from other places.

to old assign and friends. The installed version was checking the
nodes table, which is no longer updated. See, thats why rtagging the
entire tree is so nice ...

Cisco module commits.

* Add vnode0,vnode1 slots to the delays table. This will allow us to
  match ipfw pipes to nodes after swapin, hopefully allowing us to
  control the delays for lans in addition to duplex links.

* Add IPaliases to interfaces table. Used in assign_wrapper when
  constructing "emulated" links, which will share a link via the use
  of aliases on the interface. This is a comma separated list of IP
  addresses (no, I refuse to make this a separate table!).

* Add virtnode_capacity slot to node_types, defaults to zero. This is
  used in ptopgen for determining how many virtnodes fit on a real
  node. I have not thought this through completely, but it allows me
  to make progress on other fronts.

* Add linkdelays delays, which sorta resembles the delays table, only
  this table stores oneway delays links, to be set up on the endpoints
  of a link or lan. Also, instead of being based on ipfw bridge rules,
  it is based on IP address/mask rules. At some point this table may
  merge witk the delays table, but will take time to work out the
  details and I do not want to mess up existing experiments by
  changing the delays table! Anyway, a duplex link gets one of these
  for each endpoint (a xmit ipfw rule). To mimic our lan setup, lan
  links get two, an xmit *and* a recv rule. My hope is that link
  delays will look just like normal delays (a packet leaving a node
  for a lan will get the outgoing delay, and a packet arriving gets
  the incoming delay).

the RCS control file in the repository so the history is left intact.

Two new modes, which used to be in the old mkacct. There is an init
mode, which is used on new users to create the initial pub key. There
is also a write mode, which is used regenerate the authkeys files for
people after they add/delete keys via the web interface. Used to be
that addpubkey wold add the key to the DB, but mkacct would deal with
creating the authkeys files. All this functionality is now localized
in this one script.

parsing will move from php to this script, like addpubkey (ssh), but
for now all it does is invoke the script to rebuild sfs_users on ops.

time to debug a delayed LAN problem.

script, which is now called tbacct, and lives in the account directory
instead of tbsetup (all account scripts are moving into this
directory). The command line is different:

	Usage: tbacct <add|del|mod|freeze|thaw|sfskey> <name>

However, it is not really intended to be called from the command line,
but if it is, it always does the right thing based on the DB. All of
the ssh commands are localized here as well (mkproj and others will
invoke this script instead of doing pw commands themselves on ops).

My experience with this indicates a couple of things.

* We should probably not invoke these backend scripts (which are
  setuid) as the user driving them from the web. This complicates
  things, especially in light of having to deal with users with no
  accounts (say, a new user, unapproved, who wants to change their
  password). Not sure what the right model is, but since the script
  always does the right thing, it really makes no difference who
  invokes it.

* The actual pw commands should be driven from a script on the other
  side. This would make it easy to retarget to linux or whatever. I
  thought about doing that, but the shell quoting is a pain in the
  butt, and its not like I'm supposed to be doing this stuff.

Add a limit=N option for controlling how many nodes returned.

VLAN domain. This means that we have to create and delete VLANs on
all switches, instead of just the 'leader'. This behavior is
controlled through the new single_domain column in the
switch_stack_types table.

Added Alloc State machine library calls.
(NodeSetAllocState, NodeGetAllocState)

Most snmpit_cisco functions now take Cisco-specific VLAN numbers,
rather than testbed VLAN ids.

This is an intermediate step in removing dependance on VTP.

As a nice side-effect, this saves another 25% of the time to
operate on topologies for large experiments.

for large experiments. Not as much of an improvement as I had hoped,
but further improvement will probably take a lot more rearranging.

it was worth, even if that stupid query was really slow.

"active" was changed to mean experiments with reserved nodes,
regardless of its actual state. The result is that the default view
would return every single experiment, and then we would prune out
later, but not before we did a second query per experiment, which made
the admin version of the page really slow!

This query was duplicated in the menu page (which I just fixed) to
determine if a "new" icon needed to be displayed.

warnings.

It is still used when doing a debugging build.

In addition to assign, assign_wrapper and ptopgen have been modified
to output the new formats used by assign.

but eventually did not use it. Current images still make a call to it
from libsetup but will get an empty string that doesn't affect anything.