GitLab

you can do this in your NS file:

tb-set-node-loadlist $myboss FBSD73-S2,FBSDXX-FOO

Note that this does not replace tb-set-node-os(), cause that is a
little too special. So in practice, you would do:

tb-set-node-os $myboss FBSD73-STD
tb-set-node-loadlist $myboss FBSD73-S2,FBSDXX-FOO

USERNODE, shortcut to exec only, without the ssh.

"/dev/da0" notation.

(jail) on boss. A typical NS file might look like this, although note
that windows support is not finished yet. 

source tb_compat.tcl
set ns [new Simulator]

tb-elab-in-elab 1
tb-elabinelab-singlenet
tb-elab-in-elab-topology opsvm
tb-set-elabinelab-attribute CONFIG_WINDOWS 0

namespace eval TBCOMPAT {
    set elabinelab_maxpcs 1
    set elabinelab_hardware("boss") pc3000
}
$ns run

    
    OPSVM_ENABLE=0
    OPSVM_MOUNTPOINT=""

OPSVM_ENABLE=changeme
OPSVM_MOUNTPOINT=changeme

CreateVnodes().

worked in the "new" version.  (It continues to work fine in
os_setup (no new).  The new version is used primary by GENI related
and Utah projects.)

instead of two slashes, if I want a single slash in the DB.

2. Add a new inner_elab_role; boss+fs+router which makes boss the FSNODE
3. Add loadlist to both virt_node and nodes for plumbing multi image load
   through from the NS parser.

Part I of "Lessons learned from a 100K node experiment". nfree would
run for hours trying to free 100K virtual nodes due to a slow query in
Node::ReleaseSharedBandwidth(). Now it takes 5 minutes.

There were other lessons as well, but they fall in the category of
"rearranging deck chairs" since we are going to have to massively overhaul
lots of infrastructure to support O(100000+) node experiments on a regular
basis.

This script is intended for use with 'git filter-branch' - it takes
a directory, finds all files that have been moved into that directory
and where they came from, and produces a script that moves (in the
filesystem sense, not the git sense) them from their old locations
to their new ones.

Then, you give this script to 'git filter-branch --tree-filter', and
it will re-write all history so that it looks like the files were
always in their new locations.

Written specifically for the move of the clientside stuff into its
own directory, and then off into its own repository.

ClearingHouse certificate.

so make them peoper Version 3 certificate. Add a driver script to handle
all the details of updating.

Delete cached ClearingHouse credential.

certificate object.

Lahey.

Power saving turns off nodes that have been sitting in PXEWAIT (and
are thus free) for more then a set amount of time (see sitevar
general/idlepower_idletime, which defaults to 3600 seconds).

The driver script is tbsetup/idlepower.in and needs to be added to
/etc/crontab at sites that want to do this. Even so, operation is
enabled by the sitevar general/idlepower_enable. Each time it runs, it
checks for nodes that need to be turned off, and then calls power.
Note: This should be a daemon not a cron job.

To be considered for power saving, you must add an attribute to the
node_type_attributes table called 'idlepower_enable', set to 1.

Locally, I hacked up stated and power to make the state transitions
legal so that stated does not whine. I added POWEROFF as a valid
transition from any state, to opmodes NORMAL, NORMALv1, and NORMALv2.
Barry's original patch already had a state transition for PXEKERNEL.
In power, I added code to look at the actual operation, and in the
case of "on", do not send an event if the node is not in POWEROFF,
since a user can foolishly say power on anytime, and the node is on
nothing is every going to change, and the state transition would be
wrong.

node_reboot takes of powering nodes on, when they are in POWEROFF.

Barry on copyright issues:
 "I'm not sure those rights are mine to grant! Remember that this code
 came originally from Kevin Lahey (kml@patheticgeek.net) and
 originated at DETER (although he's apparently not there anymore). I
 don't foresee a problem from our point of view (but I'll double
 check, of course). Shall I try to contact Kevin try to sort this mess
 out, or do you think it's better to coordinate from your end?"

which says that any source state is okay, for the specified target
state. Basically a convenience.

capturing and returning all output. Convenient.

clientside requres it to be there. As noted in previous revision, the
full emulab build always includes the clientside subdir, but a
clientside build will not necessarily include the rest of Emulab.

This is the client reorganization! All of the code that makes up the
client side (the stuff we install on images and mfs's) has been moved
into a single place; the clientside subdir. The eventual plan is to
turn this into a separate git submodule, but for now it lives inside
of the emulab repo.

There are two modes of installation.

1. Build as part of a full emulab build; The top level configure and
   makefile operate as before; the clientside directory is built and
   the client and client-install targets do what they always do. Ditto
   the various mfs and subboss targets.

2. In the clientside directory, there is a new configure script that
   covers just this subdir. chdir into clientside, run configure with
   no options or defs file, and then build/install. This will be a lot
   esier to handle for user wanting to simply update existing images
   with the latest clientside.

Notes:

1. I used "git mv" so the history will follow the files when you do a
   git log, but only if you use the --follow option.

2. configure has been updated to 2.63 cause the old one we were using
   did not support some of the substitutions I wanted.