Check each individual counter for wrap before aggregating counters.

Use absolute mode of portstats (-c) to avoid counter resets at experiment
boundaries and at the whim of the experimenter.

images are in the image server.

locked down and didn't need to be).

If the app fails to launch after 10 tries, give up.

Refined the hack to detect the case where we wind up in progmode()
on a restart before the old child has completely disappeared.
Somewhere we still have a race that leaves behind capture zombies.
The party really starts when we run out of processes!

error.

sql/database-create.sql had the correct schema all along.

Note that the shared-node-listener should be doing this, but this
is easier.

a single choice (as per meeting last week).

what I learned doing the bridges.

clicking on their buttons. Also some fixes to hopefully prevent node
context menus from getting left behind (cause Jacks is not passing out
the event I need).

possible for people of only average intelligence, to create pretty graphs.

for all running experiments so that the web interface is updated. Normally
the event stream tells us about changes, but having some problems with
that, so put this in place until I get that figured out.

Hide announcements when seen_count exceeds max_seen or when max_seen is zero.

This is a hack. The info is extracted from the intermediate files we
generate when creating the "usermap" info.

feature on nodes that are prereserved to the same project as the
experiment, and add a 0.2 weight desire to the vtop file. Rob says it is
okay to always add this desire, so that is what I did.

experiment in the project, but without a prereserve setting, then set the
pre-reserve (as per Mike's request a few months back). But also need to
check that if the node is allocated to the project, and it does have a
pre-reserve, that pre-reserve must actually match the project and have the
right reservation name.

on the APT cluster.

A couple of Clemson machines have 47 disks each!