the case that they didn't give us a hostname.  It just mocks up a
bogus hostname of dhcp-ip.wa.emulab.net and sends that along.
The substantive change is that it now prefers extracting the IP
address from the ifconfig for the interface instead of via hostname,
which is pretty unreliable.  But it'll try both.

from the burned-on-CD register.pl script into the downloaded
netbed-setup.pl script.  Doesn't actually change anything, but
a CD burned with the new register.pl and intended for DHCP could
be fixed by a revised netbed-setup without requiring a new CD.

This is a transition commit to fix some minor brokenness and
inflexibility before I really revamp things to enable DHCP support,
which may require a v5 CD instead of incremental changes to the v4 CD.
We'll see...

whole topology, we run it once for each type of node in the virtual
topology. This way, we can do a much better job of coarsening when we
have multiple types of nodes that have different packing factors.

the NSE scheduler slop factor to be controlled from the user-specified
make-simulated code. Also added a null tb_compat file so we can
ignore any tb-* commands that get specified inside make-simulated.
These commands are anyway evaluated during the initial parse.

( ~ 4 mins on the 416 topo ). This was also driving up the load on
boss. After the fix, it takes 20-30 secs of tcl-to-tcl conversion that
happens on ops. Hardly any noticeable load on boss as well.

off linktest async via the web interface.

into a file in /proj/<pid>/exp/<eid>/logs/nse-vnodepktrate-<vnode>.log.
This proc is called from C++ if the NSE RT scheduler declares ``violation''
and before sending an NSESWAP event.

also stores previous mapping info in the lock file for debugging/data
collection. max_retries for reswap is also set high so we pretty much
go until some vnode gets mapped one-to-one and if that does not track
real-time, the experiment fails. There are also a couple of sleep
statements with times determined empirically so there is a chance for
as many pnodes to report ``violation'' as possible. All these changes
are for the ''doubling vnode weights'' method of sim auto-adaptation.

the nse instance. XXX Current hack is to do a killall since the pid file
was not stored by startnse. Will fix later.

takes a while to finish. Without this, later NSESWAP events from other pnodes
were being ignored since the event scheduler was blocked for a long time.

corresponding simhosts. For sim nodes, this is a useful way of
updating the client side scripts and nse binary during development
without updating disk images or using cvsup.

copyright

we need to make sure www/ is done before event/, because the latter
tries to install some web documentation.

their ports.

that requires some stuff in rc.local to enable them.

a disk image in case the image isn't stored on the boot CD
(needed to handle heterogenous boot CD media).  netbed-setup.pl
now tries this alternate location if the local disk image
is missing, old, or corrupt.

even if the disk is up to date.

Add a sliceNN_alt_image pointer to a web location for downloading
a disk image in case the image isn't stored on the boot CD
(needed to handle heterogenous boot CD media)

even if the disk is up to date.