With apologies to all those non-broadband trogs (use green dot if you want
to download fast) :-)

in boss/ops install scripts.

sending the request, otherwise two connections are made (doh).

bit in the exception message.

TBImageIDAccessCheck, otherwise they seem to return true all the
time...

it (thereby creating a file of that name) ...

so as to trigger an accept failure.

Also, statically link capture so we don't have to worry about installing
SSL and friends on tipserv machines.

autologin of the swap-in user, and a user sshd on port 2222.

elvind/libelvin

       column in the reserved table and not the nodes table:

Also fix a cut/paste error and renumbed some items, we went from 1.279 to
1.270 and started counting up again.

made me look bad in front everyone. I really hate it when that
happens.

and the special multicast addresses.

exist; fire up frisbeelauncher early to see if it can get the image.
This is harmless in that frisbeelauncher can be called multiple times
anyway, so when its called is not really a big deal.

download images from the outer emulab. This script is invoked from
frisbeelauncher when ELABINELAB=1 and the filename does not exist
(thus attempting to get the image file before bailing). The
frisbeeimage script uses a new method in the RPC server to fire up a
frisbeed (using frisbeelauncher on the outer Emulab), subject to the
usual permission checks against creator of the elabinelab experiment
(I assume that the creator will have access to any outer images that
are used inside the inner emulab). If outer frisbeelauncher succeeds,
its return value is the load_address (IP:port), which is used to fire
up a frisbee client to get the image file and write it out (using
Mike's new -N option that just dumps the raw data to file). Once the
image is downloaded, control returns to inner frisbeelauncher and
proceeds as normal.

I whacked this together pretty quickly. Under heavy usage it might hit
a race condition or two, but I do not expect that to happen in an
inner elab for a while.

(like stated).

getting images for ElabinElab.

- add some missing casts
- avoid void * arithmetic
- change "q" printf formats to "ll"

Still not there yet by any means.  Note that the version of the NTFS library
we use won't compile with "-pedantic -ansi" so those are not part of the
default CFLAGS.

configure so that we can localize it for inner emulab. Not good to
overlap the address space between inner and outer emulab if inner
boss is going to use frisbee client to suck images from outer frisbeed.

This patch file is part of the RHL90 image build procedure.  It probably
doesn't belong here in the `doc' tree: it belongs *somewhere*, but nobody seems
to know where exactly.  Until a better place is found, here 'tis, alongside the
build script.

Minor improvement to the node.available() method so you can get the
number of free nodes of a particular type (e.g. pc850).

  * xmlrpc/GNUmakefile.in: Add node_avail symlink.

  * xmlrpc/emulabserver.in: Add some optional parameters to
    node.available() so you can specify the class/type of nodes as
    well as project credentials.

  * xmlrpc/script_wrapper.py.in: Add node_avail class that calls
    node.available() with the given parameters.

to the list of ALLOWED_PATHS.

This is a copy of `doc/updating-RHL.txt' r1.2, modified on Nov 12 2003.

I'm committing it now, before I start trying to clean it up.

This change makes the configuration more portable between testbed sites.