Add the ability to specify the set of nodes you are intersted in to
the node.getlist() function.  Do this by adding a
"nodes=<node_id>[,<node_id>...]" argument.

connect to magent before it (magent) is ready. Use the option Jon
added to magent to make it daemonize *after* it sets up ports, etc.
The changed to auto-magaent.sh should have taken place long ago - I
don't know why it didn't.

conditions) plab nodes without actually allocating them.

Changed init-elabnodes to support this.  Elab to plab mapping is set in the
NS file via the $opt array, which init-elabnodes extracts via the XMLRPC
interface.

So to run the "really simple model":

 1. set "no_plab" to 1 in the NS file
 2. make sure you have enough nodes listed in "plabnodes" in the NS file
    to cover "num_pcs".
 3. swapin the experiment
 4. after swapin, run init-elabnodes.pl.

Also, cleaned up the NS file some: got rid of hokey tcpdump setup as it
was redundant, made tracing optional (though on by default), expect proper
return values from program agents.

arguments - we haven't been using them.

* Create a per-experiment DB user for the per-experiment DB; the user
  name is equal to the DB name.

* Add a dpdbpassword field to the experiments table; this is the
  randomly generated password for the DB user mentioned above.

* For Templates, use the above user/password in the environment,
  instead of the swapper uid/password.

* Add experiment dbname/dbpassword to the Show Experiment page.

per-experiment user and password for the per-experiment DB.

- managerclients send their measurement request commands to the manager
  through the event system.
- the automanagerclient sends its measurement requests to the manager
  using TCP.
- The manager is the only entity to send measurement requests to the bgmons
  running on the nodes.
- A queuing system on each bgmon orders requests based on test frequency,
  so that the fastest frequncy measurement is used.

the pid,eid and passing that to the XMLRPC server to be mapped to the
template.

template record to replay an instance, but all it really does is
replay the first run. There is also support to replay a specific run
in an instance, but that is not hooked up to the web page yet.

Note that the bulk of the replay code went in by mistake a couple of
days ago.

Added some extra debugging information. Better division between packets and a packet time stamp at the beginning of each packet.

 * cloudinfo.in: change to handle hybrid (split BW/delay handling)
   delay pipe setup, also a hack to allow you to run the .in file out
   of your source tree

 * init-elabnodes.pl: handle hybrid cloud/lan model, use xmlrpc to find
   type of plab nodes and setup delays accordingly

 * auto-*.sh: continued attempts to get a meaningful exit status out
   of them for the program agent--still doesn't work

 * run-monitor-libnetmon.sh: make sure initial conditions get properly
   propogated

New least squares time sensitive code. LeastSquaresThroughput now gets a single average throughput metric over the whole period of the samples and uses time as the x coordinate value for the linear regression.

Second attempt to fix the problem of duplicate log entries.  I am
99.99% sure this will get 100% of the cases, and 99.999% sure it won't
break anything.

It basically detects when the DB handle is a child and if so set
"InaciveDestroy" before the database handle DESTROY method is called.
Since the DB handle can be closed in several different places I created a
new class to override the Db Handle (the Mysql class) DESTROY method. The
other alternative is to add special code anywhere where the database handle
could be destroyed which is when every a reconnect is done and when the
module exists.  The later would have involved putting code in the END block.
I think the new class method is simpler for that reason.


Also, add a note about patching Mysql.pm in doc/UPDATING.

it.

is something to debug.

phases of a run. Eventually we want to remove dead nodes from
consideration, but this will do for now.

globbing since that was not working properly (not sure why, but this
is more efficient anyway.

Add backup routine to back up failed template swapins to .$eid-failed.

Don't completely prevent the use of plab nodes that are low on disk space.

Convert plabmetrics over to using CoMon data.

after a swapmod goes haywire.

after a swapmod goes haywire.

  info so we have a record of it.

* First attempt at dealing with nodes that do not respond to the
  synchronous events that are sent from start and stop run. Rather
  then failing, attempt to figure out which nodes are actually dead,
  and save some state in the DB associated with the run. The current
  method for figuring out which nodes are dead is the node_status
  table, since the event scheduler is the only thing that knows what
  nodes did not respond. Will probably revisit this very soon.

* Bug fixes of course.

* Start implementing a Run object so replace some of the code in the
  Instance object.

we have a record of what nodes failed to respond during the
synchronous parts of the start and stop experiment run. Data is stored
for now, but not sure how it will be used later. The main goal is to
avoid the problems we see with plab nodes where they do not respond,
and start/stop run hang and fail.