►
From YouTube: Kubernetes Kops Office Hours 20180330
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
No
I
had
everyone.
This
is
cops
last
hours,
it
is
Friday
March
30th,
as
I'm
sure
people
know,
kubernetes
110
was
released
the
other
day,
so
we
will
be
thinking
about
the
future.
I
think
a
little
bit
in
terms
of
our
release
schedule
and
we
have
a
bunch
of
things
on
the
agenda
in
terms
of
one
nine
one
ten
to
Rufus.
A
If
anyone
wants
to
anything
to
the
agenda,
please
do
so.
The
link
is
in
the
calendar,
invite
and
otherwise
shall
we
jump
into
it.
So
I
put
two
things
on
the
agenda,
which
I
think
are
good
things
to
start
with
in
terms
of
setting
the
scene,
one
of
which
is
the
proposed
release
schedule.
So
we
talked
in
the
past.
I
guess
two
meetings
about
how
we
wanted
to
the
idea
that
a
cop's
release
means
that
the
whole
ecosystem
we
believe
in
stable.
A
So
with
that,
I
drew
up
a
calendar
of
what
that
means
in
practice,
I
think
for
the
next
two
months
of
releases.
So
what
I
wrote
down
is
that
I
think
we're
ready?
Do
you
want
to
know
beta
one?
We
have
a
couple
of
issues,
but
they
look
like
they're,
not
there
at
the
right
level
to
sort
of
do
the
alpha
to
beta
transition.
A
A
We'll
aim
to
do
the
cops
111
release
about
a
month
after
that
and
when
we
do
the
cops
111
release
will
do
the
cop
with
the
cops
112
alpha
1
at
the
same
time,
and
the
other
nice
thing
is
this
sort
of
means
that
it's
pretty
predictable
and
it
also
means
that
sort
of
we
only
have
one
open,
ish
branch
at
a
time-
hopefully
so
obviously
won't
backward
backward
or
cherry-pick
things
that
need
to
go.
But
we
we
don't
have.
You
know
eight
open
branches
at
once
for
features.
B
A
A
Don't
know
that
we
need
that
I
think
we
can
probably
we'll
see.
I
think
that
we
can
probably
so
one
of
the
differences
is
that
we
believe
that
you
should
be
running
the
it
should
be
safe
to
run
the
latest
costs
against
any
version
of
kubernetes
or
any
prior
version
kubernetes,
and
so
there's
no
real
reason
right
now.
So
right
now
cops
1/8
whatever
it
is,
is
released.
A
So
so
the
we
have
less
of
a
need
to
cherry
pick.
For
that
reason
and
more
of
a
need
to
so
once
we
open
the
like
112
alpha
branch
we
will
we'll
have.
Another
will
still
want
to
put
things
on
to
the
one
11
release
that
need
to
be
going
to
a
release,
and
so
those
will
need
to
be
cherry
pick,
but
hopefully
there
will
be
fewer,
they
won't
be
features,
and
hopefully
they
will
be
smaller
things.
B
C
D
A
The
releases
are
fine
in
themselves.
The
heartbeat
is
is
doing
well
actually
doing.
The
actual
release
notes
like
synthesizing,
like
release,
notes
that
are
sort
of
human
readable
and
the
hard
bit
is
knowing
when
to
do
the
actual
release.
Release
so
like
a
new
alpha
release
or
a
new
beta
release.
Isn't
is
mostly
mechanical,
like
there's
a
judgment
calling
we
say
alpha
beta,
but
other
than
that.
A
A
And
I'd
love
to
get
the
button
and
it
works
more
or
less
I'd
love
to
get
it
fully
automated,
but
actually
yeah,
there's
actually
so
Google
cloud
build
seems
like
it's
actually
really
might
solve
a
lot
of
our
issues
there,
but
we,
our
build
process,
is
currently
probably
too
convoluted
to
working.
Google
cloud
builds,
maybe
post
basil
that
that
we'll
see
so
I
was.
E
A
Are
gonna
start
off
as
the
same
code?
Okay
and
then
yes,
they
will
deviate
rapidly.
So
in
practice,
problems
I,
don't
know
whether
I
can
actually
do
it,
but
I
would
probably
release
110
out
for
one
from
Mass
Oh.
Actually,
that's
true,
so
technically
I
would
release
once
an
out
for
one
from
the
master
branch
or
from
work
a
bit.
So
at
once
we
do
beta
one.
We
we
tend
to
be
more
judicious
about
putting
in
things
right.
E
A
B
Goes
back
to
the
question
I
had
about
cherry-picking
I
mean,
maybe
does
everybody
know
what
I
mean
by
that
like
there?
Okay,
so
does
everybody
want
to
have
a
cherry-pick
type
methodology
where
we're
choosing
what
to
move
into
the
into
the
release
branch?
Because
what
we've
done
past
is
everything
goes
basically
into
the
release
branch.
If
it's
been
merged,
we
just
do
everything
back
into
the
release
like
redo
the
release,
branch
or
update
the
release,
branch
from
master
and
that's
what's
going
into
the
beta
and
beta
is
a
really
only
bugs
like
that.
A
B
B
A
F
B
A
Kubernetes
release
notes
are
also
a
similar
Cluj
like
it's
not
like
they're
easy.
Either
we
have
a.
We
have
us.
We
have
some
script
stuff
that
will
from
the
commit
history
will
generate
unabridged
like
each
commit
like
who
wrote
it
that
sort
of
thing
so
the
links,
and
then
it's
basically
just
like
like
trying
to
organize
it
more
thematically
and
trying
to
make
it
so
that
the
most
important
things
sort
of
like
action
required
type
things
are
high
and
like
collapsing.
A
A
lot
of
the
like
I
tend
to
do
like
dogs,
typos
and
comment
type
fixes
and
do
a
general
thanks
with
a
list
of
everyone
at
the
end,
rather
than
you
know,
most
people
installing
just
want
to
see
that
some
did
it,
but
don't
care
about.
You
know
who
wrote
each
particular
one.
It's
not
really
relevant
to
someone
doing
an
upgrade.
A
A
B
Branch,
it's
a
risk
that
we
still
support
the
install
because
what's
gonna
happen
is
somebody's
gonna
go
okay!
Well,
cops
can
install
it
and
then
they
report
a
bug
on
it
and
coronations
community
says
no,
it's
not
usable
anymore.
We
should
at
least
warn
on
it
right.
One
thing
here
since
you're
interested
in
it
can
I
volunteer
you
to
start
a
PR
on
it.
B
A
Which
you
can
probably
use,
which
is
the
channels
like
the
channels
I
often
chapels,
channel
stable,
it
can
recommend
you
an
upgrade
of
cops.
You
can
recommend
you
an
upgrade
of
kubernetes
and
I.
Think
it
can
strongly
upgrade.
I
think
it
can
even
refuse
to
run
at
must
be
passed,
an
environment
variable,
so
it
like
that's
what
we
do
if
we
find
a
like
a
security
flaw
on
a
version
of
cops
like
so
your
versions
of
cops
on
and
still
like,
the
latest
QB,
NS
fixes
and
so
I
think
we
require.
A
You
use
a
certain
version
of
cops
now
and
we
can
use
that
sort
of
mechanism
and
it's
not
really
required,
because
you
can
just
pass
an
n/bar
to
get
around
it.
That
feels
like
a
fair,
like
you
really
don't
want
to
be
installing
a
grenades
one
for
today.
Without
you
know,
without
saying
yes,
I
know
what
I'm
doing.
D
A
So
that's
the
release.
Schedule
I
will
do
a
document.
If
people
don't
disagree
and
we
can
sort
of
see
how
it
goes.
I
guess
I
think
it's
great.
Thank
you
so
yeah.
It
means
that
well,
it
means
that
like
come,
1:11
we'll
have
much
will
have
basically
a
released
we'll
have
a
beta
one
at
the
same
time
as
kubernetes
is
released
and
we'll
have
an
alpha
for
112
the
next
version
of
communities
during
the
112
release
process
I
the.
B
Only
caveat
I'd
ask
for
Justin
is
that
we
put
in
there
very
cleanly
that
these
are
proposed
dates.
If
there
are
any
issues
that
we
have
any
instability,
we
are
going
to
have
them
slip.
So
that's
the
only
thing
because
people
are
gonna
be
like.
Oh,
we
have
dates,
where's
your
release,
so
I
just
want
to
make
sure
there's
a
big
old
disclaimer
in
there
that
we
care
about
stability
rather
than
the
dates.
A
And
then
the
other,
the
other
topic
for
discussion
I,
think,
is
just
what
we
want
to
put
into
the
next
few
cops
races.
So
there's,
obviously
the
obvious,
which
is
support
for
the
latest
kubernetes,
but
the
other
ones
that
sort
of
I'm
thinking
about
and
are
sort
of,
the
at
CD
manager,
getting
the
SED
management,
modularized
and
sort
of
out
out
of
out
of,
but
reusable
outside
of
cops
vein
by
a
tree
thing,
and
so
the
SED
manager
that
I'm
working
on
also
supports
at
CD
upgrades,
which
I
think
is
thing
on
everyone's
mind.
A
Similarly,
a
similar
sort
of
functionality
for
a
Dom
management
I
feel
like
we're
now
in
a
place
where
there's
enough
or
some
consensus,
around
add-on
management
that
we
can
actually
sort
of
start
to
make
that
more
standard
or
reduce
this
the
emerging
standards.
And
then
the
machines
API,
which
is
sort
of
part
of
the
custard
API,
seems
you
making
a
lot
of
progress
and
it
has
the
potential
to
let
both
work,
that
it
was
sort
of
pups
classes
work
better
with,
for
example,
the
autoscaler,
but
also
supports
other
clouds.
More.
E
A
B
B
Also,
the
two
never-ending
PRS
that
three
people
are
highly
involved
in
on
this
call,
one
of
them,
which
is
waiting
on
Justin,
our
wonderful
I,
am
profile
PR
and
are
also
a
wonderful
security
group
PR,
because
I
have
had
sidebar
requests
for
both
of
those
from
two
different
groups.
Besides
who
is
on
this
call,
sorry
Justin
guessing
those
another
horrible,
but
we
have
two.
So
what
is
the
security
use
case?
B
C
I
have
a
security
group
and
they
don't.
Let
me
create
security
groups
or
I
am
roles
they
will
give
them
to
me
and
say
use
these
so
currently,
I
do
something
really
really
ugly,
with
cops.
I
literally
spit
out
a
pair
form
template
and
then
I
have
something
that
automatically
modifies
that
for
me,
so
that
I
can
actually
use
stops
and
what
it
generates.
So
life
cycle
would
just
let
me
stop
using
terraform
as
a
shim.
A
C
Even
that
currently,
so
just
for
the
I
am
roles.
For
example,
all
the
I
am
roles
that
cops
generates
or
the
I
am
roles
that
cops
generates.
I,
don't
want
all
my
users
to
have
those
permissions
like,
for
example,
I.
Don't
want
my
users
to
even
be
able
to
create
elby's
with
kubernetes,
so
I
don't
give
them.
Those
permissions
at
all.
I
have
a
ingress
controller.
C
B
You
know
these
I
know
it
would
be
awesome
if
we
had
other
options,
but
we
from
my
perspective,
you
know
we
need
to
either
have
a
separate
release
for
these,
or
you
know
we
need
the
PRS
and,
as
as
we've
worked
on,
you
know,
we
need
the
API
support
it.
As
we've
agreed
to
this
API
Justin,
where
it's
on
the
instance
group
there's
a
security
or
list
of
security
groups
on
the
instance
group,
there's
an
iamb
profile,
so
we
we
you
signed
off
on
the
API.
What.
A
B
Well,
the
the
thing
is
that,
like
literally
in
the
notes,
release
notes,
we
should
say:
don't
use
this
feature.
I
know
it's
horrible,
but
at
times
you've
just
got
a
you've
got
a.
What
can
we
tell
them
that
perspective?
There's
enough
users
asking
for
this,
but.
A
B
And
it
has
to
have
the
crux
secure
your
roles
on
it,
but
we've
discussed
this
just
indeed
we
we've
gone
over
this
and
beat
this
to
death.
I,
don't
understand
why,
like
you
agreed,
you
know,
and
we
can
take
this
offline,
but
you
agreed
to
the
API,
but
why?
What
what
is
causing
us
not
to
be
able
to
merge
it
I.
A
B
B
A
C
B
A
B
F
B
Can
skip
the
rolls
we
can
skip
the
security
group
roles
with
the
override
flag,
so
we
can
tell
cops
not
to
do
any
security
group
Earl's
Phillip
so
that
we
already
got
fixed.
But
if
we
tagged
it
correctly,
cops
is
gonna
pick
it
up.
It's
gonna
and
if
we
tag
it
with
shared,
it's
not
gonna,
delete
it
right,
but.
A
B
We're
changing
the
way
we've
implemented
it
again,
just
by
the
way,
at
the
end
of
what
again
well,
we've
already
agreed
on
you
already
agreed
on
a
name.
You
know:
I'm,
sorry,
I'm,
beating
this
to
death;
I'm,
not
trying
to
be
an
dude,
sorry
to
say
that
online
record
it,
but
we
are
agreed
to
an
API.
Now
we're
changing
our
approach
again,
you
know
felt
like
I've
I've
redone
that
I
am
profile
here
three
times
Philips
done
at
once.
You
know
sorry
test
to
dude.
Let's.
A
B
A
C
C
A
C
B
A
B
C
Like
I,
don't
want
to
generate
security
groups
outside
of
the
thing
that
already
generates
security
groups.
For
us,
we
have
their
form
profiles
that
manage
all
the
security
groups
and
having
something
external,
that's
generating
these.
It's
just
not
manageable.
Having
I
am
profiles,
I've
generated,
that's
a
different
way
is
not
manageable.
Okay
and
again,
I
don't
want
to
use
the
cop
security.
I
am
profiles
because
they're
too
permissive
and
I
just
can't
apply
them
know.
B
A
I
mean
I,
think
I.
Think
profiles
like
using
a
different
profile
is
I,
think
allowed
I
think
the
only
or
is
I'm
controversial.
I
think
the
only
thing
that
we
that
we
need
to
do
is
by
default.
I'd
like
it
to
keep
validating
and
I
think
to
Rodrigo
you'd
have
to
specify
a
lifecycle
that
turns
off
validation,
because
your
your
I
mean,
ideally,
you
could
we'd
be
able
to
have
enough
granularity
in
the
API
to
express
your
like
more
restricted
permissions
in
practice.
A
B
C
I
bested,
both
of
them
I,
tested
the
security
group,
one
and
I
tested
the
I,
am
Roland,
yeah
and
and,
like
Justin
said,
I,
just
skipped
them
and
I'm
completely.
Fine
with
that.
F
C
B
A
A
C
A
It's
certainly
helped
people
in
the
past
and
has
taken
you
know
a
day
to
get
to
the
point
of
narrowing
it
down
to
like
the
dhcp
options.
Right
I
know,
that's
why
we
validate
everything,
I
wanna
connect
who
don't
load
it?
Do
you
see
two
options
but
in
the
sharing
case,
but
yes
like
getting
there
so
that
we're
able
to
getting
able
to
like
see,
see
the
problem
fast
and
yeah.
Like
tell
people
hey
that
you
know
sure,
greedy
groups
across
clusters
is
gonna,
an
using
ill
visas.
B
A
F
A
B
A
There's,
no,
that's
I,
think
that's
a
report
of
the
general
like
the
latest
urban
Eddie's
thing,
but
yes,
we
should
get
that
in
Shan.
Shan
said
alley:
child
support
for
cops
issue
for
one
to
seven,
I.
Think.
Yes,
let's
get
that
into
a
110
I.
Think.
Is
there
any
question
about
how
to
oh
it
says?
Should
we
do
I
know
you
get
to
me?
Yes
and
the
VFS
support
I
actually
was
debating,
merge
them
into
one
nine,
but
I
did
I
refrain.
I
was
a
good
boy
and
waited
so
yes,
don't
yes,
mom,
okay,.
A
E
B
E
B
G
A
It's
additive,
so
you
can
end
one
big
PR.
But
let
me
about
this:
how
about
I
review
the
VFS
stuff
and
the
sold
four
or
seven
one?
Nine
and
we'll
merge
it
when
we
cut
one
nine
beta
1,
so
the
minute
we
opened
1:10
right,
which
we
said
is
we're
aiming
for
about
a
week.
But
if
you
want
to
also
send
the
other
PR
I'm
happy
to
review
that
as
well
in
like
a
mega
PR
generally,
if
it's
an
additive.
So
because
it's
a
new
because
it's
a
different
cloud,
it
won't
affect
well
shouldn't
users.
G
E
C
B
A
complicated
process-
just
let
you
know
we
can
get
the
PR
in
and
I-
can
show
you
how
to
run
full
end-to-end,
ok,
but
then
we've
got
to
go
to
C&C,
yeah
blah
blah
blah.
Hang
me
once
we
got
the
code
in
take
it
as
an
action
item
or
file
an
issue
and
we
can
get
full
end-to-end
testing
right.
I'll
be
awesome.
All.
B
B
A
A
I
think
I
am
one,
is
easier:
I
I,
don't
my
concern
about
securities
like
it's
gonna
be
hard
to
stop
people?
Oh
it
second,
so
we're
gonna
say
it's
gonna
be
hard
to
stop
people
sharing
a
security
group
across
clusters.
Maybe
we
can
use
tags
and
obviously
most
people
would
opt
out
of
it.
They
would
stop
the
casual
mistake.
Let's,
let's,
let's
not
go
back
down
that
rabbit
hole.
My
friend
we've
got
15
minutes
left
all
right.
B
B
We
ran
into
an
issue
where
Blake
upgraded
at
CD
or
up
started
using
kubernetes
1/9
with
etsy
d2h
a
and
it
took
out
the
EDD
server.
I
ops
wing
completely
nuts.
We
have
chosen
to
put
in
a
flag
which
removes
which
turns
off
a
new
quorum
function
on
the
API
server.
It's
under
ete
testing.
Now
it's
passing,
but
there's
known,
there's
suppose
it
known
issues
with
quorum
and
n
CD
@
CD.
3
performance
with
form
is
much
better.
C
A
Default
changed
from
false
to
true,
and
it's
only
at
CD.
So
on
that
to
single
nodes,
it
doesn't
make
a
difference,
but
it's
actually
much
different,
but
it's
much
slower
to
run
with
quorum.
True,
so
we
said
that's
a
false,
which
is
correct.
I,
don't
think
anyone
would
argue
at
that.
I
think
the
four
I
think
it's
correct:
okay,
but
the
the
complicated
case
is
H,
a
mode
for
ed
CD
2,
where
the
performance
is
basically
like
everything
becomes
a
right.
A
Every
read
becomes
a
disc
right,
so
it
has
a
heavy
performance
cost
so
we're
effectively
keeping
the
same
1.8
behavior,
where
we
have
quorum
off
by
default
for
ED
CD
2
in
H
a
mode
there
comes
one
9
and
we're
gonna,
add
I.
Ups
Billy
did
specify
ups.
If
you
do
want
to
stay
on
at
CD
2
and
we're
gonna
make
the
XE
d3
an
easier
flag,
I
think
in
110
and
I
think
we're
going
to
well.
B
So
once
we
have
so,
we
have
n
CD
manager
coming
in
which
will
give
us
the
capability
to
upgrade
once
that
functionality
is
built,
and
this
is
really
the
mitigation
plan.
For
this
whole
quorum
thing
is
to
get
everybody
on
two
at
CD,
three
or
at
least
provide
that
capability.
Then
we'll
switch
over
to
defaulting
my
Etsy
d3.
So
that
answer
your
question.
A
And
we
actually
have
ass,
we
have
a
roadmap
with
a
schedule
on
it
and
it
is
110
and
yet
stay
in
your
quest
as
well.
Creating
trust
as
we
change
at
City
version
from
v2
to
v3
110
before
cost
create
and
I
didn't
mean
before
cuffs
created,
but
you
can
use.
You
can
currently
specify
it
with
a
slightly
complicated
flag
and
it'd
be
nice
to
make
that
flag.
A
B
Got
ten
minutes
and
they
have
four
issues:
3
right,
3
I
can't
count
I'm
in
here.
Damn
it
relax,
sarcasm,
guys,
sarcasm,
ladies
and
gentlemen,
so
validation.
We
still
seem
to
have
a
bug:
I'm
gonna
punt,
that
I'm
gonna
somebody
needs
to
bug
Blake
and
figure
out
how
he
broke
it.
Is
there
an
issue
at
all
mm-hmm?
B
A
B
Saw
so
they
so
they
sorry
to
interrupt
you
Justin,
they
seem
to
have
changed
the
mount
cats
I
think
instead
of
I
think
they're
using
/dev.
Now
it's
what
but
there's
a
gentleman
by
the
name
of
Shane.
He
used
to
work
for
I
used
to
work
with
him,
he's
super
sharp
and
he
could
actually
fix
it
in
protic
you
or
at
least
give
us
some
guidance.
I
was
hoping
that
the
folks
from
the
UK
gamble
and
forget
the
others
Schieffer
I
forget
the
other
gentleman
same
hope.
B
B
A
Yes,
if,
if
new
Kouros
clusters
are
gonna,
come
up
or
auto
upgrade
so
the
chorus
currently
auto
upgrades-
and
we
don't
turn
that
off,
I
would
like
to
make
it
possible
to
turn
that
off
we're
easy
to
turn
it
off
and
yeah.
If
every
core
OS
Custer
is
gonna
launch
and
then
silently
upgrade
to
a
new
version
that
starts
working.
That
is
not.
That
is
not
that.
A
B
B
A
It
would
never
clear,
and
so
we
have
a
test
now
that
effectively
runs
cups,
create
concept,
cups,
update
cups,
update
instead
of
running
against
real
AWS,
runs
against
our
mocked
out
AWS
and
what
it
what
the
mock
database
basically
does.
It
keeps
track
of
the
API
objects,
it
doesn't
actually
know
create
any
instances
or
anything
like
that
or
do
any
real
logic.
It
just
says,
like
you
know,
when
I
create
a
subnet.
Okay,
you've
got
a
subnet,
and
when
you
list
the
subnets
you
get
them
back,
so
it
tracks.
A
Basically,
this
is,
let
us
run
the
code,
the
real
code
without
marking
individual
call,
and
it
lets
us
write
tests
that
you
know,
are
sort
of
a
higher
level
and
actually
test
sort
of
more
lifecycle,
e-type
things,
so
multiple
repeated
operations
without
really
having
to
do
a
lot
of
work
other
than
write
the
mocks.
So
it's
certainly
something
to
consider
if
you're
looking
at,
like
supporting
a
new
cloud,
is
looking.
A
A
B
Carry
on
so
let
me
say
it
a
different
way
to
make
sure
that
we
heard
you
Justin.
We
have
complete
interfaces
for
AWS
cloud
provider
now
from
soup
to
nuts.
Pretty
much
right.
You
can
do
stuff
like
run
a
crate
cluster
update
cluster
bump.
The
version
of
kubernetes
run
another
update
cluster
so
that
you
can
see
like
we
were
able
to
test
stuff
like
I,
open,
Pandora's
box,
with
shared
tags,
mm-hmm
with
route
tables
and
that
gateways
the
middlee.
B
It
was
kind
of
busted
my
opinion,
but
we're
able
to
test
that
the
correct
shared
flags
are
being
accessed
during
during
updates.
We've
never
had
the
capability
of
fully
disconnecting
from
the
cloud
and
having
complete
unit
test
run.
So
it's
pretty
sweet.
I
am
NOT
getting
my
head
around
exactly
how
you're
doing
it
Justin
so
I'd
love
to
see
one
of
these
sessions,
where
you
can
like,
say:
okay,
we're
going
to
implement
this
and
real
simple
test
and
walk
us
through
it.
That
would
be
too
cool.
Okay,.
B
B
On
the
other
hand,
I'd
love
to
see
all
the
cloud
fully
t-test,
because
that
does
that's
a
whole
nother
level
of
testing
Oh.
On
that
note,
plans
are
to
move
GCE
full
ete
into
our
cops
blocking
test.
So
not
only
will
we
be
running
an
AWS
blocking
test,
but
we're
gonna
be
running
a
GCE
blocking
test
whenever
we
merge
code
with
cops
so
once
I
can
get
the
test
screen.
A
Of
good
stuff,
yeah
see,
I
mean
these.
Obviously,
this
tests
like
to
be
clear
this
test
is
this
is
not
comparable
to
the
ETD
test
and
it
doesn't
like
test
kubernetes
at
all.
It's
a
complete
fake
right.
It's
a
mock
test
coughs.
It
does
not
test
like
if
you
an
e
to
e
test
with
cops
you're,
also
testing
kubernetes,
which
is
nice
this.
This
definitely
does
not
test
kubernetes
it
just
tests,
whether
cups,
like
life
cycle
type
of
things,
work
and
also
we
do
not
have
a
complete
implementation
of
AWS.
A
B
A
B
G
B
Fixing
this
stuff
is
a
whole
lot
lot
easy
folks,
please
concentrate
on
testing
1-9
for
us
all
the
cni
providers.
If
you
have
some
cycles,
let
me
know
and
I'll
point
you
in
the
right
direction
really
excited
about
all
the
cloud
coming
in
by
the
way
team.
That
is,
that
is
huge
and
we're.
You
know
it.
Yeah
I
can't
express
how
excited
that's,
gonna
and
I
hope
you're
ready
for
the
amount
of
support
you
get
to
do
because
you
guys
are
gonna,
have
a
bunch
of
users.