►
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
B
Yeah,
I
just
I
just
created
a
pr
for
it.
I'm
looking
into
some
test
end-to-end
test
failures,.
A
Is
this
gonna,
be
I
see
that
it's
on
the
main
branch
do
you
wanna?
Do
we
want
to
have
this
for
0,
311
or
just
four
zero?
Four.
B
We
could
have
it
for
zero
three
yeah,
I
think
zero.
Three
is
fine
because
for
zero
four,
if
yeah
there
could
be
more
changes
that
could
affect
this
stuff.
So
yeah.
A
There
is
only
one
or
maybe
two
features
that
we
kind
of
like
wanted
to
merge
and
they
were
the
the
prs
from
external
remediation.
A
This
seems
like
a
fine
improvement.
I'm
okay,
backwarding
it
fabrizio.
A
C
Yeah,
I
think
I
would
personally
prefer
we
stick
to
like
only
bug,
fixes
or
just
really
small
changes
that
we
know
are
not
like.
Don't
have
any
risk
for
zero
through
eleven
considering
we've
already
started
merging.
You
went
off
of
course,
stuff
to
the
main
branch,
and
that,
like
I,
I
just
feel
like.
If
we
make
this
release
too
big
and
like
where's,
the
line
like
if
we
start
taking
features
like
everyone
will
prefer
to
get
their
stuff
in
quicker
and
then
we
just
add
more
risk
and
it's
a
never-ending
loop.
A
B
Go
ahead,
yeah,
sorry,
so
the
only
the
only
reason
I
suggested
it
for
for
backboarding
it
when
I
thought
about
it
was
that
this
seems
like
if
people
are
going
to
be
using
the
even
alpha
3
release
and
sticking
for
it,
sticking
with
viewing
alpha
3
for
a
while,
then
this
provides
more
benefit
to
them.
Moving
forward,
if,
let's
say
like
certain
proposal
like
the,
if
the
management
cluster
operator
proposal
goes
through,
that's
just
gonna
have
a
lot
more
changes
in
the
code
base.
B
So
like
the
cluster
curl
cli,
as
we
see
it
today
I
mean
it.
Yeah
it'll
just
go
through
a
lot
more
changes,
so
it
won't
benefit
like
even
alpha.
Four
people,
the
I
feature
of
maybe
waiting
for
provider
will
be
there
but
yeah
it's
just
who,
which
users
do
we
want
to
benefit
like
do?
Is
this
going
to
benefit?
We
want
alpha,
3
or
folks
or
not,
but
I'm
fine
either
way.
I
don't.
I
don't
need
it
to
be
no
release
but
yeah.
C
So
yeah
I'm
not
against
this
change.
In
particular,
it's
just
about
the
general
process
of
how
we're
going
to
do
this
during
this
grooming.
I
think,
if
we're
gonna
put
things
in
as
a
back
porch
to
zero
through
eleven
for
each
one,
we
do
backboard.
We
should
have
like
a
solid
like
reason
for
doing
that,
like
what
is
this
fixing?
What
is
it
bringing
to
v1
offer
three
users
and
what
are
the
risks
involved?.
A
That's
there,
I'm
I'm
tending
to
agree
and
like
let's,
let's
take
that
going
forward
so,
like
also
for
other
issues,
are
going
to
come
up.
If
it's
a
behavioral
change,
unless
we
it's
something
that
came
up
from
zero,
four
I'll
start
zero
three
like
and
we
already
agreed
to
back
port
it.
We
only
bring
this
new
changes
to
zero.
Four
that
sounds
good
to
everyone.
A
I
see
a
lot
of
thumbs
up
pretty
okay.
She
was
gonna
watch.
These
updates,
but
they're
gonna
think
that
I
don't
know
what
I'm
doing
probably
my
mind
too
much
all
right.
Let's
keep
going
guppy
k,
bootstrapping
and
failing
long
bandwidth,
environment.
E
Yeah,
so
this
one
yeah-
I
I
don't
so
th
this
could
be
argued
as
a
bug
that
we
might
want
to
fix
and
back
port.
It's
basically
like
if
you,
for
whatever
reason,
are
not
using
the
images,
the
the
container
images
that
are
baked
into
your
ami
or
ova
or
whatever
os
image
you
have
and
you're
on
a
low
bandwidth
in
a
low
bandwidth,
environment,
the
bootstrap
token.
E
We
stop
trying
to
refresh
it
as
soon
as
we
see
that
the
infrastructure
for
your
machine
is
ready,
so
power
on.
We
see
it's
powered
on.
If
it
takes
more
than
15
minutes
to
pull
the
images,
then
your
toast.
So
I
think
this
could
be
a
help
wanted
issue
and
a
backboard
candidate.
A
dear.
A
Okay,
so,
but
yeah
we
already
reversed
tokens
today.
If
the
english
doesn't
come
up
in
time.
A
Patch
helper,
which
one
is
this
one?
Oh
no!
I
fixed
this
so
this
was
fixed
in
37
30.
No,
oh,
my
gosh,
which
one
is.
A
A
F
A
E
Yeah,
I
think,
like
I
know
we
don't
have
a
a
fixed
bar
for
deciding
back
ports
or
not.
I
know
we
talked
about
it
a
few
minutes
ago,
but
this
is
the
sort
of
thing
that
just
does
not
need
to
be
backboarded.
A
I
investigate
the
use
of
uncashed
and
cash
clients.
Yes,
we
should
do
this
and
I
probably
I'd
say
like:
let's
just
do
announcements
for
and
not
backboard
this,
because
this
might
change
a
bit
unless
it's
a
bug.
There
was
some
other
bug
that
was
found
like
in
that
the
cache
was
misbehaving,
and
so
the
cluster
was
re-initialized
twice
or
initialized
twice.
So,
if
you,
if
anybody
has
time
to
investigate
this,
that
would
be
great
any
volunteers
before
I
move
on.
A
All
right,
let's
move
on
then
provisioning,
each
control
plane
machine
with
unique
credentials.
E
E
I
don't
think
so
I
mean
maybe
there's
some
overlapping,
but
this
is
like.
I
think
this
was.
I
don't
know
it
was
file
based
yeah,
so
they're
trying
to
do
credentials
as
a
file,
but
they
want
the
per
machine
so
it
even
if
it
does
overlap
with
flavors.
It's
still
the
the
fundamental
problem
is
that
there's
no
way
with
kcp
to
distinguish
between
individual
members
of
the
replicas
that
it
creates
and
configure
them
differently.
G
E
E
H
Yeah,
would
this
be
better
as
a
kept
against
the
project
because
it
seems
like
it's
maybe
larger
than
just
an
issue.
A
Possibly
I
usually
like
would
kind
of
try
either
to
discuss
in
the
issue
or
at
a
community
meeting,
which
is
what
andy
I
guess
like
was
asking
from
the
author.
Oh,
I
think
zach
is
here,
so
I
have
nadir
and
then
zach
and
mark.
F
I'm
I
might
take
this
as
one
of
the
use
cases
for
composable
bleach
strappings.
Maybe
maybe
it
can
be
solved
as
part
of
that,
whatever
needs
to
be
done
for
that
as
well,
but
yeah
I
probably
want
some-
would
like
some
more
information
on
the
specifics.
F
I
Yeah,
so
I
could
add
some
more
scenarios
and
use
cases.
I
mean
nothing's
pressing
here
it's
just
kind
of
playing
around
with
the
idea.
I
don't
know
how
it
would
be
implemented
in
kcp,
but
yeah
yeah
I'll
respond
more
on
the.
A
Third
sounds
good
okay,
so,
given
that
that's
accurate
here
can
just
move
on
for
now.
J
No,
I
thought
that
I
would
need
to,
but
I
think
fabrizio
knows
about
it
too-
it's
because
it's
very
difficult
right
now
to
either
test
the
docker
controllers
any
of
the
docker
controllers.
So
that's
basically,
this
issue
talks
about
like
injecting
fakes,
to
make
it
easier
for
for
us
to
add
those
tests
in
there
yep.
That's
what.
A
Okay,
capti
quick
start
is
broken
when
selecting
a
combination
version.
It's
not
available
from
kind.
A
Yeah
spent
here,
I'm
inclined
to
close
this.
I
don't
think
this
is
actually.
A
D
A
A
A
Okay,
a
milestone
next,
I
don't.
This
is
definitely
not
a
rock
blocker.
C
Yeah,
I
would
like
to
fix
this
soon
because
it's
blocking
us
from
bumping
calico,
but
I
have
we
haven't
gone
to
the
root
cause.
Yet
so
that's
probably.
C
D
I
I
think
that
it
is
fine
to
keep
it
into
the
zero
three,
because
if
we
get
to
the
root
of
the
of
to
the
cost
of
the
program,
this
is
a
bug
and
should
be
fixed.
It
is
difficult
to
find
the
route.
A
It's
life
cycle
active,
I
guess
more
or
less;
okay,
perfect
elite
conversion
files
before
running
conversion
gen.
This
is
a
zero
four
thing
right.
B
A
A
E
E
E
So
the
request
is
in
my
management
cluster.
I
want
to
assign
cpu
memory,
storage,
etc
quota
that
applies
to
the
vms
or
machines
being
created
for
workload
clusters.
So
maybe
I
have
you
know
like
aws
and
azure,
for
example,
have
limits
on
the
resources.
E
So
maybe
you
have
a
team
sharing
a
namespace
and-
and
maybe
the
namespaces
are
all
sharing
a
single
account,
and
you
want
to
give
a
certain
team
a
subset
of
the
total
available
assets
that
they
have
access
to
in
that
account
so
team,
a
in
namespace
a
can
only
use,
I
don't
know
10
gigs
of
ram
or
something,
and
what's
a
little
weird
about
this.
A
A
Oh
I
reopened
this
because
there
was
a
lot
of
discussion
on
it.
I
don't
know
where
we
are
landing
here
I
mean.
E
Yeah,
so
the
the
requests
as
it
has
been
reopened
not
based
on
the
original
title
and
description
is,
I
have
some
machines
or
machine
deployments
in
cluster
api.
Those
correspond
to
nodes.
E
E
C
Yeah
the
proposal-
that's
open
for
spot
instances.
Termination
handler,
I
think,
has
a
requirement
on
that,
or
at
least.
E
E
C
E
Cluster
api
has
a
fixed
label
that
it
will
sync
down
to
the
node,
and
that's
that's
scoped
to
just
that
one
label
and
not
open
to
the
to
the
user
to
change
or
do
anything
with.
A
C
C
Does
the
label
syncing
also
includes,
when
you
upgrade
kubernetes
version
and
you
your
nodes
are
replaced
by
new
nodes
like
keeping
those
existing
labels
on
the
new
nodes?
Is
that
part
of
the
scope.
A
So,
and
I
want
to
send
you
to
agree
on
a
small
design
and
open
a
different
fee,.
E
Well,
since,
if
I
can
take
like
a
minute
or
two
to
kind
of
survey,
the
folks
here,
so
we
had
said
a
long
time
ago-
no
we're
not
going
to
sync
labels,
because
potentially
it
could
be
a
security
issue
or
conflicting
owners
trying
to
manage
the
same
set
of
labels
on
nodes.
I
know
that
the
cubelet
restricts
a
set
of
labels,
so
you
may
not
register
a
node
with
a
certain
set
of
labels
that
are
protected.
E
A
One
other
question
that
came
up
like
one:
we
discussed
this
like
a
year
or
so
ago.
Like
was
who
has
the
last
word
right,
which
is
what
jason
was
saying.
Probably
and
that's
I
don't
know,
maybe
we
want
to
like
scope
it
down
like
to
the
minimum
amount
that
we
want
to
do
as
in,
like
we
add
the
labels,
but
I
don't
know
if
you
want
to
remove
labels.
A
Easier,
simplified
ux,
I'm
gonna,
keep
this
index
for
now
because,
like
it
seems
this
would,
if
you're
interested
in
ux
like.
I
would
really
suggest
to
take
a
look
at
this.
It's
just
to
provide
a
like
a
simplified
user
experience.
A
F
Oops,
I
thought
I
had
put
a
comment
on
here,
but
I
do
actually
have
a
list
of
options
that
we
probably
want
to
support,
which
is
the
union
of
cops
and
cloud
foundry
container
on
time.
So
I'll
tack
it
on
the
end.
But
it's
all
the
different
types
of
tuning
behaviors
that
people
have
ended
up
putting
in,
for
various
reasons.
E
A
All
right,
this
is
a
back
clock.
Yes,
given
that
dear
spock
up.
A
A
I'm
going
to
keep
it
in
next
then
you're
like
just,
I
guess
like
see
if
you
want
to
change
the
milestone,
but
at
immutable
market
for
fields
in
crt
definition
yeah.
We
can't
do
this
yet
that
was
not
there's
no
support
for
this
upstream.
A
Yet,
okay!
Well,
we'll
keep
it
in
the
backlog
cluster,
with
the
same
name
under
different
name.
Space
is
provision
but
no
info
created.
I
swear
that
every
now
and
then
this
comes
up.
This
is
still
a
problem.
I
don't
know
if
there
is
anything
that
well,
I
guess
like
we
could
use
some
sort
of
client
to
force
names,
but
still
it
doesn't
yeah
dear
go
ahead.
A
Maybe
currently,
like
I'm
questioning
that,
like
our
infrastructure
provider,
actually
use
the
cluster
name
or
the
aws
cluster
name
or
the
azure
blockchain
name
to
create
other
resources
that
are
then
not.
You
know
that
are
global,
like
a
vpc
name
or
things
like
that.
So
then
might
conflict
and
there's
like
also
probably
some
im
rules
for
aws.
A
So
I
have
andy
and
then
jason.
E
I
was
gonna
say
yes
to
nadir.
Last
I
checked
the
tagging
that
the
cloud
provider
implementations
at
least
aws
is
using,
doesn't
include
the
namespace
so
get
the
namespace
added
part
of
this
problem
goes
away,
at
least
for
new
clusters.
G
Yeah,
so
so
the
complication
is,
is
there's
two
different
things
that
are
overlapping
here.
One
is
the
actual
resources
that
we're
creating
on
the
provider
side
and
that
would
be
relatively
simple
to
fix
by
you
know,
migrating
to
a
new
naming
scheme
and
then
there's
the
kind
of
internal
cluster
api
integration,
where
you
know,
like
andy,
said
the
upstream
cloud
provider
and
trying
to
align
both
of
those.
So
it's
a
combination
of
both
of
those
too.
G
So
probably
you
know
codify
it
into
the
contract
that
infrastructure
providers
need
to
use
a
combination
of
the
namespace
and
cluster
name
for
creating
any
resources
to
avoid
any
conflicts
like
that,
and
then
having
the
discussion
with
sig
cloud
provider
would
be
a
longer
tail
on
it.
A
D
A
Complicated
yeah:
do
you
want
to
get
assigned
to
this
jason
and
to
possibly
draft
up
a
possible
contract.
G
F
Do
you
yeah,
I
completely
assess
because
I'm
helping
nick
turner
and
andrew
seikin
with
the
version
two
aws
cloud
provider,
so
we
we
are
writing
a
new
cloud
provider
for
aws,
so
it
could
potentially
cover
this
and
we
can
ask
cloud
provider
as
a
whole
as
well.
C
What
if
you
have
two
different
management
clusters
and
you
use
the
same
namespace
name
and
the
same
cluster
name
and
you
apply
because
I've
done
that
and
it
breaks-
and
I
even
did
this
where
I
had
two
different
configurations
and
then
I
ended
up
with
a
mixed
infrastructure
where
it
had
infrastructure
from
the
first
cluster
and
then
it
added.
So
I
had
two
load
balancers
and
it
was
like
a
cluster
monster.
G
A
C
A
Document
these
limitations-
yeah,
not
everybody,
sure
I'm
gonna,
cca,.
A
A
All
right
define
supported
combinations.
E
Go
ahead,
andy
this
one's
kind
of
gone
all
over
the
place.
I
don't
remember
what
was
at
the
bottom:
it
either.
E
I
I
think
we
either
need
to
close
this,
or
we
need
somebody
to
read
through
and
digest
everything
that's
in
here
and
see
it
through
to
completion
whatever.
That
means.
A
I
would
like
to
open
two
different
issues
for
these,
if
possible,
because
at
least
the
first
one
just
saying
like
in
the
dark
saying
like
hey,
we
need
this
to
build
os
images.
D
It
popped
out
recently
to
to
be
honest:
if
we
are
at
0
311
and
no
one
is
asking
it,
it's
not
a
big
deal.
D
A
A
All
right,
no
comment,
so
I'm
just
going
to
move
on
figure
out
long
term
interface
for
static
pod
customization,
with
cube
idiom.
A
F
Yeah,
I
added
that,
as
a
use
case
to
the
cube,
adm
library
thing
that
mentioned
on
wednesday.
I
I
mean,
but
for
brittany,
if
you,
if
I'm
not
mistaken,
this
is
not.
We
run
out
for
before
we're
not
going
to
get
around
to
having
the
library
ready
for
we've
been
out
before.
Are
we.
D
A
Oh
okay
profile
cluster
api
to
provide
guidelines
on
memory
and
cpu
requirements.
Maybe
we
should
do
this
at
some
point
actually
like
if
the
operator
work
that
warren
alphabets
are
pushing
forward.
Like
does
expose
this
information,
we
can
provide
at
least
some
basic
cpu
and
memory
requirement,
and
not
just
like
yeah,
so
just
something
to
think
about.
We
don't
have
to
profile
it
today,
but
yeah
we
maybe
beta
would
be
nice.
A
Cecil,
did
you
have
a
comment?
I
saw
that
you're
muted,
no,
okay,
better
darwin
support
this
is
for
cathy.
Is
it
still
valid.
D
Yeah,
it
is
still
still
valid,
but
I
I
think
that
that
we
can
close
it
is.
It
depends
by
the
different,
but
the
fact
that
in
on
macos,
the
docker
interface
is
different
than
on
linux,
and
so,
when
you
export
a
kuber
config,
you
have
to
manually
change
it
to
get
access
to
the
to
the
cluster.
D
But
given
that
cup
d
is
for
development,
for
me,
it's
fine
to
to
manage
this.
D
A
Okay,
perfect
refactor
machine
set
reconcile
to
be
consistent.
I'm
going
to
skip
over
this
and
this
one
andy
might
remember
this
from
a
long.
A
A
And
if
you
look
at
the
code,
we've
tried
to
make
it
like
a
little
bit
more
consistent
with
the
rest
of
the
code
base,
but
we
have
not
gotten
there
yet.
So
if
you
have
an
interest
in
kind
of
like
doing
a
little
bit
of
refactoring
and
of
these
two
reconcilers
feel
free
to
reach
out
document
approaches
for
infrastructure
provided
to
consider
for
securing
sensitive
bootstrap
data.
Yes,
we
should
definitely
document.
F
A
I'll
find
goals
again.
Do
you
have
an
issue
for
the
secure
node.
F
C
Of
go
ahead:
do
you
want
to
mark
the
two
refactors
as
hob
wanted.
E
E
I
think
we
can
probably
just
close
this
issue
like
from
the
generic
machine
perspective
like
we,
don't
really
have
anything
in
there
that
lets
you
set
an
ip
and
I'm
fine
for
now.
I
think
if
the
infra
providers
want
to
offer
support
for
it.
A
A
Configure
closer
api
manager
controllers
with
a
config
map,
I'm
actually
going
to
close
this
because
we're
going
to
have
closing
this
in
favor
of
key
management.
Cluster
operator.
A
The
component
of
this
report
is
actually
hasn't
actually
merged.
The
admin
controller
on
time
but
appear
should
be
right.
If
you
are
curious
about
it,
it's
8
91.
If
I,
if
I
remember
correctly,
okay
ability
to
disable
rolling
updates.
F
A
What
do
we
have
here?
Machine
set
finishes
thinking
replicas
with
machines
still
under
the
deleting.
A
A
G
Yeah,
I
think
I
think
this
was
related
to
their
expectations
with
how
how
things
were
operating.
A
Yeah,
I
don't
remove
the
life
cycle
fruits
and
then
ask
for
an
update
if
they
want
to.
E
A
A
I
mean
we
could
pay
64
in
gzip,
but
if
the
providers
that
do
this
on
their
own-
like
I
guess
we
could
bring
this
up,
but
I
don't
know
if
this
is
kind
of
a
headache
that
we
want
to
solve
today.
G
F
A
A
D
G
C
I
think
we
should
keep
this
open
and
I
think
if
we
define
a
process
and
a
place
to
put
the
stuff,
like
jason,
said
a
lot
more
people
would
be
able
to
step
up
and
help.
I
can
help
with
french
and
I
think
a
lot
of
people
on
my
team
can
help
with
other
languages
as
well,
but
yeah.
I
think
we
should
aim
to
do
this,
because
I
think
it's
very
important
for
inclusivity
and
like
showing
that
our
good
faith
in
having
like
documentation,
that's
accessible
by
everyone.
A
Let's
see
also,
we
need
to
kind
of
like
redo
the
the
book
structure,
probably
for
to
allow
this
and
yeah
that
might
take
a
bit.
I
do
I
do
kind
of
want
to
point
out,
though,
like
if
we
support
multiple
languages,
we
need
to
keep
them
in
sync
within
reason.
So
that's
one
thing
that
we
need
to
work
with
possible
maintenance
of
that
language,
translation.
A
Okay,
let's
move
on
automate
the
release
I'm
going
to
move
on
from
this.
This
is
one
of
my
wishes
for
some
point
in
the
future:
submit
copy
providers
for
cnn
cncf,
conformance
verification,
well,
yeah,.
E
I'd
say
probably
well,
I
was
going
to
say
per
provider
because
it's
not
like
you
can
just
take
cluster
api
and
say
here's
conformance
results
for
it
unless
it's
cap
d,
maybe
if
there's
something
we
can
do
in
cluster
api
to
facilitate
and
make
it
easier
for
infra
providers,
maybe
there's
some
work.
Did
you
have
thoughts
on
that
jason.
G
C
G
So
if
we
have
a
similar
view
of
the
various
combinations
of
providers
that
we
care
about-
and
you
know
it
runs
on-
you
know
the
latest
tagged
releases
instead
of
you
know,
just
you
know
automatic,
post-submit
or
periodic
then
we
can
take.
We
could
basically
take
the
output
of
the
conformance,
run
and
use
that
to
seed
updates
to
the
conformance.
A
Interesting
that
would
be
cool,
and
then
we
could
generate
some
docs
and
publish
them
on
automating
them,
okay,
for
which
I'm
gonna
cco.
On
this,
given
that
you're
working
on
the
okay,
that's
great
cool,
do
we
want
to
change
a
milestone.
A
Actually,
we
are
one
minute
and
qbm
control
plane
the
size,
support
policy
for
non-control,
plane,
workloads
and
control,
plane
machines-
I'm
gonna
keep
this
next,
given
that
they
know
still
assigned
to
it.
Andy.
G
A
Yeah,
okay,
I'm
I'm
gonna,
keep
this
next.
Like
I
wear
time,
we
have
done
a
page
and
a
half
so
probably
need
another
one
of
these
before
or
maybe,
hopefully
not
more
than
another
one.
If
there
is
anything
that
like
you
would
like
to
retry,
imagine
the
in
the
next
milestone.