Major Upgrades
This week will see the upgrade of all major systems and a shift in kit / location.
Due to recent changes in prices and resourcing our current data center has decided to increase their prices quoting an increase in power costs and their main justification. Given we have had a few issues recently with on-site power and that the increase was of the order of 350% we have decided to relocate all servers to an alternate data center.
This should take place the week beginning the 7th of January with completion by the 14th.
As part of the upgrade we are taking the opportunity to upgrade the servers and move to a more robust cluster-based system which is flexible and attuned to providing scalable hosted services.
All processes are moving across to XEN instances running on XEN 3.1 hypervisors while all data is being moved to the Gluster clustered filesystem.
Although we can't give the exact times services will migrate, there should hopefully be no noticeable downtime on any core services are we are running duplicate sets of kit in each physical location. (down time should be limited to service restarts)
If you do spot any problems over the next week, please let us know asap and we will investigate.
Update (Mon 2am)
DNS2 (Live, taking 90% of DNS+Mail)
ZEO (Ready, awaiting web module)
Monitor (Live, running ZenOSS)
Zimbra (Ready, awaiting DNS switch)
MySQL Cluster - online, 3 Data nodes
GlusterFS (nodea) online, running all of the above
XEN online (nodea) running all of the above
VPN online (nodey)
Complete Email switch expected tomorrow + all V3 Plone instances.
Live mail switch (Tue, 4:30pm)
All DNS for mail systems should now be pointing at the new site and all mail should be synchronised (another sync is running atm). Looks like most name servers have picked up changes immediately .. so we should be live!
Switch-over "complete" ... (!)
All looking good so far ..
Current Status (Sun 19:00)
Currently MX1 is now pointing to the new DC and is taking half the incoming email load, watching for issues at the moment. Waiting for support to add reverse lookups for the new mail servers before pointing outgoing load in the same direction.