GitLab

Our little ball icons looked like they had tentacles coming out of the
bottom right side.  Maybe it was supposed to be a shadow or something,
but it was really bizarre looking.  So, I cleaned them up.  w00t!

asymmetry.

bottleneck detection. Jay/Rob/Jon/Kevin
are welcome to take a look and send comments or make changes.

Not real happy with how it is turning out however...needs more thought.

a new client.  Mostly replace talk of elvin with that of pubsub.

should work on FC4 too).  I haven't changed the linkdelay code -- updating
the kernel modules was enough for that.  The delay node stuff is new.

I figured it was worth the extra few hours to get delays fully working,
but I hadn't banked on the virtual descent into hell that is RED/GRED
parameters.

Things seem to be working pretty well.  Of course, RED/GRED queues aren't
tested.  :-).

     www/editexp.php3 - The reworked PHP page.
     backend/{editexp,GNUmakefile}.in configure configure.in - New backend script.
     www/experiment_defs.php - Add an Experiment::EditExp class method
                               bridging to the script via XML.
     db/Experiment.pm.in - Add an EditExp worker class method for script arg checking,
                           And a missing description method.
     sql/database-fill.sql - Add to the table_regex 'experiments' checking patterns.

Now uses a different row in emulab_indicies: frisbee_index, which is simply
incremented.  We convert the index to a MC address in the script rather than
saving an encoding of the address in frisbee_mcastaddr.  The new row is
automatically added the first time and the old row is still there in case
anything goes wrong.

Other changes mostly have to do with cleaning up when interrupted, in
particular when we are downloading an image inside an elabinelab.

This gives shared BW to all destinations from each source.

partial image.

2.4 to 2.6.  PLR and delays (the two schedulers we need) seem to be working
nicely, but need more testing.  Note that the iptables patch isn't tested yet,
but it's the right one to apply...  There is still a locking bug in the PLR
sched, but it only affects stats dumping so we don't care at the moment.

The two imq patches are from the linux imq project, like the old ones.

     www/newmmlist.php3 - The reworked PHP page.
     backend/{newmmlist,GNUmakefile}.in configure configure.in - New backend script.
     sql/database-fill.sql - Add table_regex 'mailman_lists' checking patterns.

Added file to reset the pipes for the elabc cloud.

Properly reset the pipes before setting them with the new parameters.

Used to be treated specially.

     www/newgroup.php3 - The reworked PHP page.
     www/newgroup_form.php3 - Removed, form merged into newgroup.php3 .
     www/group_defs.php - Add a Group::Create method bridging to the script via XML.
     www/showproject.php3 - Link to newgroup.php3, rather than newgroup_form.php3 .
     configure configure.in  - Add the newgroup backend script.
     backend/{newgroup,GNUmakefile}.in - Add the Perl script.
     db/Group.pm.in - Update the AccessCheck method to allow TB_PROJECT_LEADGROUP.
     sql/database-fill.sql - Add table_regex 'groups' checking patterns.

tell caller about it.

as a double by the RPC::XML library.

previous attempt.

of scripts - and a lot of CVS cleanup.

Also, added a check for variance in the case of Rubenstein congestion detection
code - the results are identical to those of wavelets in my Emulab experiment.
I guess we'll carry on working with both..

table entries but no wires table entries (e.g., the pc3000 eth1's)

to it.

worth seeing.