►
From YouTube: 20190604 sig cluster lifecycle
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
A
Hopefully,
folks
can
see
my
screen
one
of
the
things
there's
a
couple
of
items
that
I
have
listed
here
and
we
can,
if
you
have
other
topics,
that
you
want
to
discuss
a
free
to
add
them.
At
the
end,
one
of
the
first
things
I
wanted
to
discuss
was
like
kind
of
lessons
learned
from
the
face-to-face
I
had
tried
several
times
before.
We
had
our
face-to-face
to
try
and
solicit
feedback
for
an
agenda,
so
I
decided
to
go
with
an
unconference
style
and
I.
Think
the
the
lesson
I've
learned
there
is
that
never
again.
A
So
I'm
gonna
I'm
gonna
come
there
with
it
armed
with
like
a
full
agenda
to
go
through.
If
people
don't
want
to
actively
work
on
an
agenda
ahead
of
time,
because
I
think
I
tried
to
solicit
agenda
items
and
there
was
kind
of
crickets
in
in
that
sort
of
weeks,
but
I
think
next
time,
I'm
gonna
try
to
do
the
same
process
for
solicitation
at
least,
but
then,
if
people
do
not
engage,
then
I
will
definitely
create
almost
a
much
more
structured
agenda.
I.
A
Think
with
the
larger
group
of
folks
on
the
sig
I
want
to
make
sure
that
we
start
to
develop
a
set
of
best
practices
for
decorum
for
making
sure
we
keep
the
conversations
focused
and
on
point,
because
there
was
a
lot
of
people
there,
and
it
was
interesting
to
me
that
we
try
to
redirect
the
conversation.
Then
it
would
continue
to
sprawl
so
I
think
having
a
documented
procedure
for
administrating
some
of
the
stuff
for
the
larger
groups.
A
You
know
different
sub
projects
have
different
size,
so
some
projects
are
a
smaller
band
of
folks
that
work
very
tightly
together
and
they
probably
don't
need
to
go
through
this.
But
if
they're,
a
larger
group
of
folks
who
are
want
to
coordinate
across
potentially
conflicting
objectives,
then
we
need
to
have
a
little
bit
more
process
that
can
be
the
process.
Indecorum.
A
B
Yeah
I
think
also
the
so
what
we,
what
we
did
still
mentioned,
the
conversation
strolled,
so
what
we
did
was
basically
to
do
a
round
of
introductions
and
instead
so
also
to
let
everybody
speak
for
for,
like
some
seconds,
at
least
when
they
introduce
themselves
and
why
they're
interested
in
in
the
space
and
how
they
would
like
to
contribute.
And
then,
after
this
it
started
thinking
more
about
the
values
we
actually
brought
down.
So
the
discussion
of
this
is
in
down
this.
B
Yes,
here
on,
Monday
20th
May
2010
cube
con,
so
there
is
in
this
dark,
more
information
on
what
we
actually
talked
about,
but
we
more
focused
on
the
values
of
trust
right.
So
all
the
people
wanted
to
leave
those
unconference
tile.
Everyone
wanted
to
talk
about
for
API,
so
we
talked
about
tools
for
API
and
we
realized.
We
need
to
take
step
back
and
focus
more
on
the
values.
What
are
the
key
principles?
We
want
to
do.
A
Hopefully
this
week
is
I'm
still
digging
out
of
email
with
regards
to
sort
of
writing
down
running
down
some
of
those
values
to
help
drive
conversations
floors,
but
also
to
document
taxonomy
or
conventions
that
used
to
communicate,
because
one
thing
that
I
found
is
we
spent
a
inordinate
amount
of
time
talking
past
each
other
when
we
basically
are
in
violent
agreement
but
but
words
matter.
So
we
should
make
sure
that
we're
consistent,
at
least
within
that
space
until
everybody's
on
the
same
page.
A
I
mean
I
mean
a
word
pack
in
the
documentation.
Are
there
other
things
that
folks
had
his
key
takeaways
or
lessons
learned
that
they
wanted
to
turn
up
on?
I
did
realize
that
the
holo
track
afterwards
was
actually
really
fruitful,
and
we
got
a
larger
amount
of
consensus
on
different
things
that
folks
wanted
to
get
done,
and
I
know
that
Andy
is
actively
working
on
I.
Think
with
Vince
as
well
adopting
me
to
try
and
address
some
of
those
key
takeaway
areas.
C
So
I
had
an
observation:
I
I,
don't
know
that
it's
really
actionable
population,
but
it's
occurred
to
me,
as
the
group
has
grown
larger
and
larger,
that
the
the
cost
of
our
words
is
more
expensive.
So
the
mathematician
and
me
thinks
about
it
like
this.
If
I
speak
for
30
seconds-
and
there
are
50
people
in
the
room,
then
that's
a
lot
more
than
30
seconds
and
because
so
many
people
want
to
speak.
C
Sometimes
it's
difficult
to
actually
have
a
linear
conversation
where
you
go
back
and
forth
and
make
progress
on
a
particular
topic,
because
quite
some
time
might
pass
before
you
have
the
urge
or
idea
to
speak
and
then
the
time
when
to
return
to
speak,
which
makes
it
again
hard
to
have
those
linear
conversations.
So
I
think
Tim
mentioned
this,
but
in
my
reflection
on
this
I
think
that
having
a
more
clear
agenda
is
one
will
help
drive
the
course
of
the
conversation
and
then
just
for
me
personally.
C
One
thing
that
I've
done
a
lot
in
the
recent
months
is
I.
Ask
myself
after
five
minutes
as
fast.
Do
I
really
need
to
respond
to
something
that
somebody
said
five
minutes
ago,
or
is
it
okay
to
let
the
conversation
proceed
now
that
we're
talking
about
something
completely
different,
though
that's
that
was
my
lesson
learned.
B
We
hope,
and
in
other
words
we
want
to
encourage,
also
contributing
code,
and
if
you
want
to
see
something
happen,
also
like
chime
in
and
take
care
of
the
issues.
If
there's
something
that
you
now
that
we're
formalizing
these
words
as
Tim
said
as
we
formalized
them.
If
there's
something
you
disagree
on
or
you
want
to
clarify,
go
ahead
and
actually
send
the
pr
to
clarify
or
retain
that
wording
instead
of
just
by
by
words,
going
back
and
forth
with
that,
all
those
people
are
the
same.
A
A
A
We
can
refine
that
particular
template
if
you
wanted
to,
but
if
we
want
a
synchronous
communication
on
feature
requests,
github
really
should
be
the
source
of
truth,
because
that's
the
only
thing
that
has
global
visibility
and
if
we
can
always
have
a
larger
discussion
on
issues
which
is
kind
of
normal
in
the
community
to
point
to
an
issue
that
have
this
separate
email
thread
or
discussed
channel
thread
regarding
that
kind
of
issue.
I.
A
A
B
D
B
D
A
D
A
A
Are
there
any
other
themes
or
things
that
folks
wanted
to
sort
of
highlight
in
the
release
notes
for
150
going
once
twice
three
times
next
up,
I
want
to
do
a
PSA
with
regards
to
because
we
talked
about
it
many
times
about
image.
Building
motion
I
had
discussed
several
times
back
and
forth
about
what
the
MVP
for
image
building
might
look
like.
We
came
to
agreement,
but
I
need
to
take
a
pass
through
the
doc
I
get
again
in
more
detail,
but
I'd
like
to
push
up
a
cap
for
a
sub-project
creation.
G
A
Yeah,
the
the
idea
was
that
the
tule
was
meant
to
be
batteries
included
with
swappable.
So
that
way
we
could
have
the
ability
to
have
a
phased
approach.
So
that
way,
if
people
had
existing
tooling,
that
could
shim
it
in
but
have
a
consistent
way
of
doing
it.
So
if
they
wanted
to
use
phase
1
versus
phase
2
very
similar
to
the
way
we
have
comedian
structured,
that
that
gives
them
the
option
to
leverage
it
in
a
piecemeal
fashion.
A
Good
yeah,
that
was
that's
our
primary
concern,
so
I
think
that
there's
a
tremendous
amount
of
words
in
the
stock.
What
how
many
pages
it
is
no,
but
we
should.
We
should
make
sure
that
it's
highlighted
that
the
goal
is
not
to
usurp
other
tooling.
The
goal
is
to
create
a
phased
approach
by
which
you
can
call
whenever
tooling,
that
you
deem
necessary
for
your
workflow.
G
A
A
But
within
those
individual
phases
it
has
a
default
action
right
and
we
were
going
to
create
a
wrapping
tool
to
allow
the
coordination
of
those
phases
right.
The
wrapping
to
will
be
very
declarative,
but
allow
you
to
have
the
hooks
to
use
whatever
truly
wanted
on
the
backend,
so
that
way
it
just
default.
Lee
creates
an
image
in
a
very
simplified
way.
We
wanted
to
allow
folks
to
be
able
to
use
phases
of
the
piecemeal
fashion.
A
So
if
people
wanted
to
basically
have
like
an
installer
phase
for
installing
the
software,
they
could
just
use
that
almost
as
a
as
a
separate
step
right
so
for
people
who
are
on
on-prem,
that's
a
common
user
story,
but
I
think
what
I
need
to
do.
My
plan
to
do
is
to
go
through
the
document
and
make
sure
that
the
images
that
we
created
earlier
on
line
with
the
narrative
that
motion
I
kind
of
agreed
upon
a
system
approach.
G
G
A
But
what
I'll
try
to
do
justin
is
I
will
try
to
clean
up
the
document
and
the
words
inside
of
there.
So
it's
kind
of
clear
from
a
user
story
perspective
and
put
a
shut
that
kept
2k
a
community
and
we
can
maybe
next
time
we
can
have
a
conversation
in
more
detail.
That's
not
good
yeah
and
we'll
do
the
standard.
You
know
X
weeks
for
common
time
out
thing.
B
H
Can
give
you
update,
we
release
Kappa
zero
point
three
point
out
for
the
other
best
provider,
there
is
a
breaking
change
and
cluster,
like
with
the
old
version,
are
not
supported
in
the
new
version,
but
there
is
a
migration
path
for
you
to
take.
H
H
I
I'm
not
saying
it
refreshed
on
in
zoom,
but
at
least
in
the
doc
I
added
a
link
to
our
discussion
on
the
scope
of
the
one
alpha
2
and
another
link
to
the
pull
request
that
you
made
mints
for
the
proposed
release
cadence
or
cluster
API.
So
if
you
have
feedback
on
either
of
those,
please
do
that
and
at
tomorrow's
meeting
I
imagine
we'll
discuss
those
in
more
detail.
E
B
Can
send
it
as
soon
as
the
points
error
or
is
out,
which
was
exactly
what
what
Ben
did
for
kind,
but
still
I
mean.
Although
Cuban
and
Congress
promotes
the
same
thing,
we
have
well
different
tests
for
Cuba
and
just
to
have
that
in
there
a
separate
product
to
now
with
its
own
logo,
which
is
gonna,
be
nice.
And
yes
just
so,
you
can
reproduce
it
with
wave
front
whatever
on
your
local
local
machine.
E
Yeah,
probably
we
can
discuss
it
tomorrow
is
kind
of
booked
with
the
CRS
report,
but
maybe
we
can
create
a
Google
Doc
for
this
animal.
Let's,
let's
talk
about
it:
offline,
okay,
so
H
is
going
doing
nicely.
We
graduated
do
one
beat
the
v1b
to
config.
It's
all.
The
changes
are
in
place.
A
I
can
add
a
link
later
to
the
doc.
You
can
check
it
out,
so
we
update
all
the
end-to-end
test,
jobs
as
well,
and
all
the
dogs
PRS
are
already
sent
to
the
website.
A
The
other
question
that
I
had
that's
type
of
mind
is
that
just
as
a
PSA,
because
it
cuts
across
projects,
we
talked
about
it
in
the
committee
and
group
in
order
for
us
to
graduate
to
the
config
to
GA.
We
want
component
config
to
be
in
a
more
healthy
state
because
it
would
drastically
affect
the
config
for
your
medium.
So,
as
we
move
forward
is,
will
probably
go
through
many
beta
release
cycles
until
we
actually
get
component,
they
convey
in
a
healthy,
healthy
state.
A
B
That's
right
basically
I'm.
Well,
basically,
we
would
need
more
help
like
actually
coding
work
to
component
standard
to
make
this
happen.
Yes,
and
also
the
one
of
the
problems
today,
is
that
less
very
long
delay
in
between
sending
PRS
and
getting
them
reviewed
and
or
merged
to
finding
consensus
in
the
API
mission
or
field
because
says
that's
just
a
few
and
that's
not
a
problem
of
the
reviewers
or
some
of
themselves.
It
is
more
like
a
structural
problem
of
having
too
few
final
approvers
for
api's,
which
is
a
thing.
B
A
We
can
like
coordinate
with
with
Jordan
and
maybe
have
a
session
where
we
can
kind
of
up
level
folks
who
have
been
around
a
while.
This
is
not
for
the
faint
of
heart.
I
would
not
recommend
like
new
people
to
go
into
this
blindly,
unless
you
have
a
lot
of
experience
in
crafting
api's
and
this
sort
of
landscape
just
because
it's
fraught
with
kubernetes
isms.
A
A
B
E
B
Can
yes,
I
I
have
some
accounts
in
there
and
I'm
actually
not
on
this
week,
so
I
can
I
can
try
to
get
get
hold
of
this
rice.
It's
probably
pretty
pretty
unlikely
that
she
happens
to
have
a
time
then,
but
I
can
always
ask
if
she
could.
She
could
join
or
I
talk
to
her,
otherwise
at
some
kind
of
fit
delegate
that
perhaps.
A
E
E
Design,
that's
the
biggest
problem
yeah,
but
yeah.
The
other
book
is
a
good
one.
Yes,
a
box
of
sorts
like
maybe
3d,
but
not
like
a
Bosch,
the
Bosch
fire
type
of
3d
book,
no
I
don't
know.
Well,
maybe
we
can
take
this
off
offline
or
like
I
can
like
I
said
we
should
create
a
Google
Doc
for
this
problem.
E
A
G
G
Policy
I
do
think
it
is
the
right
thing
for
us
to
do
to
catch
up
and
we
gave
a
talk
which
I
don't
know
the
videos
are
posted
yet
so
if
people
are
interested
in
that,
like
you
can
sort
of
U
equal
to
coping
with
change,
but
it's
sort
of
dozens
like
the
whys
and
wherefores
is
why
we
did
that
much.
This
might
be
of
interesting.
Some
maintain
errs
but
yeah.
So
that's
what
we're
we
were
hoping
s.
Ok,.
B
Step
can
plan
a
good
thing
right,
so
we
have
a
new
proposal
for
cube
controller
manager,
configuration
and
I'll
just
quickly
if
you
can
click
up
the
the
proposal,
and
so
just
together
and
scroll
down
to
alternative
to
this
is
just
yes.
There
we
go
so
just
a
drafting
myself
and
Stefan.
We're
thinking,
cube
con
and
we're
gonna
formalize
this
and
it's
a
real
thing,
but
for
context
the
cubelet
is
doing
pretty
well
keep
rocks,
it
has
a
couplet
has
beta
component
config
the
schema
and
support,
for
example,
overriding
Flags.
L
B
B
So
that
our
cube
scheduler
is
also
pretty
much
the
same
as
as
foq
proxy.
It
has
component
config
but
likes
many
of
the
the
characteristics
that
are
needed,
however,
and
what
we're
doing
the
cube
controller
man
under
consideration
here
is:
keep
control.
Managers
have
own
and
config
at
all,
so
it
has
basically
has
no
dashes
config
flag,
which
we
want
to
try
to
fix
here.
We
have
but
there's
different
problems
like
we
need
to
do
so.
Suppress
some
opportunities
belong
to
the
manager
context.
That's
managing
many
controllers.
B
Some
options
are
common
between
controllers
and
we
need
to
be
able
to,
for
example,
the
main
object
be
v1
because
but
then
we
introduced
a
new
controller
with
an
unstable
schema.
So
what
did
we
do
those
kinds
of
those
kinds
of
problems,
but
we
we
have
some
kind
of
idea
where
it
starts
and
says:
please
join
join
the
conversation
there
and
the
workgroup
meetings
bandage
meets
component
standard
working
group
and
yes,
we
found
some
new
contributors
to.
Therefore
it,
for
example,
Valerie
Lancer
II
will
help
with
proxy
Oh
for
us
also
as
something
that's
really.
A
You
know
I
think
what's
interesting
too,
is
that
which,
which
a
lot
of
people
don't
really
appreciate
or
understand
unless
you've
been
working
on
this
stuff
for
a
long
time,
is
that
it
enables
a
much
more
configurable
system
having
these
capabilities
in
place.
So
one
of
the
ideas,
long
thought
idea
is
that
what
we
wanted
to
do
is
that,
at
least
in
my
mind
and
I
mentioned
this
several
times
so
to
get
you
hyped
up
about.
The
idea
is
that
people
often
screw
up
deployments
many
times,
so
you
don't
often
get
your
first
deployment
correct.
A
You
have
to
go
through
several
iterations
before
you
actually
do
this,
but
if
you
had
dynamic
computed
config
across
all
your
different
components,
you
could
do
recon
from
the
fly
for
your
different
components,
and
that
is
a
highly
beneficial
useful
thing
that
allows
folks
to
iterate
more
fast
in
their
environments.
So
if
they
want
to
enable
certain
features,
it
should
be
a
modification
to
your
config
that
allows
them
to
basically
sing
up
their
cluster
with
it.
Config
change.
A
A
L
D
For
comic
relief,
yeah
I
can
comment
on
add-ons
there's
good
progress.
I
believe
that
there
are
kind
of
ad
hoc
approach
to
governance,
has
been
working
lots
of
asynchronous
conversation
occurring
in
the
add-on
management
document
that
we
started.
Writing
people
talking
about
various
solutions
and
as
well
as
generic
constraints,
to
the
problems
that
the
add
ons
conversation
is
attempting
to
solve,
and
many
of
those
conclusions
are
now
becoming
actionable
as
a
result
of
two
and
a
half
weeks
of
this
kind
of
async
discussion,
as
well
as
a
real
looping
with
the
group.
D
D
Justin
and
Jeff
have
done
some
great
work
on
the
coop
builder
declarative
pattern,
which
allows
you
to
build
operators
easily
around
manifests
using
them
as
a
base
for
iterative
kind
of
work.
Inside
of
the
go
once
we
have
that
kind
of
thing
merged.
We
can
start
really
collaborating
in
the
repo
on
things
like
early
terminate
of
the
operator
for
the
job
use
case,
which
seems
to
be
an
attractive
idea
for
multiple
people
talked
with
several
parties
who
liked
the
idea
of
a
single
operator
codebase,
that's
also
usable
in
an
ad
hoc
fashion.
Allah.
D
Obviously
this
is
request
for
comment
as
much
as
it
is
a
light
proposal.
There's
some
interesting
ideas
there
and
then,
as
far
as
this
lifecycle
of
add-on
goes,
we
should
have
a
talk
on
the
coop
8:00
a.m.
call
about
what
the
mechanic
said
that
are
going
to
be
like
after
some
discussion
with
a
few
folks.
D
But
this
is
something
that
you
know.
At
least
we
could
get
koopa
dam
and
cops
to
agree
upon
right
and
then
that
allows
us
to
that
and
have
a
packaging
source
for
add-ons
from
a
singing
place
for
those
two
projects
and
then
potentially
benefiting
other
installers,
whether
it's
your
enterprise
thing
or
goop
spray
or
anybody
else
who
we
haven't
had
the
chance
to
talk
to
and
that's
kind
of
the
general
update
on
what's
happening
in
that
sub
project.
Does
anyone
have
any
questions
or
comments?
Do.
A
You
have
a
question
and
I
prepared
the
Lumiere
does.
My
first
question
is
like
I
think
from
it
could
be
the
inside
read
happy
to
pilot
some
of
the
work.
I
don't
know
if
we
need
to
enable
it
by
default
right
away.
We
definitely
turn
on
CI,
but
the
question
I
have
is
broadly
about
the
cluster
directory,
because
that
is
really
the
default
CI
signal
for
most
of
the
project.
I,
don't
know
if
you
plan
to
I,
don't
know
how
we
plan
on
dealing
with
that
and.
D
I
mean
if
we
were
moving
with
the
customized
thing,
I
mean
I
could
see
the
cluster
directory
maturing
into
something
that
is
central
package
source
for
core
add-ons
or
like
kind
of
in
in
or
add-ons,
but
with
the
emphasis
that
the
interface
that
we
are
using,
you
know
be
it
get
or
OCI
images
is
something
that
everyone
anyone
can
implement.
Alright.
So
if
you
are
calico
you
do
not
have
to
you
know
you
can
host
your
add-on
externally.
If
your
court
DNS,
you
can
host
your
add-on
externally.
L
G
It's
not
just
logistics,
it's
also
like
figuring
out
what
we
what
we
want
right.
So
hopefully
we
can
agree
that
as
a
community
like
do
we
do
we,
regardless
of
the
clusters
directory
itself,
do
we
want
a
single
place
where
we
do
try
to
like
have
a
test,
the
configuration
of
coordinates,
for
example,
and
secondarily,
if
we
do,
where
does
it
live,
which
I
think
are
not
trivial
questions.
G
The
the
value
of
the
clusters
directory
as
I
see
it,
which
is
probably
separate
from
my
gross
a
Googler,
is
like
just
testing
a
broader
set
of
like
e
to
e
things.
I
think
Google
has
its
own
or
I
know
who
has
its
own
like
tests,
fit
at
least
started
that
way,
so
I,
don't
think
Google
would
care
too
much
if
we
remove
that,
but
I
think
the
community's
project
would
suffer
if
we
just
like
blanket
deleted
it.
So
I
I
hope
we
find
a
way
to
rationalize
the
cluster
directory.
G
But
there
are
some
important
functions
that
it
is
still
fulfilling
like
providing
été
signal
on
clouds
to
the
open
source
communities.
Project
I
think
we're
agreed
like
separately
they're,
like
you
know,
I,
wouldn't
necessarily
we're.
Never
gonna
run
those
on
every
PR
and
we
want
to
get
those.
It's
like
those
sort
of
PR
test
to
run
on
on
Kubb,
ADM
or
kind,
or
something
like
that.
D
A
Mean
yet
this
this
is
a
really
thorny
discussion.
Maybe
we
can
table
this
specific
one
for
like
the
next
meeting,
because
I
think
we
could
actually
have
an
entire
conversation
about
how
we
could
plan
to
manage
the
slash
cluster
directory
and
there's
other
updates
in
a
lot
more
still
to
this
agenda
for
today.
So
I
wanted
to
make
sure
that
we
yeah
can
get
to
the
other
stuff,
but
folks
I,
the.
E
G
More
or
less
correct
today,
there
is
I
believe
an
effort
to
extract
out
a
package
like
a
pkg,
even
customized,
so
that
the
functionality
would
be
more
composedly
reusable
because,
for
example,
our
declarative
library
that
Jeff
and
I
wrote
like
as
a
lot
of
very
similar
functions
to
customize
that
it's
silly
for
us.
Each
to
we
implement
yeah.
D
And
then,
as
you
know,
that's
a
really
interesting
area
of
discussion.
I
guess
it's
like
just
in
closing
for
the
head-on
thing
is,
we
would
really
love
people
to
help
join
us
and
not
just
talk
about
things
but
also
write
code.
So
we
need
people
to
help
with
proof
of
concepts
and
implementation,
and
that's
a
like
the
customized
stuff
is
definitely
in
area
or
work
needs
to
be
done.
I,
don't.
A
Well
perhaps
like
these
two
topics,
we
didn't
discuss
next
study
because
they
broadly
applied
to
the
sig
across
a
number
of
different
efforts
that
people
are
trying
to
do
I
so
want
to
keep
on
going
because
we
have
ten
minutes
so
I
want
to
make
sure
we
get
to
spray
folks,
so
that
wanted
to
give
motion
chance
to
do
a
demo.
If
you
could
there's
a
KP
table
for
now,
yeah.
A
Next
up,
coop
spray
I
do
not
see
Antoine
but
I,
see
Chad
on.
Is
there
any
update
that
he
wanted
to
go
Chad.
O
P
A
P
Cool,
so
just
a
quick
intro.
This
is
basically
part
of
the
the
image
build
work
and
it's
sub
component
or
a
phase
of
that.
So
the
software,
provisioning
and
installation,
but
that
have
been
kind
of
PR,
seeing
in
the
background
and
it's
essentially
ger
tool
to
replace
and
through
fifth
puppet
bash
for
the
use
case
of
bootstrapping,
like
you,
ladies
first,
so
to
show
you
the
specs
cookies.
So
if
you
have
a
aspect
with
the
kubernetes
version
and
container
runtime,
we
can.
P
Manaphy
that
so,
basically
we
run
through
multiple
phases.
So
the
first
phase,
who
will
take
these
high
level
declarations
or
intends
to
say
I,
want
to
run
kubernetes
I,
wouldn't
want
to
container
on
time.
I
want
to
run
whatever
else.
It
is
so
very
specific
to
kubernetes,
and
then
we
take
that
out
and
emit
packages
commands
and
Falls
if
we
can
enjoy
an
action
and
we
do
it
in
a
multi
district
manner,
so
you'll
see
that
there's
Red
Hat,
like
fedora
like
and
debian
like
basically
can
handle
young
DLF
and
packaging.
P
We
embed
the
upstream
repos
for
all
of
those
and
then
that
basically
allows
you
to
do
a
couple
of
things.
So
the
first
thing
it
can
do
is
it
will
verify
that
what
you've
declared
so
obviously
not
at
a
command
level.
But
at
the
package
level
it
can
tell
you
what
you
haven't
have
not
installed
and
then,
when
you
run
that
it
it
will
then
go
and
execute
all
of
that,
and
then
you
can
reapply
it
and
it
will
verify
that
they
are
actually
installed.
G
P
It's
there's
on
to
start
from
a
stock
image
on
any
distro
install
kubernetes.
It
can
all
say
in
McCloud
in
it,
so
it
because
it
doesn't
work
on
at
runtime.
It
builds
out
these
virtual
file
system
analysis
of
commands.
It
can
emit
card
in
it
as
well.
So
you
can
use
it
essentially
to
bootstrap
in
any
way
image
cloud
in
which
all
random
walk,
I.
F
B
B
A
D
B
B
D
A
This
is
more
proactive.
We
started
doing
this
ahead
of
time
before
Paris
was
involved,
I
went
separate
things,
but
I
think
we
found
that
with
such
a
large
group
of
folks,
that's
coming
in
across.
You
know
varied
interests
that
we
want
to
be
able
to
have
a
clean
way
and
unwrapping
people
or
cleaner
method
of
unwrapping
people,
so
that
way
folks
can
be
engaged,
feel
engaged,
feel
like
they're
powered
and
feel
productive.
A
B
B
I
can
do
so,
then,
we'll
fill
in
for
we'll
fill
in
the
links
to
the
cube
con
talks.
Here
as
we
as
we
go,
we
had
a
bunch
of
them
which
we
got
our
sub
projects
and
cig,
intro
and
and
similar
just
for
the
people
that
wear
is
there,
so
they
can
catch
up
easily
and
then
I'm
not
sure
why
the
last
item
on
the
agenda
is
strikethrough.