►
From YouTube: Kubernetes WG K8s Infra 20181128
Description
GMT20181128 163456 k8s infra 2074x1158
A
A
A
The
other
way
to
look
at
it
so
I
think
we
I
added
a
a
project
as
well.
Cuban
artists,
interest
team,
okay,
I'm
gonna
paste
the
link
in
the
zoom
chat,
so
we
have
three
umbrella
shoes,
so
we
can
look
at
those
above
all
of
all
three
of
them
are
in
the
backlog.
That's
another
one
way
to
look
at
it.
The
other
way
to
look
at
it
is.
A
A
B
Canaries
are
in
last
time
we
talked
hippy
was
going
to
be
working
on
a
test
for
the
canary
I,
haven't
seen
a
PR
there.
If
he's
he's
not
here
yet
or
maybe
not
today,
at
all
I
can
take
it
over
I,
don't
want
to
rip
it
out
of
his
hands.
If
he's
actually
interested
in
doing
it,
it
doesn't
seem
that
complicated
famous
last
words
but
I
think
that's
really
the
only
thing
that's
blocking
it
there's
a
couple
of
records
that
don't
really
have
names
attached
to
them
yet,
but
I'm
willing
to
proceed
with
that.
B
C
B
A
That's
true
a
okay,
so
maybe
we'll
update
the
issue
also
with
some
note,
with
whatever
you
are
typing
in
the
Google
Doc
Tim,
yep,
okay,
okay,
let's
close
this
out
and
the
next
one
was
GK
cluster
for
loops,
good
idea,
GK
cluster
for
BOTS
and
utilities.
So
Tim
we
wrote
a
few
things
up:
Justin
and
I.
It
is
in
the
issue:
okay,
one
five,
two.
So
what
is
so?
Basically?
Okay,.
A
E
Don't
think
so
I
think
I
was
just
trying
to
like
do
the
same
thing
as
you,
but
expressing
in
terms
of
philosophy,
rather
than
in
terms
of
like
okay
policies.
I
think
I
also
think
there
might
be
a
more
complicated
way
to
do
it
than
setting
up
fake
email
Gmail
addresses,
but
we
can
figure
that
out
like
there
is
some
sort
of
temporary
granting,
but
we
can
figure
that
out
when
we.
Let
me
ask
you,
have
something
to
try
as
well:
I
guess,
okay.
B
I
guess
I
don't
have
a
problem
creating
a
do.
You
guys
want
to
do
an
alpha
cluster
or
a
real
cluster
I'm
worried
about
accidentally
engaging
with
features
that
we
don't
really
want
to
engage
with
one
of
the
theory
of
the
Alpha
cluster
so
that
if
it
doesn't
be
destroyed,
yeah
I
mean
I,
get
that
that's
the
the
primary
benefit
of
it
and
I.
Don't
care
either
way.
B
E
A
B
So
what
we
did
with
DNS
is
we
created
a
Google
group
that
is
the
access
permission
for
club
for
whatever
the
admin,
so
we
can
create
like
a
Kate's,
infra
cluster
admins
group.
Add
your
guys
selves
to
it,
and
then
I
can
add
that
group
with
permissions
for
cluster
admin
or
cluster
whatever
cluster
or
edit
or
cluster
owner.
Whatever
it's.
C
B
B
That's
good
progress
so
far.
The
other
thing
that
we
want
to
come
up
with
is
the
billing
report
right
so
far,
we're
not
actually
doing
anything
that
needs
billing.
Once
we
turn
on
DNS
for
real
it
will,
but
it's
so
small.
It
won't
really
matter.
The
cluster
will
be
the
first
thing
that
actually
starts
to
spend
real
money
and
so
like
pretty
quickly.
I
want
to
think
about
how
we
can
figure
out
the
billing
on
it.
D
Let
me
work
on
it,
so
we
probably
can
can
also
create
something
like
the
group
of
Baron
Edmonds
wisdom
billion
management.
Permissions
like
follow,
follow
almost
the
same
ways.
We
have
it's
always
in
fragments
and
I'll.
Take
a
look.
How
can
we
create
a
billion
reports
and
send
them
to
two
dead?
Okay,
I'm.
A
B
A
A
B
You
know
anybody
else,
I
mean
we
can
start
with
that,
I'm
more
interested
in
figuring
out
the
actual
mechanism
that
we
want
to
use
to
generate
reports
or
dug
very
deep
into
you,
know:
billing
exports
and
data
studio
and
those
sorts
of
things.
So
we
should
just
if
you
just
spend
time
figuring
out
what
we're
actually
capable
of
doing
and
what
we
think
is
going
to
be
satisfying
to
the
larger
community,
because
we
really
want
to
be
transparent
about
this
and
I.
Think
that's
a
first
start.
A
B
B
A
A
Then,
let's
go
back
to
the
previous
meeting
and
see:
if
there
is
anything
so
did
we
do
the
work?
The
Charter
thing
anybody
remember,
opening
up
the
pull
request,
it's
still
open.
So
the
action
item
here
is
Erin
and
me
go
back
and
fix
this
up.
I
think
there
was
a
steering
meeting
where
there
was
some
feedback
about
it's
too
much
like
a
workgroup.
We
need
to
too
much
cig
like
so.
A
A
A
E
F
Can
you
hear
me
yeah,
yeah,
no
I,
don't
think
that
there
is
I,
don't
think
we
can
do
a
tradition,
so
I
think
we
would
have
to
discern.
You
know
these
are
the
Blessed
images
like
these
are
the
ones
that
we,
yes,
the
smaller
and
just
a
tent-like.
How
do
we
attest
it,
though?
Is
there
any
automated
only.
B
F
F
B
F
F
E
Is
a
there
is
a
trick
which
is
if
we
have
a
name,
an
image
name
whose
tag
is
the
Shah?
Oh
did
I
freeze?
No.
If
we
everybody
whose
tag
is
the
shot,
then
we
can
use
the
same
sort
of
binary
redirector
to
pull
from
mirrors
the
sha
axis
validation
on
the
all
the
way
down,
and
then,
if
we
wanted
to,
we
could
then
use
the
same
redirector
service
to
do
a
named
tag.
E
Resolution
so
we
would
have
a
canonical
location
for
the,
in
the
same
way
that
we've
a
canonical
location
for
artifacts.
We
could
have
a
canonical
location
which,
from
which
you
can
pull
images,
but
it
would
still
redirect
to
a
in
theory
anywhere
in
the
world
that
we
wanted
to.
So
we
could
reuse.
What
does
that
work?
Does
dr.
flow
respect
for
you?
I
have
looked
guests,
octuple
does
because
it
all
goes
to
CloudFlare.
E
E
The
the
piece,
a
actually
thought
was
gonna,
be
complicated.
Is
that
my
understanding
is?
We
want
each
project
to
be
able
to
push
docker
images
without
granting
cross
permissions.
In
other
words,
I
can't
push
to
the
DNS
project
from
cops,
but
we
we
believe
we
would
like
to
have
a
single
source
of
binaries.
So.
B
Yeah
I
think
we
need
to
figure
out
whether
we
have
sort
of
a
one
true
push
place
where
we
just
have
a
small
number
of
trusted
people
not
to
step
on
each
other
or
whether
we
give
every
project
of
their
own
pushed
place
and
have
a
little
bit
more
lenient
trust
there
and
just
use
the
promoter
mechanism
to
move
it
from
the
the
push
place
to
the
production
place
or
places
and
I.
Think
that
those
that's
sort
of
orthogonal
to
how
do
you
do
the
auto
mirroring.