►
From YouTube: OpenCrowbar Design Meeting 2014 07 09
Description
http://pad.opencrowbar.org/p/community-design-meeting-2014-07-09
Topics included WSMAN, in-band, out-of-band and Cloud Jig for Crowbar
A
Hello,
everybody
welcome
to
the
July
9th
crowbar
design
meeting.
We
have
a
huge
agenda
of
stuff
to
cover,
so
I
was
going
to
have
us
dive.
Give
some
brief
update
to
where
we
stand
and
then
talk
through
design
updates
and
try
to
try
to
focus
on
things
that
are
immediately.
It's
moving
to
resolving
issues
that
are
immediately
tripping
people
up
and
we'll
just
push
forward.
Think
they'll
make
it
okay
Victor.
Do
you
want
to
do
an
update
on
where
you
are
on
the
stuff?
You
did
yeah
so
last
night.
A
My
kid
woke
me
up
at
two
in
the
morning
and
I
couldn't
go
back
to
sleep.
I
started
working
more
on
wsn
code
that
I've
been
working
on
it
for
the
last
few
days
and
I
finally
got
to
the
point
where
the
pull
request
that
I
had
up
for
just
doing
power
management
via
the
vs
man
is
actually
using
the
native
Ruby
open
w
span
bindings.
A
So
it's
not
calling
out
to
whistle
or
dennis
van
CLI
or
any
of
those
other
tools
to
talk
bilious
man
at
servers
that
supported
it's
actually
using
the
native
Ruby
bindings
and
talking
via
the
open,
WS
man,
libraries,
which
reduce
nan
Batian
for
building
on
the
more
w
network.
You
know
when
you
start
actually
getting
into
talking
to
the
lifecycle
controller
on
dell
boxes,
to
flash
raid
and
flash
the
flash
the
firmware
and
manage
the
raid
controllers
and
you
mess
around
with
the
bios
settings
and
that
sort
of
thing
so
yay
progress,
Rob.
A
It
powers
inked
on
and
off
and
the
lights
blink
when
you
ask
flights
to
blink-
and
I
also
test
it
against
a
virtual
machine
because
I'm
trying
to
make
sure
that
the
ws
man
calls
don't
interrupt
our
flow
on
virtual
table
and
they
don't
so
oh
and
I
think
I
spent
up
a
bunch
of
dr.
nose
against
the
same
case
so
yeah
it
looks.
It
looks
pretty
good
for
that.
A
A
A
So
that's
going
to
stand
with
WS
million
right
now.
The
reason
just
got
so
excited
was
very
much
Jeff.
Uptown.
Sorry
I've
only
had
my
first
cup
of
coffee.
B
B
A
A
Was
that
our
current
disk,
no
troll?
Is
it
really
an
inventory,
no
troll,
and
it
it
figures
out
what
you've
got
and
we're
going
to
allow
you
to
sort
of
immediately
assuming
you
have
an
inventory,
make
some
decisions,
but
you
have
to
make
those
assumptions
or
you
can
really
do
anything.
What
we'd
like
to
be
able
to
do
is
treat
disks
more
like
assets
and
have
rolls.
A
Rolls,
and
with
them
downstream,
so
that
you
could
add
a
disk
roll,
build
a
raid
set
for
an
OS
or
build
a
cash
that
cashed
it
447
tough
deployment.
Rather
than
doing
it
all
the
way
back,
give
you
achieve
yeah
for
those
that
are
familiar
with
how
networking
and
probe
our
works
it'd
be
kind
of
like
how
we
do
late
binding
with
your
network
layout
in
chrome,
won't
actually
create
custom
interfaces
that
do
tag
vlans
and
such
a
little.
A
We
could
do
it
some
other
thing
with
this,
but
only
if
we're
doing
it
out
of
in
and
there's
a
serious
of
architectural
trade-offs.
So
you
have
to
talk
about
to
make
sure
that
all
the
crazy
you
stuff
I
just
talked
about
also
doesn't
make
it
impossible
to
do
inbound
disk
management,
because
we're
going
to
need
that
for
everything
that
you
know,
we
can't
just
talk
to
the
great
controller
directly
without
film
being
in
the
operating
system
for
having
the
Buddhist
pledge
remember
to
use
a
mega
CLI
or
that
sort
of
thing
we've.
A
Don't
know
exactly
how
to
capture
all
this,
but
it's
worth
understanding
that
you
are
I,
think
we've
got
a
short-term
workaround
you're
used
to
crowbar
one.
Basically,
for
you
install
the
OS,
you
have
to
make
a
whole
bunch
of
of
disk
provisioning
decisions
and
with
true
out-of-band
management.
We
can
actually
do
things
even
after
the
OS
is
up
or
we
can
do
things
at
different
visions.
B
C
A
A
A
A
Wrote
their
own
IP,
my
native
IP,
my
junk
in
Python,
but
there's
talking
like
lamp
luff
straight
at
the
ipmi
controllers.
To
do
it
and
I
believe
rocks,
has
this
capability
and
anything
was
talking
to
pm's
instead
of
physical
hardware
also
has
the
capability,
because
it's
talk
to
the
hypervisor
yeah.
From
that
perspective,
it
takes
you
closer
to
the
virtualized
infrastructure,
yeah.
A
Let's
cross
over
that
topic
for
a
middle
and
I
know
all
the
pike
it's
all
do
and
what
is
the
name
of
it,
not
clouds,
but
the
other
thing
to
use
much
machines
to
do
know
and
I
do
not
cloudy
things
to
do
with
a
cloudy
hpc.
Yes,
those
people
they're
all
so
crazy
and
they
wrote
free
ipmi,
which
is
an
entire
suite
of
IPL
I
management
tools.
That's
oriented
around
having
one
command
reboot
a
hundred
machines
and
was
able
to
do
yeah.
A
B
A
Why
would
they
talk
the
idea
my
protocol
and
I
can
configure
them
I,
don't
really
care
how
they
don't
really
care
where
they
said.
It
means
that
we.
A
Right
now
we
detect
that
you're
running
on
an
ID
Act.
We
switch
seal
and
interface
over
to
dedicated
loaded,
otherwise
be
coat.
We
just
leave
it
at
whatever
was
so.
You
can
send
up
to
run
over
share,
so
it's
running
on
shared
mode.
It
doesn't
have
to
get
a
new
key
phrase.
It's
going
to
keep
the
shared
Nick
mode
pinterest,
which
means
that
you're
the
board
you
ordered.
Where
did
not
you
my
a
shared?
It
should
just
work
cool.
Let
me
know
if
it
doesn't:
untradable.
Okay,
you're
out
of
town,
soon
be
kind
yeah.
B
B
A
A
A
A
A
Pass
ddd
tips
that
I,
don't
that
final
error
does
not
result,
just
especially
in
the
script.
Let's
make
it
it'll
blow
up
the
moment
it
roll
into
an
air,
but
a
lot
of
progress
on
that
so
I'm
hoping
by
next,
but
by
the
planning
session.
We
will
actually
have
gated
check
in
that.
Have
it
the
true
smoke
test
cool,
then
then
we
need
to
work
out
a
shadow
to
make
them
work
against
nam.
A
A
B
A
Request
basically
turns
on
the
alternating
lines
CSS,
so
the
table
is
displayed
having
white
and
grey
bars
the
which
is
I'm
trying
to
think
of
where
we,
where
we
have
that
place,
we
don't
so
most
of
the
screens.
We
don't
do
alternating
bars,
but
just
to
a
big
table.
And
then,
when
you
hover,
you
get
a
highlight
on
that.
That
row.
A
I
think
alternating
bars
in
general
are
better
as
long
as
you're
settled,
I
used
to
read
it
from
a
UI
perspective.
I
believe
it's
better
to
happen
to
have
the
type
of
color
contrast.
It
helps
line
up
things
when
there's
a
lot
of
data
on
the
screen.
Well,
for
instance,
looking
at
github
their
table
there.
Every
line
is
also
it's
all
shipping
lines
every
time.
A
C
A
A
So
one
of
the
things
I
would
ask
for
the
next
planning
session.
If
there
are
you
ice
creams
that
are
worrisome
from
that
perspective,
I
know
that
he's
willing
to
10
suggestions
on
which
screens
to
attack
or
what
screens
are
needed
to
be
built,
and
we
should
we
should
plan
on
bringing
that
out
the
minute.
B
A
Think
it's
it's
more
than
just
cleaning
up
a
screen
here
there
in
the
other
place
right.
What
we
need
to
do
is
evaluate
what
the
use
case
closed
should
be
through
the
through
the
user
interface
I.
I.
Think
that
I
would,
I
would
love
to
see
a
proposal
for
that
I
think
the
readystate
wizard
is
getting
going
to
help
I
hope
that
it
helps
with
that
a
fair
bit
because
it
takes
away
a
lot
of
the
pieces,
but
even
if
it
means
a
total
revamp.
Scott.
Let's,
let's,
let's
put
some
suggestions
on
the
table.
A
I
mean
this
is
the
sort
of
thing
work
with
you,
while
a
front-end,
back-end
architecture
where
the
UI
is
a
separate
thing
from
the
API
layer
might
come
in
handy
could
very
well.
Well,
that's
long
term,
that's
not
in
the
future
and
that's
I
think
at
some
point
we're
going
to
need
to
say
these
are
the
this
is
the
UI
of
the
open,
crowbar
project
and
some
of
what
what
we're
describing
would
be
funded,
UI
development,
okay,
great
and
that's
that
velvet
this.
A
Be
good
enough
to
people
to
accomplish
the
basic
goals
and
get
that
cluster
running
to
a
certain
size.
Well,
frankly,
as
the
scale
work
problems
going
to
be
possible
at
that
scale,
we
really
want
you
right
at
I.
Think
molanda
users
will
be
a
lot
more
comfortable
dealing
with
it
in
terms
of
the
api's
than
they
will
in
terms
lot.
I
think
the
way
we
use
to
you
is
going
to
be
very
different
right.
You're,
going
to
have
a
lot
more
20-foot
display
type.
You
buy,
meaning,
see
a
big
cluster
of
nodes.
A
You
don't
do
much
too
I
great
choice,
so
I'd
love
to
have
loved,
to
have
some
funds
to
actually
invest
in
a
really
UI
design.
Now
that
we
actually
understand
what
you
get
the
hundred
no
challenge
behind
us
and
really
see
how
the
UI's
are
breaking
I
guarantee,
they
look
good
Nick,
I,
don't
know
how
much
further
we
want
to
go
on
the
impact
of
the
ws
man
API
pieces,
except
to
say
that
we're
going
to
have
you
movies
and
design
consequences.
A
So
to
do
out-of-band
management
you
make
out
of
band
calls
right
now.
Those
are
synchronous
calls.
Does
this
Victor
and
I
were
talking
about
a
little
bit
beforehand
and
there's
a
pattern
now
for
doing
these
out
of
man
calls
you
won't.
He
talk
through
some
of
the
design
design
thought
that
you
have
for
out
of
and
verses
in
band
and
how
you're
managing
them.
A
Well,
it's
not
so
much
not
so
much
out
of
band
versus
in
bed
management,
as
I
think,
is
at
some
point
we're
going
to
need
just
a
general
love
to
generalize
the
chunk.
A
I
hacked
into
adam
its
abstraction
layer
for
how
to
manage
power
on
the
system
into
a
more
general
of
how
to
manage
nodes
outside
of
the
context
of
no
roles,
because,
ultimately
doing
things
like
you
know,
messing
or
less
menacing
materials,
the
bios
settings,
but
changing
the
boot
order
on
a
node
and
turning
the
note
on
and
off
and
making
it
blink
and
doing
that
sort
of
thing.
That,
ultimately,
is
something
that
lies
outside
of
the
context
of
walking
without
row
graph
to
convert.
A
A
What
I'm
talking
about
when
I'm
talking
about
I'm
talking
about
out
of
band
versus
and
band,
so
like
an
in-band
action
on
the
node
is
like
an
indent
action
is
like
deploying
a
database
on
a
node.
You
have
a
note,
roll
that's
bound
to
it
and
says
I
want
to
deploy
postgrads
and
the
Jake
eventually
comes
along
after
you
know,
annealing
the
cluster
and
discovers
at
least
some
salt
close
grabs
at
it.
Does
it
now?
A
That's
part
of
the
flow
of
you
know,
bringing
a
cluster
up
to
whatever
state
you
want
to
bring
it
up
to.
However,
out
of
band
is
doing
things
like
I
want
to
reset
the
node
it
doesn't
affect.
The
wall
may
have
effects
on
me,
no
role
state
graph,
oddly
no
dro
grass,
because
all
the
numerals
on
that
note
have
to
up
be
moved
out
of
active
when
you
do
it,
it's
ultimately
not
something
that
is
always
going
to
be
under
the
control
of
the
CRO
of
our
framework.
A
You
know
the
ultimate
arbiter
of
whether
or
not
being
notice
on
and
off
as
I
power
company
or
the
system
that,
depending
along
that,
how
far
back
you
want
to
take
it,
and
you
know
whether
or
not
I
noticed
turned
on
and
off
or
what
it
booths
to
is
something
that
crowbar
can
manage.
But
it's
not
the
only
thing
that
gets
that
can
manage
it.
A
A
A
B
A
That
matters
on
a
device
by
statement,
but
I'm
I'm
thinking
if
it'd
been
enough
in
in
terms
of
whether
or
not
be
normal
graph
is
the
is
the
thing
that
is
performed.
The
act
of
diversion
manila
graph
is
a
thing
that
is
responsible
for
doing
the
action
to
it
and
the
third
third
things
that
in
the
nodal
graph
being
done,
would
be
to
be
done
out
of
band.
A
So
yeah
win
arbitrary
choice.
There
right
the
jig
might
choose
to
do
with
your
thing
out
of
an
action,
but
maybe
prob.
If
you
would
define
what
you
mean
by
not
a
bad
action
I.
So
from
my
perspective,
I
was
thinking
out
of
band
means
that
it's
not
done
directly
through
the
OS
of
the
system.
So
so
you
don't
log
into
the
node
or
talk
to
the
chef
agent
on
the
node
to
take
the
action
it's
done
by
an
API
call
to
external
hardware,
or
so
anyway.
A
C
A
I
get
what
you
getting
that
Victor
I
think
that
what
was
confusing
me
is,
you
know
it
out
of
being
the
management
is
the
terminus
for
around
a
lot
in
the
industry.
It
is
typically
meant
to
enact
actions
on
a
subsystem
of
a
piece
of
hardware
right,
so
a
lie
low
or
an
ID
rack,
or
something
like
that
to
effect
changes
on
the
system.
This
actions.
B
A
You
know
orchestrated
external
to
the
system
and
push
to
it,
yeah
yeah
kind
of
the
concept
that
I
had
in
mind
like,
for
instance,
a
sysadmin
logging
onto
a
system
and
using
make
a
CLI
to
blow
away
a
raid
array
in
the
operating
system.
That
acros
meant
that
a
crowbar
that
for
system
is
managed
by
crowbar
would
be
considered
out
of
band.
From
the
point
of
view
of
crowbar
right.
So
basically,
I
mean
there
are
very
few
things
that
would
be
in.
A
Arm
someone
could
always
log
into
a
system
is
root
and
retract.
Some
of
the
actions
that
that
the
configuration
utility
is
put
in
place
right
and
then
concepts
I
had
none
was
more
infantis
when
the
core
of
our
framework
does
it
as
part
of
bringing
the
cluster
to
whatever
desired
state
we
wanted
it
to
be
in,
and
out
of
band
blows
everything
else.
A
Hurt
you
that
it's
some
of
what
you're
saying
is
those
actions
could
actually
be
in
band
actions
from
the
way
I
would
yeah.
They
could
look
at
their
with.
So
we
have
we
sort
of
like
we
can
reboot
a
note
by
FSA
ginseng
and
saying,
or
we
could
reap
it-
a
note
by
poking
edgy
I
to
my
controller
insane
reset
switch
yes
or
you
could
poke
at
a
powered,
pdu
and
reboot
it
that
way.
A
A
Try
and
understand
them
better,
and
then
we
talk
about
the
details
of
the
work
of
the
deep
land
in
the
next
meeting.
So
this
is
a
very
topical
conversation
trust
to
try
to
resolve
lexical
issues,
yeah,
I'm
still
having
trouble
understanding
actions
that
are
not
considered
part
of
the
node
world
graph.
B
A
Well,
as
I'm
cool
yeah,
can
we
talk
about
those
nouns
and
verbs
I
mean
is
that
is
that
useful,
that
the
noble
graph
is
a
bunch
of
basically
notes
so
nouns,
all
bunch
of
I'm,
this
I'm,
this
I'm
this
and
the
sum
of
what
your
act
we're
talking
about
are
verbs.
The
problem
is
the
jig's
actually
execute.
A
A
A
C
C
A
Via
the
via
the
power
stuff,
you
can
ask
whether
or
not
it's
on
or
off,
but
the
core
framework
doesn't
know
or
care.
Whether
or
not
a
system
is
on
or
not
all
it
cares
about,
is,
is
it
alive
and
doing
things
and
can
I
can
I
reach
out
and
control
it
or
not?
But
it
is
once
we
know
about
a
node.
We
do
track,
whether
it's
on
or
off
no.
A
B
B
There
is
there
a
medic
Rex
that
we
need
to
deal
with
you
what
he
means.
Well,
we
have
ready
in
a
managed
state
transition
crash,
but
all
the
critical
functionality
that
we
are
trying
to
maintain,
but
we
also
have
environmental
track
that
we
may
elect
to
monitor
and
transitions.
That
is
not
germane
cuz.
I.
A
A
Yeah
control
control
ends
if
the
vector
limits
that
we
ultimately
have
no
control
or
I'm
not
really
interested
in
putting
that
in
the
graph.
No,
but
it
would
be
nice
to
have
an
external
events
that
actually
marked
machine
noses
off
as
offline.
If
they
is
not
alive,
we
knew
they
were
not
a
lot.
We
do
that,
like
all
the
power
stuff
right
now,
if
you
tell,
if
you
ask
crowbar
to
turn
the
machine
off
its
going
to
market
is
not
alive.
A
A
A
Right,
so
do
we
draw
a
hard
line
between
what
is
on
the
node
graph
and
what
is
not
in
my
mind,
the
things
that
are
on
the
node
well
graph
are
things
that
we
do
to
bring
a
cluster
to
a
certain
operational
state
in
terms
of
what's
deployed
where
and
how
it's
all
configured,
things
that
are
not
on
the
Novo
graph
or
anything
external
to
that,
like
the
patent
like
managing
the
power
state
or
ensuring
that
we
have
a
network,
connectivity
are
efficient
at
the
time
or
any
one
of
the
huge
number
of
potential
diagnostic
or
monitoring
tools
that
we
might
add
to
chrome
bar
in
the
future.
A
Like
I've,
been
looking
at
console
and
trying
to
convince
myself
to
play
around
with
it
and
I
think
it'd
be
awesome
if
we
had
console
and
salt
on
all
the
notes
just
for
monitoring
purposes.
But
the
only
thing
that's
beyond
the
node
well
graph
for
that
would
be
you're
going
to
install
console
on
this
and
here's
how
it's
going
to
be
configured
to
join
its
little.
You
know
gossip
clusters
so
that
everything
knows
about
everything
else,
but
everything
that
we
do
to
actually
talk
to
actually
listen
to
that
fire.
A
B
A
A
To
me,
that's
where
the
interested,
the
thing
that
gets
interesting
with
this
is
switching
switch
configuration
right.
We've
known,
it's
always
been
the
biggest
challenge.
Well,
it's
not
a
bad.
Until
probe
are
gross
game,
bility
the
managed
switches,
in
which
case
it
would
turn
into
in
band,
so
the
other
cultures
will
return
moving
down
the
track.
Well,
because
I
mean
for
the
way
that
we
have
it
right
now.
You
know
managing
networks
is
in
that
is
in
band.
A
A
If
we
had
control
of
the
top
Iraq-
and
you
know
in
the
road
switches
or
whatever
the
next
new
hot
day,
the
Senate
apology
is
going
to
be
all
right.
You
know
we
can
make
that
all
in
band
up
to
the
part
where
we
lose
control.
There
is
another
potential
user
case
whether
or
not
it's
a
good
thing
to
do
that
as
an
entirely
different
discussion.
There
is
another
side
sweep
how
much
time
do
we
have
120
minutes
simulans
that
we
haven't
covered
yeah
hi.
A
The
purpose
here
is,
except
this
is
the
way
I
believe
design
works.
Is
you
have
to
bring
up
topics
early?
First,
you
have
to
get
people
primed
on,
though,
then
they
think
rocket.
It's
very
circular,
so
I
wanted
to
get
this
topic
started.
We
are
hitting
it
actively
and
so
we're
going
to
keep
discussing
it
and
trying
to
figure
out
how
to
make
it
work
and
I'm
and
I'm
actively
looking
for
people
in
the
community
good.
A
Thank
you
for
adding
the
on
grandpa
I'm
looking
for
people
in
the
community
who
want
to
help
sponsor
switch
integration,
work
that
I
think
is
we're
primed
to
take
on,
because
that
gets
really
interesting
and
I'm
talking
to
some
people
in
different
networking.
Networking
scenarios
about
this
is
really
exciting.
A
Roller
backer,
so
I
do
think
that
I
definitely
wanted
to
cover
I
think
we
actually
covered
the
use
cases
for
power
management.
How
to
handle,
becomes
a
good
chef,
metal
I
want
to
talk
through
and
be
wait.
How
does
it
how
to
handle
the
ends?
Yes
come
buddy,
gimme,
more
than
disgusting
that
which,
which
one
do
you
want
to
talk.
B
About
the
out-of-band
amount
of
noise,
roll
brush,
full
of
transition,
graphics,
good
question,
yep,
one
that
is
potentially
of
interest
to
us
down
the
road
is
one
of
doing
monitoring
and
introspection
on
the
operation
of
the
clutter
of
a
cluster
that
we
are
actually
managing
with
the
state
transitional
are
in
order
to
determine
if
we
need
to
bring
up
additional
nodes
or
for
SLA
management.
So.
A
My
my
expectation
on
that
is
that
we
will
provide
a
framework
that
allows
somebody
to
make
those
judgment
calls
that
that
what
you're
describing
is
would
be
an
add-on
would
be
an
add-on
or
a
separate
projects
on
top
of
what
we're
doing,
rather
than
crowbar
making
that
judgment,
I'm
trying
I'm
trying
to
crow
bar
should
be
really
good.
At
saying
I
have
a
new
nodes
spin
it
up
I.
A
Have
this
note:
I
wanted
to
become
that
and
make
those
things
happen,
taking
it
to
the
point
to
make
those
judges
the
actual
determination
of
whether
or
not
it
needs
to
do
those
thing,
that's
out
of
bounds
of
her
power.
That's
that
to
me
is
that
be
something
else
driving
crowbar
yeah,
because
I
can
see
right
now
that.
B
A
So
actually
may
be
very
specific,
I
would
love
to
see,
and
she,
the
size
of
the
chef
metal
I,
would
love
to
see
ironic
using
crowbar
as
a
way
of
vision.
It
knows
to
any
state
and
then
take
over
using
you
delivering
them
into
it
opens
into
a
cloud
as
capacity
chef
metal
has
the
exact
same
semantics:
chip
metal,
the
gold
for
that
is
not
to
provision.
The
nose
is
to
allocate
the
nodes
into
into
we
know
ropes.
So
the
idea
of
Chef
metal.
A
A
Have
those
and
there's
that
one
that
we
can
make
look
like
that,
and
so
in
the
last
meeting
there
were
some
questions:
I'm,
pretty
intent
on
spending
a
couple
days,
getting
chef
metalworking,
because
I
want
to
demo
that
at
Oscar
yeah,
I'm
kind
of
interested
in
what
exactly
your
goal
is
with
Chef
metal
there.
So
what
my
goal
is
the
chef
Mehta
late
guys
are
pretty
straightforward.
A
What
they
do
is
they
assume
that
you
have
a
pool
of
available
resources
and
that
you
can
say,
give
me
a
server
from
your
pool
and
make
it
mine,
which
basically
means
putting
a
thing
taking
meds
from
the
system
deployment
and
putting
them
into
a
chef
metaphor,
ok
and
then
that
it's,
then
you
can
take
the
OS
you
want
on
it
and
that
it
gives
you
the
chef
and
it'll
it
either.
I
think
it
expects
the
ssh
key
insult
the
chef
client,
although
I
think
we
might
be
able
to
install
chef
climate
in
pageant
widget.
A
So
that's
the
it's
a
pretty
straightforward,
API!
It's
really
designed
for
Klaus,
but
I.
Think
it's
a
great
proof
point
of
exactly
what
John
was
just
bringing
out
and
what
we've
been
saying
for
a
while.
The
nuance
is
that
you
also,
we
also
need
to
be
able
to
access
the
chef
server
and
since
we
install.
A
A
A
C
A
If
we
also
discuss
it
in
the
list
of
priority
of
other
items
that
we're
trying
to
achieve,
is
there
something
so
this
one
to
me
if
I
can,
if
I
can
get
it
for
oz
con,
it
strikes
me
as
a
big
wind
to
get
in
front
of
the
chef's
audience,
which
is
why
I've
been
prioritizing
more
so
than
say
it's
more
achievable
than
doing
OpenStack.
What.
A
A
A
A
A
Think
it's
definitely
reasonable
to
write
up
a
proposal
for
what
it
is
you
go
to
try
to
accomplish.
You
know
and
then
get
comments
on
that
and
the
sort
of
pieces,
but
I
think
we
should
also
make
sure
that
we
have.
You
know
a
reasonable
prioritized
list
of
items
that
we
want
work
right.
A
How
does
the
we
need
to
go
back
so
in
the
in
the
next
agenda?
Somebody
has
access
to
the
past,
so
next
week's
agenda
I
definitely
would
like
to
talk
through
this
market
vision
roadmap
slot,
so
that
we
for
think
that
it
in
conversations
it
occurs
to
me,
is
we're
very
good
at
talking
about
technical
things
that
we
were
delivered,
but
I
think
that
you're
looking
for
the
community
to
come
in
behind
us.
We
need
to
be
talking
through
some
broad
brushstroke
things.
B
A
Case
because
we
have
time
to
do
that
so
Scott
for
broom
the
things
that
that
we
are
looking
at
word,
progress
on
ipmi
raid
for
progress
on
gated
see
I
upgrade
it's
something
we
have
not
yet
worked
on
some
point.
We
need
to
think
about
that
and
then
the
other
thing
we
were
supposed
to
be
doing
is
enabling
workloads
the
right
stuff
would
be.
Another
reveal
iterative
for
that.
A
Good
at
it's
a
good
point,
so
let
me
look
at.
Let
me
look
at
what
it
would
take
to
start
doing
some
dusting
off
the
SAT
for
clubs,
there's
a
piece
of
core
and
that
I
broke
that
affects
the
horse.
Would
let
work
on
bare
metal.
B
A
A
A
A
B
B
A
You,
okay,
so
I
want
to
talk
about
our
p.m.
is
installed
maitenance.
We
have
people
who
are
trying
to
use
the
RPM
install
and
it's
down
as
horribly,
and
it
is
failing,
probably
in
minor
ways.
It
would
be
really
helpful
to
have
somebody
going
through
that
those
steps
so
that
we
figure
out
what
was
broken
with
it.
The
thing
that's
broken
with
it
right
now
is
not
worth
updating
the
RPMs.
That
would
be
very
that
would
be
sourced
in
that
again,
the
thing
that's
broken
about
it
right
now
that
nothing
is
updating
the
RPM
I.
C
B
A
A
C
A
A
A
Think
it
is
more
planning
work
that
we
just
needed
to
do
it,
but
I
think
what
we're
saying
is
we
are
going
to
commit
to
getting
those
rpms
yeah
got
what
I'll
do
for
you
is
I'll,
modify
the
kvm
a
transcript
to
kyndra
laterally
rebuild
RPM,
instead
of
thought,
only
rebuilding
them
on
demand
and
then
from
there
you'll
just
be
able
to
run
up
from
the
core
just
run
tool,
/
KVUE
madman,
and
it
will
try
to
spin
up
and
install
on
such
as
five
from
the
rpms.
A
A
A
Kids
here
and
so
I
tried
to
document
his
escape.
You
didn't
reply
back,
but
just
like
we
spin
up
docker
containers
for
dev
test
full
bar
could
talk
about.
Api
is
spin
up
nodes
that
it
would
then
manage,
and
so
the
idea
is
that
if
I'm
going
to
start
building
a
snap
deployment
or
a
hadoop
deployment-
and
I
have
a
clout
available
to
me-
I
could
actually
spin
up.
You
know
attend
node
set
deployment
like
crowbar.
A
Only
Pyrrhic,
where
bar
controller
nose
is
also
in
a
cloud,
and
you
don't
check
everything,
also
all
right
dhcp
from
the
clouds
IP
address
range.
That's
the
doctor
doesn't
doesn't
rely
on
DCP
to
do
its
thing
so
conceivable
that
doesn't
matter
person
who's
been
up.
An
admin
noted
this
up
at
the
issue
that
we
cannot
do
it
right
now.
Oh
so,
you're
concerned
that
you
have
to
have
the
networking
to
get
to
the
cloud,
because
the
cod
API
calls
are
routed
banned.
In
that
perspective,
true.
A
A
Most
clubs
will
even
support
ipv6
for
that
matter.
Well,
a
lot
of
them
doubles
for
not
use
IP.
You
say
so.
This
could
work
against
a
small
private
cloud
as
waiters
there.
This
could
work
against
a
private
cloud
where
you've
taken
care
of
ensuring
that
your
address
that
the
address
range
that
the
cloud
uses
internally
or
externally
overlaps
with
a
cloud
range
that
crowbar
can
use
in
training
your
external.
A
A
Yeah,
the
doctor
stuff
was
never
meant
to
be
a
production
anything
but
as
it
runs,
containers
and
privileged
mode,
which
means
it's
true
jewel
for
any
script
occasion
running
inside
the
container
tunnel
in
the
physical
box.
So
that's
there's
no
doctor
right
now
as
a
single
node
test,
dev
port
richey,
structurally
Darren
I,
can't
repeat
that
enough
times,
yeah
dr.
A
A
A
Goodnight
I
think
that
there
are
interesting
repercussions
from
krav
are
starting
to
manage
virtual
hour
through
clouds
managing
virtual
machines
on
those
another
sticky
Riggins.
Those
are
wickets
that
I
would
prefer.
We
say
we
are
not
going
to
deal
with
those
explicitly
proudly
for
at
least
a
year
and
a
half
three
reduce
those
are
the
domains
of
cloud
management's,
awful.
B
A
A
A
C
A
Yes,
at
that
point
just
use
the
cloud:
IP
is
been
a
crowbar
tirely
inside
the
cloud
that
we
saw
in
XL
16
on
amazon.
He's
like
why
I
asked
Nathan
to
clarify
his
case
and
with
that
I
think
we
leave
gone
a
little
bit
further
so
that
we've
got
some
additional
feedback
for
that
conversation
employ
keep
it
on
the
list
the
year
at
work
and
we're
out
of
time.
So
it's
gonna
close.
We
call.