►
Description
Want to view more sessions and keep the conversations going? Join us for KubeCon + CloudNativeCon North America in Seattle, December 11 - 13, 2018 (http://bit.ly/KCCNCNA18) or in Shanghai, November 14-15 (http://bit.ly/kccncchina18).
Panel: Kubernetes Office Hours Live! - Moderated by Jorge Castro, Heptio Inc. and Ilya Dmitrichenko, Weaveworks (Any Skill Level)
This session is a live office hours session about Kubernetes. We will have a panel of Kubernetes developers available for you to ask questions and have them answered live for the audience.
This session will be participatory, so bring your questions and tips and tricks to share with everyone.
About Jorge Castro
Jorge works at Heptio Inc. as a dedicated community manager for Kubernetes. You can find him in SIG Contributor Experience, where he is currently working on a slew of projects. If you're looking to get started contributing please introduce yourself!
About Ilya Dmitrichenko
DX, Weaveworks
About Bob Killen
Bob has been with the University for 15 years, and in his time there has spurred adoption of Cloud Native tools and practices. He is passionate about bringing Cloud Native Computing methodologies to Research and Academia.
About Kris Nova
Kris Nova is a Senior Developer Advocate for Heptio. She is a Kubernetes maintainer and contributor. She wrote the book on cloud native infrastructure. She spends her free time climbing mountains.
About Carolyn Van Slyck
Carolyn is a software developer based in the wilds of suburban Chicago, working remotely on the Microsoft Azure Containers team. Her passion is developer tools, and automating all the things with a sidecar of containers. She is a maintainer for the Go dependency manager (dep), a Kubernetes contributor, and Service Catalog SIG member. As an organizer for Women Who Go, she regularly hauls her cookies across the world to share her love of Go, Kubernetes, and excessive emoji with her peers.
Join us for KubeCon + CloudNativeCon in Barcelona May 20 - 23, Shanghai June 24 - 26, and San Diego November 18 - 21! Learn more at https://kubecon.io. The conference features presentations from developers and end users of Kubernetes, Prometheus, Envoy and all of the other CNCF-hosted projects.
A
Welcome
to
office
hours
live.
This
is
actually
a
live
representation
of
our
officers
program,
which
happens
every
third
Wednesday
of
the
month.
If
you
type
into
google
kubernetes
office
hours,
we
have
a
session
for
europeans.
We
have
a
session
for
the
west
coast
of
the
Americas.
We
have
written
down
how
we're
doing
this
and
streaming
this.
So
if
you
are
in
an
area
the
idea
is,
you
should
be
able
to
execute
your
own
office
hours.
Just
by
reading
all
the
stuff
I
have
like
a
whole
script
that
you
can
read.
We
have
an
intro.
A
We
have
an
outro
for
those
of
you
watching.
The
video
would
really
really
love
to
bring
this
closer
to
the
Indian
timezone
and
that
part
of
the
world
and
the
Far
East,
as
as
an
American.
What
I
call
the
Far
East.
So,
if
you're
interested
in
doing
that,
we
have
the
resources
to
help
you.
We
can
help
you.
A
We
just
need
people
to
get
together
in
those
time
zones
in
a
be
able
to
let
us
execute.
So
what
is
office
hours?
Basically
I,
I
sort
of
have
to
you,
I,
didn't
know.
Kubernetes
I
was
gonna,
write
everything
that
I
was
gonna
edit,
a
wiki
right
I
did
the
Linux
thing
and
I
was
like
you
know
it's
the
new,
it's
the
new
era.
A
I
wouldn't
do
a
radio
show
on
YouTube,
because
that's
what
all
the
kids
are
doing
and
I
talked
to
Ilya
and
he
was
like
we'd
wanted
to
do
this
for
a
long
time
and
he
comes
from
a
user
experience
background.
So
he
was
like
great.
It's
like
there's
a
safe
place
where
you
can
just
ask
questions.
No
judgement:
no
one's
gonna
yell
at
you.
A
If
you
don't
know
what
a
pot
is
Joe
doesn't
know
what
a
pot
is
just
kidding
just
exactly
exactly
so
what
we
really
do,
because
we
can't
go
in
and
like
diagnose
your
cluster
and
SSH
in
what
we
do
do
is
is
talk
about
the
concepts
and
the
basic
understandings
to
give
you
a
place
to
go
so
using
the
pod
example.
When
we
had
Joe
on,
we
didn't
just
ask
him
to
explain
what
a
pod
was
right.
It
was.
How
did
you
back
then?
A
How
did
you
come
to
the
conclusion
of
a
pod
being
the
primitive
right?
Why
does
every
pod
have
an
IP
address?
So
when
we
talk
about
the
concepts
like
that,
I
feel
it
helps,
people
and
users
understand
what's
happening
and
and
your
understanding
of
kubernetes
exponentially
grows
and
mine
has
just
I,
don't
even
I,
don't
even
answer
questions
I
just
sit
there
and
madhuri.
A
So
we
have
a
lot
of
great
companies
that
donate
engineering
time
to
to
allow
their
engineers
to
help
users,
and
they
do
this
every
month,
so
I'm
gonna
introduce
our
esteemed
panel
by
the
way
we're
always
looking
for
volunteers.
So,
if
you're
looking
for
a
way
to
give
back
to
the
Cooper
naty's
community,
this
is
a
great
way
to
do
it.
You
don't
have
to
be
a
total
expert,
because
there's
enough
of
you
up
here
that,
like
you,
can
just
George
it
which
is
fake
it
till
you
make
it
so
booyah.
A
C
D
E
Amelia
from
B
works,
working,
D,
X
team,
develop
experience
and
focusing
on
developer
experience.
Well
when
it
comes
down
to
app
developers
using
kubernetes,
perhaps
for
the
first
time
or
you
know
there,
some
some
of
them
used
for
the
first
time
and
you
but
like
there's
that
developer
experience,
stuff
up,
develop
experience
and
and
also
know
a
fair
bit
about
networking
and
and
things
related
to
that
I'm.
G
H
C
G
E
Yeah
a
bit
of
working-class
lifecycle
stuff,
so
it
really
depends
on
the
style
you'd
like
to
go
about
this.
So,
for
example,
one
one
option:
if
you're
interested
is
it
could
be
Linux
kit,
you
could
actually
bake
images
which
include
all
the
beds
with
Linux
kit
and
make
them
sort
of
it's
like
I.
Don't
know
if
you
heard
of
Linux
good,
but
it's
basically
like
a
toolkit
for
building
your
own
container
optimized
operating
systems
right
and
you.
E
You
could
also
consider
it
Google's
container
optimized
device
as
well,
but
if
you
want
to
use
a
more
traditional
or
less,
that
soap
is
the
least
kid.
You
could
basically
build
VM
images
that
you
just
deploy
for
a
master
or
node
or
whatever,
and-
and
there
is
very
little
configuration
you
just
need
to
pass
the
metadata
in.
E
So
that's
one
option
like
there's,
also
a
question
of
whether
you
want
to
like
use,
VMware
storage
or
anything
like
that,
if
your,
if
your
Furies
in
a
paid
version
of
VMware
a
free
version,
obviously
it's
also
the
question
whether
you
have
that
added
capability
like
storage
or
anything
light
up
and
another
thing
I
know.
There's
some
supporting
cops
but
I
think
it's
still
early
days.
E
So
you
know
if
you,
if
you
prefer
to
go
the
route
that
that
would
basically
you
would
basically
use
some
of
the
tools
they
used
in
the
past,
whatever
you
used
in
the
scripts
to
install
Linux
for
for,
like
that
that
you
mentioned
right.
If
you
want
to
build
up
on
that,
then
then
there's
probably
a
way
to
do
that
but
like
if
you
want
to
do
something
completely
new.
G
H
J
K
J
There's
a
ton
of
projects
out
there,
the
one
that
that
I've
been
working
with
a
lot
and
that
we're
continuing
to
invest
in
is
cube.
Atm
now,
qadian
cube
ATM,
we're
gonna
be
working,
it's
it's
marked
as
beta,
but
it's
pretty
stable
and
a
lot
of
folks
are
using
it.
If
you're
just
doing
a
single
node,
a
a
single
node
control
plane,
it's
really
easy
to
get
going
right
away.
There
are
instructions
for
doing
H
a
it's
a
little
bit.
More
sort
of
you
know
elbow
grease,
but
you
can
get
their
cube.
J
And
so,
as
we
start,
you
know
getting
more
clarity
around
those
layers.
I
think
that
we're
gonna
we're
gonna,
see
more
mix-and-match
of
these
different
types
of
components
and
more
flexibility
and
new
features
like
cluster
auto-scaling
in
a
more
Eric
way.
So
I,
don't
think
anybody
using
those
things
has
I
haven't
seen
anything.
That's
like
here's.
The
instructions
boom
boom
boom,
but
if
you're
gonna
start
looking
at
technologies
and
investing
in
understanding
them
understanding
them,
that's
the
stuff
that
I
would
start
looking
at.
B
Yeah,
like
additionally,
two
to
six
Lassa
life
cycle.
There
is
now
a
work
group
on
cluster
API,
so
which
is
very
new
this.
What
like
Joe
was
talking
about
when
talking
about
Cuba
and
Cuba
corn.
So
there's
like
going
the
API
approach
to
spinner
cluster,
if
you
want
to
do
it
right
now
and
you're
on
Korres,
try
like
pixie,
booting
VMware
is
very
easy.
We
do
that
on
prymatt
some
customers
and
then
try,
maybe
ignition.
There
is
a
tool
by
Korres
or
now
Red
Hat,
which
is
called
matchbox.
B
E
Just
wanted
to
clarify
on
my
first
comment
regarding
Linux
get.
Let
me
skid
is
like
an
alternative
to
Korres,
but
it's
more
DIYs,
you
fully
in
control
of
what
goes
in
the
image
and
it's
built
on
cube
ATM
already
as
well.
So
a
take
a
look
at
interview
for
you're
curious
yeah.
It's
like
you,
said
of
using
an
upstream
image
from
Korres.
You
would
build
it
yourself,
so
you
don't
get
the
updates,
but
it's
a
different
model
right.
Just
something.
I
worked
on
I
thought
that
bench
in
it.
E
L
I've
so
hi
I'm
Stephan
from
Express,
and
we
run
all
justice
and
everyone
knows
the
set
up
the
untested
in
all.
This
is
a
pretty
complex
topic
and
I
fear
that
this
gets
more
and
more
complex
and,
of
course,
there's
tools
like
you
button
that
help
if
the
complexity,
but
the
complexity
is
still
there
just
hidden
from
you
and
in
the
end,
if
something
goes
wrong,
you
still
need
to
deep
dive
and
I
think
if
you
have
set
up
a
new
class.
F
So
I
mean
I
think
the
important
balance
you
want
to
strike
is
the
balance
between.
Like
is
the
right
amount
of
complexity
right,
because
if
you
have
too
little
complexity,
if
you
go
to
too
much
effort
to
hide
complexity,
either
you
end
up
with
something
that
is
very
constricting
and
doesn't
meet
people's
use
cases
or
like
will
hit
a
real
world
use.
F
Works,
works
fine
for
a
small
cluster
and
then
hits
a
real
word
use
case
in
a
big
cluster,
a
big
organization
that
doesn't
work
and
having
like
way
too
much
complexity,
where
it's
not
needed
so
like
in
the
case
of
certificates.
In
the
case
of
CAS,
there's
there's
actually
a
really
good
reason
for
all
those
difference.
F
Cas
Joe
agree,
disagree
with
me,
but
there's
a
good
reason
for
a
few
of
them,
but
I
I,
think
yeah,
but
but
no
I
think
like
I,
think
there's
places
where
we
could
simplify
and
I
think
there's
places
where
we
could
do
a
lot
better
at
explaining
like
what
it
is.
That's
actually
going
on
and
like
the
reasoning
behind
what
we're
doing
and
less
of
just
like,
generate
these
CAS
and
put
them
in
this
place
or
set
these
options
and
you're
good
like
explain.
Okay,
this
is
this
is
why
this
is
here.
M
Hi
I'm
Carol,
Willing
I'm,
actually
on
project
Jupiter
and
Jupiter
hub
and
we've
been
using
in
production
kubernetes
for
a
little
bit
over
a
year,
and
it's
great,
we
love
it.
It
is
also
a
really
steep
learning
curve
for
even
technical
people
who
are
developers
making
applications
I'm
a
core
Python
developer
and
it
was
painful
to
set
up
instructions
to
a
to
kind
of
go
break
through
the
jargon,
even
with
like
the
books
that
are
out
there.
M
The
podcasts
friends
that
I
can
ask
questions
of-
and
it's
almost
like,
is
there
like
a
master
visual
of
like
the
map
of
kubernetes
or
what
resources,
because
I'm
gonna
work
with
University
people
and
how
do
I
explain
to
high
end
biology,
researchers
or
chemists
or
whatnot
why
they
want
to
use
kubernetes
and
what
the
jargon
means
like
they.
They
know
what
disk
is
a
disk,
but
if
it's
named
X,
Y
Z
or
the
set
the
other,
it's
confusing.
E
On
point,
why
I
think
I
think
one
good
reason
for
that
could
be
that
all
the
new
use
cases
such
as
tensorflow,
for
example,
as
we've
seen
cube
flow
demos,
work
very
well
with
kubernetes,
so
that
is
yeah
right,
so
yeah.
So
exactly
so
because
I
mean
I,
just
can't
see
what
it
would
take
to
to
take
something
like
tensorflow
into
machine
learning,
applications
into
some
kind
of
legacy
infrastructure
without
communities.
C
Once
we
get
past
a
few
of
the
big
things
like
a
lot
of
times
when
they
see
a
service,
they
think
like
a
system
D,
you
know
service
or
you
know
your
classic
service
and
not
you
know
something.
That's
like
port
or
service
discovery,
so
just
sort
of
like
going
through
those
primitives
and
like
I've
I
have
a
couple
slide
decks
and
things
like
that,
but
like
I've
specifically
targeted
for
like
that
sort
of
audience,
maybe.
M
I
misspoke
a
little
because
part
of
it
is
understanding
what
kubernetes
is
doing,
but
the
other
part
that
was
hugely
painful
was
getting
to
the
point
of
helmand
it
on
every
single
different
cloud
provider,
because
everybody
has
different
jargon
of
how
to
set
up.
You
know
just
to
even
get
to
the
point
where
I
can
use.
Kubernetes
was
incredibly
painful
and,
like
you
know,
David
Aaron
tech
saw
our
zero
to
Jupiter
hub
guide,
where
we
get
people
to
the
point
where
they
can
do
helmet
and
he's
like.
M
B
I
mean
that's
the
whole
kind
of
thing
that
we
did
was
micro
services
for
the
last
I
can
know
how
many
years
it's
we
remove
complexity
from
the
software,
but
we
moved
it
into
the
infrastructure
sack
right.
So
someone
needs
to
take
care
of
that
still
and
I.
Don't
think
we
need
to
like.
We
should
bother
too
many
people
is
actually
taking
care
of.
B
Carini's
communities
should
become
at
some
point
a
commodity
either
on
the
cloud
or
by
some
people
that
run
it
for
you
or
have
automation
that
can
run
it
for
you
on
Prem
or
in
different
environments.
We're
not
there
yet.
Definitely
we
still
need
lots
of
work
to
make
it
easier
like.
Even
if
there
is
a
team
that
knows
companies
in
and
out
those
guys
also
need
to
have
a
life
and
and
sleep
right.
They
don't
want
to
be
woken
up
at
3:00
a.m.
B
so
there's,
and
the
problem
is
that
it's
not
done
with
communities.
There
is
so
much
more.
There's
like
you
need
to
take
care
of
your
networking
plugins,
your
dns
that
may
change
change
like
in
two
months.
There's
other
things
that
might
change.
Did
the
whole
cloud
native
stack
on
top
that
will
at
some
point
help
you
deploy
stuff
easier
like
SEO,
someone
needs
to
manage
that,
and
it
shouldn't
be
the
end
user.
It
should
be
dedicated
teams
and
hopefully
dedicated
teams
that
you
don't
need
to
have
in-house
I.
C
Know,
at
least
in
our
world:
it's
usually
the
end,
usually
that's
ping
it
up.
They
don't
have
the
infrastructure
people
to
handle
it.
That's
definitely
where,
like
you
know,
gke
and
a
lot
of
those
services
come
in
and
as
other
things
like
cube,
ATM,
mature
and
installing
imaging
and
stuff
on
prem
becomes
an
easier
experience.
I
think
you
know
that
pain
point
will
will
go
down
a
bit.
D
But
if
you
imagine
like
back
like
early
2000s
when
you
when
you
looked
at
like
this,
is
what
it
took
to
run
a
web
server
or
an
email
server,
and
then
actually
people
bothered
to
run
their
own
email
server
until
they
realize
that
sucked
right
like
we
don't
we
none
of
us
for
the
most
part
deal
with
that
anymore
and
we
focus
on
our
higher
level
domain
that
we
actually
cared
about.
Shipping
I
think
that's
going
to
happen
with
Cooper
days
as
well.
D
So
if
you
like,
wearing
many
hats,
you're
like,
of
course
I
want
all
the
toggles
I'm
gonna
be
my
own
cluster
operator,
but
then
you're
gonna
see
more
and
more
people
who
are
like
I'm,
a
researcher
I
just
want
to
run
my
notebooks
they're,
going
to
either
end
up
with.
You
know,
Caltech
or
something
like
that.
Giving
them
clusters
that
already
have
Jupiter
hub
installed
on
it
or
I
think
a
juror
has
stuff
with
like
notebooks
and
things
like
that
as
well,
where
you
can
just
jump
on
and
start
doing
that.
F
So
I
see
this
right
now
very
much
like
installing
Linux
all
right.
So
so
you
can
totally
you
can
go.
You
can
grab
the
guide
for
Linux
from
scratch.
You
can
grab
the
kernel
sources.
You
can
compile
it
with
GCC.
You
can
install
all
the
infrastructure
around
the
kernel
that
you
need
to
run.
You
can
install
your
gnu
toolchain.
You
can
install
X
from
scratch.
You
can
configure
your
graphics
drivers.
F
Gen
2
I
always
need
to
install
Jan
Timman
like
Gen
2
is
dear
to
my
heart,
but,
like
I,
think
most
people
don't
want
Jen
to
write
like
most
people
want
Fedora,
Linux,
Mint
and
I
think
we
need
to
get
better
a
at
giving
that
to
them
in
the
form
of
like
opinionated
distributions
of
kubernetes
that
are
easy
to
set
up
and
I
think
we
also
need
so
like
I
think
as
a
community,
people
need
to
start
realizing.
It's
ok
to
run.
You
know
fedora
and
Linux
men.
That's
fine!
F
E
Just
wanted
to
to
to
odd
I
think
cluster
API
will
become
very
important
in
this
area.
For
example,
the
cloud
providers
right
right
now
Jiki
has
its
own
CLI.
Are
the
coop
provider
says
that
our
own
CLI,
as
you
mentioned,
they
have
all
the
different
bells
and
you
have
to
make
sure
you
get
the
right
ones.
I
Can
I
just
disagree
a
little
we
purchase
I
think
at
the
moment
the
we're
still
in
that
adoption
phase
that
that
we
need
it
to
be
Kannada
tea,
I
think
pulling
from
the
DevOps
like
background
of
of
getting
people
up
to
speed
in
my
last
company
we
basically
ran
workshops
like
full-day
workshops
of
right.
You've
got
an
app
it's
running
on
an
instance.
Here's
here's!
What
apology
is
here's?
What
a
demon
said
is
here's
what
not
demon
said
because
we
just
get
all
of
that,
but
you
know
here's
what
the
deployment
is.
I
Here's,
what
the
service
is.
Here's,
what
an
ingress
controller
is
this
is
this
is
what
you
need
to
know
and
then
the
rest
of
the
the
day
was
like
here:
let's
get
your
application
into
a
container
and
into
a
helm
chart
and
then
because
you're
on
call,
you
need
to
know
some
of
these
things,
because
you'll
need
keep
CTL
to
be
able
to
to
kill
a
pod.
I
Just
you
know
deal
with
it
and
I
think
there
is
still
that
that
that
ramp
on
at
the
moment
that
yes,
we're
still
in
that
zone,
where
we
do
need
to
explain
those
concerts
and
I,
think
that
that
will
go
away
with
time
and
with
serverless
fit
functions
or
service
all
that
kind
of
stuff.
But
at
the
moment
I
think
we
do
need
that
with
with
the
training.
That's
a,
although
so.
D
Think
I
just
wanna
make
it
clear,
though,
that
like
there's
a
there
when
you
think
about
I'm,
sorry
I
used
to
be
a
p.m..
So
when
you
think
about
personas,
for
example,
if
we're
specifically
talking
about
scientists
at
a
university
who
are
interested
in
actually
just
running
their
simulations,
I
really
disagree
that
they
need
to
be
aware
of
what
a
pot
is
and
in
it
containers
and
anything
like
that
at
all
yeah.
D
N
M
I've
got
a
deploy.
Cooper
net
like
I've,
got
like
deploying
kubernetes,
isn't
the
problem
it's
getting
to
kubernetes
and
then
also
sizing
things
and
that's
being
done
by
grad
students.
It's
not
necessarily
being
done
by
research,
computing
they're,
supporting
from
the
standpoint
of
some
knowledge.
But
if
you
don't
have
money
to
pay
research,
computing
you're,
not
getting
it,
and
you
know,
and
the
thing
with
grad
students
and
this
persona.
Is
they
don't
want
the?
Let
me
just
run
a
script
and
it
being
magic.
They
want
to
know
like
okay.
B
And
I
think
we
like
we
gained
so
much
like
we
gave
so
much
power
into
developers,
hands
with
docker
and
communities
like
with
all
these
manifests
and
resources
that,
like
I,
would
I
would
feel
sad.
Taking
that
away
again,
like
good,
like
right
and
and
I
think
we
just
need
to
do
a
better
work
with
documentation
with
better
tutorials
like
the
documentation.
The
there
is
different
goals,
like
the
ones
upstream,
like
the
the
Carini's
documentation
about
parts.
B
It
needs
to
tell
you
everything
about
parts
all
right
and
then
there's
tutorials
that
show
you
a
bit
but
there's
still
like
some
parts
where
we
can
do
more
and
we
try,
like
all
of
us,
try
to
write
more
blog
posts,
write
more
dogs,
write,
more
stuff
and
I,
see
more
and
more
dogs
contributions.
I
like
I,
think
it's
awesome.
It's
like
it
shouldn't
be
just
lines
of
code.
Lines
of
markdown
should
be
twice
as
twice
the
points
I.
M
You
know
thank
you,
because
part
of
our
problem,
with
keeping
up
with
the
academic
folks
and
the
research
folks
is
because
you
guys
did
a
great
job
with
kubernetes.
It's
like
once
you
get
to
kubernetes.
They're
like
this
is
amazing.
We
can
do
this
ourselves,
so
I,
guess,
if
I'm
leaving
you
with
anything,
it's
like
thank
you
for
what
you
have
done
and
we'll
continue
to
do
and
as
vendors
take
it
back
to
your
companies
that
creating
like
the
the
managed
like
kubernetes
offerings.
Don't
mess
them
up
so
that
they're
different
in
every
circumstance.
J
So,
just
with
respect
to
that,
don't
mess
them
up.
There
are
a
bunch
of
efforts
around
conformance
and
things
like
sanam
boy
to
try
and
keep
everybody
honest
so
that
we
don't
screw
a
good
thing
up
and
I
just
wanna,
I
wanna,
agree.
I
want
to
agree
with
Carolyn
a
little
bit
here.
This
idea
of
creating
experiences
on
top
of
kubernetes
that
are
actually
more
curated,
I,
think
that's
gonna,
be
a
critical
thing.
J
J
They
don't
have
to
wrap
kubernetes.
They
can
actually
it's
not
an
either/or
thing.
You
can
both
use
those
higher-level
platforms
and
then
drop
down
to
the
lower
level
when
you
need
to
so
to
like
extend
the
the
Linux
analogy
like
maybe
kubernetes
is
the
kernel
and,
and
people
have
to
you
know,
learn
how
to
do.
They
want
to
build
it
or
use
the,
but
then
you
might
have
you
know
an
X
11
GUI
window
manager
on
top
of
that
with
different
flavors.
J
O
O
But
what
you
do,
if
you
ever
want
to
do
this,
is
you
start
with
a
small
amount
of
tests
and
you
get
all
the
vendors
on
board
because
they're
all
able
to
pass
yeah
you
reel
them
all
in
you'll
hook
them
in
cuz
once
you
get
them
all
in,
it's
like
the
old
road
tells
roaches
check
in.
They
can't
check
out
okay.
O
So
we
got
everybody
checked
in
with
about
156
conformance
tasks.
We've
hired
through
the
fountain
cloud
native
computing
foundation.
We've
we've
got
a
couple
testers
from
Google,
a
person
who
works
for
me,
who's
testing
and
two
contractors
as
well,
and
we
are
in
a
very
aggressive
schedule
to
get
these
targets
of
40%
covers
and
then
80%
coverage.
So
look
for
a
very
aggressive
coverage
as
we
ramped
up
the
amount
of
conformance
tests
and
improve
that
even
more
but
I
just
went
under
people
understand
why
we're
doing
it.
Oh
thank
you.
E
N
So
definitely
a
switch
of
topic
if
I'm,
spinning
up
a
cluster,
a
lot
of
progress
has
been
made
with
cube
ADM
in
in
being
able
to
just
install
a
master
and
then
then
spin
up
a
bunch
of
nodes
and
add
the
nodes
super
easy.
My
question
is
like
what
about
what
about
for
the
the
rest
of
H
a?
Where
are
you
wanting
to
have
like
a
fully
redundant
cluster?
You
know
we're
not
thinking
about
communities
as
a
service
on
eks
or
whatever.
How
is
that
progressing?
What's
the
roadmap?
How
what?
What?
J
J
If
you
make
a
bunch
of
assumptions
like
something
like
cops,
where
you
have
an
s3
bucket,
that
you
think
is
secure,
that
that
makes
a
lot
of
this
stuff
easier
when
you're
trying
to
make
it
be
that
experience
I've
just
run
a
command
on
each
machine
and
boom.
It's
done.
That's
gonna,
take
a
little
bit
longer
and
so
I
think
it's
going
to
be
a
little
bit
more
manual
for
now,
but
hopefully
over
time.
J
J
We're
gonna
be
working
really
hard
to
get
cube
admin
into
GA,
which
means
that
for
a
single
node
master
or
you
know,
control,
plane,
making
sure
that's
secure
and
that
all
the
features
that
it's
depending
on
things
like
the
TLS
bootstrap,
is,
is
in
GA
and
then
I
think
you
know
starting
to
chip
away
at
making
the
the
the
H
a
experience
better
is
going
to
be
starting.
You
know
now
and
then
continuing
it's
going
to
be
an
evolving
story
over
the
next
couple
of
versions.
But
it's
you
know
eye
on
the
prize.
C
L
So
I've
definitely
an
from
fixed
Pacific
Christian
about
the
release
process
and
and
one
example
years
is
not
sure.
I
think
your
proxy
starting
from
one
point,
seven
point:
you'd,
hey,
come
online,
fax,
are
duplicated,
please
use
this
config
file
and
then
there
was
no
documentation
in
config
file.
So
you
read
the
source
code.
Okay,
this
how
it
works,
get
a
config
file
there
and
I
think
in
1.9,
then
the
config
file
format
changed
so
M
inverse.
L
C
J
The
goal
with
kubernetes
is
to
have
a
deprecation
policy
where
once
something
says
v1
on
it,
it's
gonna
take
like
a
year
to
get
deprecated
the
whole
effort
of
moving
things
into
config
files,
as
component
is
called
component
config
with
the
idea
to
make
it
easier
to
distribute
these
things
and
configure
from
within
the
cluster.
So
you're,
not
in
a
situation
where
you
have
to
reach
under
the
cluster
to
start
configuring,
stuff,
I
empathize,
with
the
lack
of
documentation
on
this
stuff
and
I.
Think
this
really
emphasizes
that
we
do
need
a
cluster
operator
guide.
J
J
Yeah
one
I
mean
one
of
the
one
of
the
things
that
we're
trying
to
do
is
create
more
communication
patterns,
more
ways
for
folks
to
to
to
make
sure
that
they
that
they
do
all
the
things
necessary
beyond
just
writing
the
code
right
we've
had
okay,
there
was
this
example.
The
other
day
somebody's
like
hey
I,
want
to
be
able
to
take
a
specific
value.
I,
don't
know
and
push
it
down.
It
was
like
a
downward
API
and
so
he's
like.
Oh,
we
added
that
two
versions
ago.
J
We
just
forgot
to
document
it
right,
and
so
we
see
more
and
more
of
that
stuff.
So
it's
this
tension
between.
How
do
we
get
features
done
and
make
progress
versus?
How
do
we
make
those
things
accessible?
Usable
documented,
yeah?
Well,
and
it's
like,
and
it's
you
know
and
documentation
is
not
just
more
documentation.
It's
the
right
documentation
in
the
right
place
that
people
can
find
it.
J
It's
a
difficult
thing
and
I
know
the
docs
folks
are
working
super
hard
on
it,
but
there's
a
lot
of
ground
to
cover
and
in
honestly
everybody
has
to
chip
in
to
cover
that
and
we
don't
I,
don't
think
we
have
the
right
incentives
as
a
project.
You
had
to
really
make
sure
that
we
all
problems
like
that
in
a
holistic
way.
A
F
We
we
do
actually,
we
have
I
think
most
of
our
deprecation
policy
written
up.
It
is
actually
in
the
documentation.
I
can
never
remember
exactly
where
it
is,
but
I
can
usually
find
it,
which
means
I,
think
we
need
to
move
the
link
but
yeah.
So,
like
I,
think
we
have
a.
We
have
a
slightly
different
deprecation
policy
for
api's
versus
command
line
flags.
The
API
one
I
know
off
the
top
of
my
head:
alpha
can
change
release
to
release
beta
needs,
deprecation
for
a
release
and
then
can
be
removed
and
anything.
F
A
Have
a
quick
one
before
we
go
I
guess
no
time!
Thank
you.
If
you
like
this
format,
it
actually
has
a
sister
program
run
by
Parrish
Pittman.
We
just
called
meet
our
contributors,
which
is
basically
this
and
Carolyn
sits
on
both
where
you
could
say
same
format,
different
questions.
How
do
I
get
my
first
patch
in
I?
A
Did
a
PR
and
a
bot
yelled
at
me,
like,
although
ton
of
questions
how
things
are
organized,
what
labels
we
use.
So
this
is
kind
of
focused
on
on
user
using
kubernetes
if
you're
interested
in
moving
up
to
the
next
level,
just
know
that
we
also
have
programs
and
mentorship
programs
in
place.
If
you
need
a
one-on-one
with
someone
for
an
hour
to
figure
something
out,
we
have
all
sorts
of
stuff
and
if
you
just
type
kubernetes
mentorship
into
Google,
that
will
take
you
to
the
right
place.