern version of PHP from the recent upgrade. This result in PHP errors which are reported via cron as an email to the user that runs that job. These emails were approximately 1MB in size each, and over the course of several days, filled the volume with 2.6GB of errors.
The immediate resolution to the problem was to remove the emails, restart the database server and disable the automated job. The final resolution will be to rewrite the automated job to use the API rather than screen scraping the HTML.