Upgrade of the primary CloudLab/Emulab servers

[Mike Hibler]

The time has come to upgrade the hardware of the two main CloudLab/Emulab
server machines ("boss.emulab.net" and "ops.emulab.net"). In the process,
we will be upgrading the OSes as well.

This is a big task that will take a long time just to move the data
(10,000+ user/project filesystems on the ops node), and as such I would
like to dedicate two days to doing it. Hopefully, it will take less than
one, but...stuff happens.

During the time that these servers are down, experimentation on the testbeds
will be affected. The primary CloudLab/Emulab portals will be inaccessible
and you will not be able to start new experiments or terminate existing
experiments. Experiments already running on the Emulab cluster nodes will
likely hang due to loss of the shared /proj filesystem or other centralized
services. Running experiments at the Clemson/Utah/Wisconsin clusters will
likely continue to run, but you will not be able to use the web interface to
gain console or shell access, or power cycle the nodes. You should still be
able to ssh in directly from a remote machine.

Anyway, I am proposing to do this on Wednesday and Thursday, December 19th
and 20th (two weeks from yesterday).

Let us know if you have serious objections to these days.

Mike