pulse: transition to new hardware

[Jonathan Griffin]

Currently, pulse is running on an unmanaged VM with limited capacity.
Dustin has been working hard to bring up a new managed cluster for pulse
in the PHX datacenter, which will provide us with some redundancy,
puppetization, better capacity, the ability to implement nagios
monitoring, and some other IT goodies.

The new cluster is currently undergoing testing. Assuming that no
problems are uncovered, Dustin will point pulse.mozilla.org to the new
cluster on Thursday, May 17th. We don't anticipate any downtime per se,
but the DNS switch may interrupt publishers and especially consumers,
which may need to be restarted after the change.

Any messages remaining in durable queues at the time of the switch will
not be transferred to the new server. However, at this time, there are
no such messages, and if usage patterns remain consistent for the next
10 days, there won't be any at the time of the change. (Currently,
messages accrue in durable queues only when the consumers for said
queues are disconnected.)

If you'd like to test your own code against the new server, it is
available at pulse-new.mozilla.org; all other details remain the same.

I'll send out another notice closer to the switch. If you have any
questions or concerns, please ping me.

Jonathan