►
From YouTube: Kubernetes Community Meeting 20190110
Description
We have PUBLIC and RECORDED weekly meeting every Thursday at 6pm UTC.
See https://github.com/kubernetes/community/blob/master/events/community-meeting.md for more information!
A
B
It
hello
there
I'm
gonna,
share
my
screen
now.
Okay,
can
everyone
see
my
screen?
Okay,
okay,
thank
you.
So
much
hello,
folks,
I'm
gonna
talk
about
the
crew,
the
package
manager
for
cube,
Caudill
plugins.
Let
me
go
into
the
full
screen
mode.
B
Okay,
so
we
developed
the
package
manager
for
cube
cutter
plugins
at
Google
and
today
I'll
be
giving
a
demo
of
that.
My
name
is
Amida
balkan.
You
can
find
me
on
github
and
twitter
as
ahmet
be
so
in
30
seconds
or
less
cube.
Cutter
plugins
are
stable.
As
of
1:12.
It
makes
it
possible
to
extend
the
cube
Caudill
plugins
with
make
commands.
Let's
say
you
have
an
executable
named
cube,
cuddle
few
in
your
path
and
then
you
can
walk
it
like
Q
cuttle.
Space
foo
is
a
sub
command
of
Q
cuddle.
B
So
this
is
what
Q
cutter
plugins
are.
Why
do
why
are
plugins
useful
I
want
to
give
two
examples
of
scenarios.
So
let's
say
you
want
to
extend
cube
cuddle
with
extra
functionality
that
still
feels
like
part
of
cube
cut'.
Oh,
so
you
imagine,
we
had
a
plugin
named
cube,
cuddle
service,
catalog
or
svgs,
or
something
like
that.
So
this
could
be
an
addition
to
Q
cuddle
functionality
without
adding
any
code
into
Q
cuddle
or
let's
say
you
work
at
a
company
that
wants
to
encapsulate
their
work
flows
into
cube
cuddle
instead
of
running.
B
You
know
a
bunch
of
shots
across
so,
for
example,
I
have
an
example
here
that
says:
cube
cuddle
delete,
orphaned,
pods.
Maybe
you
have
like
a
three
four
line
bath
script
that
does
this
or
you
can
just
package
in
a
cubicle
command
and
provide
it
to
your
developers
by
installing
this
on
every
single
developer
machine.
B
So
when
we
looked
at
the
cube
cuddle
plug-in
space
early
on
the
one
of
the
few
areas
that
we
realized
is
when
this
feature
goes
out
goes
out
to
the
cost
to
the
people
using
cube
cuddle,
so
users
will
want
to
discover
plugins
and
then
they
will
probably
need
a
way
to
install
these
plugins.
This
could
be
their
package
manager.
They
could
be
just
copy
pasting
some
binaries
around,
and
they
probably
want
to
also
keep
these
plugins
up
to
date.
Similarly,
developers
probably
will
want
to
make
their
plugins
discoverable.
B
So
if
you
don't
know
any,
you
know
five
cube
cuddle
plugins
today,
that's
not
your
fault!
That's
because
we
don't
have
a
good
discoverability
story
around
the
plugins
and
the
walkers
will
probably
want
to
figure
out
how.
How
can
I
package
my
plugin
for
multiple
platforms
like
Windows,
Linux,
Mac
OS,
and
so
we
developed
crew
at
Google
in
the
summer
of
2018.
The
project
was
started
by
me
and
our
intern
look
Burchard
his
LBB
on
github,
so
crew
is
a
plugin
manager
for
cube
cuddle.
B
It's
like
home
brew
the
brew
command,
but
it's
for
cubicles,
so
that's
the
hence
name
and
what
crew
does
is
searching
the
plugins
installing
plugins,
upgrading
plugins
and
you
know
removing
plugins
from
your
system.
So
we
took
a
lot
of
shortcuts
to
avoid
being
an
actual
package
manager,
so
we're
trying
to
intentionally
avoid
a
lot
of
the
aspects
of
you
know,
being
the
non-trivial
parts
of
being
a
package
manager,
so
we're
trying
to
not
invent
a
new
package
manager,
but
rather
a
in
a
more
minimalistic
version
to
distribute
plugins.
So
cream
has
been
already
open-source.
B
Be
then
announced
officially
and
so
today-
and
there
are
about
two
hundred
stars
on
github
repos.
If
you
go
to
github.com,
slash,
google
container
tool,
slash
crew
you'll
be
able
to
find
the
crew
source
code
there
and
you
know
play
with
it
right
away,
and
today
we
have
over
20
plugins
distributed
through
crew,
and
that
number
is
pretty
huge
actually
because
there
are
not
a
whole
lot
of
plugins
today.
So
I'm
gonna
quickly
give
a
demo
when
I
searched
my
terminal
here.
So
I
have
installed
crew
to
my
computer.
B
Crew
itself
is
a
plugin,
so
if
I
do
to
TTL
crew
and
we'll
be
getting
crew
commands,
so
then
I
type
Q,
CTL
crew
search
I'll
be
able
to
see
all
the
plugins
that
are
available
on
my
system
or
on
the
plugin
index,
not
on
my
system.
Sorry,
so
this
means
I
can
choose
any
of
these
plugins
and
install
into
my
computer
as
long
as
it's
supported
plugin.
So
if
I
Coop's
et
al
crew,
install
I'm
just
gonna
choose
well
I,
don't
like
our
back
view
here
when
I
hit
return.
B
This
is
going
to
the
plugin
index
figuring
out.
What
the
plug-in
is
downloading
it
and
then
telling
me
hey:
you
can
now
run
this
command
cube.
Ctlr
back
view
to
you
know,
do
something
whatever
that's
plug.
It
does
actually
don't
know
what
it
does
so
I'm
not
gonna
run
it,
and
so,
as
you
see
here,
I
just
installed
the
plug-in
I
can
choose
any
other
plugins
up
there
and
install
it
to
my
machine
as
well.
B
So,
similarly,
when
it
comes
to
upgrade
the
plugins
I
can
just
type
Q,
cpr,
creve
upgrade
and
I
will
be
upgrading.
All
my
plugins
or
I
can
just
do
cube.
Ctl
crew
remove
our
back
view
and
then
the
plug-in
goes
away.
So
this
is
what
you
would
expect
from
a
simple
package
manager
that
would
install
upgrade
and
such
commands.
So
I'm
gonna
go
back
to
the
slides
to
explain
a
little
bit
about
what
crew
does.
B
Alright,
so
why
would
you
you
know?
Why
would
you
have
something
like
crew
to
package?
Your
you
know,
do
the
plug-in
management
for
you.
If
you
know
all
it
takes
to
add
just
binaries
to
your
path,
environment,
variable
right,
so
you're
totally
right,
you
can
totally
use
app,
gets
brew
or
go
again.
However,
you
want
to
install
plugins
or
your
machines,
but
crew
actually
gives
a
single
way
to
install
plugins
and
keep
them
up-to-date
in
your
machine.
B
Perhaps
some
they
can
go
into
queue
cuddle
in
a
form
of
plugin
or
something
like
that.
So
I
want
to
talk
a
little
bit
about.
Where
does
murder
all
these
plugins
come
from
right
so
right
now
we
have
a
packed,
centralized
package
index
that
is
great
for
discoverability,
because
all
the
plugins
are
in
one
place,
but
that's
not
so
great
because
you
need
to
enforce
certain
stuff
like
is
this
plugin
good
enough?
Actually,
like
is
this
plug-in
securities?
Are
you
trying
to
steal
people's
credentials?
Is
your
name
descriptive
enough?
B
Are
you
just
trying
to
put
you
know,
put
dibs
on
a
name
or
something
so
we're
trying
to
figure
out?
How
will
that
work
like
in
a
centralized
environment?
But
we
also
really
want
to
add
a
third
party
package
index
like
you
know,
apt,
repos,
that
you
installed,
you
add
to
your
computer
manually
or
real
taps
that
can
be
just
like
any
other
github
repo,
so
this
plug-in
index
I
want
to
give
a
quick,
quick
preview
about
that,
but
before
I
go
there
publishing
your
plugins
on
crave
is
actually
pretty
easy.
B
B
Okay,
so
this
is
this:
is
a
package
manifest
I
know
it's
a
bit
complicated,
it's
still
in
early
stages,
but
it's
been
working
for
us
fine
for
the
past
six
months
or
so,
and
it
we've
been
able
to
package
20-plus
plugins
with
this.
So
what
we're
looking
at
here
on
the
top
there
is
the
name
of
the
plugin.
This
determines
is
the
actual
name
that
the
queue
kill
command,
that
you're
gonna
have,
and
then
we
basically
specify
a
bunch
of
platforms.
B
So
in
this
example,
I
have
the
one
platform
spec
here
and
then
the
platform
that
spike
has
a
selector.
So
it
says
which
OS
in
architecture
you
want
to
install
this.
So
in
this
case
we're
installing
and
when
those
windows,
x64
machines
and
now
for
this
platform,
specify
where
do
I
get
this
plug-in
from
and
then
here
I
have
a
torval
URL.
That
is
a
Polock
URL
and
then
I
have
a
shot
that
just
verifies
that
integrity
of
that
and
then
I
in
the
file
section
I'm
describing
okay
inside
this
tar
file.
B
When
I
extract
this
tar
file,
which
files
should
I
copy
out
of
that
tar
file
to
the
install
directory
and
then
I'm
copying
all
the
bin,
slash
Exe
files
here,
as
you
can
see,
and
then
in
the
bin
path.
I'm
saying
this
is
the
entry
point
to
my
plug-in
executable,
so
we're
gonna
go
and
link
this
thing
as
coddled.
B
Our
back
look
up
Exe
in
this
case,
so
this
person
actually
developed
a
plug-in
ago
so
and
now
they're
able
to
distribute
this
plug-in
on
both
Mac
and
Windows
and
Linux,
true
crew.
So
what
doesn't
crew
do
it?
Doesn't
install
external
dependencies
such
as,
if
you're
writing
your
plug-in
and
Python,
and
you
don't
have
Python
on
your
machine.
Curry's
not
gonna,
be
able
to
do
anything
for
you.
So
far,
most
of
the
plugins
are
written
and
going
batch.
So
this
actually
helps
a
lot
helps
us
a
lot.
B
But
if
you
have
external
dependencies,
we're
going
to
just
show
you
a
post,
install
message
just
like
when
I
install
that
plug-in
earlier.
That
says
it
says
you
hey,
you
need
this
JQ
tool
or
you
need
an
Oakland
SSL
to
run
this
thing.
So
right
now
we're
relying
on
that
and
we
don't
know
what
to
do
about
the
version
skew
between
cube
shuttle
and
the
plugins
themselves.
For
example,
a
lot
of
plugins
call
back
into
cube.
Cut',
oh,
and
this
is
a
problem.
B
If
you
know
you
have
an
older
version
of
cube
cuddle
after
you
know
one
or
two
years,
and
we
don't
do
anything
regarding
security
scanning.
If
you
install
a
plug-in
today,
you
are
entirely
responsible
for
installing
a
third
party
software
on
your
machine
that
you
haven't,
fed
it
and
we're
not
letting
those
things
either,
and
we
have
a
centralized
package
index
today,
and
this
is
bringing
a
lot
of
problems
we
have
to
go
and
discuss
with
the
plugin
developers
like
we
have
to
tell
them.
Your
name
is
not
good
enough.
B
You
somebody
will
be
able
to
understand
what
your
plug-in
does
from
your
name
and
stuff
like
that.
So
and
this
won't
work
for
vendors
like
say,
Google
or
Red,
Hat
or
Microsoft
they
want
to.
They
may
want
to
distribute
plugins
because
they
will
want
there.
They
want
to
have
full
control
over
where
their
packages
come
from.
So
we
need
to
give
them
it
not
centralized
packaging.
That's
basically
so.
B
So,
in
conclusion,
if
you're
interested
in
crew,
please
come
to
the
six
CLI
meetings
we're
discussing
about,
you
know
where
to
take
the
plugins
and
how
we're
gonna
do
this
package
management
thingy
and
crystal
being
do
a
lot,
although
it's
there
ready
to
try
it
out.
If
you
want
to
contact
me,
you
can
find
me
on
Twitter
as
I
might
be,
or
I
might
be
at
google.com.
This
will
be
all
my
presentation.
Thank.
A
A
C
Hi
everybody
I
am
Aaron
Upstate
beard,
I,
am
your
114
or
at
least
lead.
Today's
cat
t-shirt
represents
my
state
of
mind
again.
I
have
no
idea.
Oh,
can
you
see
through
the
hair?
I
have
no
idea
what
I'm
doing
okay,
that
is,
that
is
where
we're
at
so
now
that
you
have
that's
pretty
much
all
you
need
to
know
really,
but
just
in
case
that's
not
enough.
I'm
gonna
share
a
Google
Chrome
window,
which
has
the
meeting
notes,
which
I
will
be
using
as
my
talking
notes.
So
we
are
at
week
one
for
1:14.
C
The
upcoming
milestones
are
that
by
tomorrow
the
end
of
week,
one
I'd
like
to
have
the
release
schedule
finalized
and
by
January
18th
the
end
of
week.
Two
I
would
like
to
have
the
release
team
finalized.
That's
all
release
shadows.
We
already
have
all
the
lead
roles
taken
care
of
and
then
looking
a
little
bit
further
out
at
the
end
of
month
that
week
four
I
would
like
for
enhancements
for
use
to
happen.
So
this
is
all
proposed
in
our
release.
C
Schedule
here,
I'll
go
to
the
files
change
section,
I
added
everybody
as
owners
and
then
I'll
go
ahead
and
view
the
file
to
talk
about
the
TLDR
is
here
so
you'll
start
hearing
warnings
about
these
in
the
weekly
updates,
but
code
freeze,
I'm
forecasting
as
March
7th
and
in
the
release
itself.
I'm
forecasting
is
March
25th.
If
you
have
any
questions,
comments
or
concerns,
this
PR
is
the
place
to
discuss
them.
C
What
are
we
gonna
do
differently
this
time
around
I've
got
it
listed
here,
but
I'm
actually
gonna
go
back
to
the
meeting
notes
to
talk
about
it,
but
I
will
also
say
that
I'm
trying
to
document
what
the
different
phases
of
the
release
are.
I,
don't
know
if
this
markdown
table
is
really
useful
as
a
consumable
artifact
that
everybody
reads
and
understands
so
I'm
trying
to
sort
of
describe
what
enhancements
freeze
means.
C
What
code
freeze
means
stuff
like
that:
okay,
going
back
to
the
community
meeting
notes,
so
in
terms
of
finalizing
the
release
team
steven
augustus
helped
put
together
a
questionnaire
on
so
bunch
of
people
have
volunteered
to
be
shadows
on
this
issue.
The
next
step
was
for
Steve
Augustus
to
put
out
a
questionnaire
and
send
that
out.
Anyone
interested
should
be
posting
on
this
issue.
C
Okay,
let's
see
here.
So
what
will
we
do
differently?
Everything
must
have
a
cap
I
talked
about
this
last
week,
I'm
talking
about
it
now
I'm
going
to
talk
about
it
at
sega
architecture.
In
about
45
minutes.
I
talked
about
it.
It's
in
tribute
or
summit
a
coupon.
Everything
must
have
a
cap.
If
you
want
something
to
land
in
this
release,
you
can't
just
have
an
enhancement
tracking
issue.
It's
got
to
have
a
cap,
even
if
it
didn't
have
a
cap
before
it
must
have
a
cap.
C
C
C
Why
am
I
blanking
on
his
name
anyway?
The
brace
yourselves
mean
it
looked
for
that
to
come
out
about
a
week
before
code
freeze
code
slush
has
historically
meant
that,
like
it's
a
bad
idea
to
try
landing
a
really
large
feature
or
a
really
large
refactoring
a
week
before
code
freeze,
because
everybody
else
who
is
trying
to
get
their
stuff
in
will
not
like
you.
So
we,
the
release
team,
have,
at
our
discretion
the
ability
to
reject
PRS
that
do
that.
C
So
we're
going
to
ask
you
all
to
be
mindful
if
you
want
a
date
to
tie
to
to
consider
yourself
like
what
should
I,
when
should
I
decide
if
I
can't
get
it
in
by
now,
it's
not
going
in
until
next
release.
Take
a
look
at
the
release
schedule.
We
start
our.
We
start
increasing
the
cadence
of
the
release
team
burned
down
meetings
a
week
before
code
freeze,
something
we've
done
in
some
releases
past,
but
we
hadn't
done
in
113
and
112
I'm
going
to
be
using
a
milestone
to
track
our
work.
Thank
you.
C
It
was
Ned
Stark.
You
said
brace
yourself,
I
appreciate
that
chat
so
I'm,
using
a
milestone
in
those
big
release.
Repost
talk
about
the
release
team's
work,
we'll
see
how
this
works
out.
I'm,
considering
doing
things
like
code,
freeze,
enhancements,
freeze,
different
builds
of
the
release
being
issues
that
are
in
here
that
we
can
link
to
when
we're
talking
about.
This
is
blocking
that
stuff.
Like
that
other
thing,
new
year,
new
team
we're
gonna
go
ahead
and
purge
the
kubernetes
miles.
C
Then
maintainer
x'
team
stevens,
taking
the
lead
on
this
and
we'll
be
going
through
this
on
friday.
The
milestone
maintains
team
is
the
team
that
lets
you
use
the
slash,
milestone,
command
and
means.
We
trust
that
you're
kind
of
generally
keeping
up
with
what
you're
sig
is
doing
or
what
is
landing
in
the
release.
C
As
such
membership
criteria
into
this
team,
are
your
estate
release,
lead
or
chair
your
release,
team
member
or
you
are
a
chair
or
technical
lead
of
a
sink
if
somebody
else
needs
to
be
on
this
team
like,
for
example,
if
you
are
a
sick
chair-
and
you
don't
have
time
for
this,
but
you
want
to
delegate
this
to
somebody
else.
Please
come
talk
to
us.
A
Thank
you,
I'll
just
go
over
the
patch
releases.
Unless
somebody
from
the
patch
release
team
wants
to
speak
Tim,
do
you
did
you
want
to
take
this
or
ping
Fei
or
okay?
So
first
thing
is
the
patch
release
team,
which
is
now
a
team
is
trying
to
actually
scheduled
patch
releases,
barring
unexpected
security
updates.
Of
course,
you
can
see
a
link
there
from
the
notes
to
a
document
where
they
are
trying
to
assemble
a
schedule,
so
we
will
have
some
more
predictability
in
the
future.
As
part
of
that
schedule,
1:13
is
due
out
today.
C
I'm
back
again:
ok
now
I'm
your
contributor
tip
of
the
week
guy,
not
your
release.
Lee!
So
I
want
to
talk
to
you
a
little
bit
about
project
boards
and
again
I'm
going
to
share
my
screen
while
I
do
this
and
I'm
gonna
use
the
meeting
notes
as
my
my
notes
and
links
to
click
thing
again:
Wow
zooms,
acting
really
weird,
sorry,
just
a
second!
C
So
I
just
wanted
to
point
out
to
you
that
anybody
who
is
an
ORAC
member
of
kubernetes
can
create
an
oral
because
projects
can
include
issues
or
PRS
from
any
repo
I
think
anywhere.
But
it's
kind
of
helpful
if
they're
tied
in
the
same
org,
because
you
can
link
to
them
pretty
nicely
so
I'll
just
walk
through
one
example.
Here
the
contributor
experience
forward
I'm
walking
through
the
example
of
their
permissions,
their
permissions
are
pretty
pretty
familiar
to
anybody.
Who's
familiar
with
repo
permissions.
C
I
do
have
to
do
this
weird
dance
through
a
couple
of
different
things
to
get
to
settings,
but
every
project
board
has
visibility
settings
and
it
has
Oracle
permissions
for
members.
So
in
this
case,
contributor
experience
has
locked
their
board
down
so
that
org
members
can
read
the
project.
They
can
see
the
issues
on
there.
They
can
see
where
the
cards
are,
but
they
cannot
move
them
from
place
to
place.
C
Next.
There
are
team
level
permissions,
so
the
community
admins
team
and
the
community
maintainer
x'
team
have
admin
and
write
privileges
to
this
project
and
then,
finally,
you
can
add
direct
collaborators
to
this
project
board.
So
caleb
has
have
been
access
to
this
board.
For
example.
This
is
really
really
similar
to
the
way
repo
Commission's
are
set
up,
so
it
seems
kind
of
intuitive
to
me
so
I
walked
you
through
sig
contributed
up
so
the
other
thing
there
is
the
order
in
which
I
walked
through
everything.
C
Higher
permissions
higher
in
the
list
override
lower
permissions
lower
in
the
list.
I
couldn't
make
somebody
explicitly
have
read-only
access
if
they
happen
to
be
an
organ
member,
and
you
set
your
board
up
such
that
work.
Numbers
have
admin
access
to
your
org
contributor
experience,
uses
their
board
on
a
recurring
basis,
their
meeting
so
start
of
every
meeting.
C
They
sort
of
walk
through
this,
and
this
is
how
you
know
what
the
sig
is
up
to,
what
it's
blocked
on,
what
it's
planning
to
do,
what
it's
actually
doing
right
now,
that's
pretty
helpful
in
their
case,
they've
locked
it
down,
so
that
only
the
state
leadership,
aka,
chairs
and
TLS
can
really
keep
track
of.
What's
going
on,
it
can
be
a
little
weird
to
have
to
look
at
the
activity
tab
to
keep
tabs
with
like
who's
done.
What,
by
the
way,
now
that
I'm
looking
at
the
activity
tab.
C
This
is
sort
of
weird
Kate,
see,
iRobot
is
doing
stuff
and
it's
got
this
little
automation
tag
next
to
it.
So
it
turns
out
if
you
go
to
different
columns.
I'll
start
with
the
to-do
column.
I
click
on
manage
automation,
github
provides
a
couple
automation
presets
for
columns,
so
it's
a
little
bit
limited
and
what
it
offers
it
only
offers
to
do
in
progress
and
done,
but
in
the
to-do
case,
any
new
issue.
C
That's
added
to
this
project
will
automatically
get
added
to
the
to-do
column,
which
is
kind
of
convenient,
I'll
start
mucking
around
with
settings
and
something
that
doesn't
happen
just
so.
You
can
see
that,
like
the
in
progress
setting,
for
example,
could
say
things
will
move
to
that
column
once
you
know,
there's
a
pending
approval
or
approval
by
reviewers
and
stuff
I'm,
not
sure
how
well
this
will
play
with
our
project.
C
Considering
that
we
kind
of
do
the
LG,
TM
and
approve
dance
out
of
band,
but
it
looks
neat
and
then
you
can
also
have
issues
like
move
to
closed
or
emerged
on
issues
move
to
a
done
column
when
they're
closed
or
when
PRS
are
merged
automatically.
This
seems
like
it
could
be
useful
for
some
of
you
who
want
to
keep
track
of
work
organically.
So
another
way
we're
going
to
try
and
do
this
is
I
talked
a
couple
times
with
a
couple
folks
about
deep,
linking
kubernetes
and
to
end
tests
at
coop
con.
C
If
you're
interested
in
this
we'd
love
to
have
you
come
help
out.
This
issue
is
the
umbrella
issue
or
the
trailhead.
For
this
it
was
something
we
talked
about
at
the
113
retrospective
and
maria
natal
and
CI
signal
lead
for
114
has
taken
a
lead
in
creating
a
project
work
which
she
could
do
because
she
just
happens
to
be
an
org
member
and
we've
added
everything
that
happens
to
have
the
area
d
flake
label
to
it.
So
we
can
chew
through
this,
to
sort
of
create
a
prioritized
backlog
and
then
figure
out.
C
What's
the
most
important
thing
to
work
on
and
any
order
who
wants
to
about
can
drag
it
to
in
progress
so
on
and
so
forth.
So,
finally,
that's
cool,
but
that's
a
lot
of
manual
clicking
around.
It
would
be
great
if
we
had
some
automation
to
help
us
out
here.
So
we
have
one
community
member
who's
working
to
see
if
we
can
add
issues
via
prowl
command
since
we're
so
used
to
slash
mile
stating
stuff.
Maybe
we
could
slash
project
something
that
seems
like
that
could
be
neat
I.
C
Think
another
really
useful
thing
could
be
the
idea
of
automatically
adding
issues
or
PRS
to
a
project
board
based
on
a
github
query
like
I
said
everything
in
the
deep
linking
thing
happens
to
have
an
area
label,
but
that's
really
just
to
help
a
human
triage
by
like
issuing
a
github
query
and
seeing
what's
not
on
the
project
board.
So
we
have
somebody
there's
some
discussion
of
how
this
could
be
done
and
we
have
somebody
who's
linked
to
PR
as
a
work
in
progress
down
here.
C
A
D
Okay,
okay,
hi
hi:
this
is
a
time
for
all,
subscribing
so
I'm
going
to
give
our
introduction
about
the
cost
gathering
here,
yeah
I
think
the
cost
can
be
made
from
from
some
patch
work
out
for
lamppost
man
here
drop
some
training
table
of
Mussolini
so
for
the
requirement,
the
default
income
and
the
chooser
scheduling
a
desired
system
ho
that
will
I
get
to
gather
it
all
together.
So
that's
install
the
word,
that's
mean,
oh,
nothing,
so
the
mo.
D
If
one
job
want
to
get
attention
CPU,
they
need
to
get
all
of
them
all
of
them
to
run
several
poles.
So
if
they
don't
care
enough,
this
house
the
whole
several
polar
will
now
to
start
it
together.
So
it's
this.
This
is
a
requirement
to
choose
the
batch
workload
because
the
this
workload
and
you
to
change
date
her
or
you
change
some
information
between
each
each
each
other,
so
they
need
to
be
starved
together.
I
think
this
is
the
original
requirement
of
the
of
the
list
key
mister.
D
When
we
try
to
discuss
our
achieve
this
target,
we
have
several
case
and,
as
they
were
discussion
here,
yeah
in
the
key,
so
here
we
go
so
it
is
a
battered
you
cost-cutting.
We
also
how
to
kids
here
now.
The
first
one
is
is:
is
cannibal
hearted
comment,
for
example,
for
MPI
job
for
them,
or
a
venir
to
tense
the
attend
CQ
and
how
to
impose
one
CPU
for
each
and
another.
One
is
something
like
spark
or
something
like
a
streaming
drop.
It
suggests
a
need
to
start
a
part
of
for
a
part
of
poll.
D
Well,
that
boy
is
just
a
need
start
of
one
to
our
work
with
part
of
it
is
cuter.
The
trouble
came
wrong
yeah.
So
this
is
a
different
case
in
the
cost
scheduling
we
also
support
of
such
candle
for
kiss
to
wrong
to
run
a
different
were
close
yeah
in
the
cow
scheduling
we
introduced
a
new
pad.
You
can
here
we
call
it
a
pod
group.
We
have
back
import
group,
we
have
post
back
and
opponent
state
her
selected,
the
other
other
tabs
in
the
whole
group
of
Spector
I.
D
Think
the
major
the
major
member
is
min
were
members
of
this
protocol
is
defined.
How
many
people
should
be
starred,
at
least
four,
to
run
the
whole
draw
as
their
read
pardon.
This
is
a
example
of
how
how
to
use
a
protocol
for
for
the
existing
tab
for
the
more
in
this
case,
I
use
a
job
as
a
example
for
you
job
tablets.
This
will
how
a
new
we
introduce
a
new
annotation.
D
D
D
So
here
is
here
is
the
the
if
some
hope
to
cover
the
cowards.
They
are
two
different
at
tap
now,
the
first
one.
If
this
a
minimal
member
is
8,
so
it's
a
ecoach
user
to
the
total
number
of
qualified
another
one.
The
minimal
member
can
be
less
than
the
total
number
so,
where
I
can
well,
we
are
going
to
make
sure
the
sakes
Apolo.
D
We
will
start
together,
so
the
other,
the
other
two
poll
will
be
on
demand
for
them
over
there
and
how
enough
resources
it
will
get,
even
if
they
don't
have
enough
that
don't
have
enough
resources.
The
other
two
poll
may
be
pending
we
talk.
Another
topic
is
that
we
needed
to
work
together
with
Kota
yeah.
Because
look
you
know
if
controller
create
poll
create
a
poll
based
on
the
color.
This
doesn't
mean
scheduler,
it's
done
mean
is
always
got
the
resource.
Yes,
in
this
case,
we
were
too
powerful.
D
Part
of
full
power
will
be
created
for
the
to
drop
because
of
the
color.
Oh,
there
is
restructuring
from
the
Cata
so
because
the
only
part
of
jog
I'll
create
so
in
the
scheduler
part.
We
cannot
do
anything
because
the
new
polar
will
not
be
created,
yeah
and
each
row
job
come
out
how
enough
pending
work,
lower
the
truth
bar?
D
D
D
The
other
one
yeah,
the
other
feature
is
that
full
some
trouble,
such
as
MPI,
if
one
holder
is
a,
is
a
field.
Oh
kill
the
best,
because
some
some
reason
the
hotel
should
me
should
be
star
because
those
risk
after
restart
one
one
poll,
the
whole
job
cannot
be
work
except
they
have
some
checkpoint
here,
but
he
in
general
we
sing
the
whole
job
need
to
be
restart.
So
we
are
going
to
also
to
introduce
a
PUD
group
controller
here.
So
we
can,
if
one
whole
field.
D
Yes,
for
now,
I
single
call
scheduling
basic
feature
of
cost.
Gabbing
is
implemented
in
cool
batch,
a
sub-project
overseeing
scattering
in
the
otaku
at
least
sold
out.you.
We
have
start
kind
of
implementation
in
crew
batch
and
for
the
polar
group
controller,
such
kind
of
lifecycle
measurement
for
them
both
but
the
to
care
so
I
described
Lobo.
It's
a
handled
by
whole
group
controller
answer
is
ongoing.
We
are
trying
to
try
to
do
that
in
you
know,
favor
it
is
yeah
another
one.
Is
a
cult?
D
A
D
I
think
just
a
while
to
page
yeah
yeah,
another
one
is
the
starvation
about
the
huge
trouble
because
of
for
now
we
we
reserve
resourceful
for
the
trout,
so
so,
let's
who
can
be
in
in
impact
their
associated
aj
utilization,
if
we
keep
back
for
some
at
some
small
tomorrow,
but
they
were
gonna,
be
a
starvation
of
the
huge
drop.
So
we
are
going
to
also
discussed
such
careful
proposed
new
confidence,
one
yeah.
No,
we
also
have
some
other
features
in
the
crew
battery
for
the
bachelor
clone.
D
A
Thank
you
very
much
and
if
people
are
interested
in
Co
scheduling
and
have
feedback
on
this
feature,
please
join
the
cig,
scheduling,
meeting
or
Channel
and
provide
feedback
to
Klaus
and
the
team
on
the
feature.
So
thank
you
very
much
for
for
your
presentation
and
bring
us
up
to
date
on
the
coast
gradually
effort.
Okay,.
A
D
E
Alright,
so
hey
I
am
Steven
Augustus
I
am
the
urban
today's
product
management
chair
going
to
give
you
a
quick
update
about,
will
be
if
it
working
on
the
last
few
cycles
and
kind
of
what
goals
are
moving
forward
in
2019,
so
just
a
few
slides
will
be
dead
last
cycle
we
also
by
the
way
we
also
have
intro
and
deep
dive
talks
from
Cuba
on
Seattle,
which
are
available
on
YouTube,
along
with
the
rest
of
the
cute
cone
talks.
So
those
links
are
down
there.
E
I
will
also
post
the
link
for
this
slide
deck
in
the
chat,
or
rather
in
the
agenda
yep,
so
yeah,
so
sick
cam.
We
kicked
out
a
survey
just
to
kind
of
query
what
was
going
on
in
the
kubernetes
community.
What
people
are
looking
to
do,
how
they're
implementing
kubernetes
the
size
of
companies?
You
know
whether
or
not
they're
using
public
cloud
private
cloud
on-premise
just
kind
of
getting
a
getting
an
idea
of
what
the
space
is.
So
we
actually
talked
about
that
in
the
sig
p.m.
E
intro
for
witches
linked
below
the
the
presentation
is
available
on
that
link.
That's
that's
right.
There
again,
I
will
I
will
post
this
in
the
agenda
the.
So
we
also
worked
on
heavily
worked
on
the
the
kept
process.
Joint
effort
between
myself,
Caleb
chase
singer
to
Mars
so
kind
of
a
sick
p.m.
sick
architecture,
mashup
so
kept
one.
A
is
the
medic
app
implementation?
So
it's
basically,
how
do
we
implement
the
idea
of
and
and
put
the
process
and
procedures
in
place
for
how
we
actually
use
caps?
E
So
we
started
off
on
that
and
around
that
we
kicked
out
a
project
board,
an
umbrella
issue
for
caps,
so
the
things
that
we
actually
need
to
do,
one
of
them
included
so
you'll,
see
below
one
of
them
included.
Actually
migrating
Mike,
Wow,
okay,
migrating
kept
away
from
the
community
repo,
so
we
did
a
huge
forklift
of
that
into
the
enhancements
repo
we
renamed
the
features
repo
into
enhancements.
E
So
for
anyone
who
is
not
aware
and
I
think
that
might
be
a
few
people
we,
we
came
up
with
this
idea
of
what
enhancements
are-
and
you
know
so.
Caps
are
essentially
created
to
capstan
for
kubernetes
and
answering
proposals,
and
caps
are
essentially
created
to
track
multi
release
efforts
right.
So
it's
it's
essentially
the
aggregate
of
multiple
features
right
things
that
you
can't
things
that
you
can't
get
done
in
just
one
really,
so
you
know
and
examples
that
might
be
something
like
a
sure
availability
zones
right.
E
So
that's
you
know
our
topology
we're
scheduling
right.
So
it's
apology,
we're
scheduling,
you
know
the
one
bringing
forth
the
idea,
and
then
you
know
to
being
able
to
implement
that
cross,
multiple
clouds
and
and
on
and
on
premise
right.
That's
not
something
that
can
simply
get
done
in
one
cycle
right,
so
so
things
that
can't
get
done
in
one
cycle.
We
call
an
enhancement
right.
So
those
you
know,
the
idea
is
that
they
would
be
bundled
as
multiple
features
right
and
those
feature.
E
So
we
have,
we
have
said
for
a
while
and
that's
in
the
documentation
which
I'm
going
to
be
working
on
changing
that
caps
are
kind
of
in
beta
stage,
and
you
know,
and
and
it's
it's
of
you
know-
of
a
six
own
solution
to
to
decide
whether
or
not
they
pick
up
the
cap
process
that
is
going
away.
Caps
are
required.
E
You
should
always
be
finally
kept
for
anything
that
has
crossed
sig
cross
project
impact,
so
we'll
also
be
discussing
that
in
architecture
in
a
few
minutes,
a
few
minutes
so
yeah,
so
the
the
kept-kept
1a
is
mostly
implemented
at
that
point.
At
this
point
you
can
check
that
out
on
the
umbrella
issue.
The
lea
enhancements
idea
has
has
come
forth.
E
We
have
started
to
categorize
the
open,
the
open
design
proposals
and
caps
as
kind
kept,
so
there's
new,
there's
a
new
label
for
that
that's
available
and
in
group
names
community
as
well
as
kubernetes
enhancements,
so
I
think
the
next
step
for
sig
p.m.
is
we're
going
to
start
revamping
the
sig
p.m.
Charter
and
the
CPM
charter
is
one
of
the
the
earlier
charters
before
there
was
an
idea
of
charters.
E
E
A
E
Going
to
start
laying
down
documentation
for
that
stuff,
we're
going
to
move
ahead
with
more
of
the
more
of
the
migrations,
Yeso
caps
to
kubernetes
enhancements.
That
has
already
happened
from
from
my
standpoint.
I
think
the
next
piece
is
to
move
design
proposals,
so
we
have
a
design
proposals
repository
that
is
in
kubernetes
community
right
now.
That
is
a
maje
of
sale
proposals
as
well
as
things
that
are
still
actively
being
worked
on
things
that
have
actually
already
been
been
implemented.
So
we
have
to
kind
of
start.
E
You
know
getting
rid
of
the
craft
over
there
and
deciding
what
we
can
and
can't
move
or
what
needs
to
be
migrated
into
kept.
So
we're
gonna
be
talking
about
that
this
year,
as
well
as
so
within
the
design
proposals
repositories
as
well
as
kind
of
scattered
around
a
few,
a
few
other
directories,
incriminated,
community
or
architectural
designs
and
diagrams-
and
we
want
to
bring
those
sorts
make
them
more
discoverable
and
and
available
to
the
community.
An
easy
place.
E
Will
that
be
urban
annex
enhancements,
I,
don't
know
yet,
but
possibly
so,
we've
got
the
capsule
ISIL,
which
we
talked
about
in
the
seriously
it
go,
watch
the
acute
context
there
I
think
they're
great,
but
we
have
a
kept
CLI
library
that
we've
been
working
on
so
hats
off
to
hats
off
to
Caleb
for
crunching,
11,000
plus
lines
of
code.
For
that
that
essentially
allows
you
to
build
a
workflow
around
around
tooling
right
so
being
able
to
propose
initialize.
The
cap
propose
the
cap.
E
Have
the
cap
approved
right,
so
the
idea
is
that
eventually
we'll
be
able
to
move
this,
this
workflow
of
like
reviewing
caps
and
generating
the
metadata
for
caps
and
figuring
out
what
the
next
kept
number
is,
or
maybe
doing
away
with
that
altogether,
bring
that
in
to
bring
that
into
automation
right
and
bring
that
into
tooling,
because
we
like
tooling,
we
like
automation
where
we're
engineers,
we
have
engineer
mindset,
so,
let's,
let's
figure
out
a
way
to
make
our
life's
life
easier
in
product
management
as
well
right.
So
eventually
this
is
a.
E
E
That's
you
know
easily
viewable
and
consumable
for
for
not
just
people
who
are
actively
working
on
caps,
but
also
people
who
are
just
stopping
by
and
want
to
understand
like
how
how
the
kubernetes
project
is
evolving
and
what
features
and
enhancements
are
coming
out
in
the
you
know
in
the
next
few
years.
Right
so
bringing
that
to
the
contributor
site
will
is
a
huge
step.
We
want
yeah
Steven.
E
Up
we're
running
low
on
time,
awesome
all
right:
okay,
cool,
so
caps
going,
hopefully
GA
in
115
that
incentive.
We
want
to
get
rid
of
the
release,
the
the
release,
enhancements
and
feature
tracking
spreadsheet,
literally
everyone
that
I've
talked
to
has
said
that
that
is
not
a
great
way
of
doing
things.
I
agree
as
someone
who
has
done
that
before
and
bring
forth
bring
forth
cross
sig
project
tracking,
so
release
planning
events
as
well
as
cross
project
roadmaps.
E
A
F
About
we
are
responsible
for
all
the
components
that
and
adjust
your
pods
that
infrastructure
to
match
the
current
needs.
So
in
particular,
we
are
responsible
for
horizontal
path,
autoscaler
cluster
autoscaler
and
the
new
vertical
path
autoscaler.
So
what
has
been
going
with
this
components?
Residency
in
the
couple
customers?
We
significantly
improved
the
algorithm
of
horizontal
path
of
the
spinner,
so
now
it's
significantly
faster.
If
she's
react
faster,
it
should
not
have
any
block
period.
So
as
soon
as
your
spike
comes
in,
we
should
add
you
more
pods.
F
In
the
next
couple
months.
We
will
expand
the
HP
API
so
that
we
will
be
able
to
adjust
the
speed
of
scanning
hub
and
the
speed
of
scanning
down.
We
have
a
community
member
that
is
working
on
the
proposal
right
now
and
should
be
a
purple
into
the
public
zone
in
cluster
autoscaler.
We
are
basically
catching
up
with
all
of
the
scheduler
new
features
and
bugs
we
have
been
working
with
basic
scheduling
on
resolving
all
of
the
problems
with
priorities.
They.
F
However,
there
are
a
couple
of
pending
issues
that
means
that
need
fixing
vertical
auto
scale,
graduated
to
better
it's
available
in
kubernetes,
/
autoscaler
repository.
So
please
go
and
try
it
one
more
thing:
contrast
autoscaler.
We
have
one
new
cloud
provider
available
algobit.
Also,
if
you
are
using.
F
Also
give
it
a
try
on
the
organizational
note,
Solly
Rose,
who
has
been
a
cyclic
since
very
beginning,
decided
to
step
down.
So
I
would
like
to
take
a
moment
to
thank
him
for
all
the
contribution
he
made
during
the
last
years.
We
haven't
just
selected
a
replacement
yet,
but
we're
working
on
it.
Just
a
small
reminder.
We
have
a
meeting
every
Monday
early
in
the
morning.
Pst
is
please
strike.
Rewards'
different
timing
was
the
right
time
in
your
time
zone.
A
G
All
right,
everyone
see
the
slides,
yes,
okay,
I'm
assuming
you
can
hear
me
because
you
responded
hi,
I'm
Bowie
from
cig
network
to
give
a
community
update
like
this
short
and
quick.
So
we
have
currently
several
projects
in
progress
on
some
for
114,
and
these
are
things
that
have
an
implementation,
have
a
cap
and
basically
are
moving
from
one
stage
to
another,
so
ipv6
support.
G
We
have
a
cap,
that's
open!
That
needs
work
to
move
it
forward
to
an
implementation
of
a
dual
stack
implementation
for
supporting
both
ipv4
and
ipv6.
We
have
a
cap
out
for
no
local
and
topology
where
services
also
we're
looking
at
revamping
some
pieces
of
service
and
endpoints,
making
them
more
scalable,
making
the
interface
more
rational,
I
think
we
were
called
out
by
Brian
grant
in
the
contribs
Summit.
So
that
was
definitely
a
pointer.
G
We
have
a
sort
of
a
proposal
out
for
specifying
how
multicast
will
work
with
kubernetes
and,
of
course,
supporting
Windows
is
going
to
be
a
big
thing
that
cuts
across
all
the
networking
stuff,
and
all
of
these
are
in
more
detail
in
the
cube
con
community
update,
which
is
in
a
2018
slides
in
terms
of
where
to
find
us,
we
have
biweekly
meeting
every
other
Thursday.
The
links
are
there,
we're
also
on
cig
network
slack
and
I
talked
to
some
people
at
the
contribs
Ummat,
and
they
were
asking.
G
A
Wow
well,
thank
you.
That's
a
quick
summary
of
particularly
given
I,
particularly
given
dual
Network
the
so
thank
you
very
much
for
bring
us
up
to
date
on
what
cig
network
is
doing
from
here.
We
have
a
few
announcements
which
I
would
get
to
if
Google
Docs
hadn't
is
frozen
on
me.
There
we
go
so
Steven
to
summarize
the
purge
of
milestone,
maintainer
x'.
C
C
E
C
Christoph's,
not
here
I,
now
I'm
wearing
that
Kristoff
hat
as
a
member
of
a
github
management
sub
project,
which
is
part
of,
say,
contributor,
experience
and
is
populated
by
six
admins
one
of
the
things
we
try
to
do
to
make
our
lives
easier,
so
we
can
service
you
hundreds
and
hundreds
of
people
is
to
use
automation
to
manage
things
today.
The
automation
supports
automatically
managing
our
membership,
we'd
like
to
do
the
same
thing
for
team
membership.
C
What
this
means
is,
if
you
are
used
to
being
a
team
maintainer
and
using
the
github
UI,
to
add
people
to
your
team
by
clicking
things
that
will
no
longer
be
possible
or
actually
what's
gonna
happen
is
when
we
run
our
tool,
it's
going
to
remove
people
from
teams
who
aren't
in
a
file
that
says
they
should
be
in
that
team.
We
probably
have
some
work
to
do
around
documenting
this
process.
We
apologize
if
this
is
bumpy,
but
please
come
file
issues
on
kubernetes,
slash
org.
C
A
Paris
is
reminding
us
that
the
CFP
for
khoob
khun
barcelona
closes
on
January,
18th
I,
believe
the
one
for
Shanghai
closes
on
January
31st,
so
new
year,
new
coupe
cons,
even
more
coupe
cons,
there's
also
a
recently
scheduled
I
think
there's
a
reasonably
scheduled
coupe
day
in
the
India,
but
I
don't
actually
have
the
schedule
for
that.
So
look
for
it.
The
I
were
revamping
the
developer
guide,
see
the
link
in
the
notes
for
that
and
next
week's
meeting
we're
going
to
have
updates
from
sig
COI
and
sig
No.
A
So
if
you
are
a
member
of
one
of
those
SIG's
I,
please
take
that
back
to
your
cig
meeting
and
prepare
something
for
next
week.
That
would
be
great
and
then
the
other
thing
is.
We
only
have
demos
for
this
meeting
scheduled
through
the
rest
of
January
and
as
a
result,
if
you
are
working
on
a
project
in
association
with
kubernetes
or
a
special
new
feature
or
a
product
that
encapsulate
kubernetes
or
something
pleased
valent
ear
to
do
a
demo.
A
We
want
to
know
what's
going
on
in
the
ecosystem
I
and
get
yourself
in
the
schedule
and
then
last,
but
certainly
not
least,
shout
outs.
Although
I'm
going
to
point
out
again,
if
you
have
a
shout
out
to
somebody
who
did
something
cool
in
the
kubernetes
community,
that
you
want
to
bring
attention
to
which
you
should
do,
because
with
lots
of
cool
people
who
do
cool
things,
please
post
it
in
the
shout
outs
channel
in
slack,
I,
think
so
shout
outs
and
discuss.
A
Isn't
there
any
way
so
that
we
can
include
that
this
week
we
actually
only
have
one
shout
out
coming
from
both
myself
and
from
Paris
to
Noah
Kantrowitz,
both
for
making
Linux
making
last
week
in
kubernetes
development
I
continue
to
publish
over
the
holidays
and
I
for
fielding
contributor
questions
over
the
holidays,
I,
both
of
which
he
continued
to
do
well.
A
lot
of
the
rest
of
us
were
mediating
family
problems.