But opsrecv and friends were never officially installed, so took care of
that too (conditional on PELABSUPPORT in the defs file).

from the GNU Radio tutorial to the usrp map.

that filters on 'usrp' for GNU nodes.

The monitor now correctly distinguishes between different connections on the same pairs of nodes and interfaces with dummynet to implement this. It is now a good citizen and resets all of the pipes except those used for initial conditions when a connection ends.

c. Add some logging and remove a complication from CircularTraffic.cc while it i
s under study for a bug.

This speeds up queries and gives a fully-connected set more readily,
but there is no guarantee the nodes returned actually have data specific
to their node name (only site).

Add federation workshop talks.

Change start event name to start_real and stop to stop_real.

Add support for SKIPLIST.

- automanagerclient
- opsrecv
- manager

defined (for now) as data exists within 6 hours (code param) in the
past for latency and bw (forward and backward) between all nodes in
the set.


example run:
perl goodnodes.pl -e tbres/pelabbgmon 10
where 10 is the number of nodes to ask for.
The -e option limits the nodes to look at to those listed in that experiment.

archived.  Rather then a special "archive" directory in the experiment
directory, we know archive the entire experiment directory.

This change should be backwards compatable, but let me know if not.

Note that the nsdata directory is gone; the nsfile comes from the
tbdata, but I know place a copy in nsfile.ns so that the name is well
known.

The packagedir is now passed to {ops,boss}-install with a -p argument instead of
as a PKG_PATH envar.  If the old way is done, PKG_PATH is ignored and packages are
all built from sources.  This failed in many ways (due to sync problems between the
6.1 packages and ports tarballs?)  In any case, to prevent others falling down the
same rabbit-hole, building from sources is now off by default, requiring a -s
argument to be given to do it.  Libinstall now shows PhaseSkip reasons, to make
it more clear what's going on.

Also tweak dhcpd_makeconf to handle a null DHCPD_DYNRANGE in the defs file,
to prevent a bogus dhcpd response from an inner emulab, half set-up for testing.

Removed duplicate declaration of make_timestamp in event-sched.h (also
declared in util.h).

that cannot resolve even their own name.

Put a check in the generated rc.plab to see if the node can ping itself
(i.e., that its hostname resolves).  If not, we put an entry for ourselves
in /etc/hosts.

This does not solve the general problem of a bad nameserver, but allows
our infrastructure to completely setup.  Whether this is a step forward or
not is debatable, but it makes me feel all warm inside.

Change timestamp output of event-sched to something human readable.

Also add a small amount of debug output to "$ns report" to help me
track down why it is being called at seemingly random places.

nodes have been removed from the topology.

graph info from a given path.

number of fake planetlab nodes - this allows us to remove planetlab nodes that
fail from the experiment, and *maybe* everything else will just work.

(@prefix@/etc/plab/plab.pw).

When a run is over changed:
  Template: 10424/23 (Current Run: bt-R54)
to
  Template: 10424/23 (Last Run: bt-R54)
Rather Than displaying
  Template: 10424/23 (Current Run:)

Also changed "Instance Bindings" to "Last Run Bindings" since that makes
much more sense, at least to me. (FS # 128:
https://users.emulab.net/flyspray/?do=details&id=128)
The function to show Instance Bindings still exists so it can easily be
changed back.