►
From YouTube: Kubernetes Community Meeting 20190808
Description
See this page for more information! https://github.com/kubernetes/community/blob/master/events/community-meeting.md
A
Howdy
and
welcome
everybody
to
the
August
8th,
kubernetes
community,
needy
first
things.
First,
as
always,
we
could
use
someone
to
take
notes
during
the
community
meeting.
This
helps
people
who
are
not
able
to
attend
because
of
Clem's
owns
or
whatever
else
get
a
quick
capsule
of
what
happened
at
the
community
meeting
and
what
they
might
want
to
fast-forward
through
the
video
to
watch.
So
if
somebody
is
able
to
take
notes,
please
just
type
your
name
in
under
note
taker
right
there
thanks
a
lot.
So
then
second
I
need
to
tell
everybody.
A
This
meeting,
like
other
meetings
of
kubernetes
I,
adheres
to
the
kubernetes
code
of
conduct.
So
please
be
a
responsible
adult
also
we
are
going
to
be
recording
it.
So
you
have
two
reasons
to
be
a
responsible
adult
because
you
are
in
public.
So
don't
say
anything
on
this
meeting
that
you
would
not
want
to
be
part
of
an
archival
recording.
A
B
B
So
a
quick
overview
of
Pulu
me
basically
expose
a
bunch
of
open
source
SDKs
to
create
cloud
infrastructure
and
applications.
The
programming
model
is
a
declarative,
steady-state,
so
very
similar
to
kubernetes.
You
tell
it
what
you
want,
not
exactly
how
to
do
it.
The
engine
figures
that
out
for
you,
the
steady
state
declarations
built
using
so-called
real
programming
languages,
so
typescript
JavaScript
Python,
it's
not
DSL,
so
no
yamo,
each
SDK
covers
all
the
major
public
clouds
plus
kubernetes.
B
So
first
a
couple
of
motivating
examples.
So
the
first
one
provisioning
kubernetes
clusters.
So
often
you
want
an
easy
way
to
provision
that
kubernetes
cluster
on
the
public
cloud
of
your
choice,
so
that
could
be
AWS
or
Google
or
Azure
or
even
locally.
Typically,
this
is
fairly
complex
when
you're
setting
up
all
the
identity,
roles
and
kind
of
the
different
resources
necessary
and
it's
different
for
each
of
the
cloud
providers.
B
So
this
this
ends
up
being
kind
of
hard
to
integrate
with
a
CI
CD
system,
so
that
ends
up
being
a
barrier
for
creating
clusters
on
demand.
So
a
lot
of
times
people
will
end
up
spinning
up
one
long-running
cluster
and
then
just
using
that
motivating
of
these
example.
Two,
sometimes
you
want
to
deploy
an
application
to
multiple
clusters,
so
each
each
cloud
providers,
kubernetes
service,
has
subtle
differences.
B
So
if
you
want
to
avoid
the
kind
of
the
lock
into
one
cloud,
it's
nice
to
be
able
to
test
if
your
application
can
deploy
to
different
clouds
like
that
and
then
the
second
part
of
that
is
how
can
you
easily
configure
the
connection
settings
for
each
cluster
so
the
way
that
you
couldn't
solve
this
with
Pulu
me
so
I've
written
a
typescript
program
that
declares
my
desired
state
here.
So
in
this
case,
I'm
I'm
provisioning,
3,
3,
public
cloud,
kubernetes,
cluster
e+.
B
I
have
the
dr
docker
desktop
cluster
running
locally
and
I'm
deploying
the
the
QWERTY
app,
which
includes
a
deployment
plus
a
service
on
each
cluster.
So
I
write
the
program.
I
run
a
plumie
update
that
creates
a
plan
if
I
accept
that,
then
I
execute
the
plan
and
provisions
that
for
me,
so
with
that
this
this
code
is
available
publicly
on
github,
with
this
link
and
I'll
jump
over
to
that.
B
B
So
then,
down
here,
it's
just
creating
references
to
these
clusters,
selecting
an
image
tag,
and
then
here
I'm,
just
using
a
for
loop
looping
over
these
clusters
and
deploying
this
demo
application
here,
so
I'll
hop
over
to
the
demo
app.
So
the
demo
app
here
is
just
a
class.
So
in
my
constructor
I'm
declaring
my
deployment
to
plus
my
service.
So
up
here
you
can
see
it's
just
a
kubernetes
app
v1
deployment.
We
exposed
the
the
entire
kubernetes
API
surface.
That's
our
SDK
is
mostly
auto-generated
from
the
open,
API
spec.
B
So
you
could
deploy
any
version
of
the
resources
you
want
and
then
the
the
actual
spec
looks
just
like
you
would
expect
from
working
with
the
yam
all
or
JSON
with
kubernetes.
So
I've
declared
my
application
here
and
my
service
I'm,
exposing
exposing
the
endpoint
here
and
then
down
here
at
the
bottom.
I've
got
this
app
URL,
so
it
will
take
the
address
that
comes
out
of
the
service
and
create
a
URL
for
me.
So
I'll
hop
over
to
my
terminal,
real
quick.
B
So,
just
because
the
the
clusters
take
a
while
to
provision
I've
gone
ahead
and
stood
those
up
already
and
deployed
the
application.
So
this
is
what
the
output
looks
like
once
once
the
updates
complete.
So
you
can
see
I've
got
endpoints
across
the
the
four
different
clusters
here,
so
I'll
just
jump
into
that
real
quick.
So
you
can
just
see
I've
got
my
Cordy
app
live
and
I
can
click
around
to
here.
B
B
So
I'm
skipping
the
preview
step
here.
If
you
just
run
an
update,
it'll
show
you
a
preview
of
the
the
diff
that
it's
planning
to
run
and
then
you
can
select
whether
or
not
you
want
to
actually
run
that
so
just
take
a
few
seconds
and
then
what
it's
going
to
do
is
go
through
that
for
loop
and
deploy
the
the
updated
tag
to
each
one
of
my
clusters.
Here.
B
Well,
I'm
waiting
for
that
all
I'll
jump
back
over
here,
so
I
guess
I'll
go
ahead
and
look
at
what
some
of
the
actual
cluster
code
looks
like.
So
you
saw
up
at
the
top
where
I
was
creating
the
managed
clusters,
so
this
is
all
done
in
code
as
well,
so
you're
able
to
provision
all
the
resources
that
you
need
for
the
cluster.
B
So
here
on
the
ETS
case,
it's
provisioning,
VPC
and
all
the
resources
you
need
for
networking
and
identity
and
it
creates
a
cluster
and
you
can
set
all
of
these
parameters
like
how
many
nodes,
like
which
instance
types.
So
this
looks
a
little
bit
different
for
each
one
of
the
providers,
this
one's
gke.
B
You
can
see
I'm
setting
up
a
cube,
config
here
and
then
the
AKS
one.
So
you
can
do
everything
from
generating
passwords
to
setting
up
all
of
your
service
principles
and
identity,
networking
and
then
here's
the
kubernetes
cluster
and
the
nice
thing
about
doing
this.
Is
you
can
you
can
write
libraries
like
this
to
abstract
all
that
complexity
here?
So
your
app
developer?
All
they
need
is
just
this.
B
One
line
function
call
and
then
you
can
provision
yourself
a
cluster,
so
it
makes
it
really
easy
for
testing
okay,
so
back
over
here,
this
finally
finished
you
can
see
that
it
tells
you
what
it's
updated,
so
it
changed
the
spec
on
each
one
these
deployments
and
gave
me
these
URLs.
So
now,
if
I
hop
over
here,
you
can
see
that
this
is
now
running
the
green
instead
of
the
blue,
that
it
was
previously.
B
Guess
one
other
thing
that
might
be
of
interest.
This
is
this:
is
our
service
here,
so
I
can
come
in
here
and
look
at
kind
of
detailed
logs
of
everything
that
it
was
doing.
So
this
is
what
it
would
have
looked
like
had
I
previewed
it,
you
can
look
at
details
about
exactly
what
it's
going
to
change
and
then
details
I
wanted
to
change
all
of
the
the
logs
here
in
one
place.
B
I
guess
one
one
thing:
that's
interesting
to
note
about
this
balloon
water:
this
would
automatically
figure
out
dependencies.
I
do
referencing
things
here,
so
you'll
notice
I
was
able
to
create
this
app
URL.
This
one
actually
have
useful
data
filled
out
until
these
things
are
forbidden.
So
first
I've
got
my
deployment
and
then
I've
got
my
service.
B
So
then
the
service
is
referenced
down
here
and
I'm,
able,
to
just
say,
status,
load,
Balasore,
ingress
and
then
here's
here's
the
difference
between
gke
and
eks.
So
it's
either
hostname
or
IP,
but
then
based
on
that
I
can
pull
that
out
into
a
to
app
URL.
So
it
figures
out
the
dependencies
automatically.
B
A
B
A
We
next
up
is
the
release
update
Oh
infirmities
watch
the
demo,
excite
others
and
link
to
the
slides
and
a
link
to
his
example
demo
there
in
the
notes.
So
next
up
we're
doing
the
release
updates,
Jeffery
Sita,
one
of
our
shadows
for
the
116
release.
It's
going
to
update
us
on
the
many
things
going
on
with
releases
this
week
and
last
week,
hello.
C
Everyone
currently
we
are
in
week
6
of
the
1/16
release,
things
that
have
just
happened,
I
believe
it
was
two
days
ago
yesterday
we
released
116
alpha
3.
That
is
going
to
be
the
final
alpha
release
for
116
the
big
things
coming
up.
We
have
the
first
beta
release
coming
up
next
week
on
Tuesday
August
13th,
and
the
ever
impending
code
freeze
is
coming
is
at
the
end
of
the
month
at
August
29th.
So
keep
that
in
mind.
There
have
been
many
patches
that
have
been
released.
C
They
were
addressing
one
CVE
that
sorry
one
was
in
incomplete,
fix
for
a
CVE
and
then
an
actual
fix
for
cube,
cuddle,
CP
directory
traversal
vulnerability,
as
well
as
API
servers
allowing
access
to
custom
resources
via
the
wrong
scope.
So
those
are
bad,
but
they
have
been
patched.
Please
update
your
clusters.
All
of
these
are
announced
on
the
kubernetes
security
announced
mail
as
well
as
kubernetes
unknowns.
We
have
additional
patch
releases
coming
up
targeting
August
19th
and
those
will
be
115,
3,
114,
6
and
113
10.
C
A
Since
none
of
this
is
being
updated
for
112,
which
has
been
you
know
for
a
while
now
so
we're
going
to
move
into
our
next
section,
which
is
the
cig
updates.
We
actually
have
two
SIG's
today:
updating
us
on
current
status
and
activities,
we're
going
to
start
out
with
cig
contributor
experience
with
chair,
Paris,
Pittman,
hello,.
D
D
Hi
everyone
all
right.
Let
me
share
my
screen
all
right,
all
right
as
usual:
contributor
experiences
old
fire,
meaning
we
are
cranking
out
work
y'all.
There
are
30
31,000
contributors,
the
last
time,
I
checked
over
190
repositories.
That
means
there's
a
lot
of
y'all
and
a
little
bit
of
us,
so
I'm
going
to
present
some
of
the
work
that
we
have
been
accomplishing
and
our
undertaking
for
both
the
last
three
months
and
the
next
three
months
cool.
So
first
things.
First,
we
have
taken
on
the
cluster
lifecycle
approach
to
meetings.
D
We
have
a
lot
of
sub
projects
now
that
I'll
go
through
during
the
presentation,
but
we
are
taking
these
sub
project
approach
and
doing
what
and
if
you
can
see
my
screens
I,
don't
know
if
I
just
shipped
I
just
share
the
presentation
or
did
I
share
the
browser,
okay
cool,
so
you
can
see
on
our
meeting
agenda
that
we've
broken
out
by
sub
projects
which
helps
us
do
the
work
and
also
helps
you
figure
out
what
it
is,
a
we
do
and
B
how
you
can
get
involved.
So
we
have
events.
D
Community
management,
github
management,
slack
in
for
a
contributor
documentation
and
mentoring,
and
that's
full
steam
ahead.
Oh
and
also
we've
moved
on
our
weekly
meetings
to
bi-weekly
meetings
so
that
the
smaller
sub
projects-
it's
not
like
we're
always
meeting
and
it'll
help
us
capture
that
a
little
bit
better
second
thing
about
meetings
and
I
wanted
to
bring
up
is
that
our
AIPAC
team
is
rolling.
D
D
So
the
first
thing
I
always
want
to
talk
about
is
mentoring
and
contributor
growth.
That
is
the
most
important
part
of
us
at
this
point
and
the
most
important
part
of
the
project,
at
least
in
my
opinion.
No
really
this
is
actually
super.
Important
burnout
is
a
thing
that
we've
all
heard
about
and
know
about,
and
the
only
way
to
the
only
way
to
get
help
with
that
is
to
raise
people
up
with
you.
That's
literally
the
only
way.
D
So
what
we're
doing
in
contributor
experience,
we
actually
broke
down
and
have
a
mentoring,
sub
project,
and
we
had
our
first
meeting
this
morning.
We
are
basing
our
our
programs
around
contributor
personas
things
like
students,
code,
reviewers,
approvers
and
other
needful
areas
of
the
project.
Some
of
the
things
that
we're
seeing
that
are
really
really
really
helping
and
really
helpful.
Right
now
is
succession
planning
via
teams.
So,
for
instance,
one
team
that
you
all
are
very
known
and
accustomed
to
is
the
release
team.
The
release
team
is
awesome.
D
Y'all
have
built
shadow
rolls
that
is
really
a
great
way
to
go
and
open
source.
So
we've
taken
this
concept
and
applied
it
in
multiple
other
areas,
including
the
events
team,
which
I'll
show
you
in
a
minute
for
contributors
for
contributor
summits,
but
also
we're
taking
that
a
little
bit
farther
and
doing
triage
teams
that
have
role
books
and
things
like
that.
Anything
that
you're
sig
is
doing.
Where
is
a
clear-cut
role?
D
That's
a
role,
and
that
can
be
explicit
and
we
can
advertise
for
that,
and
other
contributors
really
like
that,
because
it's
clear-cut
it
gives
them
an
on-ramp
and
it's
not
ambiguous,
Lee
jump
in
and
help
other
mentoring
at
scale.
Initiatives
and
I
say
it's
scale,
because
one
on
one
with
31,000
contributors
doesn't
scale
learning.
By
doing
does
so,
we've
been
concentrating
and
getting
other
SIG's
on
board
of
doing
things
like
code
base
tours
live
sessions
like
API
review
codes
triage.
D
This
is
this,
gives
us
the
ability
to
put
it
on
our
YouTube
channel
and
then
syndicate
that
and
put
it
on
our
official
documentation,
so
that
contributors
can
all
board
faster
and
at
the
quality
that
we
need
them
to
on
board
and
it's
not
just
bland
documentation.
We've
also
learned
that
from
both
academia
and
our
real
world
experiences
that
not
everybody
reads:
contributor
documentation,
some
people
actually
watch
videos.
D
So
why
are
we
catering
to
one
persona
when
there
are
multiple
personas
here
that
we
could
help
out
and
grow
new
contributor
office
hours
is
also
a
thing
that
is
really
helpful.
So
if
you're
say,
gets
a
lot
of
new
contributors
that
come
into
your
slack
channel
and
say
things
like
hey,
how
do
I
get
started
and
it's
you
know,
that's
that
can
carry
a
lot
of
time.
D
Burden
have
a
once
a
month,
new
contributor
office,
hours
and
point
folks
to
that
and
then
get
them
acclimated
there
and
get
them
an
on-ramp
and
things
along
those
lines.
Cluster
lifecycle
host
on
this.
We
do
this
sort
of
with
meet
our
contributors
as
well,
so
any
of
those
things
are
really
really
great
for
SIG's
to
do
so.
If
you
are
in
a
cig,
please
think
about
adopting
some
of
these
things.
D
Another
thing
that
we're
taking
advantage
of
is
cube
con
face
to
face
mentoring
sessions.
I
would
say
about
60%
of
the
people
that
show
up
actually
show
up
with
contributing
questions.
Most
people
think
that
it's
and
user
questions-
and
that
is
definitely
a
myth.
So
this
is
an
opportunity
for
us
to
to
get
in
front
of
people
face
to
face,
and
then
group
mentoring.
D
This
is
where
ten
people
to
get
together
with
three
mentors
in
a
group
and
they
all
peer
mentor
each
other
to
a
goal,
and
that
goal
is
our
contributor
ladder
and
we
need
other
six
to
adopt.
This
I
can't
go
into
super
details
right
now,
just
because
of
time,
but
if
you're
saying
is
like
wow,
we
really
need
more
approvers
route.
Mentoring
is
the
best
way
to
do
it
at
scale
so
reach
out
to
us
and,
of
course,
meet
our
contributors
here.
Stephens
lovely
face
right
in
front.
D
This
is
a
monthly
YouTube
that
series
that
we
do,
that
sees
at
least
I'd,
say
thirty
people
live
end
to
end
and
then
at
least
a
hundred
recording
and
end.
So
this
is
reaching
massive
people
at
scale,
and
this
is
really
only
being
talked
about
during
on
contributor
based
advertising
channels,
which
means
that
we
are
saturating
the
contributor
market,
which
is
exactly
what
we
are
looking
to
do.
D
This
is
why
I'd
feel,
like
most
of
you
know
us
from
and
had
talked
to,
people
like
Kristoff
and
Nikita
and
Bob,
and
so
many
other
of
our
awesome
folks
that
take
care
of
our
hundred
and
ninety
plus
whatever
it
is
at
this
point,
I
feel
like
I,
can't
even
keep
tabs
on
it.
One
hundred
and
ninety
repositories
and
they're
doing
all
things
automation.
D
This
is
where
they're
creating
repos
for
you.
We
have
more
than
over
30
30
new
repositories,
I
think
it's
something
like
the
last
three
to
six
months.
If
I
can
recall
my
data,
my
data
range
there
and
then
they're
also
processing
is
kind
of
membership
requests,
which
is
great.
The
membership
requests
give
a
gives
the
contributor
access
to
talk
to
the
bots
a
little
bit
more
clearly
and
interact
with
them
and
in
better
ways.
D
So
if
people
are
saying
hey,
but
the
bots
aren't
wonderful
ask
them
if
they're
a
member,
if
they
are,
they
might
have
a
little
bit
better
of
an
experience
and
then
we've
also
built
out
some
project
websites.
I
think
Bob.
Mr.
Bobby
tables
had
a
huge
play
in
that.
So
thanks
to
him-
and
that
is
a
you-
can
get-
that
information
in
the
kubernetes
slash
or
issue
template
area
upcoming
stuff.
D
We
are
renaming
fado
bot,
I,
think
it
is
I,
don't
know
if
it's
decided
somebody
yell
out
if
it
is
but
I
think
it's
going
to
be
Cates
triage,
robot
I'm,
not
not
sure
how
official
that
is.
But
that's
that's
what
it
is
at
this
point
on
this
slide
and
then
we're
also
going
to
be
doing
a
lot
of
auditing
of
owners
files.
So
if
you're
in
a
cig
and
your
sig
has
some
stale
owner's
files,
this
is
for
you
and
any
sake
at
any
time
can
do
owners
files,
hygiene
feel
free.
D
D
Contributor
documentation:
this
is
one
of
the
projects
that
we
have
that's
living.
That's
always
going
that
always
has
issues
files,
as
always
says,
broken
links
things
like
that,
because
it
is
a
living
living
living
beast.
We
have
so
many
things
going
on
within
within
contributor
documentation.
At
this
point
we
have
the
contributor
guide
itself.
We
have
the
contributor
chichi,
which
George
édouard
from
outreach
e
and
so
many
other
people
in
contributing
to
which
is
so
lovely.
D
You
don't
necessarily
need
to
know,
go
to
do
and
then,
of
course,
our
contributor
site
that's
been
in
progress
for
a
while,
but
we
are
chugging
chugging
along
now
and
think
that
we
have
and
think
that
we
have
a
good
MVP,
we're
probably
going
to
go
out
the
door
with.
Could
the
contributor
guide,
the
events
and
the
calendar
for
the
first
go?
You
can
see
the
net
la
vie
link
there
and,
if
you're
interested
in
learning,
what
kind
of
content
is
coming
and
if
you
can
get
involved
and
how
you
can
get
involved.
D
There's
the
issue
link
right
there.
We
also
have
a
contributor
documentation,
some
some
project
meeting
at
this
point
and
would
love
to
see
people
there
and
would
love
to
get
more
people
and
invested
and
involved
in
this
area
of
the
project
for
sure
I
think
I
have
some
in
chat.
Oh
and
then,
of
course,
community
management.
D
D
Infra,
one
of
the
things
that
we
are
still
continuing
to
do
is
reduce
the
noise
for
chairs
and
tech
leaves
get
them
the
information
that
they
need
very
fast
because,
as
everyone
knows
in
the
project,
there's
75
different
communication
platforms,
so
we
need
to
make
sure
that
they
get
the
information
that
they
need
immediately
and
we've
also
built
a
slack
channel
for
them.
It's
called
chairs
and
tech
leads,
and
it
is
highly
moderated
just
so
they
can
have
a
place
to
play.
D
Thank
you
special.
Thank
you
to
mungus,
thank
you
actually
to
all
of
our
moderators.
We
have
so
many
who
moderate
both
zoom,
YouTube
slack
mailing
list,
discussed
kubernetes,
IO
and
so
many
more
if
you're
interested
in
policies
guidelines
moderators
how
to
how
to
be
a
moderator,
etc.
Everything
you
need
to
know
is
in
that
link
right
there
on
that,
on
that
deck,
special
shout
out
to
Katherine
as
well
give
her
shoutouts
all
the
time.
She's
amazing
she's
been
building
a
slack
automation
that
helps
us
keep
this.
D
That
really
helps
us
keep
this
at
bay
and
this
bed
and
this
at
bay,
meaning
this
like
80,000
person,
beast
of
a
slack
instance
and
which
is
completely
blown
the
scope
of
contributor
experience.
But
we
won't
talk
about
that
right
now,
but
this
is
an
awesome
reporting
function
that
I
always
want
to
remind
people
is
they're.
D
D
This
is
I
feel
like
the
first
event
of
any
kind
that
I
have
been
involved
with.
That
is
this
open,
y'all
I
just
want
to
take
a
quick
shout
out
to
everybody.
That's
involved
with
this.
This
is
amazing.
You
have
a
project
board
meetings,
I
mean
even
see
the
content
you
can
see
like
down
to
the
dotted
I
of
what
lights,
which
we're
going
to
switch
off
when
for
what
and
it's
so
so
great.
But
this
is
also
a
call
for
proposals
and
talent.
You're
gonna.
D
Ask
me
why
talent
in
a
second
I'm
sure,
but
call
for
proposals.
So
if
you're
out
there
and
you're
like
wow
I,
really
wish
all
of
our
contributors
knew
how
to
do
performance
tests
better
or
conformance
or
I,
really
wish
they
could
do
my
code
reviews
a
little
bit.
Better
teach
them
call
for
proposals
and
then
the
talent
piece
on
Sunday
night.
We
have
the
opportunity
to
potentially
do
a
talent
show
anyway,
so
call
for
talent.
If
you
play
instruments
sing
a
song
teach
karate
we
want
to
hear
about
it.
D
I
want
your
whole
selves
anyway,
almost
now
hiring
two
for
Amsterdam
for
all
leads.
If
you
are
interested
in
working
a
contributor
summons
next
year,
please
reach
out,
and
then
the
next
major
event
that
we
are
operating
right
now
is
a
steering
committee.
Election
steering
committee
is
due
for
their
yearly.
Their
yearly
ballots
PR,
they
are
current.
D
The
steering
committee
is
currently
picking
our
election
officers
and
that
should
be
coming
in
any
second
now
and
then,
once
that's
set,
everybody
will
see
even
more
information
about
how
you
can
be
a
steering
committee
member,
how
you
can
nominate
someone
etc.
So
please
look
out
for
that.
I
hope
everybody
rocks
the
vote
here.
D
So
how
can
you
contribute
join
our
meetings,
of
course,
be
a
mentor.
Yes,
you,
even
if
you
form
with
a
project
for
two
days,
spread
our
communications
far
and
wide.
We
have
a
large
project.
Ask
your
cigs.
Did
you
know
that
this
thing
was
happening,
use
good,
first
issue
and
help
on
it
labels,
because
that
helps
us
help.
You
join
our
triage
team,
build
triage
teams
and
think
about
triage
for
your
SIG's
as
we
go
forward,
and
we
also
need
bloggers
and
lots
of
you.
Why?
D
Because
we
have
a
lot
of
contributor
information,
clearly
I
just
took
up
almost
15
minutes
of
your
time
and
I
probably
could
take
up
another
30
if
I
went
into
like
hardcore
details
about
what
we
do,
this
stuff
needs
to
be
discussed,
needs
to
be
talked
about,
needs
to
have
a
lot
of
light
of
eyeballs
on
it,
and
we
want
to
put
some
of
the
human
touches
to
some
of
the
work
that
your
sakes
are
doing.
Reach
out.
There
is
also
some
blog
some
blogging
issues
and
the
community
repo
reach
out
to
us.
D
We
are
having
a
goal,
this
time
of
200
respondents
and
we
actually
take
the
we
actually
take
some
of
the
stuff
you
say
and
create
actionable
results
with
it,
and
I
want
to
actually
show
you
some
stuff
that
we
did
from
last
year
from
last
year's
survey.
So
I'm
gonna
go
off
the
script
a
little
bit
here,
just
to
prove
to
you
that
we've
done
some
cool
stuff
in
this
area
hold
on
one.
Second,
I
am
switching.
D
About
to
see
Jeff's
head
really
big
I
think
alright,
alright.
So
these
are
the
some
of
the
really
cool
things
that
we
got
from
the
survey
the
last
time
and
this
time
we
intend
to
take
the
entire
survey
and
get
a
designer
and
do
something
along
the
lines
of
a
Python
developer
report.
If
you
saw
there
from
jetbrains,
it
was
so
gorgeous
and
beautiful,
and
that
would
be
amazing
if
we
could
do
that
for
our
contributors.
D
D
So
clearly
our
contributor
audience.
We
have
a
lot
of
folks
on
Twitter
all
right,
so
does
your
employer
support
your
contributions
to
kubernetes
66%
at,
and
this
was
last
September
y'all,
so
note
note,
last
September
66%
of
the
people
said
yes,
it's
a
part
of
my
job,
I'm,
very
curious
to
run
this
again
and
63%
of
the
people
said
that
they
contribute
at
least
several
times
a
month.
So
what
this
says
about
the
survey
is
we
had
a
lot
of
active
a
lot
of
active
people
respond
and
then
would
to
the
tooling.
D
Do
you
find
useful?
This
obviously
wasn't
that
great
of
a
question-
we're
gonna,
probably
remodel
this
a
little
bit
so
that
we
get
better
results,
but
you
seem
up
for
the
most
part
everybody
likes
Birds
automatic,
merging
of
their
approved
PRS.
Why
like?
Why?
Wouldn't
you
so
I
feel
like
that?
That
was
sort
of
an
obvious
deal,
but
we'll
probably
remodel
that
question
a
little
bit
better
next
time,
this
one's
a
little
bit
about
the
contributor
ladder.
D
D
Here's
a
really
good
Blacker's
by
contributor
level
chart
that
came
from
where
we
talked
about
what
are
some
of
the
major
brought
major
blockers
that
you're
having
like
number
one?
It's
not
a
problem,
five
I'm
having
a
really
hard
time
with
this,
and
so,
for
instance,
approvers,
said
they're
having
a
really
hard
time.
Debugging
test
failures:
reviewers
are
having
a
really
hard
time.
Debugging
test
failures,
no
idea
this
was
even
a
thing
setting.
D
So
no
idea,
this
is
even
a
thing,
is
the
membership
level,
so
that
means
there
was
people
that
took
it,
but
did
not
know
that
the
contributor
ladder
was
even
a
thing.
So
those
folks
said,
setting
up
their
dev
environment
was
the
hardest.
So
they're,
probably
like
our
newer
folks
right
that
needs
some
help
there.
So
all
of
this
stuff
we
can
take
and
make
better
programs
with
this
data.
D
What's
the
best
part
of
the
community
meeting,
this
is
where
we
like
lost
the
dev
stats,
where
we
we
said
no
to
dev
stats.
It
didn't
do
so
well
on
this.
We
actually
took
it
off
because
it
didn't
do
so
well,
but,
as
you
can
see,
most
people
like
a
lot
of
the
a
lot
of
things
in
the
community
meeting,
they
find
it
useful
different
person.
D
I
even
did
it
by
persona
too,
but
you
can't
see
it
on
this
graph,
but
I
was
very
curious
to
see
which
persona
likes
what
and
it
was
pretty
much
every
persona
likes
everything.
So
congratulations
to
the
community
meeting
y'all
think
this
is
fruitful
and
then
do
you
think
slack
adds
value
to
this
project.
That
goes
without
saying,
80.
82
percent
of
you
said.
Yes,
it's
useful
for
both
users
and
contributors.
So,
let's
see
I,
think
I'm
pretty
much
done
with
the
slides.
D
D
A
E
Thank
you
so,
and
thanks
and
huge
shout-outs
to
see
contributor
and
paris
for
their
effort.
Unfortunately,
my
updates
are
not
as
interesting
but
I
I'll.
Try
to
give
you
a
couple
of
good
news
about
new
features
that
we
are
building
in
scheduling
in
115,
we
built
a
the
first
version
of
what
we
call
the
scheduling
framework.
This
is
basically
an
attempt
to
make
the
scheduler
eye
a
more
pluggable
to
suffer
more
pluggable
components.
E
So
far,
scheduler
has
been
like
a
monolithic
piece
of
software
that
has
been
hard
to
change
if
you
wanted
to
add
something
to
discard
or
if
you
wanted
to
customize
it
for
your
own
cluster.
So
the
scheduling
framework
is
an
effort
to
change
that
and
make
the
scheduler
a
pluggable
architecture
so
that
various
users
can
customize
it
customize
it
to
their
needs
and
also
it
helps
us
build
more
complex
features
without
putting
all
the
code
in
in
a
core
of
the
scheduler.
We
built
very
early
version
of
this
in
115.
It's
an
alpha
version.
E
E
Another
another
feature
that
we
we
actually
released
in
with
115
was
non
preemptable
priority.
So
what
what
is
this
about?
So,
as
you
know,
as
you
may
know,
pot
priority
helps
you
identify
positive,
are
more
important
than
others.
So,
when
your
cluster
is
out
of
resources
and
some
of
these
parts
arrive,
the
scheduler
will
remove
some
of
those
less
important
ones
in
order
to
schedule
more
important.
E
E
Basically,
if
there
are
a
lot
of
pending
pods
on
waiting
to
be
scheduled,
those
which
have
higher
priority
go
to
the
head
of
the
scheduling
to
a
non
pre-empting
priority
that
has
basically
among
a
pilot,
has
non
preempted
by
early
and
as
a
high
priority
goes
to
the
head
of
the
scheduling
to
you,
but
it
does
not
preempt
any
of
the
existing
or
running
workers.
This
is
useful
in
a
lot
of
postures
that
run
bad
workloads,
because
in
those
clusters,
people
do
not
want
to
remove
existing
workloads.
E
They
just
want
to
specify
that
some
of
their
workers
shouldn't
be
scheduled
first,
as
soon
as
the
cluster
gets,
some
resources
were
SMS
among
the
existing
workloads
need
the
pasture.
So
this
is
about
that.
It
exists
in
115.
We
also
design
event
for
the
spreading
I.
Wouldn't
talk
about
this
picture
in
my
next
slide.
This
is
also
a
new
feature
for
this
scheduler.
E
We
managed
to
design
this
in
115,
but
due
to
some
API,
and
we
do
bandwidth,
we
could
not
land
this
feature
in
115,
but
a
good
news
is
that
116
is
going
to
have
it.
Also,
we
monitoring
and
reliable
it
meaning
of
the
scheduler
and
115,
nothing
very
notable
there
more
metrics
to
monitor
this
table
and
what's
coming
in
116.
So
as
I
said,
one
of
the
important
features
that
we
are
releasing
with
116
is
partly
spreading.
What
is
this
about?
E
So,
as
you
may
know,
a
lot
of
people
want
to
have
higher
higher
reliability
for
their
workloads
and
they
achieve
this
by
spreading
their
workloads
among
multiple
failure.
Newman's,
for
example,
people
schedule
their
services
in
multiple
zones.
If
one
zone
goes
down
the
others
all
stay
up,
how
do
you
achieve
this?
So?
First
of
all,
the
scheduler
has
some
default
way
of
scheduling
the
by
departing,
based
on
the
spreads
workloads
and
multiple
failure
domains,
but
those
are
limited
to
no
its
end
zones
only
and
it's
not
guaranteed,
but
even
for
spreading.
E
E
We
had
an
try
if
anything
in
the
past,
so
you
couldn't
specify
an
to
infinity
for
your
parts,
but
the
problem
with
on
the
Infinity
is
that
if
you
basically
enforce
it,
if
you
have
an
affinity,
the
morning
puts
one
part
for
thing.
They're
doing
somebody
spreading
removes
that
elimination.
You
can
specify
how
you
want
to
spread
your
parts.
E
This
was
a
common
request
from
a
lot
of
users
and
the
good
news
is
that
they
have
engine
1/16
and,
as
I
said,
we
felt
the
very
first
version
of
this
kind:
entering
working,
1,
6,
115
116.
We
are
going
to
have
all
expansion
points,
so
the
scheduling
framework
is
going
to
be
ready
to
use,
although
we
decided
not
to
promote
it
to
beta
in
order
to
get
higher
confidence
or
our
API.
So
we're
going
to
keep
it.
E
You
know
in
alpha
feature
or
in
one
sixteen,
but
hopefully
we
can
get
some
impact
on
users
and
promoted
to
data
in
117
and
finally,
we
are
also
implementing
some
resource
being
high
clean.
This
is
also
featured
and
help
from
users
achieve
better
resource
utilization
in
their
clusters
by
putting
packing
things
with
more
thoughts
on
fewer
number
of
nodes.
This
is
important
if
you
want
to
achieve
higher
resource
utilization
and
reduce
the
cost
of
your
cluster.
E
How
these
plans
affect
you
generally,
it
should
not
affect
most
of
our
users.
All
these
new
changes
are
expected
to
be
fully
backward-compatible.
The
only
thing
is
that
the
scheduling
framework
allows
people
to
then
all
sort
of
custom
plugins
for
the
scheduler.
Some
of
these
custom
plugins
may
not
work
with
cluster.
Obviously
we're
trying
to
base
make
it
possible
for
all
in
scaler
try
to
identify
all
these
parties
and
use
them,
but
the
scheduler
is
more
new
schedule.
E
Framework
is
more
flexible
than
B,
then
the
simulation,
but
anyway
so
autoscaler
has
some
simulation
in
the
simulation
part
of
autoscaler,
certainly
being
seven
new
features
of
this
kind
of
framework
are
not
implementable,
so
there
is.
We
are
really
gonna,
basically
document
everything
and
tell
users
which,
which
one
of
these
features
will
be
supported
in
cost
value
scale
and
which
ones
cannot
be
supported,
and
finally,
where
ever
you
can
find
us
here
is
some
information
for
you.
The
home
page
were
scheduling
and
we
have
a
slack
channel
which
is
very
active
in
just
winning
question.
A
A
Okay,
well,
so
thank
you
so
we're
now
on
to
our
announcements,
our
first
announcement
as
Paris
mentioned,
we
are
soliciting
session
proposals
for
the
2019
San
Diego
contributor
summit.
If
you
are
going
to
be
at
the
contributor
summit,
if
your
contributor
consider
leading
or
even
requesting
a
session
for
pre
program
sessions,
we're
going
to
have
both
we're
gonna,
have
presentation
sessions,
we're
gonna
discussion
sessions
and
we're
gonna
have
some
AI
contributor
focus
workshops
in
addition
to
the
new
contributor
intermediate
contributor,
day-long
workshops.
So
we
are
looking
for
all
of
those
things.
A
Speaking
of
coop,
con
I
wanted
to
remind
any
sig
leaves
who
are
on
this
call
or
who
are
watching
the
video
that
the
cig,
intro
and
deep
dive
session
proposals
are
due
August
16th,
your
SIG's
should
decide
on
what
sessions
you're
going
to
propose
for
your
six
sessions
and
submit
them
I
soon,
you've
got
like
a
week
to
figure
those
out
and
submit
them.
So
please
do
that.
A
A
Okay,
apparently
not
so,
go
ahead
and
check
out
the
CN
CF
blog.
The
same
thief
has
published
the
results
of
the
Cuban,
a
security
audit
with
the
whole
bunch
of
findings.
Documents
has
already
had
the
effect
of
triggering
a
security
release
to
fix
some
of
the
things
that
they
found.
There's
a
whole
bunch
of
other
useful
information,
particularly.
A
Again
for
contributors
who
work
on
all
the
components
of
kubernetes,
there's
a
long
discussion
in
there
about
how
the
complexity
of
configuration
for
kubernetes
leads
to
exploits
of
production
implementations
because
operators
are
actually
configuring
their
clusters
incorrectly
because
they
just
can't
figure
out
how
to
configure
them
correctly.
The
so
please
read
that
section.
Hopefully
that
will
inform
thinking
on
our
API
and
management
design
in
the
future,
and
there
are
lots
and
lots
of
other
stuff
in
there.
So,
let's
go
to
shout
outs
from
Jason.
A
We
have
shout
outs
to
Justin
and
MRI
sorry
I
actually
know
who's
handled.
It
is
forgetting
the
AWS
credentials
created
and
updated
for
the
cluster
API
provider,
AWS
testing
using
Bosco's.
So
the
automated
testing
is
now
fixed.
There
Paris
says
thanks
to
the
meet
our
contributors,
Mentors
Alaina,
hashman
and
Nikita,
who
give
great
advice
on
caps
and
getting
started
and
a
bunch
of
other
things,
and
also
thanks
to
all
right
Jeffrey
for
making
fixing
YouTube
for
meet
our
contributors.
A
Jim
angel
shouts
out
to
dims
and
don't
know,
Samuel
that
is
College
CEO
dyc
for
existing
new
contributors
navigate,
get
issues
in
sigdoc,
slack
so
eye
shadows
for
everybody.
Remember
you
can
post
shout
outs
in
the
shadow
at
slack
channel
at
any
time
the
and
we
will
collect
those
for
the
week
and
it's
also
a
good
way
just
to
show
people.
You
appreciate
what
they
did
so
thanks
everybody,
and
that
is
our
community
meeting
for
the
week
as
usual.