The cause of this unusually ~30 minutes long outage was due to intermittent nature of a harddive fault on our  secondary cluster node. This negatively affected nodes' communication and consequently performance of the primary while trying to resolve and recover.

Our system is resilient enough to handle any hardware failures, however in this instance a speedy recovery which normally lasts less than five minutes, dragged on for a lot longer.

A replacement for an entire RAID array discs is now in progress.

We value your custom and we are sorry for any inconvenience caused.

UPDATE 21/08/2009
==================
RAID array was now replaced.