►
From YouTube: Kubernetes Community Meeting 20180419
Description
See this page for more information! https://github.com/kubernetes/community/blob/master/events/community-meeting.md
A
Shoutouts
help-wanted
anything
into
the
section,
go
ahead
and
add
that
now
all
right,
let's
kick
this
away,
hi
everyone,
my
name
is
Paris
I
am
a
co-chair
of,
say,
contributor.
Experience
today
is
April
19th.
This
is
the
Thursday
edition
of
kubernetes
community
meeting.
We
do
have
a
full
schedule
today
of
demo
release
updates.
We
have
an
interesting
graph
of
the
week,
which
is
our
YouTube
channel
stats
from
George.
We
have
these
zig
CLI
update
an
AWS
update
and
a
GCP
update,
and
then
we
also
have
our
analysis
section.
A
B
So,
hey
guys,
I,
just
posted
the
link
to
the
repo
in
the
chat.
You
guys
could
take
a
look
at
it
while
I'm
talking,
but
I
am
that
I
am
a
software
engineer
at
Google
on
the
container
tools,
team
and
I
will
be
demoing
some
scaffold
today.
First
I'll
talk
a
little
bit
about
it.
Can
everyone
see
my
screen
alright,
this?
B
Scaffold
dev
is,
does
it
build
and
deploy,
but
also
watches
the
dependencies
of
your
docker
file
or
whatever
kind
of
artifact
you're
building
scaffold
actually
supports
building
things
and
in
a
few
different
ways,
one
of
the
kind
of
main
ideas
behind
scaffold
is
plug
ability.
So
that
means
that
it's
pluggable
in
the
sense
of
builders
and
deployers,
so
an
example
of
pluggable
builders
would
be
building
your
docker
images
with
with
basil
instead
of
docker
and
an
example
of
a
kind
of
a
pluggable
deployer
is
deploying
with
home
instead
of
coupe
control.
B
So
let
me
just
dive
into
the
getting
started
demo,
so
you
start
off
with
yeah
I'm,
just
in
the
scaffold
repo
in
the
getting
certain
example.
I
have
a
very
basic
go
program
that
just
prints
hello
world.
Every
second
I'm
deploying
it
just
in
a
very
simple
pod:
there's
no
restriction
on
resources.
You
can
deploy
it
scaffold
and
then
here
just
a
very
basic,
multi-stage
doctor-doctor
file
here
is
the
scaffold
demo.
B
B
You
can
use
with
scaffold
so
I'm
just
gonna
go
ahead
and
change
this
that'll
trigger
another
read
though,
and
redeploy,
and
you
could
see
that
it
had
to
exclamation
points
here
and
now
we're
streaming
logs
from
the
redeployed
getting
started
container
and
that's
pretty
quick
because
we're
doing
this
all
locally
if
I
wanted
to
I
could
do
something
like
so
this
is
the
GK
cluster
now
so,
as
part
of
this
flow
it'll
actually
push
the
image
instead
of
just
load
into
the
docker
demon.
So
I
could
do
something
like
this.
B
This
will
take
slightly
longer.
I
mean
this
is
a
very
minimal
image.
So
you
know
it's
it's
not
that
crazy,
but
you
can
see
that
it's
it's
it's
deployed
now
and
if
I
make
it
change,
basically
the
same
thing
will
happen,
but
remotely
so
that's
nice
for
for
switching
between.
We
want
to
support
workflows
that
are
both
local
and
kind
of
remote
with
the
kind
of
minimal,
add
I'm,
not
gonna.
Let
this
complete
I'll
just
move
on
to
another
quick
example:
I'll
go
back
to
mini
keep
here.
B
B
We
have
a
basil
build
here,
so
it's
it's
using
the
rule
of
docker
repository
and
rules
to
to
build
a
go
image
here.
So
this
is
basically
gonna,
be
a
very
similar
example,
except
there's
no
docker
file
and
in
this
folder
and
it
with
basil.
So
if
I
go
you
as
scaffold
of
this
will
also
be
pretty
quick
because
it's
already
been
built
and
cache,
but
you
can
see
that
it's
a
very
similar
flow
here.
B
B
Another
kind
of
nice
thing
about
scaffold
is
here:
I
have
just
the
microservices
example
here
we
actually
have
a
scaffold
EML
with
two
images
and
two
deployments.
So
the
idea
is
that
a
lot
of
times
your
development
environment
is
not
just
one-to-one
with.
You
know,
image
two
most
likely
you're
deploying
multiple
projects,
you're
developing
on
multiple
images
at
once.
That
all
depend
on
each
other.
So
the
relationship
here
is
that
I'll
just
pop
into
these.
These
are
these-
are
also
very
basic,
apps
Leeroy
web.
B
B
So
this
one's
a
little
different,
because
I
haven't
actually
exposed
anything
I
see
so
if
I
hit
the
Leroy
web,
endpoint
I'll
get
Leroy
up
here.
So
if
I
change
is
Leroy
appcode,
it
should
trigger
the
rebuild
here,
and
this
will
rebuild
the
worker
and
then
I
should
be
able
to
fit
this
endpoint
and
get
the
kind
of
new
might
take
a
little
longer
to
deploy.
B
My
laptop
is
heating
up
with
all
these
services
mini
Cuban
and
zoom
so
possible,
but
the
okay
there
we
go
I
mean
so
the
idea
is
that
you
can
use
this
to
develop
with
kind
of
multiple
services
and
multiple
apps
and
there's
still
a
lot
of
work.
We
need
to
do
with
that,
but
the
the
kind
of
the
framework
is
there.
B
I've
been
working
on
that,
but
the
idea
is
that
we
want
to
support
tools
like
that.
We
want
to
support
tools
like
image
and
build
up
and
because
I
think
in
the
future.
Your
your
development
environment
is
going
to
look
a
little
different
and
you're
going
to
want
to
use
these
tools
that
have
the
faster
dev
loop
on
them.
Just
a
simple
docker
build
yeah,
that's
pretty
much.
All
I
got
I
can
take
questions
or
for
if
we
don't
have
any
time
for.
C
Hey
I
have
a
question:
what's
your
what's
your
plan
around
integration
with
the
main
kubernetes
community
ci,
but
or
maybe
what's
the
state
of
it
at
the
moment,
so
I
asked
Cooper
entities.
New
kubernetes
versions
are
coming
out.
How's
your
how's,
your
kind
of
qualification
testing
working
for
that
yeah.
B
That's
a
great
question
right
now:
we're
pinned
to
110,
you
know
we
we
just
kind
of
you
know
the
project
released
a
few
weeks
into
March,
I'm
says
fairly
new,
so
we
we
have
integration,
test
coverage
for
all
of
our
current
examples,
but
we
don't
have
any
kind
of
version,
skew
testing
yet
but
I
think
going
forward.
We
probably
like
to
stick
with
you
know
the
kind
of
kubernetes.
You
know
two
versions
ahead
or
two
versions
behind
promise:
I
think
that
would
be
a
reasonable
thing
for
schedule.
B
C
B
I
mean
that's
also
a
good
question:
I
mean
you
know
it's
we're
trying
to
keep
it
as
unup
enya
nated
as
possible.
For
now,
there's
really
no
ties
to
I
mean
we
do
have
Google
Cloud,
though
their
support,
but
we're
not
designing
it.
You
know
essentially
with
you
know
it
being
tied
to
GCP
or
a
certain
cloud,
or
anything
like
that.
So
you
know
maybe
in
the
future.
If
people
feel
like
it
would
be
better
as
a
community
project
I
think
we
would
be
open
to
that.
But
there's
no
plans
right
now.
Yeah.
D
Bob,
this
is
Matt
Farina.
Real,
quick
I
can
jump
on
this
a
little
bit.
There's
was
a
number
of
discussions
on
this
in
places
like
the
mailing
list
and
other
places
and
developer
level.
Projects
like
this
in
the
ecosystem,
where
people
are
expecting
there
to
be
an
explosion
of
tools
and
competitors,
just
like
there
are
competitors
to
scaffold
today,
is
to
live
in
the
ecosystem
and
kubernetes
doesn't
want
to
take
on
the
management
of
those
as
a
project
or
to
even
be
opinionated
on
saying,
which
ones
to
choose.
D
That's
kind
of
one
of
the
things
that
the
steering
committee
has
communicated
as
a
goal.
Here
you
know
kind
of
let
these
ecosystem
things
be
in
the
ecosystem.
One
of
the
nice
things
is
because
they
all
have
to
talk
to
the
kubernetes
api,
it's
api
versioning,
and
so
that
means
it's
not
as
painful
as
being
directly
linked
because
it
talks
to
the
public
api.
The
way
other
tools
do.
A
B
Yeah
so
right
now
it
only
supports
Basel.
We
are
working
on
support,
which
is
is
kind
of
similar,
but
for
for
Java
I,
don't
know
if
you
guys
have
seen
that,
but
I'll
post
a
link
to
that
in
the
chat
we
we
kind
of
have
the
infrastructure
now
on
to
support.
You
know
arbitrary
formats
to
some
degree,
the
only
requirements
for
adding
it
are
is
that
you
need
a
way
to
query
whatever
file
for
its
dependencies.
B
So
for
basel
we
can
do
something
like
a
basil
query
on
the
build
file
to
to
understand
what
the
source
code
dependencies
are
and
for
a
docker
file.
We
actually
do
some
parsing
of
the
docker
file
for
ads
and
copies
to
figure
out
what
your
real
source
code
dependencies
are,
but
other
than
that
implementation
of
you
know.
Other
builders
should
be
pretty
easy
and
we're
accepting
those
entry
right
now,
but
if
it
gets
to
a
point
where
we
have
so
many,
we
might
move
to
something
like
a
plug-in
model
or
community
model.
Thank.
D
A
E
Yep,
so
we
have
scroll
down
so
first
of
all,
most
important
thing
feature:
freeze.
Is
this
upcoming
Tuesday?
So
please
file
issues
for
your
expected
features
in
the
futures.
Repo
make
sure
that
any
features
your
state
is
working
on
and
thinks
you're
going
to
complete
in
the
1:11
time
frame
are
represented
there.
E
The
second
of
all,
currently
our
test
status
for
111
for
actually
for
master,
is
not
looking
really
good.
The
Asia
actually
put
out
the
CI
signal
report,
which
is
linked
from
the
notes
here.
Also
on
kubernetes
dev
will
be
issuing
those
every
week
and
then
more
frequently
as
we
get
towards
code
freeze
were
basically
about
a
third
of
the
blocker.
Jobs
are
failing
and
about
two
thirds
of
the
upgrade
actually
more
than
that
like
like
90
percent
of
the
upgrade
jobs
are
failing
so
I.
E
The
CI
signal
has
opened
the
issues
for
all
of
the
failing
tests,
which
various
people
said
should
have
been
CC
done.
Please
take
a
look
at
those
I
will
remind
everybody,
her
last
week's
discussion,
our
plan
to
shorten
code
freeze
and
give
everybody
more
time
to
develop
on
kubernetes
is
really
dependent
on
getting
clean
signal
and
for
that
matter,
ultimately
releasing
on
time
is
dependent
on
getting
clean
signal.
So
if
we
can
fix
those
test
failures
now
and
things
will
be
worlds
easier
as
we
get
towards
the
release.
E
Also
later
today,
the
release
notes
document
the
empty
release
notes
document,
which
we
call
a
draft
I
will
be
up,
watch
kubernetes
dev.
Obviously,
everybody
drops
their
release
notes
in
to
their
issues.
Don't
let
that
prevent
you
from
adding
appropriate
stuff
to
the
release,
notes
doc
by
PR,
and
that
is
it
for
111.
For
this
week,.
F
F
Things
like
that
and
what
we
do
put
on
our
YouTube
channel
is
meetings
like
this,
and
we
also
have
an
archive
from
other
sigelei
to
record
their
meetings
and
we
put
them
on
the
channel
as
well.
We
have
about
7000
subscribers
and
it
continues
to
grow.
So
we
have
a
full
stack
of
analytics
for
the
channel,
so
I
wanted
to
just
share
a
few.
That
I
feel
are
interesting
and
enable
us
to
use
YouTube
a
little
bit
better.
So
here
the
list
of
our
more
popular
videos,
I
think
one
thing.
F
One
thing:
you'll
notice
as
I've
gone
through
this,
the
more
popular
videos
are
the
ones
that
either
have
been
around
a
really
long
time
or
a
very
descriptive
or
generic.
This
is
what
kubernetes
is,
or
the
ones
that
expect
high
user
engagement
during
a
stream,
let's
say
in
office
hours
or
meter
contributors
itself.
Unfortunately,
most
of
the
decisions
and
things
like
that,
but
you
expect
don't
really
have
like
a
long
tail
of
a
lot
of
views
right.
F
So
we
have
some
interesting
stats
here.
Most
of
these
are
generally
trending
up,
I'm,
not
really
going
to
show
you
some
specifics,
but
the
ones
that
I
really
watch
for
are
watch
time
in
minutes.
This
is
basically
how
sticky
a
video
is
of
someone
you
know
goes
to,
and
they
want
to
see
some
about
kubernetes
and
they're
only
around
for
30
seconds.
They
probably
decided
not
to
bother.
F
F
One
thing
I
did
notice,
so
I
don't
know
if
this
is
if
this
is
a
result
of
how
they
gather
the
information,
but
apparently
we
have
zero
point:
zero
females
watching
our
videos
ever
so
I,
don't
I,
don't
know
what's
up
with
this,
if
anyone
has
any
insight
on
why
that
is
or
if
guessing
demographics
and
browsers
is
still
like
a
black
art
or
something
like
that,
what
did
I
want
to
show
with
this
one?
Oh,
this
is
where
people
are
coming
from.
F
So
most
the
majority
of
these
come
from
YouTube
itself
and
YouTube
search
so
leaves
or
tech
leads.
If
you're
putting
your
videos
on
YouTube,
it
behooves
you
to
put
descriptive
things
into
the
description
that
help
it
get
found
in
YouTube.
So
in
our
channels,
I
almost
always
put
a
kubernetes
kubernetes
and
a
cloud
native
tag
on
there.
So
I
encourage
you
to
put
the
links
to
your
notes.
You
know
maybe
a
little
description
of
what
you
talked
about
or
whatever
something.
F
If
it's
something
specific
that
you
know,
people
might
be
interested
in
feel
free
to
put
that
in
description.
That
will
help
people
find
the
video
easier
I
thought
devices
with
was
interesting.
I
was
expecting
mobile
to
crush
everything
else,
but
I
guess
we're
all
just
a
bunch
of
nerds
that
sit
on
our
computers
and
actually
watch
kubernetes
videos.
I
also
found
thought.
It
was
interesting
that
people
watch
this
on
TVs
and
TV
devices
more
than
they
do
on
tablets.
F
So
that's
all
I
watch,
that's
usually
how
I
watch
the
videos
like
during
a
lunch
or
something
I'll
plop
on
a
Syrian,
Committee
video
or
something
a
pro
tip.
When
you
are
catching
up
on
the
videos,
you
can
watch
videos
on
YouTube
at
1.5
speed
which
really
helps
you
catch
up.
That's
a
pro
tip
from
joe
and
the
last
bit
is.
We
do
do
live
streams
and
normally
we
live
stream.
This
meeting,
but
I
had
technical
issues.
F
So
what
you
see
here
is
when
we
do
something
like
an
office
hours
or
we're
expecting
the
users
to
interact
with
us.
We
do
get
about
about
40
concurrent
people
at
one
point,
the
the
difficult
things
with
doing
things
like
live
streams
and
meet
our
contributors.
Is
they
start
off
pretty
slow
because
it
takes
time
for
people
you
know
to
eventually
check
their
phone
and
I
think
they're
subscribed
to
gives
them
hey.
F
You
know
the
thing
that
you
care
about
is
streaming
and
then
they
have
to
find
a
place
to
sit
down
and
stuff
so
I'm,
starting
to
think
about
where
there
might
be
worth
extending
these
out
a
little
bit
more
to
try
to
kind
of
grab
more
of
the
audience.
So
with
that.
Unless
anybody
has
any
questions,
that's
pretty
much
it
and
I'm
planning
on
sharing
these
stats
people
at
least
once
a
year
to
kind
of
show
you
how
things
are
going
so.
A
G
A
G
So
from
the
6e
live
front,
there
are
two
significant
updates
that
I
would
like
to
discuss.
One
is
something
that
will
be
a
user
facing,
which
is
the
server
side
print.
So
over
the
past
couple
of
releases,
along
with
the
SiC
API
machinery,
will
work
on
unifying
this
and
moving
the
printing
of
objects
to
the
server
so
that
any
consumer
of
the
kubernetes
api
has
the
exact
same
output
when
printing
objects.
So
this
feature
is
currently
in
beta
and
in
110
we
enabled
users
to
opt
in
for
this
feature
and
keep
CDL.
G
Currently,
if
you
do
coop
CTL
get
the
output
that
you're
getting
is
actually
the
one
presented
by
the
server.
There
is
a
option
to
opt
out
of
this
functionality.
There's
a
server
print
flag
that
you
just
need
to
turn
to
false,
and
in
this
case
you
will
fall
back
to
the
previous
functionality,
but
it
is
on
by
default
in
111
and
that's
yeah
from
a
regular
user
point
of
view.
G
G
The
other.
The
other
thing
that
I
wanted
to
discuss
is
that
we're
working
hard
on
refactoring
the
majority
of
the
commands
inside
of
QT
PL.
So
this
is
more
towards
all
of
the
developers
of
kubernetes,
and
especially
all
the
commands
developers
within
the
cube.
Cdl
is
that
we
had
many
different
patterns
across
the
entire
cube,
GTL
repository
and
we're
trying
to
unify
by
introducing
printing
flag
so
that
every
single
command
has
a
identical
set
of
flags
and
based
on
the
flags.
G
The
output
of
the
command
is
always
the
same,
so
no
matter
if
you're
doing
get
if
you're
getting
create
if
you're
doing
apply
or
any
other
command.
Given
that
you're
providing
the
same
input
Flags,
the
output
will
be
the
same,
and
this
is
one
one
approach
and
we're
trying
to
unify
the
entire
flock
handling
business
within
the
the
QC
deal,
with
the
hope
that
first
of
all,
this
will
help
to
unify
the
codebase.
G
A
All
right,
AWS,
you
are
cute
Justin,
you're,
wrong,
I.
H
Am
hi
everyone
me
so
I'm
gonna
be
giving
the
city
address
update,
but
for
people
that
don't
know,
I
actually
joined
Google
as
of
about
two
weeks
ago.
So
I
am
now
Googler,
but
I'm
still
involved
in
together
us,
at
least
from
the
technical
point
of
view.
But
in
general
in
cigarettes
were
we're
sort
of
doing
the
normal
support
for
new
AWS
features
like
nvme
and
an
LP
as
they
as
they
come
out
on
native
OS,
and
so
a
shout
out
to
Mike
a
house.
H
And
the
steering
committee
recently
defined
a
new
type
of
repository
called
a
cig
repository
which
is
still
I,
guess
part
of
kubernetes
and
hosted
by
a
sig
and
it
lives
in
the
github
comm,
slash,
kubernetes,
SIG's,
org,
and
so
the
great
news
is
that
we
have
our
first
project,
spinning
up
in
that
github
org
committee
SIG's,
ad
West
of
encryption
provider
and
that
implements
the
API
server
support
for
encryption
arrest
inside
of
EDD.
That
actually
has
a
sort
of
interesting
history
and
that
it
was
originally.
H
There
was
originally
an
implementation
entry
for
Google
cloud
platform
backed
by
Google's
PMS,
and
when
we
realized
that,
of
course,
that
would
be
putting
more
and
more
into
the
main
repo.
The
decision
was
made
to
pull
that
out
of
the
criminals,
Caminos
repo
and
instead
have
a
G
RPC
interface
instead,
and
so
our
first
cig
repository
project
is
to
implement
that
to
your
PC
and
face
for
Amazon
AWS.
So
it's
great
in
that
we're.
H
So
you
gave
us
and
for
kubernetes
and
I
hope,
we'll
see
a
lot
more
activity
outside
of
Korea's
communities
in
the
new
cig
repositories.
Avenue,
you
know,
there's
a
ton
of
stuff.
We
wanted
to
do
in
the
past.
I
have
tio
hasn't
needed
rest
identity
project
which
we'd
love
to
see.
You
know
go
through
that
process
and
there's
also
a
generally
huge
backlog
of
things
that
we
want
to
tackle.
H
It
looks
like
I
have
erections
from
mica,
yes
yeah.
We,
we
actually
had
a
kind
of
ad
hoc
meeting
around
the
AWS
cloud
provider
to
get
that
broken
out.
So
that
is
is
happening.
Great
news.
We
still
have
to
figure
out
the
non-technical
issue.
So
yes
that
but
it'll
be
great
to
see
you
know
hoping
we
can
all
forgive
them
out
together.
I
I
More
recently,
we've
had
someone
volunteer
to
try
and
be
a
co-lead
for
this,
which
is
great
news
and
so
I'm
going
to
try
and
pursue
that
and
maybe
get
bit
more
material
together,
I
think
it
kind
of
suffers
a
little
in
the
sig
PCB.
The
fields
vary,
organizationally
appropriate
and
you
need
a
seeks
to
own
the
kind
of
in
the
car
provider.
Contacts
add
to
the
code
base
that
obviously
relate
to
TCP,
it's
appropriate
for
topics
to
be
thrown
out
there.
So
everybody
would
agree,
it
makes
sense,
but
nobody
actually
is
kind
of
interested.
I
Oddly
sees
it
has
a
problem
that
it
would
soul.
I
guess
has
discussions
that
they
need
to
bring
their
the
people
working
on
DCP
related
matters
are
busy
doing
that.
Mostly
google
kubernetes
team
here
and
there's
been
quite
a
few
people
interested
in
lurking
and
interested
in
what
might
be
happening
in
DCP,
but,
like
I
said,
there's
been
havin
had
a
lot
of
luck
in
forging
the
community
around
it.
I
So
I
guess,
if
I'd
use
this
little
slot
with
the
wider
community
to
encourage
people,
if
you
do
have
any
questions
or
thoughts,
desires
related
to
TCP
or
information
like
TCP,
please
don't
hesitate
to
use
the
mailing
list
and
I
can
find
speakers
for
almost
any
topic,
and
if
people
are
interested
in
that
and
I,
hopefully
will
have
a
few
more
and
things
going
forward
as
well
with
them.
This
new
co-lead
I'm
asking
too
much
I've.
A
Right
that
ends
our
setup
dates
for
today.
Moving
on
to
the
announcement
section
first,
we
like
to
cover
shout
outs
and
thanks
to
our
active
contributors
first
and
foremost,
but
we
also
call
out
specific
folks
in
our
shout
out
slack
channel.
So
if
you
see
someone
doing
great
things,
helping
out
especially
issue
triage
and
other
chopping,
water-carrying
wood
rolls.
This
is
the
opportunity
to
highlight
those
folks.
First,
we
have
Masek
patel
for
providing
some
nuance
and
clarity
around
no.2
autoscaler
and
then
Christoph
Blocher
for
feeling.
So
many
issues
and
PRS
thanks
again
awesome
work.
A
A
This
week
is
with
sig
UI
they're,
actually
looking
for
active
contributors
as
well
and
they're
looking
to
revitalize
the
dashboard,
please
join
their
communication
channels
and
attend
the
next
meeting
to
announce
your
interest
in
this
super-important
again
get
the
word
out
say:
UI
is
looking
for
active
contributors
for
the
dashboard
and
then
last
but
not
least,
cube
Khan
update
the
current
contributor
track
session
voting,
which
happens.
The
current
contributor,
assuming
the
contributor
summit,
happens
on
May
1st.
It's
going
to
be
email
today
to
attendees
and
also
please
RSVP
for
the
contributor
summit.
A
It
is
separate
than
Hugh
Khan
attendance
and
it
is
free
though
so
no
it's,
it
does
not
cost
anything
to
go
to
the
contributor
summit.
There
will
be
a
new
contributor
workshop,
so
this
is
not
closed
off
for
just
current
contributors
and
we'll
also
do
a
dock
sprint.
Some
sig
updates
hallway
track
and
some
more
fun
things.
Does
anyone
have
any
last
minute
dates
that
they'd
like
to
share
or
announcements
just.
F
J
Yep
one
announcement
from
me
today
of
send
message
to
communities
death
about
meet
the
maintenance
move
that
organizing
for
our
projects
so
specifically
for
kubernetes.
This
call
is
specifically
for
communities,
community
and
forgiving
artists
project.
Is
that
please
sign
up
for
for
above
duty
at
keep
conclude
native
call
in
Copenhagen?
Please
read
the
best
time
when
can
hear
organize
and
provide
some
feedback
and
have
the
conversations
with
with
the
people
who
can
be
interested
in
contributing
to
communities
or
any
or
have
any
aspirations.