►
From YouTube: OpenCrowbar Community 2014 09 03
Description
No description was provided for this meeting.
If this is YOUR meeting, an easy way to fix this is to add a description to your video, wherever mtngs.io found it (probably YouTube).
A
C
B
Much
the
core
members
online
and
we'll
be
reviewing
our
little
request.
Some
new
features
that
we
have
to
support
the
varieties
of
hardware,
implementations
and
ipmi
implementations
for
our
new
quick
quarks
feature
and
we'll
talk
a
little
bit
about
the
salt.
Jig
I'll
demonstrate
so
my
chef,
metal
progress
and
we'll
review
some
of
the
crowbar
versus
cobbler
talking
points,
but
first,
let's
review
a
little
bit
of
our
agenda,
see
what
folks
that's
done.
If
there's
anything
on
the
previous
agenda
that
needs
attention
and
it
actually
looks
all
good
that
we've
covered
we
covered
everything.
B
The
in
band
and
out
of
band
features
and
just
get
that
known,
review
that
nomenclature
and
how
it
relates
product
again.
So
everybody
is
super
clear
because
it's
that,
as
a
point
of
confusion
and
also
how
JEWS
interact
with
on
activities
that
are
on
the
graph,
that
is
Indian
dealer
and
activities
that
are
out
of
vendor
or
not
in
India
or
just
and
just
work
out
all
that
terminology
again
to
beat
a
dead
horse.
B
D
B
D
B
So,
let's,
let's
review
the
Cork's
feature,
you.
B
E
A
B
Is
having
issues
getting
your
outside,
so
quirk,
one
of
the
things
that
I
added
over
my
enjoyable
vacation
was.
B
Currently,
the
only
places
that's
being
used
is
in
the
ipmi
stuff
and
we
have
a
works
in
place
or
if
a
system
is
a
dull
system
with
a
dedicated
nick
for
the
BMC,
then
whether
or
not
to
use
that
is
being
managed
as
a
quirk
and
the
and
whether
or
not
and
how
we
set
up
the
username
or
the
crowbar
username
or
the
ipmi
interfaces
for
various
cancel.
Various
ipmi
control
is
including
some
super.
B
Those
include
general
icmi
power
on
power,
ops,
stuff,
you
do
we
find
corks
in
there
as
well.
If
we
do
we'll
have
the
we'll
have
the
infrastructure
to
handle
them.
For
instance,
I
know
that
some
ipmi
controller,
even
when
you
touch
even
when
you
tell
them
to
do
way,
a
hard
power
on
or
hard
power
off.
They
still
go
through
the
full
acpi
OS
notification
cycle.
So
you
wind
up
getting
a
soft
power
off
moving
is
a
bad
thing.
B
If
you
want
to
do
things
like
use
ipmi
as
your
fencing
techniques
brownstone,
it
I
think,
and
so
that's
support
for
handling
that
or
and
working
around
it.
If
a
workaround
is
in
fact
possible
would
also
be
handled
as
a.
B
But
in
order
to
advocate
that
you
need
to
have
detailed
information
on
you
know
who
made
the
ipmi
controller,
what
product
ID?
Does
it
have?
What
model
ID
will
manufacture
ID?
That
sort
of
thing.
B
B
B
B
It
looks
like
razor
is
completely
coincidental.
I
actually
got
my
fish
for
how
to
do
that,
based
off
of
work
that
I
did
for
back
when
I
was
the
maintainer
of
p.m.
utils
and
I
had
to
manage
a
quick
database
for
how
to
appropriately
touch
all
the
weird
video
cards
that
exist
in
such
a
way
that
they
don't
explode
when
you're
suspending
and
resuming
your
system.
E
D
B
That
in
our
code,
base
top
sprinkling,
everything
with
special
k,
exceptions
that
machine
my
machine
basis,
I'm
trying
to
set
it
up
in
such
a
way
that
we
do
it.
As
that
we
have
ample
opportunity
to
generalize
and
force
better
documentation
just
by
the
fact
that
we
have,
you
know
quarks
available
for
that
that
specify
what
exactly
we're
going
to
do
for
any
given
model
of
hardware.
B
B
Yeah
I
might
have
missed
it
when
I
drop
up,
but
this
is
not
a
heuristic
pattern
to
match
particular
roles,
but
rather
just
for
operational
control.
The
hardware
itself.
Those
roles
can
be
applied
in
it
in
a
uniform
fashion.
In
a
way,
it's
working
right
now
is
pretty
much
whenever
we're
doing
stuff.
That's
harborough
specific,
that's
generally
split
into
at
least
two
phases:
there's
a
discovered
phase
and
then
there's
an
actual
configure.
The
hardware
phase
and
during
discover
phase
is
when
we
figure
out
what
set
of
works.
B
D
D
C
D
D
C
B
B
C
B
D
D
B
You
have
any
focusing
in
practice,
certainly
mixes
hardware
causing
anomalies
that
don't
exist
on
that
hardware,
when
any
one
of
the
pieces
is
used
with
another
skates
say.
The
NICU
used
with
a
Knick
and
motherboard
combination
exposes
an
issue
with
a
Knick
we're
on
another
motherboard.
It
doesn't
so
we
could
potentially
need
some
way
to
to
express
that.
B
Yes,
yes,
I
have
seemed
at.
I
have
in
fact
seen
that
all
the
way
down
to
firmware
revision
of
any
given
componentry,
where
a
buzz
will
only
happen
if
this,
if
you
know
this
niche
with
this
row
rather
than
this
system
with
this
range
of
fios
red
and
the
storage
card.
With
this
former
arrest
and
good
irq
shirt
like
that,
I've
seen
that
happen.
B
Will
this
scoring
system
allow,
you
know
quirks
like
that
to
surface?
So,
if
we're
only
seeing
you
know,
failure
with
of
this
Nick
with
this
firmware.
E
B
Okay,
okay,
so
when
we
we
articulated
by,
is
explicitly
designed
to
be
that
flexible,
because
I've
needed
a
little
flexibility
in
the
past
when
I
didn't
have
yes,
so
a
quark
entry
itself
would
be
any
number
of
blobs
of
JSON
and
if
all
or
most
of
those
blobs
meet,
then
that
particular
quirk
is
applied
and
thus
blocks
of
JSON
could
be
could
be
any
anything
about
the
hardware
or
the
firmware.
B
I
mean
there's
no
reason
that
it
couldn't
I,
just
don't
have
a
solidly
defined
Hughes
case
for
yet
and
I
do
have
probably
defined
use
cases
for
it
and
in
ipmi
and
WS
man
and
great
and
bio
for
that
matter.
What
about
the
classic
example
of
your
virtualization
heart?
We're
changing
your
mac
address
out
from
under
you,
and
you
have
not
responding
to
that
very
well
and
having
to
go
in
and
remove
you
dead
entries.
That's
totally
out
of
scope
for
this
okay.
B
That's
all
the
questions
I
got
this
gem
is
that
have
you
written
it
as
a
separate
upstream
contribution
or
resist
structure
maps?
Is
the
first
structure
mapped
on
Ruby
June.
B
B
E
B
B
B
B
There's
no
more
questions.
Let's
move
on
to
the
the
salt
jig.
D
D
So
the
work
for
the
salt
jig
is
distributed
to
try
and
basically
install
a
minion
and
hook
it
hook
it
up
to
an
existing
salt
server
and
make
that
work
out.
Although
Greg's
initial
pass
that
that
was,
did
a
lot
more
in
the
capabilities,
who's
actually
building
a
jig
and
doing
some
other
learning,
learning
crowbar
and
doing
some
other
work
as
far
as
that
goes,
but
it
is,
it
is
really
going
for
us
to
be
able
to
have
installed
assault,
assault,
minion
and
then
be
able
to
let
that
bat.
D
This
lets
people
manage
a
node
V
assault
after
we
set
it
up,
and
so
that
that's
that's
where
that
work
is
going.
If
people
are
interested
in
doing
it
its
purpose.
This
call
is
to
get
community
participation
in
that
effort
and
I
know
and
I'm
actually
trying
to
bring
this
all
team
into
it.
So
they
can.
They
can
give
us
some
guidance
on
Victor.
You
looked
at
gregs
tolls
and
I
know
that
he
wasn't
expecting
him
to
be
right.
The
first
time
yeah.
You
have
any
additional
comments
on
that.
B
B
B
B
It's
closely
just
a
matter
of
making
sure
that
the
salton
Indians
wind
up
getting
himself
properly
on
the
nose
that
we
have
currently
a
first
chef
and
puppet.
We
just
rely
on
the
fact
that
we
won't
need,
like
a
public
client
or
a
chef
client
role
on
a
node,
because
we
have
a
role
on
that
note.
That
requires
puppet
or
chef,
and
then
we
can
automatically
pull
in
the
client
role
as
a
side
effect
of
that
and
then
Destin
and
the
logistics
around.
D
Ok,
oh
and
there's
more
infrastructure
to
set
up.
If
we
did
that,
if
we
have
that
case
and
I'm
not
sure
how
much
salt
infrastructure
we
actually
need
to
set
up,
I
think
that's
something
we're
going
to
want.
Community
feedback
on
just
died
in
one.
D
It's
very
logical
to
me
for
him
to
do
that,
because
in
order
to
test
it,
he
needs
the
whole
stack
/
for
the
use
case
that
we're
trying
to
prove
you
don't
but
I
think
what
he's
doing
is
is
probably
the
easier
to
test
and
that
validate
and
then
we'll
have
to
go
back
and
add.
In
the
special
case,
there's
already
an
existing
salt
server.
D
B
It
looks
like
it's
modularized
well
enough,
so
that,
even
though
we're
doing
a
full
stack,
install,
it'll
work
kind
of
like
the
chef
server
and
chef
clients
up
in
that
the
assault
master
role
can
either
insult
master
and
record
that
everything
should
talk
to
it
or
just
have
information.
That
said,
you
need
to
talk
to
something
out
there.
D
D
C
D
There
is
one
I
think
it
has
a
salty
name
to
it
like
minor
or
shaker,
or
something
like
that
for
great
out.
Grains
I
believe
is
the
data
collection.
I
think
it's
called
grains
or
shaker
I
think
the
I
think
it's
a
shaker
and
the
details
are
called
brains
that
got
cringes
over
the
tortured,
analogies
that
were
pushed
into
these
communities.
D
C
B
B
E
B
I
made
a
bunch
of
progress
yesterday
on
getting.
B
Chef
metal
hooked
up
into
the
work
that
Rob
has
done,
Andrew
crowbar,
be
able
to
deploy,
I'll,
actually
allocate
and
then
make
ready
the
machines
that
are
discovered
by
crowbar.
What
chef
metal
lacks
is
a
discovery
process.
B
It
is
just
assuming
that
whatever
driver
you're
using
and
a
driver
could
be
any
source
of
virtual
machines
like
anything
supported
by
fog,
which
would
be
a
to
the
US
or
Rackspace
or
OpenStack
or
vagrant,
which
a
vagrant
fox
resource
would
have
to
be
created
if
you're
using
vagrant
and
chef
metal
and
they
run
box
resource,
would
point
to
a
vagrant
box
with
a
year
precise,
64,
dot,
box,
vagrant
box,
and
then
it
was
that
considers
that
sort
of
an
open
pool
or
a
limitless
pool
to
create
as
many
vagrant
boxes
as
you'd,
like
out
of
that
precise
bak,
the
there
are
no
boundaries,
it
seems
that
I
know
there
are
no
guides
for
maximum
number
of
boxes
or
anything
already
built
into.
B
Chef
metal
and
I
was
I
was
starting
to
understand
it
and
see
a
pattern
that
might
emerge
where,
if
crowbar
is
now
the
driver-
and
we
have
a
few
nodes
that
have
been
discovered
in
crowbar
the
size
of
fact
of
deciding
which
notes
to
bring
into
an
allocated
and
then
ready
state
it's
how
many
of
these
nodes
will
we
be
a
maximum
number
of
these
nodes?
So
a
particular
deployment
doesn't
run
off
with
all
the
all
of
the
hardware
I
thought.
Making
a
guard
instantly
like
that
would
be
would
be
really.
B
E
B
Giving
developers
and
DevOps
engineers
a
feeling
of
where
they
exist
or
how
many
resources
they
have
or
or
what
type
of
object
this
is
that
they
can
play
with
so
I've
got
those
basics
set
up
and
I'm
now
just
working
on
the
the
kinect
sequence.
So
so
these
chef
recipes
and
the
driver
code
can
actually
talk
to
the
crowbar
box.
Just
getting
up
to
the
point
yesterday,
where
I
was
making
the
call
to
crowbar
and
authenticating
and
then
the
cafe
closed
and
I
ran
out
of
time.
B
The
rav's
initial
design
pattern
and
remain
of
moving
assets
and
moving
nodes
appear
in
the
system.
Deployment
moving
them
into
some
other
deployment
say
an
allocated
employment
as
an
initial
step,
but
I
can.
The
obvious
next
step
is
that
DevOps
engineers
are
going
to
want
to
name
this
deployment,
something
or
at
least
add
to
have
more.
E
B
A
frame
they
name
for
it,
so
they
can
create
it.
Fleming
crowbar
calls
I
test
system
and
match
that
deployment
in
chef
metal
by
calling
the
the
crow
our
deployment
resource.
They
passed
my
test
system
to
match
that
name
and
we
can
go
ahead
and
start
deploying
nodes
into
that
and
applying
whatever
roles
are.
B
B
B
C
B
D
B
He
could
and
I
was
very
concerned
about
that.
I
wanted
to
also
drop
it
in
in
Victor's
head
to
pass
him
about
how
we
would
manage
that
if,
if
a
chef
metal
is
is
running,
chef
client
on
the
note
to
apply
certain
cookbooks
and
run
and
rolls,
but
it's
chef
roles
that
data.
If
we
point
it
back
to
our
chef
server,
the
chef
server,
that's
that's
being
run
through
the
annealer.
B
What
is
that
relationship
there,
and
what
do
I
need
to
look
out
for
is
the
anneal
or
smart
enough
to
look
at
the
chef
server
and
say:
oh,
these
roles
have
already
been
run,
or
these
recipes
has
already
been
run
on
this
note,
they
already
on
the
note
I.
Don't
really
need
to
do
anything
except
maybe
try
again
and
ensure
that
it's
item
potently
correct.
B
B
B
B
D
D
D
D
B
C
E
D
E
A
B
C
B
The
Machine
administratively
not
available,
for
instance,
because
no
a
system
administrators
doing
weird
things
to
it
manually
and
we
don't
want
crowbar
to
step
all
over
what
the
admins
doing.
D
Right
now,
I
think
it's
a
little
bit
of
an
abuse
of
the
flag
of
the
intent
of
the
flag.
I
think
that
it's
actually
what
you
described
is
it's
exactly.
What
we're
doing
we're
saying
hey
this
is
machines
is
no
longer
for
crowbar
to
manage,
but
it's
in
one
case
it
was
the
assumption
that
it
was
for
administrative
work
in
this
case
we're
handing
it
off
to
a
different
control
system.
B
B
D
B
D
C
B
B
B
B
And
you
know
want
to
try
to
run.
You
know
a
crowbar
workload
on
it
and
there'll
be
some
problem,
but
otherwise,
as
long
as
the
chef's
metal
roll
is
that
the
terminus
of,
or
at
least
is
just
as
long
as
it's
have
determined
that
that
knows
no
role
craft.
There
shouldn't
be
any
problem,
because
crowbar
will
just
do
as
a
kneeling
thing
and
then
once
it's
finished,
it'll
just
can't
control
back
over
to
chef,
metal.
D
D
D
I
understand,
victory,
I
know
it's
not
a
Herculean
task,
it's
just
from
it's,
not
one
of
the
tools
in
jugs
toolbox
right
now
so
shut.
The
chef
metalwork
is
pretty
much
a
one-way
action
right.
You
pull
it
out
of
the
system
pool
or
an
existing
deployment
you
meet.
We
need
to
walk
it
through
the
deployment
and
its
end
is
the
chef,
no
troll,
we're
actually
I.
D
Think
for
now
at
the
OS
deployed
at
the
OS
installed
no
troll,
and
then
we
need
to
add
maybe
moving
in
notable
to
add
in
ssh
key
that
the
chef
server
expects
you
might
not,
because
I
think
we
have
an
ssh
key.
We
just
need
to
share
it
with
the
server
matter
of
fact
that
nodes
already
registered
in
the
server
all
you're
really
doing
is
saying
is
handing
the
information
back
to
the
chef's
metal
infrastructure.
D
E
D
D
E
B
Alright,
so
the
I'm
done
there,
the
we
should
move
on
to
the
crowbar
versus
cobbler
talking
points.
D
D
What
I've
been
what
I've
been
coming
up
with
in
my
reviews
is
that
I
think
that
we
need
to
figure
out
what
what
it
would
take
to
make
crowbar
a
preferred
option
over
cobbler.
You
know,
I've
looked
at
cobbler
enough
to
think
to
see
that
you
know
it's
much
more
limited
than
what
crib
or
does,
but
it's
also
more
popular
and
I.
Think
we
may
need
to
find
things
to
simplify
or
streamline
as
a
against
cobbler
position.
D
Right,
it's
just
it's
an
entrenched
base
and
it's
it's
got
a
pretty
broad
install
and
what
I
want
to
do
is
come
up
with
a
product
list.
I'll
coordinate
with
Scott
a
little
bit
about
this,
come
up
with
a
feature
list
to
say
all
right.
These
are
features
that
if
we
had,
we
would
displace
cobbler
and
then
I'm
planning
to
engage
the
community
and
and
figure
out
what
their,
what
their
feelings
are
on
this
effectively
do
some
some
market
studies.
D
D
D
D
D
B
C
A
D
D
C
D
C
A
B
D
D
B
I
guess
required
to
power
and
I
love,
wish
everybody
a
wonderful
day
and
we
got
through
our
agenda
and
signing
off
from
the
cove
our
community
developer,
design
call
you
can
find
us
online
and
and
next
week
for
planning
call
any
other
commitments
for
folks
what
folks
are
going
to
be
working
on
for
this
coming
week,
I'm
going
to
be
working
on
this
jeff
metal,
stuff.
B
B
E
B
B
B
The
hardware
detected
see
in
the
system
deployment
because
they're
not
really
designed
to
be
modifiable,
but
things
like
the
raid
configuration
and
biota
config
and
whatnot
should
really
not
be
in
the
system
performance.
Oh,
we
would
need
some
mechanism
to
either
not
find
them
in
this
is
to
deposit
or
move
them
somewhere
else
as
needed.
D
Could
we
also
be
splitting
them
so
that
the
data
like
your
originally
suggesting
the
data
for
them
would
be
would
come
from
downstream
rural
fullback,
it's
a
little
twisted,
but
that
would
make
that
would
be
be
the
other
way
a
handle
at
the
role
that
actually
did.
The
configuration
was
in
system
deployment
that
the
the
data
that
you
wanted
for
it
was
in
a
downstream
role,
little
backward.
D
B
D
B
D
B
E
B
E
B
A
E
B
So
it
goes
we're
way
past
the
top
of
the
hour
shut
down
the
chat
and
see
you
all
next
week
and
I'll
see
you
online
already
have
fun.