Frontend services now on new replicated infrastructure

I’ve switched our frontend services (IMAP, POP, Authenticated SMTP, FTP, Web) over to two new machines in a replicated infrastructure.

Each service has two IP addresses - which are prioritised to run on different servers - but if either of the servers fails or is shut down for any reason, the IP address will migrate quickly (maximum timeout, 20 seconds) to the other host.

These machines will be placed (eventually, still waiting on cabinet migration plan to complete) in different cabinets on different power supplies, giving us much greater safety against power failures in future (though our IMAP servers aren’t replicated yet - so there are still failure modes that take us totally down)

Anyway - nice to be started on the path to full replication!

Bron.