pretty simple at the moment, there is a line for each agent and dots
on the line where an event was fired.

thought it would be cause of our current hacked up source code. I also
chased a cookie bug that is caused by domain confusion between inner
and outer emulab. No solutions for this yet, but its not likely to
bother anyone, except the dope that is trying to debug the inner
flyspray web site.

so that collab tool accounts get created.

warnings.

Changed the ipassign METIS assigner source to allow compilation under gcc 3.*. This should be enough for compilation under FreeBSD 5.* and 6.*

fetched yet.

Update description and defaults for the watchdog/rusage sitevar
(now in terms of seconds instead of minutes).

Plab connectivity monitoring changes:

* tmcd/tmcd.c

Updated connectivity log file names to include the date, and hence automatically
roll every 24 hours - much more friendly to the filesave, and easier to archive
off.

* tmcd/common/watchdog

Undo the conversion to minutes for the rusage time period (back to seconds)
to get the finer granularity Jay wants.  Not deployed yet.  Note that this
means the watchdog/rusage field will specify the number of _seconds_ not
minutes as it currently does.  I will update it's description to reflect this.

valid.

Built With Emulab logo.

redraw the ping-pong balls in openoffice so we have an easily
changeable figure; I got tired of pushing bits around with Photoshop.

The logo.sxd is just a big version of the ping pong balls. I then took
that and shrank it down to size to create builtwith.sxd, and added the
"Built with Emulab" text. Still didn't look quite right cause of
aliasing. So, I found the easiest thing to do was export it as a .eps
file, and then use ImageMagick's "convert" utility to create a .png
file:

	convert -antialias builtwith.eps builtwith.png

The antialiasing option makes it look much much nicer.

I have also changed the page footer so that when $TBMAINSITE=0, the
new logo is added to bottom of every page.

See http://www.emulab.net/builtwith.png

Add call "plabeventkeys" for listing all event system keys for the
experiments on a particular plab pnode.  Can only be called by the
service slice.

Returns tuples, one per line, for each experiment on the node like so:

PID=<pid> EID=<eid> KEY=<eventsys_key>

in the exports file to export the local filesystem to localhost.

source does not fix it; it just removes the -T argument. So that is
what I am doing.

public stuff needed on ops (cvsd).

pseudo users, pull next_uid out of the db instead of relying on the
next avail unix uid.

We used to do this (circa FreeBSD4.3) with a kernel hack, since then
we have just been logging console output again.  Cease and desist!

* Add mailman to the install process. This includes the package
  install, all the setup needed in ops-install, and the little patches
  we apply locally.

  Note: Only under 6.0: no point in going backwards since its a pain
  in the ass to build these packages for the 4.10 image.

* Push more of the package install back down into the ops/boss install
  scripts. This improves testing within those scripts, and avoids
  duplication. There are some changes in rc.mkelab that will temp
  mount the package dir on /packages for the benefit of the scripts.

scripts happy.

Fix tied STDIN and STDOUT with perl 5.8 in libtblog.

Fix a phototype warning in  os_load.in

Creating the groups for main group is enough for testing.