►
From YouTube: Kubernetes sig-aws 20180209
Description
No description was provided for this meeting.
If this is YOUR meeting, an easy way to fix this is to add a description to your video, wherever mtngs.io found it (probably YouTube).
A
A
Is
Friday
the
9th
of
February
2018
and
this
is
sick
AWS.
We
relatively
the
agenda
for
the
day,
lots
of
great
stuff
on
there
and
so
I
suggest
we
jump
right
into
it.
A
reminder
this
meeting
is
recorded
and
will
be
put
on
the
internet
and
everything
you
say
will
be
used
against
you.
So
with
that
the
first
item
on
our
agenda
by
when
do
you
want
to
plug
in
and
how
that's
work,
networking
how
that's
going.
A
A
C
A
Sake
it
is
the
leader
of
a
sig
is
a
is
a
burden,
not
an
honor.
It
confers
no
on
no
privileges.
It
is.
It
is
very
important
that
it
can
confers
no
like
privileges
or
anything
like
that,
like
I,
think
we're
trying
to
avoid
as
a
community
falling
into
the
ante
pattern
of
it
being
something
that
everyone
wants
to
put
on
their
CV,
for
example,
or
their
resume,
for
example.
So
their
responsibilities
are
too
maintain
order.
A
If
such
a
thing
is
a
problem,
fill
in
act
as
I
guess,
compare
or
MC
for
the
what
I'm
trying
to
do
right
now
in
Java
I
think
there
is
some
communication
type
requirements
so
that,
for
example,
during
release
released
planning
of
release
features.
You
know
making
sure
that
we
communicate
up
the
features
that
we
want
to
get
into
a
race
to
the
release
team
and
approving.
You
know
when
a
last-minute
thing
comes
up,
so,
for
example,
during
the
last
release,
Amazon
AWS
released
c5.
A
Your
wanted
one
of
the
ones
that
one
of
the
instances
that
like
needed,
special
handling
and
it
came
up
just
outside
of
feature
freeze,
and
so
we
we
sort
of
the
the
leaders
of
the
sig
coordinate.
You
know
just
triaging
where
there's
something
important,
something
is
to
the
community
and
whether
it
needs
an
exemption
from
feature
freeze
and
it's
almost
the
leaders
I
think
make
the
final
call.
A
So
it's
that
sort
of
it's
administrative
it
is
is
and,
like
you
know,
from
other
communities,
we've
seen
a
problem
where,
when
it's
too
big
of
an
honor
that
people
want
to
create
more
SIG's,
so
there's
more
leaders
and
things
like
that,
and
so
we're
trying
to
avoid
that.
It's
not
supposed
to
be
a
power
of
position
of
great
power.
It
is
a
facilitator
or
type
role.
B
B
E
So
I'm,
sorry
for
walking
late
I
think
it's
not
take
to
actually
decide
how
they
want
to
run
things
I
think
that's
kind
of
the
the
position
of
the
steering
committee
I
mean
we
are
trying
to
standardize
for
the
interfaces
stakes
to
the
rest
of
the
the
community,
but
sort
of
within
that
interface.
I.
Think
it's
up
to
up
to
six
to
figure
out
sort
of
how
they
want
to
run
themselves.
The
splitting
of
the
roles
is
not
necessarily
a
done
deal.
I.
E
Think
Brian's,
like
I,
don't
see
why
we
need
this,
but,
like
I,
think
some
other
folks,
like
I
personally,
think
that
separating
on
the
tech
lead
from
sort
of
the
organizational
lead,
whatever
we're
gonna
call.
This
thing
makes
sense
because
the
you
know
some
six
is
the
same
as
seeing
people
to
do
both
roles
and
other
other
things.
Definitely
split.
Those
other
things
out
more
opportunity
for
people
without
really.
B
D
I
would
like
to
I
would
definitely
like
to
nominate
Bob.
This
is
my
first
time
on
a
sink
all
in
several
months.
After
my
my
long
lives
time
at
Microsoft,
where
it
didn't
really
make
sense
for
me
to
join
the
Amazon
calls
and
now
I'm
sort
of
back
in
Amazon
land
again,
so
I
would
like
to
start
reconvening
to
the
Sagan
hope.
You
know
if
there
is
a
void
that
I
can
help
out
with.
If
there's
not
that
I
definitely.
A
A
F
A
F
Yeah
absolutely
I
would
happily
+1
it,
because
Bob
is
actively
going
to
be
driving
the
ek
s,
efforts
here
and
it's
really
important
that
one
of
us
stay
involved,
Bob
and
I.
We
talk
quite
regularly
some
happy
for
Bob
to
be
the
leader,
because
a
lot
of
these
discussions,
as
I
see,
are
very
engineering
driven
and
since
Bob
actually
has
those
resources
as
well.
He
has
much
better
influence
in
terms
of
you
know,
working
with
them.
So
absolutely
there
is
no
doubt
in
my
mind
that
bob
is
the
right
person
to
lead
CB
AWS
here.
A
Savannah
and
if
anyone
else
wants
to
nominate
themselves
or
nominate
others
or
do
so
privately,
then
you
know
feel
free
to
reach
out
to
I.
Guess
any
of
the
the
three
of
us
well
Chris
Justin
on
sack
or
whatever
it
means
you
want.
I
mean
it
doesn't
have
to
be
whatever
you
feel
comfortable,
I.
Think
it's
very
a
good.
C
Way
to
do
it,
I
really
happy
to
help
with
you
know.
Some
of
the
scheduling
stuff
I
might
be
busy
for
the
next
month
or
so,
but
yeah
I
I
would
be
happy
to
help
out
with
scheduling
in
terms
of
the
regular
meetings.
Awesome.
E
B
B
E
B
D
That
brings
up
a
really
good
point,
which
is
some
of
the
clerical
items.
As
a
single
you
I
think
the
Google
Doc
was
under
my
private
domain.
I
think
I
own
the
calendar,
invite
we
should
sort
of
revamps
look
at
things
like
passwords,
look
at
things
like
the
Google
Doc
and
give
like
all
of
our
paperwork
kind
of
a
once-over
so
that
it
fits
community
standards.
Okay,.
B
A
Okay,
okay,
so
yes,
I
think
when
why
don't
we
crisp
oven
myself
with
one
we
conference
afterwards
and
figure
out
like
whether
we
send
an
announcement,
this
AWS
mailing
list
and
use
the
Lazy
consensus
process,
which
is
I,
think
what
happened
was
they
go
off
and
my
guy
I
think
have
to
figure
out
whether
you
want
to
be
a
lead
or
a
leader
in
training
or
I?
Don't
think
he
can
be
a
distinguished
member
yeah?
A
H
H
Everybody
so
here
like
to
talk
about
some
proposal.
We
have
is
basically
to
unable
cube
scheduler
to
manage
the
node
available.
Ip
addresses.
So
you
see
in
the
problem
sections.
We
have
this
problem
that
today
the
cube
scheduler
is
not
aware
of
number
available
ipv4
addresses
on
the
node.
So,
even
though
we
ran
out
of
the
IP
address
on
the
node,
the
cube
schedule
can
still
schedule
apart
until
node,
so
node.
Basically,
the
particular
part
basically
goes
to
the
crash
loop
because
we
don't
have
you
know
IP
address,
so
that's
the
problem.
H
We
try
to
solve
and
there's
issue
tracking,
so
I
think
this
issue
here.
This
is
the
original
issue
for
this
problem.
So
the
proposal
we
have
here
is
basically
we
want
to
use
the
kubernetes
extended
resources
which
is
support.
Part
of
the
community
1.8
feature
to
enable
the
you
know
to
let
the
cube
scheduler
to
manage.
H
You
know
to
manage
the
resource
of
this
particular
resource,
so
we'll
define
a
new
resource
name.
Vp
3
pcs
are
Amazon
AWS,
comm,
/apb
4
and
for
all
the
part
which
do
not
use
a
host
networking.
They
will
specify
this
particular
one.
The
VP
see
ipv4
addresses
to
1
and,
and
the
schedule
will
do
the
accounting
for
these
and-
and
here
is
a
over
overview
of
the
flow,
how
it
works.
The
first
thing
is
we're
going
to
introduce
a
Nrap
controller
component.
H
What
it
does
is,
it
will
watch
the
node
object
in
the
class
and
when
the
node,
when
a
worker
node
joins,
and
it
will
get
a
new
watch
now,
a
new
note
notification
and
and
then
he
will
update
the
node
resource
with
based
on
the
today,
based
on
the
instance
type.
We
know
the
maximum
number
of
EPC
addresses
available
for
this
node
and
will
update
the
the
node
with
this.
H
H
You
come
into
the
right
spot.
Yes,
so
so
what
happened
is
today?
Oh
the
pod.
You
know
they
won't
have
these
particular
spec.
So
how
do
we
inject
us?
So
the
idea
is
we
like
to
have
a
initializer
so
basically
for
the
wire
for
watch
for
pod,
object--
or
deployment
object,
so
user
specify
a
pod
or
different
payment.
We
will
get
a
notification
and
the
initializer
will
inject
these
injector
part,
a
small
part
with
this
specification
to
we
inject
this
smaller
containers
back
into
the
pot.
H
Here
are
a
few
other
alternative
solution
considered,
but
we're
not
really
going
forward
with
those
you
know,
I
think
the
first
one
we
went
to
should
we
have
a
custom
scheduler
for
this
I
think
we
you
know
the
pro
is
we
can
do
whatever
we
want,
but
then
there's
so
many
other
thing
we
have
to
do
so.
The
comm
is
more
bigger
than
the
pro
and
the
second
one
is.
We
thought
about
how
about
because
today
and
every
worker
node
we
have
a
long-running
daemon.
H
We
call
a
local
IPM
agent
and
that
one
is
basically
aware
of
number
available
current
available.
Ip
v4
addresses.
So
we
could
have
that
one
ten
to
the
note
when
they
run
out
the
ipv4
addresses,
but
the
problem
is
they:
maybe
they
maybe
have
some
base
condition,
so
you
could
have
scheduler
already
scheduled
a
part
on
the
way
to
the
node
and
in
the
meantime,
we
try
to
tense.
So
I
still
have
some
corner
cases
not
covered
the
third
option.
We
also
winds.
H
Basically,
due
to
the
to
prioritize
to
see
whether
the
part
can
be
scheduled
on
the
node-
and
one
thing
we
we
thought
about
is
it
is,
it
can
be
a
performance
hit,
because
every
part
you
have
to
call
twice
to
the
extent
of
web
hooks
and
plus
another
thing.
This
is
this:
extender
have
to
do
also
do
the
managing
and
perform
counting
for
the
number
available,
IPV
source,
and
so
does
that's.
We
have
considered
I
think
and
also
for
the
issue
here.
There's
also
some
discussion
about
what
about
why?
H
E
E
The
only
other
comment
is
that
initializers
are
in
alpha
and
I.
Think
more
effort
has
been
going
into
mutable
admission
controllers
these
days
instead
of
initializers
and
then
also
could
prevent
like
if
somebody
goes
and
tries
to
say
change,
one
of
these
resource
requests
or
limits
to
two
like
it
I
assume
you
like,
say,
hey
link.
If
somebody
tries
to
set
these
things
like
and
you
little
slam
it
back
down
to
one
or
you'll
reject
that
change,
or
something
like
that.
You
can't
do
that
with
initialize,
but
you
can
do
that
with
with
admission
controller.
A
G
E
A
A
Do
you
know
how
that's
the
counterparty
that's
a
very
similar
type
of
thing,
it's
in
the
scheduler
and
there's
a
hard-coded
limit
it.
Doesn't
it's
I
think
it's
what
caters
like
40
or
something
like
that
and
I
think
it's
over
rideable.
So
if
you
want
to
run
more
because
you're
crazy,
you
can
do
that.
Okay,
but.
E
It
assumes
that
so
some
of
this
is
that,
like
not
all
nodes
are
created
equal,
so
some
actually
have
these
more
EBS
volumes
or
whatever
attached
versus
others.
So
this
really,
you
know
so
so
I
think
it
might
be
worthwhile
I
also
talked
to
the
scheduler
folks
because
they
might
be
looking
to
be
like
how
do
we
actually
sort
of
tear
off
this
hat
can
go
to
something
more
general,
because
there
is
between
networking.
This.
G
E
I
think
you
know
this
definitely
seems
like
the
same
plan
to
get
things
up
and
working
right
now,
but
it's
definitely
worth
starting
the
conversations
with
those
folks
to
see
if
there's
sort
of
you
know
we
can
solve
this.
For,
for
you
know
everybody
and
and
then
maybe
start
looking
at
pod
level
limits
and
resources.
Okay
and.
A
E
A
B
F
B
E
I
think
it's
totally.
This
is
all
in
the
open.
This
stuff
is
all
like
out
there.
Anybody
can
read
it
I
think
that
you
know
that's
appreciated
by
folks.
The
solution
here
is
very
specific
to
the
Amazon
CNI
plugin,
and
so
that's
totally
great
I
think
you
know
taking
this
and
going
to
something
like
cig,
networking
or
say
scheduling
and
I
don't
participate
in
those.
So
I,
don't,
like
you
know,
have
a
lot
of
geek
knowledge
there
and
say
like
here's.
E
How
we're
looking
at
solving
this
for
us
is
there
a
general
problem
that
we
can
help
with,
and
people
like
yeah,
that's
great,
then
it
may
be
worthwhile
to
put
together
a
temporary
design.
Doc
can
work
through
there
and
you
can
just
like
copy
and
paste
a
bunch
of
stuff
over
but,
like
I,
think
here's
like
you
know
we're
trying
to
use
the
extension
mechanisms,
but
we
think
that
we
can
do
some.
E
A
E
E
E
So
so
so
the
first
one
is
that
we've
been
trying
over
time
to
actually
better
align
our
code
structure
with
our
people
structure.
So
one
of
the
overriding
goals
is
we
want
to
make
sure
that
every
piece
of
code
that
is
as
part
of
the
kubernetes
project
has
some
sort
of
you
know,
owner
in
terms
of
a
safe
and,
and
sometimes
those
mappings
are
obvious,
other
times,
they're
less
obvious,
and
so
we're
trying
to
get
a
clean
map
of
this.
E
You
want
to
essentially
make
this
be
directional
between
some
of
the
community,
repo
and
stuff
in
other
files.
There's
a
there's
a
first
cut
in
this,
but
I
think
there's
there's
need
for
six
to
actually
claim
over
ownership
of
all
the
sort
of
bits
of
code
that
are
spread
across
all
the
organs
and
repo.
So
that's
thing,
one
and
thing
two
is,
and
this
is
Brendan's
proposal,
but
but
we
all
think
it's
pretty.
E
Sane
also
is
to
sunset
the
the
incubation
process
for
kubernetes
and
replace
it
with
something
where
we
have
three
different
types
of
code:
repos
there.
There
are
things
that
are
in
the
kubernetes
organization.
That
are,
you
know
core
to
some
definition.
Of
course,
we're
gonna
hold
those
things
to
certain
standards
in
terms
of
the
sort
of
breakfast
responsibilities
and
automation
and
stuff
like
that.
We're
not
going
to
take
anybody
out
of
that
organization.
You
know
at
this
point.
E
This
is
really
all
about
moving
forward,
so
we're
not
going
to
like
you
know,
go
through
and
try
and
try-
and
you
know
totally
rationalize
everything.
So
that's
the
first
thing
is
the
things
that
are
in
the
kubernetes
and
typically
those
things
will
be
things
that
go
into
the
domain.
Kubernetes
release
and
tooling
and
stuff
around
that.
E
That's
kind
of
a
fuzzy
thing,
I
say
Gorka
texture
is
the
ones
are
that
are
going
to
improve
intensity.
Okay,
I'm
missing
here:
okay,
yeah,
sorry,
life
in
a
start-up
crazy,
so
the
Sigma
architecture
is
the
one
that's
the
folks
are
gonna
prove
new
projects
in
the
in
the
main
organization.
The
next
thing
is
that
we
have
this
new
Ward
called
kubernetes
SIG's
and
in
any
any
city
can
create
new
code
there.
E
Also,
alright,
so
cloud
providers
would
stake
or
I
think
that's
a
separable
issue,
I
think
I'm
not
totally
up
on
that,
but
I
think
right
now
the
you
know
moving
those
things
to
well,
obviously,
for
the
cloud
providers
are
owned
by
the
cloud
specific
States
at
this
point
adds
they
can
move
out
to
be
separate
binaries,
whether
those
things
are
shipped
as
part
of
the
release
or
whether
they're
shipped
out
of
band
or
whether
they're
owned
by
the
Stig
or
I.
Don't
think
anybody's
really
really
made
any
concrete
decisions
on
that
I
think.
A
That's
part
of
the
broader
like
what
is
a
release
of
kubernetes.
Do
we
do
it
every
three
months
to
to
continuously?
Do
it
every
LTS,
all
that
discussion,
which
is
yeah?
That's
where
baby
steps
all
right,
where
you'll
get
there
like
I,
think
for
the
first
we
know
at
some
stage
we're
gonna
split
out
the
core
code
into
a
different
repo,
so
it
won't
be
in
the
communities
for
the
date
it
was
time
provided,
won't
be
in
the
company's
communities.
Reaper.
A
Whether
we
go
further
is
future
future
guesses
yeah,
so
the
I
think
I
think
the
I
guess
the
two
action
items
that
I
wanted
to
whether
talked
about
it.
You
don't
have
to
this.
Doesn't
you
know
one
is
required
to
do
anything
they'll.
Give
us,
for
example,
is
welcome
to
keep
that
you
know
I
plug
in
and
you're
welcome,
to
keep
whatever
you've
been
working
on
separate.
A
So
I
think
that
should
be
uncontroversial
or
we
should
well
I,
see
some
smiles
or
we
should
have
like
blogs
going
to
cloud
watcher
or
whatever
it
is,
or
an
Etsy
facade
on
DynamoDB
might
be
funny
right.
You
know
crazy
ideas
and
whatever
it
is
that
people
want
to
do
and
I
think
I
think
if
people
are
interested.
You
know,
if
there's
a
group
of
two
or
three
people
that
want
to
work
something.
Then
we
can
find
each
other
chatting
together.
A
Yes,
I,
don't
think
we've
entirely
defined
the
procedure,
but
my
guess
is
we'll
send
it
to
the
mailing
list
some
sort
of
announcement
and
operate
on
lazy
consensus.
If
someone
says
no,
this
is
the
worst
idea
ever
then
we
will
rethink,
but
otherwise
I
think
it's
relatively
relatively
easy
for
us
to
then
spin
up
projects
and
collaborate
on
them
in
that
venue
and
then
there's
a
question
in
chat.
Those.
A
I
hope
it's
like
so
I
mean
we
have
to
do
kms
integration.
As
far
as
I
know,
there
is
no
AWS
kms
integration
at
all,
so
that
would
be
one
which
in
my
mind,
when
I
wrote
it
down.
I
was
like.
Oh
that's.
The
one
we
should
do
is
like
our
first
test
bed
because
I
don't
know
of
anyone,
that's
doing
it
and
it's
should
be
uncontroversial,
unlike
some
of
the
other
ones.
A
E
G
A
A
I
mean
well
I
mean
what
I
was
proposing
and
is
literally
just
like
ideas.
That's
what
my
headstrong
man
is
it's
use.
If
we
can
have
two
or
three
people
that
are
interested,
so
they
find
each
other
somehow
right
so
talk
about
a
key
or
talk
about
on
slack
and
then
I
I
feel
like
we
should
do.
A
sort
of
announce,
like
incubator,
does
to
the
city,
West
mailing
list
and
I
think
I
think
we
should
operate
on
lazy
consensus.
D
I
just
think
it
would
be
good
if
we
like
have
the
same
bits
of
data
for
each
one
of
these
sub
project
proposals.
You
know
like
what
is
it
solving?
What
is
the
high-level
view?
Why
do
we
need
it
and
just
sort
of
making
that,
like
at
least
get
that
written
down
somewhere,
so
that
people
can
see
it
so
that
we
all
kind
of
understand
what
these
things
are
before?
We
have
a
chance
to
say?
D
A
B
E
First
of
all,
the
cap
doesn't
have
to
be
that
heavy
second
of
all
is
that
inside
the
cap
directory
feel
free
to
create
a
cig,
AWS
directory
and
create
an
owner's
file
there,
and
so
you
can
use
that
as
much
or
as
little
as
you
want
for
internal
business.
For
the
cig.
There's
no
sort
of
must
be
that
internal
business
has
to
go
through
that,
but
if
it's
useful
then
then
go
for
it.
E
E
So
you
can
still
move
forward.
If
you
have
somebody
that
is
objecting
the
big
question
there
is
who
gets
to
vote?
Who
has
standing
and
that
comes
down
to
sort
of
membership
with
a
cig?
And
so
you
know
that's.
You
know
all
of
a
sudden
you're
down
this
rabble
of
governance,
we're
trying
to
sort
of
tease
through
some
of
those
things
with
with
the
the
sort
of
state
charter
stuff
and
help
people
wrap
their
heads
around
that,
but
hopefully
that
won't
be
necessary
register
to
vote.
E
I
After
some
of
these
get
on
boarded
and
just
kind
of
expect
out
of
Peru
by
the
community
as
far
as
visibility,
how
does
that
work
as
far
as
I
think
that'd
be
pretty
to
recruit
people
to
bring
them
in
and
get
them
start
on
that
project?
Would
that
just
be
covered
in
this
meeting,
or
should
we
push
people
towards
doing
something
like
that.
A
I,
don't
think
we
know
yet
that
we
would.
We
would
certainly
I
think
that
there
will
be
a
bias
for
sakes
to
discuss
projects
that
are
inside
their
sig
umbrella,
SiC,
sub
projects
versus
ones
that
are
not
and
I.
Think
I,
certainly
with
you
that,
like
it,
is
easy
people
contributors
have
found
it
easier
to
contribute
to
other
two
sub
projects
into
the
kubernetes
community's
repo,
and
so
we
should
encourage
that
and
I
would
like
to
see
us
be
encouraging
of
projects
and
we'll
have
to
figure
out
a
process
by
which
we
will
retire
them.
I
Something
absolutely
I
think
I
think
that,
and
just
having
a
tool
like
that,
or
just
at
least
a
place
to
point
people
to
I,
know
there's
a
lot
of
food.
I
know
a
bunch
of
people
that
would
want
to
pay
on
a
bunch
of
these
things
as
they
come
up
and
so
just
a
good
kind
of
meeting
place
to
triage
those
and
engage
interest.
It
would
be
really
helpful.
I
know.
A
C
cluster
lexical
has
a
list
at
the
top
of
their
meeting
notes
of
their
sub
projects
sub-groups,
and
so
we
should
do
this
working
groups.
We
should
do
something
similar
for
the
CEOs
think
it's
not
least
cuz.
It
serves
as
an
organizational
jumping
off
points
you
want
to.
You
know,
keep
searching
from
various
docks
and
random
places
anywhere
notice.
We
have
a
lot
on
the
agenda
now
so
I
think.
If
there's
nothing
else,
we
should
probably
move
on
yeah.
G
A
J
A
J
H
J
J
J
Right
and
let's
say
it's
hello
worlds,
argument
one
argument:
two:
if
I,
if
I
daughter
eyes
that
you
know
I
can
still
pass
arguments
into
my
doctrine,
container,
yeah,
but
anybody
if
I
put
that
in
a
pod
as
a
as
a
docker
running
in
a
kübra
needing
slave
I,
can
no
longer
pass
their
arguments
into
the
pod
in
that
the
pod
is
not
what
configurable
or
I
can't
pass
arguments
to
a
pod.
Okay.
So
if
you're.
E
J
An
example
that
I,
you
know
tells
the
pod
that
they
can
have
arguments.
However,
in
practice,
when
I
inspect
the
results
inside
the
container
or
inside
the
pod,
I
wouldn't
accept
them.
So
I've
I've
included
a
descriptor
of
a
pod
with
arguments,
but
I
found.
You
know,
with
with
actual
testing
that
I
can't
actually
pass
the
arguments
to
the
pod,
even
though,
even
though
we
described
in
the
pod
file
pod
yeah.
A
J
D
J
J
J
A
J
A
J
A
There's
definitely
a
lot
of
people
doing
server
lists
own
communities
itself,
I,
don't
know
if
I
don't
know
if
anyone's
doing
lambda
with
kubernetes
I
think
that's
what
I'm
suggesting
maybe
in
the
future.
That
might
happen.
I
just
don't
know
yet.
Oh
well,
I
think
I
think
your
your
github
is
quantum
fusion
right.
So
presumably
people
couldn't
follow
you
on
there
and
hopefully
find
you
when
they
want
to
do
that
right,
Zoolander
with
kubernetes.
A
J
A
I
think
I
think
that
means
I
think
that
you
know
that
this
is
I,
don't
think
anyone
knows
it
seems,
and
so
let's
I
think
you
may
be
the
furthest
along
and
if
there
are
people
that
also
want
to
do
this
I
think
they
know
how
to
get
in
touch
with
you.
So
thank
you
for
those
resources,
though
it's
great.
A
K
Right,
you
know
yes,
all
right,
so
I
already
sent
out
a
issue
and
a
dr.
slack,
so
I
got
plus
people
read
it
mica.
Thank
you
and
a
few
other
folks,
so
I
just
want
to
throw
it
out
to
this
group
before
I
submit
a
pull
request
to
make
sure
that
this
doesn't
already
work
somewhere
else
being
able
to
this
is
be
able
to
especially
specify
an
I
am
role
that
we
use
to
talk
to
AWS.
K
A
To
me
at
least,
this
seems
like
a
good
thing
to
do.
I
feel
like
the
is
a
good
stepping
stone
to
I,
think
where
we
want
to
go
and
the
the
cloud
configuration
file
is
relatively
sort
of
like
a
back
door
with
a
lot
of
things
in
it
that
are
relatively
specific
and
I.
Think
we
actually
support
very
similar
things
already
in
there
and
most
people
it's
one
of
those
things
or,
if
you
don't
need,
you
have
to
touch
it
so
I,
don't
know
how
other
people
feel.
K
Into
work
plugs,
no,
this
is
run
like
controller
manager
with
a
specific
I
am
role
so
that
whenever
AWS
api's
is
gonna
use
that
role
instead
of
the
the
role
of
the
ec2
instance,
that's
running
on
okay,
so
you
would
specify
this
with
secrets,
then,
or
something
like
that.
You
just
need
to
pass
in
the
name
of
role.
So
as
long
as
we're
the
a
the
cameras
running
as
long
as
ice
permissions
to
be
able
to
assume
that
role,
then
it
they're
all
a.
A
And
we
actually
already
have
support
for
like
running
the
control
plane
in
a
different
I
think
account.
But
you
know
we
certainly
have
this,
but
I
think.
Currently
it
relies
on
you
passing
in
the
hard-coded
access
key
or
the
static
access
key
and
secret
access
key,
and
so
this
would
be
an
improvement
over
that
it's
existing
functionality,
we're
making
more
secure
and
easier,
and
it's
not
required.
C
So
a
quick
question
about
this:
technically,
if
you
wanted
to
and
you're
using
something
like
cube,
do
I
am
you
could
if
you
were
running
this,
the
controller
manager
in
a
pot
itself?
You
could
do
this
sort
of
thing
already,
but
cube
Diane
would
be
doing
it
for
you.
This
is
just
introducing
that
sort
of
SCS
functionality
into
the
the
controller
itself.
Yeah.
K
E
Just
a
minute,
sir,
on
this
odd
stuff,
the
there's
work
going
on
with
sig
off
and
the
container
identity
working
group
to
provide
ways
to
get
more
secure
jobs,
and
there
would
be
a
new
jaw
API
being
added
that
also
let
you
specify
the
audience
so
that's
gonna
be
a
more
scoped
way
to
actually
be
able
to
authenticate
and
create
some
equivalency
between,
say,
a
kubernetes
service
account
and
something
like
like
an
I
am
role.
It's.
E
A
But
I
don't
think
anyone
has
any
objections
to
this
functionality.
It's
not
gonna
be
huge
right.
It's
gonna
be
like
50
to
a
hundred.
Nice
quote.
Yes,
it's
small
yeah,
so
I
think
I
certainly
think
it
should
I
I
think
it
sounds
a
great
idea
not
just
to
see
it
go
in
I
think
we
we
do
wanna
get
to
the
I
am
and
do
it
integration
or
tighter,
Dritz
and
they're
integrated,
but
I
think
this
is
a
stepping
stone
along
the
way.
Maybe
would.
D
Venturi,
yes,
so
in
the
name
of
reading
these
kept
lights
and
just
understanding
what
the
stake
is
currently
responsible
for,
can
somebody
I'm
happy
to
do
this,
but
before
the
next
meeting
come
up
with
a
spreadsheet
or
just
a
repo
or
some
way
to
start
tracking,
all
the
different
bits
and
pieces
of
code
that
the
sync
could
potentially
be
responsible
for,
so
we
can
start
understanding
who's
owning
it
who's
in
charge
of
it
and
just
mapping
out
what
we?
What
were
responsible
for.
A
D
We
want
to
do
I
just
want
to
know
like
at
a
glance
what
project
sub
projects
bits
of
code.
Is
this
thing
responsible
for
and
who's
kind
of,
like
in
charge
of
those
bits,
a
coder
who
has
you
know
as
a
stakeholder,
everything
from
cops
cloud
provider
to
like
they,
the
be
piece
demo
for
the
scene.
Decorator
we
just
saw
like
come
on.
Those
desisted
care
is
that
our
responsibility
is
enough,
like.
A
An
inventory
of
all
all
eight
of
us
code
that
may
be
large,
but
okay,
I,
like
we
don't
go
through
and
start
tracking
it
right.
I
think
that
is
a
good
idea.
Yeah,
if
you,
if
you
start
a
doc
or
why
don't
we
figure
out
where
we
should
start
the
doc
that
isn't
properly
owned
and
then
we
can.
We
can
encourage
people
to
add
their
project
into
it
and
it
would
be
nice
Alyssa
to
reflect
the
the
structure
right.
A
B
Maybe
this
is
more
of
a
pet
peeve,
but
I've
found
that
the
dependence
on
spreadsheets
for
things
like
X,
you
know
print
spreadsheets
that
are
managed
external
to
the
project.
For
things
like
the
release,
stuff
has
been
pretty
like
hard
for
people
to
find,
and
then
you
have
to
like
go
through
permissions
dances
and
all
this
sort
of
thing
I
I,
would
want
for
one
like
really
strong.
We
won't
this
be
something
we
do
pr's
against
to
do
updates.