GitLab

tree, cause /usr is kinda small.

Munge the defs file so that WWWHOST=www instead of boss; we do
this on elabinelab cause there are no CNAMES for www, but there
will be a www CNAME on the racks.

Minor fixes to how we invoke mknewconfig.

/usr is kinda small as it is.

of the A slot, which is card=5 for us. Not sure if all racks
will be this way.

It constructs an experiment including every (available) experimental PC,
and every relevant link, so that during swap-in linktest will exercise
as much of the testbed as possible.

Also add state for tracking individual object attributes.

DB is fully populated with nodes. Just need to power them on.

Fix typo in update 333

Conflicts:
	install/genirack/GNUmakefile.in

These were always getting out of date otherwise.

Most instances of this had been fixed, but not the one in the OpsJail case.

mknewconfig.in now invokes newscript.in directly on the files it generates.

Uses the "5.0" package set (like FBSD 9.0, but unlike FBSD 8.2) which
includes perl 5.12, python 2.7 and apache 2.22. This is what will be
installed on our boss and ops later this month.

Some additional updates to the meta ports as well to make them "work better".

Change the logging code so that we get better output on early failure.
Openvz nodes do not have a "console" so early errors are lost. This is
a real pain, cause before this point, we have no idea what went wrong.
Later versions of openvz are supposed to have a console.

the control network, and the packets were getting nat'ed.