►
From YouTube: Kubernetes Community Meeting 20170803
Description
We have PUBLIC and RECORDED weekly video meetings every Thursday at 10am US Pacific Time.
https://docs.google.com/document/d/1VQDIAB0OqiSjIHI8AWMvSdceWhnz56jNpZrLs6o7NJY
B
B
A
C
Okay,
so
I'm
sharing
my
screen.
Hopefully
you
guys
see
that
you
guys
see
that
we
do
all
right
cool,
so
I'm,
going
to
demonstrate
the
integration
between
the
recently
launched
after
container
instances
and
kubernetes
as
your
container
instances
is
I
got
in
trouble
in
hacker
news
for
calling
it
server
list,
but
I'm
just
going
to
keep
calling
it
server
list.
C
It's
service
container
infrastructure
where
you
can
run
in
the
container
and
you
don't
see
the
VM,
you
don't
see
the
operating
system
underneath
you
can
just
give
us
a
container
image
and
in
a
few
seconds
it's
up
and
running
it
available.
It's
really
low
level
infrastructure,
though,
and
that's
why
I
want
to
show
today
how
that
mobile
infrastructure
interacts
with
higher
level
Orchestrator
like
kubernetes
and
so
for
a
start
I'm
going
to
come
over
here
and,
oh
so,
cute
Oh.
C
C
And
now
you
can
see
that
it's
running
and
if
we
go
back
over
here
and
we
list
nodes
again,
we
now
have
this
virtual
node
that
has
been
up
and
around
for
10
and
seconds,
and
so
what
we're
doing
is
remodeling
as
your
container
instances
as
effectively
a
really
large
node
with
near
infinite
capacity
inside
of
kubernetes.
This
is
actually
a
little
bit
different
than
a
normal
machine
and
actually
some
ongoing
discussions
with
the
scheduling
folks
about
exactly
how
we
want
to
model
this
and
have
the
interactions
with
the
scheduler
work.
C
It's
up
there,
it's
pending,
it's
the
connector
is
taking
care
of
creating
the
container
up
on
Azure
container
instances
and
with
any
luck
and
there
you
go
so
18
seconds
in.
We
have
a
container
up
and
running
and
a
public
IP
for
that
container.
And
if
we
go
over
here,
we
can
curl
that
IP
address
and
with
luck
as
the
network
gets
set
up,
it
should
actually
not
that
I
feel
fitter
straight
with
luck.
As
the
IP
address
gets
set
up
and
packets
start
getting
routed
to
it.
There
we
go.
C
The
engine
axe
is
up
and
serving
on
that
public
IP
address
that
we
created,
and
so
you
know
we
went
from
with
no
BM
infrastructure.
We've
gone
from
no
nodes
in
our
cluster
to
this
virtual
node
in
our
cluster
to
a
container
up
and
running
on
the
public.
Internet
in
you
know
order
30
to
40
seconds,
so
I
think
that's
pretty
cool.
It's
definitely
experimental.
If
you're
interested
in
playing
around
with
it.
C
D
C
Wow,
so
a
taint
so
taints
and
taints
and
seller
Asians
are
actually
a
concept
that
are
in
the
scheduler
in
general,
and
so
basically,
if
a
node
has
a
taint
by
default,
the
scheduler
won't
place
a
pod
onto
it.
Unless
the
pod
says
that
it
tolerates
that,
and
so
basically
it's
in
both
cases,
it's
more
or
less
than
Ana
or
relatable.
Sort
of
the
note
is
labeled
with
the
tanks
and
the
pod
has
an
annotation
that
says
I'm
willing
to
tolerate
that
paint
and
then
the
scheduler
just
takes
care
of
things.
Okay,.
C
C
F
C
G
C
So,
as
your
container
instances
is
a
multi-tenant
service,
but
we
use
hypervisor
isolation
to
ensure
that
the
containers
are
separated
from
one
another,
so
that
that's
what
that's?
What
about
what
I'll
say
about
that?
We're?
Definitely
not
relying
on
container
container
isolation
as
a
security
boundary
between
tenets,
where
you're
reusing
hypervisors
around
each
container
as
the
as
the
security
boundaries.
So.
F
C
Yeah
absolutely
yeah
yeah
exactly
and
that's
actually
an
interesting
question.
That's
one
of
the
reasons
I
was
discussing
this
in
sig
scheduling
is
that
you
know
right
now.
The
scheduler
kind
of
feels
like
a
node
is
a
unit
of
failure,
but
of
course,
with
admin
container
instances.
It's
not
like
that
virtual
note.
C
F
It
all
right,
maybe
I,
didn't
ask
the
question:
I,
there's
some
basic
stuff,
I
didn't
understand.
You
scheduled
one
pod
onto
this,
and
you
put
it
in
the
potty
animal
with
tanks
and
televisions
that
it
should
go
under
the
azure
container
is
just
can
I
have
any
size
of
pods
like
what
happens
when
I
need
a
bigger
pod,
not.
C
As
much
yeah,
so
you
can,
you
can
put
anywhere
from
one
two.
You
have
to
start
there's
a
minimum
of
one
core,
but
you
can
go
from
one
core
up
as
well
as
flexible
memory
sizes,
and
so
you
can
request
whatever
resources
you
want,
if
it
doesn't
support
dynamic,
ree-ree
sizing
right
now.
So
if
you
had
one
core
and
you
wanted
to
go
to
four
cores,
you
wouldn't
have
to
create
a
new
instance
not
not
scale
up
the
existence
and.
C
H
C
Well,
so,
as
your
container
instances
are
really
low
level
right,
they're,
just
one-off
container
groups,
there's
no
there's
no
orchestration,
there's
no
replication,
there's
no
there's
no
notion
of
like
replica
set
or
a
service,
or
anything
like
that.
Alright,
or
even
things
like
secrets
right,
it's
really
intended
to
be
a
piece
of
low-level
infrastructure
and
we're.
H
C
Yeah,
it
could
I
guess
but
I
guess
my
right
now.
For
example,
you
can't
access
things
like
GPU
and
I.
Don't
know
how
that
we
ever.
Would
you
know
or
SSD?
You
know
there
beyond
which
HDI
on
a
CPI
right,
yeah
so
like
I,
think
there's
a
lot
of
use.
Cases
where
having
dedicated
nodes
makes
sense,
so
I
think
that
either
for
cost.
C
I
think
if
you're
running
lots
and
lots
of
long-running
services-
and
you
can
saturate
virtual
machines-
you're
better
off
running
ACS
and
virtual
machines.
But
I
expect
that
people
will
run
effectively
hybrid
clusters
where
they
have
some
dedicated
nodes
and
they
have
some
use
of
a
jar
container
instances.
C
F
I
C
C
E
C
J
J
J
K
This
distance
seems
very
reminiscent
of
the
ouzo
song
kubernetes
on
lasers,
work
Emma
I'm,
not
did
you
take
a
look
at
that
or,
and
that
didn't
actually
seem
to
work
very
well
in
the
long
run
because
of
the
impotence
mismatch
between
running
one
scheduler
on
top
of
another.
This
seems
very
reminiscent
to
that
kind
of
approach.
Could
you
make
some
comment.
C
Yeah
I
think
there's
a
difference.
I
mean
I
think
that
there
are
think
that's
exactly.
There
are
some
challenges
associated
with
it
for
sure,
I
think
that
that
had
actually
more
to
do
with
trying
to
model
lots
of
like
lots
of
things.
There
I
think
that
this
is
or
maybe
trying
to
model
the
entire
world.
I
think
one
of
the
things
I've
tried
to
be
really
explicit
about.
It's
like
I'm,
not
sure
that
everything's
going
to
work
right,
I
think
that
there
may
very
well
be
a
bunch
of
things
where
it's
like.
C
L
Hey
gran
question,
so
I
have
reboot
issues
so
I
may
miss
it
to
start.
There
are
certain
things
that
we
do
right
now
that
sort
of
sits
underneath
and
decide
doctor
and
the
cubelet
for,
like
logging
and
metrics
and
stuff
like
that,
is
there
any
support
which
I
haven't
looked
at
ACI
in
detail
for
that
type
of
stuff?
There's.
C
Well,
there's
logs
important
in
the
API,
but
if
you
wanted
to
use
like,
but
it
doesn't
have
the
sort
of
like
you
can't
launch
a
daemon
set
right
with
the
fluid
D
agent.
Next
to
it
right,
you
have
to
run
a
sidecar
or
using.
Are
you
guys
using
the
cubed,
or
are
you
actually
sort
of
running
your
own
thing?
So
if
there's
no
cubelet
in
there,
it's
all
I
just
wrote
the
connector.
The
connector
makes
all
the
right.
Api
calls.
Basically,
it's
sort
of
interfacing.
C
Yeah
I
looked
at
I
looked
at
sort
of
the
cubelet,
it's
got
funds
I
looked
at
I,
looked
at
trying
to
do
it
as
a
CRI
plugin.
Actually,
okay,
but
like
CRI
is
AK,
because
that
seems
logical,
but
actually
see
our
eyes
is
to
low
level,
like
the
lifespan
of
CRI,
expects
you
to
be
able
to
like
create
a
container
and
then
start
the
container
and
then
like
it's.
C
It's
really
kind
of
wrapped
around
the
fact
that
had
things
on
the
machine
with
you
not
a
restful
api
for
creating
containers
and-
and
we
could
have
like
made
the
api
more
compatible
with
CRI,
but
then
the
user
experience
of
the
api
by
itself
would
have
been
lousy
because
you'd
be
like.
I
have
to
create
a
container,
and
then
I
have
to
start
a
container
and
then
I
have
to
like
it.
Wouldn't
it
wouldn't
make
sense
and.
L
L
M
C
B
N
N
We
did
the
initial
feature
walk
through
today
and
had
some
six
representing
there's
some
six
missing,
so
we
didn't
actually
make
it
through
the
whole
list
and,
as
we
went
through
that
process,
bob
wise
had
some
really
interesting
ideas
about
making
sure
that
we
need
to
track
all
the
work
and
make
sure
that
we
have
a
handle
on
all
of
the
things
that
are
happening.
So
we're
going
to
have
a
discussion
about
that.
The
next
p.m.
meeting.
N
So
if
you're
really
interested
in
how
we
track
work
and
what
the
visibility
looks
like
and
all
that
please
show
up
at
the
next
Figg
p.m.
meeting,
it's
on
the
community
calendar
and
just
quickly
the
preliminary
work
breakdown
for
this
1.8
release
looks
like
35
al
features:
16
beta
and
9
stable.
So
just
a
this
is
an
initial
cut
super,
not
scientific.
We
haven't
done
reviews
and
just
as
a
general
note,
we
are
trying
to
make
this
stabilization
release.
N
So
having
a
vast
majority
of
alpha
features
is
something
that
we
need
to
look
really
hard
at
and
see
if
those
make
sense.
So
please
come
to
the
sig
team
meeting.
If
you
are
interested
in
its
process
and
just
go
through
the
quick
one,
73
was
released
this
morning,
the
geotech
and
1
6
8
is
in
process
and
Thank
You
Anthony
again
for
all
your
service
to
the
1
6
patch
manager
role.
That's
it
and.
F
K
I
think
that
that
was
really
the
specific
specific
thing
that
was
the
discussion
this
morning,
which
is
I.
Don't
I,
don't
think
that's
a
a
well.
You
may
be
right
and
the
fact
that
we
can't
answer
that
question
is
really
the
problem.
My
sense
is
that
there
is
actually
quite
a
lot
of
stability
work,
but
it's
all
being
done
outside
of
the
feature
repo,
and
this
I
think
is
a
gap.
We
have
to
figure
out
a
close
now
so.
O
I
think
that
there's
a
really
great
opportunity
for
1.9,
so
the
ability
release
I
mean
the
times
been
banding
around
a
bunch.
I
remember
you
know
over
the
last
few
releases
in
kubernetes
and
I,
don't
know
how
well-defined
it
is.
I've
heard
different
things
and
I've
also
heard
that
you
know
different
things
feel
like
they're
participating
in
a
stability
release
more
than
others.
I
think
we
have
a
really
good
opportunity
with
1.9.
If
we
start
thinking
and
planning
now,
you
know
to
make
that
a
release.
N
Think
that's
great
and
for
sodium
I
just
quickly
as
a
release
manager,
you
don't
know
what
those
what
it
looks
like
until
it's
actually
in
the
features
repo
in
our
current
process.
So
that
puts
you
at
a
distinct
advantage
in
terms
of
beating
the
drum,
because
the
drum
actually
doesn't
isn't
been
filled
yet
so
we
need
to.
We
need
to
work
our
way
around.
That.
L
One
last
icon
of
architecture
we've
been
talking
about
trying
to
have
her
other
proposals,
process
and
I.
Think,
as
that
happens,
the
the
features
repo
which
is
currently
use
for
both
tracking,
whether
we
should
do
something
or
not
versus
when
and
how
we
actually
track
it.
Separating
those
things
out,
I
think
will
help
make
it
clearer
in
terms
of
tracking
stuff
for
our
base,
an
implementation
of
a
future
versus
the
design
process
of
that
feature.
And
how
do
we
actually
discuss
that
so
so
something
to
be
aware
that
there's
conversations
going
there
in
students
so.
F
Think
the
question
is:
what
do
we
do
given?
What's
in
the
futures?
Repo?
Now,
that's
that's
one.
Secondly,
I'm
not
sure
that
we're
following
this
whole
process
of
splitting
features
into
the
futures
reposes,
despite
those
like
the
future,
closed
beta
tests
coming
and
gone,
it's
not
clear
that
everything
is
in
the
futures
tree
well,
so,
and
the
purpose
of
the
features
reposed
to
provide
visibility
across
the
community
and
so
that
we
have
that
transparency
and
any
kind
of
dependency
which
endicott.
So
those
are
my
two
questions
to
the
group.
H
Yes,
I
would
say
that
what
enforce
means
is
a
larger
question
of
governance.
What
a
preacher
means
also
I
believe
in
the
leadership
meeting
that
we
had
I
believe
there
was
consensus
that
this
should
have
been
a
stability
released
that
if
that
meeting
did
not
include
all
of
the
signal
reads,
then
I
guess
for
the
next
time.
We
can
be
better
about
that,
but
we
did
get
together
there
as
a
fairly
large
group,
and
that
did
seem
to
be
the
consensus
coming
out
so
I
think.
Certainly
there
is
our
questions
about
SIG's.
H
You
know,
following
on
a
agreed
consensus
and
what
that
even
means
how
that
relates
the
overall
governance
structure.
I
think
most
of
these
problems
all
relate
to
and
in
an
unclear
governance
rule
for
the
steering
committee
that
has
yet
to
be
formed.
I,
don't
know
how
much
headway
we
are
going
to
be
able
to
make
until
the
governance
process
provides
more
structure
for
the
project
to
work
with.
H
H
H
Getting
merged
in
the
Ricoh
today
so
they
say
that
is
purely
like
managing
a
bunch
of
issues
over
here
has
no
authority
or
way
of
stopping
code
from
getting
merged
in.
We
see
this
a
bunch
of
times
when
it
comes
to
release
burndown
when
things
are
trying
to
sort
of
like
sneak
in
past
that
lines,
the
excuse
is
often
that
will
look
like
90%
of
the
code
is
Rp
in
here.
What
are
you
going
to
be
like
take
up
on
my
campus
and
invert
them?
You
can't
do
that.
H
So
we're
not
trying
to
top
this
problem
here,
but
yeah
I
want
to
point
it
out,
like
I.
Think
that's
something
that
was
coming
up
as
you're
saying
it.
We
had
people
in
a
room
or
if
we
had
everybody
together,
but
certainly
I
think
it
sounds
like
the
gap.
H
Is
the
communication
from
that
room
to
buy-in
and
consensus
like
there's,
two
parts
of
lazy
consensus,
actually
making
sure
you
guys
write
according
this
events
and
then
ensuring
that
it's
distributed
widely
enough,
that
everybody
who's
impacted
is
aware
of
it,
and
maybe
the
gap
is
little
bit
on
that
side.
For.
N
We
need
a
restaurant
that,
because
honestly
I
knew
this
was
a
stabilization
release
at
one
six,
when
we
said
the
even
releases
we're
going
to
be
targeted,
a
stable
position,
I,
don't
I,
don't
think
the
communication
lack
of
communication
thing
is
necessarily
the
problem.
It's
what
actually
is
a
feature
and
what
is
what
does
stabilization
means?
It
means
label
izing
in-flight
features
that
are
output
of
a
debate
at
this
table.
That
is,
it
MIMO.
O
Sig
that
feels
like
the
crux
of
it
to
me
very
much
so
when
I'm
trying
to
communicate
now
to
the
things
that
I'm
involved
in
it's,
you
know
what
what
do
we
have
that
in
you
know,
obviously
I'm
I'm
not
in
touch
enough
and
I
know
stuff
or
something
but
I
miss
the
communication.
But
the
question
is
now
that
I
know:
where
is
that
definition?
O
J
H
Nothing
that
I
don't
like
this
is
maybe
like
a
steering
committee
issue,
because
to
me
this
comes
back
to
who
has
the
authority
to
say
no
too
often
in
this
project
that
belike
for
want
of
a
document.
We
don't
do
anything
until
that
document
magically
appears,
and
then
we
just
kind
of
tear
down
what
status
quo.
I
F
F
We
can
have
stance.
Igloo
can
put
together
a
proposal
making
smart
proposal
that
says
here's
what
it
means
to
be
a
stability
release
and
the
three
things
in
there
is.
You
know
you
should
have
only
features
moving
to
beta
or
stable
and
any
hassle
features
that
you
have
should
be
in
support
of
the
features
that
are
moving
to
beta
or
stable
and
then
anything
else
that
alpha
needs
to
go
through
an
exception
process
by
the
release
team
and
the
sig
leader.
That's
a
proposal
poke
holes
at
it.
N
A
Next
one,
your
field
house
scheduled
for
810,
it's
remarkably
difficult
to
get
seven
of
us
to
have
a
two-hour
block
of
time
to
talk
right.
So
the
agenda
for
that
is
to
define
a
fine
line
to
seating,
a
steering
committee,
a
all
the
action
items
to
you
get
between
here
and
said.
Speaking
of
the
steering
committee,
a
revision
of
v2
revision
on
the
steering
committee
charter
and
then
some
sort
of
conversation
about
the
question
that
was
bandied
about
around
ratification
of
the
steering
committee
charter
or
not.
H
Sounds
great,
oh,
how
many
hold
up
sorry,
Gloria,
kill
I
think
my
only
whole
is
that
power
without
discharge
is
not
which
I
think
is
our
current
current
predicament.
If
no
one
has
time
to
do
any
of
the
things
digitus
charge
their
authority,
then
that
they
do
not
actually
possess
that
authority.
I,
don't
believe
in
a
practical
matter.
P
So
I
would
make
a
comment.
My
name
is
Jay,
go
I'm
fairly
new
on
this
project.
I've
been
here
for
about
three
months
now:
I'm
an
engineering
manager
and
I
work
with
the
groups
that
correspond
to
sit
ups
and
sig
workloads,
API
apps
life
cycle
and
perhaps
due
to
the
fact
that
I'm
fairly
new
and
naive
I
took
the
signal
that
this
is
a
stability
release,
as
this
is
a
stability
release
and
I
would
just
remind
everyone
here.
P
A
F
L
One
thing
to
note
is
that
if
you
want
to
make
progress
on
a
feature
you
can
do
that
outside
the
core
using
extensions
Brennan
introduced
some
really
fascinating
stuff
that
a
demo
this
morning
without
changing
Corrections.
So
not
almost
innervations.
You
know
they
shouldn't
have
to
go
in
the
main
release,
the
release
valves
for
doing
stuff
and.
N
Lastly,
I
think
it's
important
if
you
rewind
to
the
sentiment
in
the
room
when
we
had
the
leadership
some
movement,
the
whole
idea
of
stabilization
when
we
were
in
that
room
was
not
introducing
more
technical
debt
and
not
getting
in
a
situation
where
we're
not
paying
down
existing
technical
debt.
I
think
that
we
we
all
want
a
stable,
viable
project
and
the
more
technical
that
we
accrued
the
harder
it
is
to
maintain
that
from
70
perspectives.
N
So
I
think
that
if
the
rule
of
thumb
is,
is
we're
not
introducing
the
technical
debt
and
we're
not
in
a
situation,
we're
not
paying
it
down,
then
I
think
that's
the
that's
consistent
with
what
a
civilization
release
is
so
I
think
that
that
should
be
part
of
the
calculus
in
terms
of
how
we
move
that
decision
forward.
So.
O
O
No
one
disagrees
with
Jase
I
haven't
seen
that
happen,
so
my
question
now
is
like
given
where
we
are
1.8
and
the
fact
that
we
I
think
the
number
was
we've
got
like
50
items
in
the
futures
repo
and
about
20
of
them
are
our
new
features?
Are
they
you
know?
What
do
we
do
like
this
today?
Do
we
look
at
what
do
we
do
it
today.
N
L
Love
to
see
sig
p.m.
actually
move
its
time
to
something
that's
more
accessible
to
folks
who
actually
a
commitment
in
the
morning
I
mean
we
brought
that
up.
I
can
never
make
it
personally,
so
I
can
be
great
to
actually
have
it
before
or
after
or
join
with
this
meeting,
because
there's
a
ton
of
overlap
between
the
audiences.
H
F
N
L
N
N
Q
Still
here
still
waiting
okay,
so
this
is
David
II
from
Cynthiana
machinery.
I
will
go
ahead
and
describe
what
it
is
we
plan
on
doing
in
1/8,
so
in
1/7
we
introduced
several
new
features
right.
We
introduced
custom
resource
definitions
and
evolution
of
third-party
resources.
We
introduced
initializers
and
admission
webhooks
we're
going
to
be
working
on
enhancing
and
moving
all
of
those
forward.
I,
don't
I,
don't
know
how
likely
it
is
actually
achieve
beta
we're
still
assessing
what
is
what
needs
to
be
done.
Q
We're
making
our
lists
about
which
features
are
required
to
be
able
to
move
things
like
initializers
in
a
mission
whether
it
looks
the
data,
but
we
will
be
executing
on
that
list.
We
have
we've
already
started
identifying
some
features
that
are
missing
and
we're
talking
with
the
sig
CLI
about
how
to
visualize
this
uninitialized
things
and
other
internal
caveats.
Like
that,
the
custom
resource
definitions
we
are
looking
at
adding
validation
to
them.
That
is
moving
ahead.
There's
a
work-in-progress
full
that
will
happen
in
1/8.
Q
We're
going
to
have
a
design
out
for
sub
resources
on
custom
resource
definitions
in
particular
status
is
one
that
people
ask
about
a
lot
right
aspects
that
a
split
it's
common
as
a
piane
and
we're
going
to
see
what
we
can
do
in
custom
resources
about
that
again,
not
sure
it's
going
to
make
one
eight,
but
the
design
will
definitely
be
open
and
we'll
start
gathering
input
there.
We
also
have
a
one
feature
around
HD
on
Aging.
It's
a
scaling
problem
that
we
have
today
in
the
1/6
and
1/7
branches
of
the
code.
Q
Where
you,
you
ended
up
with
some
really
key
performance,
we're
going
to
be
looking
at
what
we
can
do
there.
Our
goal
in
one
page
is
to
actually
look
at
our
measurements
and
see
whether
taking
advantage
of
some
existing
sed
functionality
is
going
to
help
us
or
hurt
us,
and
the
last
big
item
of
note
is
garbage
collection
right
now.
Garbage
collection
doesn't
support
custom
resources,
it's
a
major
shortcoming,
and
so
we
have
started
introducing
the
ability
to
have
that
work
in
1/8
for
custom
resources,
which
will
be
good
for
everyone.
Q
B
N
E
N
We
go
a
new
law
yeah.
She
was
a
time-limited,
so
I
have
three
announcements
today.
The
first
is
we're
doing
our
first
over
ever
kubernetes
office
hours,
and
these
are
basically
going
to
be
in
a
style
similar
to
this,
where
we
have
kubernetes
developers
available
to
answer
user
questions
and
it's
going
to
be
a
mix
of
doing
a
live
stream
on
YouTube,
where
people
can
ask
live
questions
and
we'll
actually
be
fielding.
Questions
from
Stack
Overflow
as
well
I've
got
a
whole
document
that
I've
linked
there.
N
The
first
one
we're
going
to
do
is
going
to
be
August
23rd,
and
this
is
going
to
be
at
a
European
friendly
time
zone.
We
figure
we're
going
to
start
there,
because
we've
got
a
lot
of
great
volunteers
there
who
have
already
committed
so
we're
going
to
give
that
a
shot,
see
how
it
works
and
adjust
from
there.
So
if
you've
got
time
to
spend
either
help
out,
we
need
help
everything
from
organizing
to
helping
sort
questions
and
things
like
that
and
and
marketing
all
sorts
of
things.
N
So
if
you're
not
an
engineer,
I
can
also
use
your
help
and
today
also
I.
Do
we
announced
Arkenstone
ibly?
There
are
two
open-source
tools
that
we're
hoping
that
the
community
would
check
out
and
dig.
Ark
is
a
utility
for
doing
a
disaster
recovery
on
your
cluster
and
it
currently
supports
through
the
major
cloud.
So
we
hope
to
check
that
out
and
sonobuoy
is
something
that
you
run
in
your
cluster
and
it
gives
you
diagnostic
information
and
runs
conformance
tests
in
a
non-destructive
manner,
and
things
like
that.
N
D
A
quick
question
sure
so:
will
there
also
be
office
hours
for
trying
to
get
new
contributors
as
they're,
trying
to
get
up
to
speed
and
learn
things
and
learn
from
the
more
senior
developers?
Is
that
a
separate
meeting
or
the
same
meeting
sure
so
one
thing.
N
We
weren't
sure
about
is
who
the
target
audience
is
going
to
be.
Was
it
going
to
be
users
or
new
contributors?
We
figured
for
this
first
one
that
we
would
start
with
users,
because
it
feels
like
there's
always
tons
of
questions
and
users
and
on
Stack
Overflow,
and
things
like
that.
I'm
not
opposed
to
having
and
we're
also
doing
it
once
a
month
currently
right
now,
so
we're
basically
going
to
hold
meetings
as
as
often
as
we
have
contributors
so
yeah,
we
would
love
to
have
sessions
that
are
totally
only
contributor
based.
N
You
know
your
very
first
pull
requests
right
or
you
know
how
to
file
your
first
feature
submission.
You
know
things
things
of
that
nature
we
figured
user
support
would
be
first
because
we
know
there's
an
obvious
need
there
and
we
figured
it
would
kind
of
help
us
test
test,
how
we
run
the
show,
we're
not
quite
sure
the
technicalities
of
running
a
live
stream
with
live
questions.
There's
a
question
of
lag
things
of
that
sort.
So
we
figured.
N
D
So
this
is
an
absolutely
fantastic
idea.
Just
a
quick
question
would
be.
How
are
you
advertising
it
in
other
ways,
with
some
of
us
that
have
pretty
decent
Twitter
followers?
Perhaps
is
there
something
we
can
tweet
out
just
to
help?
Get
you
more
advertising
when
we
do
to
help
get
you
more
advertising
and
both
on
the
thing
and
and
then
you
know
what
you
need
volunteer
wise
would
be
fantastic
sure.
So
the.
N
Initial
of
the
initial
announcement
was
just
to
get
volunteers,
because
if
I
couldn't
get
into
volunteers,
it
wasn't
going
to
work.
Luckily,
we've
got
a
lot
of
great
people
signing
up
I'm,
currently
working
on
tweets
that
I'll
be
sending
out,
and
you
it'll
be
more
obvious
on
the
user
mailing
lists
and
on
social
media
when
I've
announced
it
actually
to
the
end
users,
because
well,
we
want
the
appropriate
questions
ahead
of
time
and
people
good
I
will
be
making
announcements
on
all
the
appropriate
lists.
I've
got
something
more
consumable
for
the
end
user.
K
I
also
have
a
question.
This
was
a
you
said.
The
word
conformance.
Is
there
a
conformance
working
group
meetings
as
being
done
under
the
the
conformance
being
done
under
the
auspices
of
one
of
the
existing
SIG's?
It's
possible
I'd
miss
something
in
kubernetes
dead.
In
terms
of
announcements.
Can
you
describe
where
that
conformance
work
is
being
done?
Yes,
nothing.
Okay,
go
ahead,
George
Berger,
I
know.
L
You
got
it.
You've
got
all
the
details.
Yeah
it
kept
me
out
of
the
fake
testing.
I.
Think
there's
this
a
little
bit
of
an
odd
split
that
there's
a
lot
of
folks
that
are
interested
from
the
CNCs
side,
and
so
there's
been
a
series
of
ongoing
meetings.
There
and
I
know
that
Tim
Sinclair
from
our
side
has
been
trying
to
sort
of
represent
that
from
the
from
the
kubernetes
IBSA
testing.
So
that's
a
place
to
look
there.
R
There's
good
conversations
going
back
and
forth
and
they
believe
Brad
is
also
on
that
meeting.
So
there's
there's
a
bunch
of
people
in
this
room
who
are
actually
in
Clayton.
Was
there
too,
who
are
both
from
the
kubernetes
community
involved
in
different
areas,
but
they're
trying
to
drive
things
through
suggesting?
But
you
know
they're
I
think
there's
going
to
be
some
communications
going
back
and
forth
between
the
working
group
and
in
this
sega's
well
mystic
testing,
as
well
as
Sagarika
tech
chure,
to
define
what
is
the
state
space
of
what
is
conformance
right
well,.
K
Okay,
I
think
that's
the
I
think
that's
what
we're
not
doing
a
good
job
either
on
the
community
side
or
the
CNCs
side
and
I'll
just
take
the
CN
CF
I
can
take
an
action
on
the
CNC
outside
I.
Don't
think,
we've
I
I,
don't
think
we
have
announcement
that
they
believe
their
admin
would
invite
participation.
Yeah.
K
Thanks
for
the
thanks
for
the
link
there
Joe,
that
was
a
good
explanation
as
well
front
of
the
text
and
if
we
have
communicated
that
widely
in
this
meeting,
consider
it
now
communicated.
Yes,
that
was
part
of
my.
That
was
part
of
my
reason
for
asking.
The
question
was
just
to
make
sure
that
we
are
communicating
that
this
is
an
effort
going
on
and
if
you're
interested,
please
get
involved.
H
I
have
a
one
quick
shout
out
to
Aaron:
he
dropped
in
a
link
on
in
the
chat
window.
He
is
trying
to
document
how
the
owners
files
are
currently
used.
So
please
take
a
look
at
that
PR
and
comment
if
you
believe
that
is
an
inaccurate
state
of
the
of
the
world.