►
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
Please
make
sure
that
you
abide
by
the
cncf
code
of
conduct
be
respectful
with
everyone
and,
if
you'd
like
to
speak,
we
use
the
raise
hand
feature
in
zoom,
so
feel
free
to
raise
your
hand
and
I'll
make
sure
that
you
get
to
your
turn
yeah
and
then,
if
you
can,
please
add
your
name
to
the
attendee
list
and
any
discussion
or
question
topics
that
you
have
in
the
agenda
with
that.
Let's
get
started
so
okay,
so
we
do
have
one
big
psa
today
v0.4.0
has
been
released.
A
So
this
is
the
first
v1
alpha
4
release
for
cluster
api.
It's
a
really
really
big
milestone.
It's
been
a
long
time
coming
and
a
lot
of
people
worked
hard
to
get
us
here
so
congrats
to
everyone.
Thank
you
to
everyone
who
helped
participated.
A
The
release
notes
are
uber
long,
so
we
try
to
summarize
a
little
bit
and
bring
out
the
highlights
of
things
that
are
important.
But
if
you
have
any
questions
parsing,
it
feel
free
to
reach
out.
A
In
slack-
and
we
can
discuss
a
few
things
to
say
about
the
release,
first
thing
is:
if
you
are
using
an
infrastructure
provider
which
has
not
yet
released
their
v1
alpha
4
release,
which
is
the
case
for
I
think
most,
if
not
all,
infra
providers
right
now,
you
cannot
use
this
release
yet
so
you'll
have
to
stick
with
zero
three
until
your
infra
provider
has
released
zero,
four
or
sorry
their
own
version,
for
it
is
availa
like
the
docker
provider,
was
released
with
this,
so
you
can
test
it
out
with
the
docker
provider
right
now,
but
if
you're
using
providers
like
azure
or
aws,
I
think
those
are
still
working
on
getting
the
release
out.
A
A
The
problem
with
that
is,
if
you're
running,
that
from
a
cluster
ctlv0.3,
it
will
try
to
fetch
the
latest
release,
which
is
now
v0.4.0,
which
means
it's
incompatible
with
the
v1
like
v1.
Alpha
4
is
incompatible
with
z3
cluster
ctl.
So
you'll
get
this
error
message
saying
you
can't
use
it.
A
The
workaround
for
this
is
to
specify
explicitly
which
version
of
the
core
providers
to
use
until
you
can
use
0.4
we're
also
working
on
patching
this
in
the
release,
so
that
the
latest
release
will
be
the
latest
compatible
release,
instead
of
like
just
the
latest
github
release,
but
for
now
make
sure
you
use
the
workaround
and
make
sure
you're
still
using
cluster
ctl
v0.3.x,
because
v0.4
is
only
compatible
with
v1
alpha
4,
which,
as
I
mentioned,
is
not
available
for
most
infrastructure
providers
right
now,
so
I
know
that's
a
lot
of
info
to
digest,
but
does
anyone
have
any
questions
or
any
comments
about
the
release
in
general,
I'll,
open
up
to
the
floor.
A
All
right,
I
don't
see
any
hands
raising
so
we'll
keep
moving,
but
if
anything
comes
up,
if
you
have
any
questions,
feel
free
to
open
an
issue
and
or
reach
out
in
slack
all
right.
So
I
don't
think
we
need
to
go
through
release
blocking
issue
updates
this
week,
since
we
just
released
mostly
we're
just
trying
to
get
the
provider's
release
and
then
we'll
keep
going
from
there.
But
let's
move
on
to
discussion
topics
cigar,
you
want
to
talk
about
the
cluster
class
proposal.
B
Yeah,
oh
thanks
cece,
so
yeah.
I
just
wanted
to
like
talk
about
the
cluster
class
proposal,
so
I
think,
like
I,
I
went
through
the
proposal
yesterday
and
I
happened
to
resolve
all
of
the
outstanding
comments
on
on
the
proposal
so
and
then
I've
seen
that
the
activity
on
this
proposal
has
died
down
in
the
in
the
past
a
couple
of
weeks.
B
So
I
was
just
thinking
that,
if,
if
there
are
no
major
blockers
for
this,
can
we
like
maybe
go
ahead
and
give
it
like
a
final
look
and
try
to
merge
this
in?
If,
if
again
like,
I
said
like,
if
there
are
like
no
major
blockers
left,
that
would
be
great
and
like
if
people
here
haven't
looked
at
the
proposal
or
would
like
to
go
and
re-review
the
proposal
please.
B
It
would
be
very
helpful
if
you
could
do
that
yeah
and
I'm
happy
to
answer
all
any
questions
on
slack
as
well
as
on
the
proposal.
Pr.
A
Okay
in
general,
in
terms
of
merging
proposals
this
week
might
be
a
bit
slower
just
because
we
just
released
v04.
So
that's
really
the
focus
right
now
and
also
I'm
the
only
approver
around.
So
I
I'd
like
vince
and
fabrizio
to
take
a
final
look
as
well
before
we
merge
it.
But
that
being
said,
let's
just
make
sure
that
every
comment
is
resolved.
A
C
Yeah
hi,
so
when
we
were
doing
the
v1
alpha
4
planning,
there
was
this
big
conversation
which
justin
was
running
around
the
low
balances
and
there
was
a
big
proposal
and
I
know
it
didn't
make
it
into
be
one
alpha.
Four.
This
is
becoming
more
relevant
again
for
us,
and
so
you
know
I
did
volunteer
back
then
to
help
out
with
it.
C
I
have
now
got
some
time
to
help
out
with
it,
so
I
wondered
if
any
advancement
had
been
made
on
that
and
where
I
can
help,
but
I
noticed
that
justin
isn't
on
the
call
today.
So
this
maybe
I'll
have
to
ask
again
another
time.
A
Yeah,
I
think
you
mean
jason
right,
jason,
jason,
yeah
yeah.
I
think
he's
not
here,
probably
best
ways
to
reach
out
on
slack.
I
know
there
was
a
thread:
oh
neither
go
ahead.
D
Yeah,
so
I
think
jason
admitted
that
he
didn't
really
have
the
time
to
take
this
forward,
so
we're
basically
looking
for
volunteers
to
take
it
forward.
Vmware's
still
interested-
and
I
was
also
speaking
to
someone
who
I
don't
know
where
they're
based
but
they're,
working
on
some
sort
of
rancher-based
control
plane,
who
are
also
interested.
So
I
think
now's
the
time
to
try
and
find
reconvene
a
group
who's
willing
to
take
it
forward.
C
Yeah
perfect,
so
we're
definitely
interested
in
this
and
not
just
our
like
cluster
api,
regular
people.
There
were
some
interest
in
this
idea
from
like
another
team
that
works
primarily
on
networking
stuff,
so
we
might
be
able
to
write
them
into
help
as
well.
So
I
will
circulate
that
internally.
But
if
we
know
where
the
enhancement
is,
then
I
can
start
taking
a
look,
certainly
and
see
where
we
can
go
with
the
comments
and
you
know
getting
it
revived.
D
Sure
yeah,
let's
chat
tomorrow,
then
I
guess.
A
Awesome
all
right:
let's
move
on
then
daniel.
You
have
the
next
one.
E
Thanks
I
was
like
I'm
seriously
typing
into
the
document
as
we're
discussing
so
yeah.
I
I
ran
into
a
kind
of
interesting
situation
with
cluster
cuddle
move,
there's
a
set
of
pre-flight
checks
and,
and
one
of
the
pre-flight
checks
goes
over
all
the
the
the
current
machine
objects
and
checks
that
each
machine
object
has
a
node
ref
and
if
it
doesn't
have
a
node
ref,
the
move
operation
fails,
and
first
I'm
wondering
why
this
is
why
this
is
necessary.
E
I,
and
and
like
open,
open
question.
I
don't
I
I
don't
know
I
don't
have
any
pretty
like
preconceived
notions
so
so
yeah.
That
would
be
great
to
know
about
that
and
then
assuming
it
is
necessary.
E
I'm
wondering
you
know
like
what
do
we
recommend
to
if,
if
cluster
cuddle
move
is
part
of
some
some
sort
of
automation
right,
do
we
recommend
that
so
so?
For
example,
you're
you're,
you
know
creating
a
cluster
and
you've.
You
know,
you've
you've
deployed
you've
deployed
er
er,
all
the
machines
have
been
deployed,
but
then
you
know
as
as
like
as
you're
doing
the
move.
Maybe
one
of
the
machines
is,
you
know,
fails
and
and
or
is
terminated
and
then
is
is
replaced
right.
E
So
at
that
point
a
new
machine
object
gets
that's
created
and
it
might
not
have
a
node
wrap
like
that.
That's
just
something
that
may
happen
at
some
point.
You
know
and
then
you're
running
the
move
and
the
move
fails.
So
you
know
what
what
do
we
recommend
do?
E
We
do
we
recommend
just
retrying
the
move
and
relying
on
this
pre-flight
check
or
do
we
need
to
pause
reconciliation
and
then
confirm
that
not
just
that
we've
paused
reconciliation
and
no
new,
you
know
no
new,
I
guess
individual
machine
objects
are,
you
know
like
they?
They
won't
be,
I
guess
created,
but
we
also
have
to
check.
I
think,
maybe
if
there
are
scaling
operations
that
are
underway
like
so
anyway,
so
like
thinking
out
loud
a
little
bit
but
but
if
anybody
has
any
any
feedback,
I
would
appreciate
that.
A
So
this
is
one
where
I
wish
fabrizio
was
here
to
answer,
but
I
might
guess
like
just
looking
at
the
comments.
There's
a
note
that
says
note.
Ref
is
considered
a
better
signal
than
infrastructure
ready
because
it
ensures
the
note
is
up
and
running.
So
I
believe
that
was
the
original
intent
behind
checking
for
node
ref,
because
you
want
to
make
sure
that
the
node
has
joined
the
cluster.
A
A
So
that
I
think
maybe
I
don't
know
if
you've
already
done
this,
but
I
would
start
by
looking
through
issues
to
see
if
this
is
something
that's
been
discussed
before.
That
has
come
up
and,
if
not
maybe
start
a
new
discussion.
So
we
don't
forget-
and
I
see
major
nadia
has
his
hand
up
so.
D
D
E
D
A
Open
an
issue,
I
think
that's
more
okay
resilient,
but
yeah.
I
think
we
do
check.
Control
plane
is
initialized,
but
we
don't
check
for
the
control
thing
nodes,
specifically
yeah,
but
they're.
I
think
it's
it's.
This
is
code
that
is
like
nader's,
that's
pretty
old-ish,
so
it's
possible
that
it
hasn't
been
updated
to
use
the
latest
before
we
have
arun.
A
F
So
in
any
cluster,
it
is
not
guaranteed
that
a
node
is
active
for
any
particular
point
of
time.
As
in
when
you're
trying
to
move
a
node,
the
source
node
may
die
before
the
move
ends
and
it's
not
guaranteed
that
you
create
a
destination
node
and
it
is
up
before
the
source
node
right.
So
there's
a
lot
of
variation.
F
What
is
the
thought
process
in
actually
ensuring
that
a
node
is
movable
for
a
control
plane?
It
is
we
need
a
minimum
quorum,
so
we
actually
will
want
a
set
of
nodes
to
be
available,
but
for
the
worker
nodes
is
there?
Are
we
going
to
guarantee
that
the
nodes
in
the
original
cluster
and
the
destination
cluster
the
numbers
will
match,
because
that
is
a
fluctuating
number,
both
in
the
source
and
the
destination
at
any
particular
point
of
time,.
A
Point
sorry:
go
ahead.
F
Yeah,
I
mean
basically
there's
a
desired
intent
and
as
long
as
the
destination
etc,
node
knows
that
that's
the
desired
intent
and
the
move
is
successful
right
after
that,
it's
the
role
of
the
cloud
providers
to
service
that
intent,
but
should
we
actually
expect
it
to
be
serviced
before
the
move?
That's
the
question.
F
A
Yeah,
I
agree.
I
think
this
is
like
a
way
of
trying
to
ensure
that
move
isn't
started
while
the
infrastructure
is
still
provisioning,
while
that
initial
provisioning
is
still
happening,
but
I
think
maybe
we
can
try
to
improve
and
like
refine
that
logic
to
make
it
so
that
it
only
checks
for
like
initial
provisioning,
but
it
doesn't
block
on
later.
Like
you
said,
I
don't
think
having
the
nodes
be.
Like
the
exact
same,
I
don't
think
that
should
be
a
hard
requirement,
so
yeah,
let's
bring
that
into
an
issue.
A
Awesome
and
yeah-
I
think
this
is
the
last
discussion
topic
that
we
had
anyone
else
have
anything
that
they
want
to
bring
up
or
any
questions
in
general,
any
provider,
questions
from
providers
about
adopting
v1
alpha
4.
F
F
Currently,
all
the
cloud
cloud
providers
are
in
six
dot,
k8
dot
io,
but
I
saw
a
recent
cloud
provider,
I
think
think
her
bill
or
something
which
was
outside
yeah,
and
I
I'm
not
sure
about
the
overall
strategy
in
the
long
term
about.
A
So,
there's
no
like
a
requirement
to
put
your
provider
in
six
dot,
kids,
dot,
io,
actually,
a
bunch
of
providers
or
a
bunch.
Several
providers
are
not
in
six
out
kids
that
I
owe
like,
for
example,
the
one
daniel
linked
in
the
chat,
the
ones
that
are
in
six
dot
cake
dot
io.
Those
are
the
ones
that
are
sponsored
by
sig
cluster
life
cycle,
which
means
they're
part
of
the
sig,
and
so
it
takes
a
certain
like
like
it.
It
takes
like
I'm,
not
exactly
sure
what
the
requirements
are.
A
I'd
have
to
check
with
the
sig,
but
there
are
some
requirements
you
need
to
fulfill
in
order
to
be
like
part
of
the
sig
repos,
but
that
doesn't
stop
you
from
having
your
own
provider,
and
it
also
doesn't
stop
you
from
adding
it
to
this
list
in
here.
So
there's
a
few
things
you
have
to
do
when
you
start
a
new
provider.
First
thing
is:
make
it
discoverable
by
adding
it
to
this
list
in
the
docs.
This
pretty
much
anyone
can
do
as
long
as
it's
an
existing
repo
and
it's
being
maintained.
A
A
And
then
I
guess
the
last
thing
is
to
move
it
into
c
cluster
life
cycle,
which
may
not
be
something
you
want
to
do
at
all.
Depending
on
what
your
objectives
are.
A
Cool
anything
else,
anyone
want
to
add
anything.
A
Okay
and
then
I
guess
we
usually
do
a
little
like
welcome
at
the
end
for
anyone
who's
new.
So
since
we
have
extra
time,
is
there
anyone
who
is
new
to
this
meeting
who
hasn't
been
here
before
and
who
just
wants
to
say,
hi
and
introduce
themselves
jacob.
G
Yeah
hi,
I'm
not
really
new.
I've
been
here
a
few
times
already,
but
I
never
introduced
myself
and
also
I
left
you
hanging
two
weeks
ago.
So
sorry
for
that.
But
after
you
did
do
it
in
the
beginning,
I
then
went
to
the
kitchen
and
prepared
some
something
I
couldn't
leave,
but
I
felt
bad
about
it.
So
I'm
doing
it
now,
I'm
jacob
or
jacob
I'm
also
working
or
I'm
working
for
deutsche
telekom
same
as
max
link
and
we're
in
the
same
team.
G
Yeah
and
we're
looking
to
contribute
more
to
cluster
api
and
to
some
providers
vsphere,
especially
in
metal
three,
maybe
currently
mostly
interested
in
in
the
cluster
class
stuff,
because
that's
something
we
could
really
use
and
yeah,
oh
and
my
github
handle,
is
s-e-h-r-e-j
or
shresh
or
shrey,
depending
on
how
you
pronounce
it
yeah
hello.
Everyone.
A
All
right,
I
guess,
you're
the
only
one
for
this
meeting,
but
thanks
everyone
and
see
y'all
in
slack
and
online
thanks.