►
From YouTube: Kubernetes Community Meeting 20180426
Description
See this page for more information! https://github.com/kubernetes/community/blob/master/events/community-meeting.md
A
All
right,
everyone,
it's
one
o'clock
and
I'm
kicking
up
the
stream.
Please
remember
this
community
meeting
is
public,
so
everything
that
you
say
will
be
streamed
and
recorded,
live
on
YouTube.
So
please
be
cognizant
of
what
you're
saying
today.
We're
gonna
have
a
demo
of
get
cube.
Some
release
updates
graph
of
the
week,
some
sig
updates
from
cluster
life
cycle,
auto
scaling
and
sig
p.m.
and
then
announcements.
It
is
one
week
to
cue
cards.
So,
let's
get
to
work
Shaheed,
you
want
to
start.
B
B
We
are
working
on
a
platform
on
kubernetes
for
application
developers.
So
get
cube
came
out
of
our
work
at
essaouira,
so
we
are
slowly
open
sourcing,
more
and
more
parts
of
Asuna,
so
I'll
jump
right
into
the
demo,
plus
x-play
I'll
jump
right
into
the
demo,
so
I'm
going
to
share
my
screen.
I
hope
it's
visible.
B
Okay,
so
get
cube,
is
a
tool
which
will
let
you
build
and
deploy
docker
images
onto
a
kubernetes
cluster
using
just
get
push
the
only
dependency
on
the
client
side
is
get
and
everything
else
is
done
on
the
cluster
itself.
So
I'll
start
with
the
demo
so
that
I'll
explain
the
architecture
alongside
the
demo.
So
there
we
have
a
repo
called
get
cube
example,
and
in
that
we
have
a
couple
of
steps
listed
out
to
go
through
a
complete
workflow.
B
So
the
first
step
is
to
create
certain
with
kubernetes
resources
on
any
kubernetes
cluster.
So
we
have
we
using
a
custom
resource
definition.
We
are
calling
it
remotes
and
it's
names
based,
so
you
can
have
our
back
based
authorizations
for
all
of
these
objects
and
there
is
another
deployment
called
get
cube
D.
This
is
a
doctor
in
docker
image
which
builds
the
docker
images.
B
There
is
another
deployment
which
is
the
get
cube
controller.
This
is
the
controller
for
the
CR
D,
which
operator
for
C
or
D,
which
watches
the
objects
and
act
accordingly.
So
I
already
have
a
kubernetes
cluster
created
on
GK.
Here,
I
have
set
up
my
cube
CTL
to
connect
to
this
cluster.
It's
a
single
node
cluster,
as
of
now
I
have
not
live
load.
Anything
on
this,
so
the
first
step
is
to
create
this
get
cube,
CRD
controller
and
other
objects.
It's
just
a
cube.
Ctl
create
command,
so
I
will
create
this
resources.
B
We
can
see
that
a
couple
of
resources
are
getting
created.
The
CRD
service
account
cluster,
all
pointing
to
deployments
and
the
config
map
the
config
map
stores
certain
configuration
required
by
the
cube
d
deployment.
The
next
step
is
to
expose
these
get
cube
d
deployment.
This
is
so
that
you
get
a
remote
I
remote
on
the
cluster
with
which
you
can
push
get
push
to
the
cluster.
So
we
have
a
example
poor,
which
is
destined
nginx
nginx
deployment.
B
So
the
first
step
is
to
create
the
nginx
deployment
and
service
objects,
so
I'm
going
to
create
it
now
so
parallely.
What
I'm
gonna
do
is
to
run
a
cube
CTL
proxy,
so
they
tell
you
how
to
provision
one
more
load,
balancer
and
all
and
I
am
going
to
look
at
what's
deployed
right
now:
oops,
ok!
So
the
this
is
just
a
plain
engine.
X
image
on
kubernetes
cluster.
B
So
the
next
is
to
create
the
remote
CRD
so
that
we
can
start
gate
pushing
to
the
to
this
to
update
this
deployment.
So
for
that
the
authentication
is
based
on
our
public
keys.
So
I'm
going
to
add
my
public
key
to
the
CRD
definition,
so
this
yard
is
called
remote
dot.
Eml.
If
you
look
at
this
it's
it
says,
the
deployment
to
update
is
called
nginx.
The
container
is
called
nginx.
The
path
is
the
current
directory.
B
The
path
at
Whitaker
build
has
to
happen,
and
this
is
the
docker
file
to
use-
and
this
is
my
SSH
public
key,
which
is
used
to
authorize
me.
So
I
will
create
this
separate
message.
Public
key
already
I
will
create
the
remote
object
now,
so
the
remote
object
is
created
now
I
will
look
at
the
status
for
this
remote
object
so
get
remote.
This
is
gonna.
Give
me
a
remote
URL.
This
is
circuit
remote
e
water,
so
I'm
gonna
set
this.
B
B
So
what
we
have
right
now
is
just
an
engine
X
plane
basic
image,
so
I'm
going
to
push
this
code,
which
is
nothing
but
some
HTML.
So
as
we
can
see,
the
gate
push
is
triggering
a
pre
receive
hook,
which
is
building
out
the
docker
image
inside
the
cluster.
It
is
as
well
as
updating
the
kubernetes
deployment
using
cube,
CTL
cube
control
set
image.
B
It
is
also
showing
us
their
all-out
status,
so
it
says
build
has
been
completed
so
when
I
go
and
refresh
now
we
have
the
new
files
up
and
running
on
the
cluster,
so
any
change
I
make
I,
can
look
at
the
HTML
and
they
change
I.
Make
here
will
be
and
and
whenever
I
commit
and
push
this
it
will
be
live
on
the
cluster
again.
B
So
I'm
going
to
push
again
so
every
time
I
push
a
new
docker
image
is
built
and
rolled
out.
The
image
is
tagged
with
commit
hash
and
I
can
see.
The
changes
are
live
again.
So
hello,
coop
net
is
community,
is
life.
So
that's
what
get
cubed
us?
We
are
working
on
more
features
like
customizable
hooks,
all
sorts
of
get
hooks
and
other
other
sub
cooks
like
web,
hooks
from
which
can
be
triggered
from
github,
etc.
So
going
through
the
architecture.
This
is
what
it
looks
like.
B
There
is
a
remote
CRD
which
is
managed
by
this
controller.
There
is
a
gate,
cube
d
deployment
which
builds
out
the
docker
images
and
updates
the
deployment.
The
gate
pushes
happened
to
this
deployment.
This
is
the
git
remote
listening
on
422
and
every
of
every
every
time,
a
user
gate
pushes
the
cube,
dibbles
a
docker
image
and
updates
the
deployment.
B
A
B
Okay,
so
shy:
do
you
want
to
tell
them
about
our
target
sequence?
Yeah,
so
we'll
be
a
tube
corn
feel
free
to
catch
us
during
queue.
We
are
also
speaking
about
get
push,
workflows
or
communities
generally
about
key
tops
and
how
we
are
using
get
upset,
consider
to
make
application
development
with
kubernetes
easier,
so
yeah
feel
free
to
catches
at
coop,
corn,
pyramid
and
I
will
be
there,
along
with
in
my
and
awesome.
A
C
Hey
there
so
111
release
features
are
now
frozen,
as
in
we
have
our
complete
list
of
features.
If
you
actually
look
at
the
link
there,
you
will
find
a
sorry.
I
can't
actually
share
my
screen
via
zoom
on
this
computer,
but
I
supplied
links
there.
So
you
can
see,
we've
got
a
list
of
44
file
features
in
the
features
repo.
C
Unfortunately,
our
CI
signal
is
not
looking
so
good
if
you
actually
click
through
to
test
status
right
now
about
a
fifth
of
the
jobs
for
general
testing,
the
ones
that
we
consider
master
blocking
are
failing
more
seriously,
pretty
much
all
upgrade
tests
are
failing.
This
is
actually
one
job,
that's
passing,
but
everything
else
is
failing
all
right
and
that's
actually
kind
of
important,
because
our
next
deadline
is
to
do
our
first
beta
release
on
May
15th
and
if
we
don't
have
a
lot
of
improvement
in
passing
tests
etc.
C
C
Issues
been
filed
against
all
the
test
failures,
some
cigs
are
being
much
more
responsive
than
others
in
terms
of
working
on.
Those
I
need
to
remind
everybody
on
this
call
that
if
your
sig
is
responsible
for
a
test,
you
are
responsible
for
debugging.
Why
that
test
is
not
passing.
So
please
respond
to
those
issues.
I
would
really
like
to
see.
E
A
F
Everybody,
okay:
George!
Can
you
hear
me
so
I'm,
great
okay,
cool,
all
right,
we're
going
to
share
my
screen
here
and
so
on
a
quarterly
basis.
Contributor
experience
is
going
to
start
releasing
some
stats
George
last
week,
for
instance,
that
YouTube
this
week
will
do
flat.
So
you
can
see
what
a
Goliath
our
slack
in
senses.
Right
now
we
have
35,000
users
of
slack
with
168
public
channels
and
our
weekly
active
users
keep
creeping
as
well
we're
over
5,000
and
just
to
give
you
some
comparison
at
the
bottom
of
factors.
F
Active
users
chart
I,
put
the
mailing
list
numbers
so
for
kubernetes
development
we
have
about
3000
and
for
super
Nettie's
users.
We
have
about
three
three
point:
three
thousand
and
just
in
the
kubernetes
users
group,
where
we
have
now
roughly
34,000
people.
So
we
can
reach
a
large
audience
this
way,
but
there
are
some
other
graphs.
Of
course,
I
included
the
rooms
with
the
most
emojis,
but
that's
on
there
as
well
and,
like
I,
said
we'll
start
producing
this
on
a
quarterly
basis
and
we'll
find
a
home
for
it
in
the
community
github
repo.
F
G
So
hopefully,
folks
gonna
see
my
screen
real
quick
I
just
want
to
give
a
quick
brief
shout
out
for
some
information
here.
So
for
those
who
are
you
of
you
who
are
not
aware
lucas's
on
conscription,
so
he's
gonna
be
out
for
a
while
and
I've
been
taking
over
a
lot
of
the
functioning
for
sequester
like
cycle.
Also,
why
Robert
is
out
too
as
well,
so
one
of
the
things
that
I
found
luckily
from
Erin
is
the
auto
labeling
feature
that
now
existed
for
owner
files.
That
is
highly
highly
useful.
G
It
has
allowed
us
to
have
a
consistent
burn
down
against
our
backlog,
our
backlog.
Currently
it
has
about
59,
more
items
to
burn
down
against
111
and
some
of
the
other
bugs
some
of
them
are
just
minor
things
or
feature
additions,
but
we'll
we'll
deal
with
that
as
we
reach
closer
to
release
date
and
the
auto
labeling
is
highly
helpful
for
actually
getting
accurate
labeling
for
determining
what
PRS
are
inbound
for
the
cig.
So
we
can
actually
review
them
in
a
timely
order.
G
So
I
just
want
to
say
thank
you
to
the
testing
for
folks
for
helping
us
get
that
in
place
and
if,
for
other
folks,
who
don't
have
in
place
either
have
been
doing
it
now
for
the
status
update.
We're
like
I
mentioned
before
we're
we're
burning
down
against
111.
The
cig
is
currently
working
on
some
major
feature
items,
one
of
which
is
the
master
joint
workflow
scenario.
G
It
would
be
nice
to
have
a
consistency
for
when
we
do
a
release
to
make
sure
that
all
the
tests
are
passing
before
we
cut
the
release
or
have
a
bar
that
we
agree
upon
outside
of
that
particular
sub
project
or
coop
idiom.
The
second
sub
project
I'll,
give
a
quick
update
on
is
the
the
cluster
API
courtesy
of
chris
nova.
There
is
a
new
repo
that
is
now
available
for
the
cluster
api
underneath
the
new
organizational
structure.
E
Alright,
so
everybody
can
hear
me
so
for
sig,
auto
scaling
for
this
next
release.
We
have
kind
of
two
major
focus
areas,
so
the
first
is
some
additional
improvements
to
the
HPA
v2
beta
api's.
So
one
of
the
things
that
users
have
been
asking
us
for
a
while
is
kind
of
a
way
to
control
the
specificity
of
their
metrics
better.
So
a
lot
of
metric
systems
that
people
are
using
the
custom
metrics
functionality
of
BH
being
with
have
the
ability
to
focus
queries
down
to
additional
labels.
E
The
different
metrics
sources
and
then
the
other
major
source
of
work
that
we're
focusing
on
is
we're
hoping
to
get
out
a
alpha
for
the
vertical
pod
autoscaler
soon,
hopefully,
by
the
end
of
this
month
or
next
month.
We're
going
to
be
doing
a
demo
of
that
actually
at
coupon
during
our
deep
dive,
I
believe
so,
if
you're
interested
in
the
vertical.
E
To
be
a
coupon
I'd
recommend
you
swing
by,
but
those
are
and
then
finally,
we
we
are
looking
to
kind
of
get
the
end-to-end
tests
for
the
HPA
custom.
Metrics
functionalities
a
little
more
robust
to
the
point
where
we
can
move
them
over
to
walking
so
that
we
don't
accidentally
break
anything
and
that's
about
it.
We're
cute
contour
to
got
us
going
for
this
release.
H
For
letting
us
present,
so
it's
8
p.m.
is
in
the
middle
of
re
sort
of
Rio
straightening.
What
we
do
and
essentially
we're
trying
to
find
the
things
that
we
can
apply
the
most
effort
to
with
a
fairly
limited
set
of
people
who
are
contributing
right
now.
The
big
focus
that
we've
decided
on
for
the
moment
is
to
try
and
work
on
getting
a
good
feedback
mechanism
for
user
suggestions
and
input,
so
I'm
playing
around
with
this.
H
This
idea
of
a
form
that
could
just
be
widely
distributed
and
shared
to
present
feedback
I'm,
also
looking
at
some
ways
that
we
can
implement
some
interesting,
automated
analytics,
maybe
using
some
machine
learning
on
the
back
into
that.
So
we
can
create
more
unbiased
user
reports
and
sentiment
reports
around
the
product.
So
we'll
see
how
that
goes.
I'm
quite
excited
about
that.
We
are
going
to
be
presenting
a
coupon
on
Thursday
at
4:35
p.m.
that'll,
be
a
horn
Amarna,
so
that'll
be
a
really
good
session
to
sort
of
learn
more
at
what
we're
doing.
H
And
lastly,
we
are
working
on
a
charter
draft
that
codifies
a
lot
of
things
that
I'm
talking
about
right.
Now,
we're
looking
at
the
fact
that
it's
8
p.m.
essentially
the
the
tea
and
PM
stands
for
more
than
just
product.
It's
really
product
project
and
program,
and
so
each
of
those
things
will
essentially
work
out
into
some
projects
and
ID.
H
The
projects
is
that
we
really
want
to
help
SIG's
get
through
some
of
the
minutiae
of
planning
and
I
have
a
successful
idea
of
what
a
roadmap
might
look
at
because
oftentimes
when
we're
really
focused
on
the
details
and
we
get
a
little
myopic
about
the
long
term
arc
of
features
and
implementations.
And
what
we'd
like
to
do
is
see,
if
maybe
a
longer
horizon
view
of
what
is
possible
within
the
committee's
ecosystem
and
within
Corinne
any
score
so
roadmaps
and
sharing
ideas
about
long
term
vision
and
having
some
agreement
about.
H
That
is
actually
a
great
way
to
do
it.
So,
hopefully,
sig
p.m.
I
can
help
facilitate
those
and
again
sig
p.m.
is
not
trying
to
dictate
or
to
be
an
authoritative
voice.
On
this,
we're
really
trying
to
just
use
all
of
our
expertise,
managing
product
project
and
program,
to
help
things
be
really
successful
in
those
environments,
so
they
can
focus
on
coding
and
implementation
and
not
necessarily
on
all
the
minutiae
in
history.
Yet
so
same
team
really
wants
to
be
an
enabler
and
that's
that's
not
it
any
questions.
A
All
right,
okay,
moving
on
to
announcements,
just
a
reminder,
cube
Khan
is
next
week
in
Copenhagen.
There
is
no
community
meeting
so
have
I
have
a
good
time.
You've
earned
it.
It's
also
the
last
chance
to
register
for
the
contributor
summit,
which
starts
the
day
before
cube
con.
Remember
you
do
need
to
register.
Even
if
you
register
for
Q
con,
you
need
to
register
separately.
Otherwise,
you're,
not
gonna
get
food
or
the
cool,
awesome
and
unique
kubernetes
contributor
t-shirt
so
do
make
sure
to
register
we're
gonna
cut
off
registrations
tomorrow
at
7
p.m.
A
and
then
Paris
and
I
on
our
plane
and
by
then
it's
too
late.
So
please
register
if
you're
planning
on
attending
cigs
and
leads
and
tech
leads,
I
remember
to
put
yourself
down
for
a
cig
update
sheet
as
part
of
the
contribs
Amit.
That's
the
quick
five-minute
things.
You
need
to
know
that
you
want
to
communicate
to
other
sins
across
the
project,
so
we
can
kind
of
foster
a
cross-leg
communication.
A
A
So
if
you
see
someone
at
any
time
doing
something
really
great
above
and
beyond
the
call
of
duty,
you
can
mention
them
and
hash
shoutouts
on
slack
and
we'll
read
them
out
as
a
community
meeting
so
Timothy
say,
Claire
would
like
to
thank
Peter
Tsou
for
steadfast
peer
review
and
contributions
to
say,
cluster
lifecycle
and
Chuck.
I
would
like
to
thank
Lika
pillow
at
stuffing
box
for
being
welcoming
to
new
folks
as
a
closer
lifecycle.
You
are
welcoming
and
helpful
and
it
keeps
our
community
healthy.
A
C
Pretty
simple:
it's
just
a
tracking
if
you
were
moving
some
labels
around
and
I
just
important
that
we
get
that
out
to
all
the
developers
in
case
we
might
be
breaking
some
piece
of
automation
or
process
that
we're
not
aware
of
so.
If
you
care
about
that,
please
look
at
and
comment
on
the
tracking
issue.
C
A
Muted
Thank
You
Cloud
Foundry
wants
to
welcome
folks
to
their
summit
on
May
1st
the
day
before
cube
con
cloud
native
con.
So
I
guess.
If
you're
not
going
to
the
contributor
summit,
you
can
attend
there.
I
don't
see
a
link
so
whoever's
typing
in
there.
If
we
can
get
a
link
for
folks,
so
we
can
plan
it
sweet
awesome
and
the
last
bit
before
we
go
someone's
using
kubernetes
to
water.
Their
lawn
I
thought
it
was
the
most
interesting
think
I've
seen
in
a
while.