relational database.  For now, this is done using some GROUP_CONCATS
and hierarchies of VIEWs, since mysql 5.x does not yet support subqueries in
VIEWs.

connections and parsing mig-generated packet specs.

. Force reboot, or at least sleep, when attempt to run  rc.reboot gets fork: Resource temporarily unavailable.
. In rc.ifc generated script, check both netsh and ipconfig IP addresses for 0.0.0.0 .

descriptor each connection is using - this will aid debugging if
we discover that I'm somehow leaking file descriptors. An old
version of the code seemed to, but I believe it's fixed now.

croak(). Instead, the client is dropped, and an error is logged. If
we get too many errors (10), we bail. But, if we go a while without
getting any errors (60 seconds), we reset the error count. This is
so that if we have any bugs (or the node is just totally fubared),
we'll exit, but long-running daemons won't just accumulate the error
counter forever.

Also, handle a rare (but, on plab, who knows!) DOS-type condition with
blocking accept() noted in Stevens.

No idea why that fixes things, but it does...  CSS, from hell's heart,
I stab at thee

After being asked for the 10th time "The Cyclades Z serial mux is discontinued
so what do I buy instead?" I decided I had better go out and update our
recommended hardware list for the first time in 5 years.

of date.

needed (reparsing of NS file with current bindings).

Mostly I decided that I made a huge mistake in using the exptidx as the
main cross index. I started to change that, which took a bunch of time.
Still more tables to fix up, but I think all of this code is gonna need
a major reorg once I am no longer under the gun to get it done.

turned out to be another file I committed by accident.

anyway due to lack of IsTemplateInstanceExperiment() function.  The function
exists in libTemplates.pm, but is not exported.

Since I'm not sure what is going on, I have just changed the "if (Is...)"
to "if (0 && Is...)".

and they appear to install and work okay).

iperf: 'k' = * 1000, 'K' = *1024, etc.

Also, some command line args were not calling the correct function to
parse their integer arguments.

assign.

it removes, and decrements the available slot count for every node
in a pclass that is being removed.

After running prune_unusable_pclasses(), we now do another type check,
minus the pruned pclasses.

This should enable us to detect more free node problems up front -
let's say there are 8 vnodes that all want 2 interfaces, but only
a couple pnodes that have 2: most have 1. The second type check will
now catch this, since all of the 1-interface pnodes have been pruned.

as I think: if the proto is foo($$;$$), can it be called with three params,
or must it be 2 or 4?

Use that option in the undoFW function to make sure we don't try
to cleanup virtnodes.

Continuing work on the creation of experiment runs. Web interface in
now in place (create and display), and the backend part that deals with
setting up the DB records. Nothing actually happens on the nodes yet.

now in place (create and display), and the backend part that deals with
setting up the DB records. Nothing actually happens on the nodes yet.

conversation with Rob and Kirk yesterday.

an alternate "filename" for the dhcpd.conf file.  "Inner plabs" are likely
to need this (to boot pxelinux for emulating the plab boot CD).

dhcpd_makeconf picks up the DB field and sets the filename field accordingly.

nfree will invoke dhcpd_makeconf if any node being freed has a non-NULL
pxe_boot_path value, in order to reset it.

like templates won't be possible in elabinelab unless its a 6.0
elabinelab (which is my default these days).

Add a 'Syntax Check' button like the one on beginexp so that you
can syntax check without loosing all your edits if you had
syntax errors.

Add a file upload box and an option to use an NS file on ops,
like the beginexp page has.

Maybe we should try to find a spiffy ajax editor for this page?