►
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
Right,
hello:
everyone
welcome
to
the
cluster
api
provider
aws
office
hours
on
the
20th
of
september
2021,
a
reminder
that
we
abide
by
the
cncf
code
of
conduct
so
generally
just
be
kind
to
each
other
and
respect
each
other.
A
I've
so
definitely
share
the
meeting
notes
and
I'm
in
the
chat
window
and
I'm
also
sharing
them
on
the
screen.
A
Any
topics
you'd
like
to
cover-
please
add
them
to
the
agenda.
Now,
if
you
want
to
to
talk
during
the
meeting,
can
you
use
the
the
raise
hands
feature
as
well
bring
the
participants
up
cool,
as
this
is
our
first
meeting
in
a
while?
I
guess:
isn't
it
are
there
any
new
people
that
would
like
to
unmute
and
say
hello
and
just
just
introduce
yourself.
C
Yeah
hi
everyone,
I'm
mother,
madhu
agarwal
and
I
have
recently
joined
the
tkg
cap
provider,
india,
team
and
so
I'll
tell
a
bit
about
myself.
So
I
have
a
total
experience
of
three
years
and
before
joining
vmware
I
was
with
amazon
working
on
the
india
payments
team
and
I
have
a
few
like.
I
had
recently
started
contribution
in
cluster
api
and
cluster
api
aws
provider.
So
so
now
I
will
continue
with
that
in
this
team
through
vmware,
so
pretty
excited
to
learn
now.
Thank
you.
A
Awesome
thanks,
madam,
is
anyone
else
that
would
like
to
have
me.
D
Yep
hi
there
I'm
suraj
deshmukh,
I
work
for
microsoft
and
I'm
currently
working
on
this
pr
on
the
upstream
or
the
cluster
api
repo
for
ignition
support
and
yeah.
This
is
my
first
time
here
thanks.
A
Is
that
is
you've
got
your
hand
raised?
Did
you
want
to
argue.
E
Yeah,
I
just
want
to
say
hi
hi,
I'm
kilo,
I'm
in
the
same
larger
team
as
suraj,
and
I'm
also
here
for
the
flatka
linux
team
and
want
to
learn
a
little
about
the
timeline
for
beta1
cool.
A
Yeah
welcome
and
then
woody
you've
got
your
hands
raised
as
well
go
for
it.
F
Hello,
hey,
my
name
is
winnie
khan,
so
I
work
for
vmware.
I've
been
working
in
coordinated
space
since
2019,
but
until
now
I've
been
working
mostly
on
our
commercial
product.
I'm
really
excited
to
start
contributing
to
kappa.
A
Cool
and
winnie,
did
you
get
your
organization
membership
through
okay,.
A
Take
too
long,
of
course,
so
there
are
no
no
psas
this
week
and
there
is
one
action
item
from
the
last
meeting
and
that
was
around
issue
2662,
where
I
was
just
going
to
check
with
some
of
our
customers.
If
they
were
it's
going
to
cause
issues
and
it
wasn't,
but
it's
it's
merged
anyway.
So
so
that
that's
good.
So
I
don't
think
we
need
to
cover
that
unless
anyone
has
any
questions
on
that.
B
Thanks
yeah,
I
I
noticed
there
was
one
one
outstanding
pr
that
yeah
that
added
some
some
validation
to
aws
machine
and
into
this
machine
template
and
it
looked
like
it
would.
It
would
be
a
breaking
change
in
that
it.
It
like
it
changes
the
existing
api
behavior,
but
at
the
same
time
it
it
appears
that,
if
you
so
previously,
even
though
the
validation
was
absent,
if
you
introduced
an
invalid
value,
you
couldn't
have
created
a
cluster.
B
So
on
the
one
hand,
it's
adding
this
validation
is
a
breaking
api
change.
On
the
other
hand,
it
in
practice
should
not
affect
any
users.
If
we
may,
you
know
like
based
on
the
assumption
that
invalid
values
would
would
not
have
given
you
a
working
cluster,
and
so
this
this
was
an
existing
pr
and
then,
some
days
later,
I
actually
had
cause
to
submit
a
a
sort
of
a
similar
pr.
B
This
one
it
looks
like
we,
so
we
we
have
support
for
different
elb
or
the
the
elastic
load,
balancer
schemes
and
the
aws
cluster
object.
So
we
can
the
kappa
controller.
The
cluster
controller
can
bring
up.
It
can
configure
a
different
elb
scheme
and
it
appears
that
we
had
a
typo
and
capitalizing
internet
facing.
B
So,
if
you
that
value
is
not
is
not
valid
and
again,
this
is
so.
This
needs
to
be
fixed
in
order
to
allow
users
to
to
make
configuration
changes
to
the
to
the
load.
Balancer,
there's
there's
a
there's,
a
field
with
with
some
subfields
for
the
load,
balancer
configuration
and
if
you
are
yeah,
if
you
make
any
changes
to
those
fields,
then
this
this
value
gets
defaulted
and
it's
capitalized,
and
then
that
is,
that
is
that
that
becomes
an
in
sorry.
B
It's
not
an
invalid
value,
but
the
the
the
cluster
the
cluster
deployment
fails,
because
the
actual
you'll
be
in
in
ec2.
Its
scheme
is
internet
facing
with
a
lower
case.
In
any
case,
this
is
sort
of
the
same
situation
where
I
I'd
like
to
fix
this.
This
is
a
you
know
what
I
think
we
would
consider
breaking
api
change.
B
However,
if
you,
you
know,
if
you
were,
if
you
were
using
internet
facing
capitalized,
you
would
not
have
been
able
to
deploy
a
cluster
successfully
so
kind
of
two
two
prs
in
this
sort
of
the
same
question
around
them
like.
Do
we
yeah?
What
what
do
we
do?
Do
we
do
we
merge
them?
Do
we
do
we
bump
an
api
version?
So
that's
that's
why
I
wanted
to
have
a
conversation.
A
Yeah,
that's
an
interesting
one,
so
the
yes,
the
actual
the
field
name
stays
the
same,
but
it's
actually
the
value
isn't
it
that
is
going
to
change
or
the
behavior
based
on
that
value
is
changing.
Could
we
I
guess
to
stop
things
breaking?
Could
we
support
the
old
value
and
change
it
to
to
the
new
value?
If,
if
we
encounter
it,
so
we
don't
break
any
assumptions
people
have.
I
don't
know.
What
do
you
think
to
that
idea?
Daniel.
B
That's
that's
interesting.
I
hadn't
actually
considered
it
yet.
I
I
know
that
anyone,
you
know
if
you,
if
you
define
an
aws
cluster
object
and
you
explicitly
put
internet
facing
with
the
capital
capitalize,
which
is,
which
is
what
I
think
the
only
valid
value.
So
you
can
only
have
internal
and
then
internet
facing
with
capital.
I,
if
you
use
the
the
capital
I
internet
facing,
you,
won't
be
able
to
create
a
cluster
because
aws
the
aws
cluster
controller
will
never
be
able
to
find
the
the
actual
elb
that
it
has
created.
B
H
Yeah,
I
just
shared
an
issue
that
I
worked
on
six
months
ago,
or
so
so
the
issue
was
the
actually
non-capital
internet
facing
was
not
working.
H
So
if
I
don't
remember
wrong,
like.
H
That's
why
I
made
the
enum
yeah:
why
don't
we
check
offline
and
follow
up
asynchronously
on
the
issue
that
daniel
created.
B
Okay
and
and
then
so,
I
think
there
are
the
two
issues
that
are
kind
of
in
this.
In
the
same,
they
have
the
sort
of
same
question
for
both
of
them
right,
one
is
the
that
first
one,
the
2740
with
the
validation.
B
I
Yeah
thanks
daniel
for
bringing
this
here
actually
with
this
issue.
There
is
an
issue
with
omit
empty
tag
like
with
the
optional
fields
in
some
of
the
structs.
We
are
using,
omit
empty
and
with
some
required
fields
also,
we
are
using
omit
md
that
we
shouldn't
use.
So
that's
why
I
created
a
separate
issue
which
can
be
an
api
breaking
change,
where
we
have
to
make
the
package
wise
validation
as
required
and
have
to
mention
optional,
with
only
the
optional
fields
so
maybe
like.
I
I
think
we
can
go
ahead
with
this
change,
because
it's
not
breaking
the
existing
users
as
well
as
well
as
the
old
users,
because
if
you
don't
pass
the
instance
type
initially,
then
it
will
fail
further
with
the
error
that
the
instance
type
provided
is
wrong.
I
So
it's
like,
like
other
people,
can
suggest
like
what
we
should
go
ahead
with
that.
But
it
won't
be
that
breaking
change
and
we
have
a
separate
issue
to
track
all
those
changes
where
the
api
can
break.
H
Also,
if
we
have
more
changes,
we
can
make
a
new
release
for
v1
alpha
4
types,
because
it's
been
a
while,
since
our
last
change,
anyways.
I
Yeah
there
are
like
lot
of
fields
I
have
seen
where
we
need
to
change
the
types
and
make
remove
the
omit
empty
thing
and
make
other
things
as
required.
A
Sorry,
I'm
just
taking
notes
and
does
anyone
else
have
any
any
opinions
on
this
one.
A
E
Yeah
we
have
a
number
of
changes
currently
in
flight
that
we'd
either
be
like
to
understand.
The
the
impact
of
it
looks
like
most
of
the
changes
that
we
have
in
flight
aren't
even
much
of
a
of
an
issue
for
the
code:
freeze,
the
upstream
code
freeze
and
we're
we're
trying
to
better
understand
that
for
every
provider,
so
we're
joining
the
the
meetings
which
arguably
we
should
have
done
a
lot
earlier.
E
A
H
No,
but
I
think
it
we,
we
need
to
follow
a
similar
code
phrase
with
cluster
api.
Again
there
was
a
threat.
Today
I
just
saw
discusses
about
our
plans
for
v1
in
the
cluster
api
channel.
I
I
think
we
can
follow
up.
E
Yeah
pretty
much
so
I
believe
the
other
providers
go
go
kind
of
in
the
same
direction.
I've
attended
the
azure
provider
meeting
and
I
was
kind
of
similar
there.
Maybe
suraj
do
we
have
anything
in
flight
that
may
impact
the
the
apis
in
any
form.
Are
you
aware
of
anything
there
that
we
should
discuss.
D
Oh
sorry,
yeah
I
was
saying
not
that
I'm
aware
of
in
terms
of
breaking
changes.
The
stuff
on
the
aws
side
is
pretty
much
all.
I
think
almost
done.
It's
just
that
the
upstream.
The
capi
work
is
what
is
blocked
on
the
e2e
tests
and
feature.
B
I
K
So
I
just
wanted
to
bring
this
to
notice
that
me
and
other
last
week
have
reviewed
the
test
grid
test
failures
and
we
found
few
issues
there.
So
I
can
tell
briefly
about
some
of
the
issues
so
there's
two
seven,
six,
four
okay
yeah
so
2764
is
about
the
eks.
E2E
periodic
test
is
failing.
It
is
failing
sporadically,
so
sometimes
the
cluster
is
not
getting
deleted
because
of
some
of
the
resources
are
not
getting
deleted
properly.
K
I
have
not
dig
in
deep
more
to
that,
but
I
think
this
has
to
be
look
looked
up.
I
also
checked
that
this
week
we
are
getting
the
green
reports,
but
there
were
failures,
mini
failures
last
week,
so
it's
worth
checking
this
one,
and
then
we
have
2
763.
K
There
is
a
failure
related
to
multi-tenancy
test
as
well,
so
that
I'm
checking
there
is
again
some
timeout
issues
happening
related
to
the
control
plane
creation.
So
we
are
not
actually
waiting
for
the
enough
time
in
the
test
cases.
So
that's
why,
even
before
the
control
plane
is
up
and
running,
our
test
case
is
getting
timed
out.
So
that's
that's
another
issue
and
then
there
is
one
more
issue
where
we
need
some
cleanup
from
v1
alpha
3
release.
K
So
there
were
some
test
cases
which
now
they
said
that
are
not
much
relevant
to
us
now,
so
we
can
get
rid
of
them.
So
for
that
also,
I
have
raised
a
pr
already.
Maybe
you
guys
can
check
that
out,
and
apart
from
that,
there
was
also
a
conformance
test
issue
which
was
getting
timed
out
due
to
the
log
collection
was
taking
too
long.
So
I
think
this
has
happened
before
as
well.
K
There
is
something
wrong
with
the
image
builder,
where
the
system
d
is
getting
immensely
large
in
size
because
of
the
audit
logs.
So
that's
why
it's.
Our
test
cases
are
finally
running
successfully,
but
at
the
end,
it's
failing
because
the
log
collection
didn't
succeed.
K
So
I
think
this
has
to
be
fixed
in
image
builder,
but
as
a
hack
for
now
we
have
provided
a
fix
in
a
cap
a
so
I
have
raised
pr
for
that
also
so
yeah.
I
think
we
have.
We
need
to
have
a
follow-up
issue
for
the
image
builder.
I
think
there
is
already
an
issue
in
image
builder,
but
that
was
closed
so
now
that
has
opened
it
now
again,
so
yeah
that
that's
what
I
wanted
to
bring
it
to
them.
A
Cool
so
yeah,
I
see
you've,
you've
signed
those
first,
two
ones
haven't
you
and
then
you've
got
the
multi-tenancy
one
and
the
eks
failure
I
can
I
can.
I
can
have
a
look
at
this
one
or
if
you
want
to
look
at
it
and
ping
me
it's
up
to
you,
I
can
do
either
all
we've
had.
We've
had
issues
with
deletion
before
in
the.
B
A
So
there
is
a
there
is
some
code
in
there
that
checks
for
child
child
resources.
So
I
wonder
if
it's
that,
but
if,
if
it's,
if
it's
flaky
it's
passive,
sometimes
I'm
not
sure
but
yeah,
we
can
have
a
look.
B
Thanks
sneak
a
question:
have
we
ever
considered
adopting
an
existing
elb
so
today,
when,
when
you
create
an
aws
cluster,
the
controller
generates.
B
An
elb
with
a
with
a
specific
name.
I
think
that
is
anyway.
It's
it's
it's
a
it
acts
I
think,
is
the
primary
key
in
ec2,
and
so
it's
it's
not
really
possible
to
guess.
Yeah.
Basically,
it's
not
possible
to
you
know
to
create
an
elb
that
aws
cluster
controller
will
will
recognize
and
and
and
and
and
adopt.
So
I
don't
know
if
we've
ever
discussed
that
if
anybody's
ever
had
a
use
case
for
that,
I
was
just
curious.
A
Yeah,
I've,
not
I've,
not
I'm
not.
I
don't
remember
any
conversations
around
that
area.
Sadef
or
anyone
else
have
you.
H
There's
a
load
balancer
provider
kept
in
with
cappy
this
probably
a
loss
using
any
other
logo
monster,
but
it
is
still
in
very
early
stage.
H
I'm
gonna
put
the
link
to
the
cat
other
than
that.
We
didn't
discuss
supporting
it
in
kappa
itself,.
B
Do
you
know
if
there
would
be
so
suppose
that
I
I
I
were
to
to
pr?
You
know,
change
the
reconciliation
that
that
allows
for
for
an
existing
elb
with
do
you
know
if
there
would
be
any
strong
objections
against
that?
I
just
just
just
I
don't
know
philosophically
or.
H
I
don't
think
so.
Can
we
start
with
an
issue
and
use
cases
so
that
we
can
discuss
on
the
issue?
If
no
objections,
we
can
go
ahead
and
we
can
check
lord
one
answer
proposal
and
where
it
is
act.
If
there's
a
common
work
there
that
we
can
like
adult,
we
can
maybe
wait
or
implement
ourselves.
J
I
Actually,
daniel
I'm
working
with
this
on
this
proposal
with
nadir,
so
it's
pretty
much
very
in
the
initial
phases
right
now
like
this
week
and
in
coming
week,
my
target
is
mainly
collecting
the
use
cases.
So
do
you
have
any
particular
use
case
in
mind
like
for
which
you
are
asking?
Maybe
we
can
add
it
in
the
proposal
itself
and
discuss
it
with
the
company
further.
B
Yeah,
I
can,
I
can
add
it
to
the
issue
I
mean
or
if
we
have,
if
we
have
time
I
can,
I
can
talk
about
a
use
case,
but.
A
B
Yeah
sure
absolutely
I
anybody
else
have
oh,
okay,
all
right
I'll
I'll
keep
working
the
yeah.
B
The
use
case
is
to
be
able
to
adopt
a
a
cluster
that
has
been
created
using
kube
adm,
but
not
kappa
and
and
it's
the
cluster
runs
in
in
aws
of
course,
and
it
has
very
similar,
topology
and
yeah
there,
as
as
far
as
like
we
we've
been
able
to
adopt
the
the
control
plane
and
the
worker
machines,
but
the
the
eel
the
elb
is,
is
the
place
where,
where
we've
run
into
yeah
into
this
issue,
where
you
know
right
now,
our
way
that
we're
working
around
this
is
we
are
cap
is
creating
its
own
elb.
B
We
we
have.
The
cluster
has
its
elb
from
its
own
topology
from
its
own.
You
know,
network
resources,
and
then
we
are
modifying
or
we're
updating
the
the
certificate
of
the
api
server
so
that
it's
valid
for
both
elbs
right
for
for
the
names
for
the
the
dns
names
of
of
both
elbs
and
then
requests
can
come
both
on
both
the
old,
but
it
would
be
great
to
yeah,
have
kappa
adopt
the
the
existing
elb
there?
B
Are
there
are
some
minor
differences
in
the
way
that
the
el
the
the
elb
is
named,
and
but
I
I
believe
that
that
it's
possible
to
like,
I
believe
that
it's
possible
to
guarantee
the
you
know
the
the
the
they're
to
provide
the
same
guarantees
while
you
know,
while
using
a
a
slightly
different
name
but
anyway,
that's
that's,
that's
more
of
the
implementation
details,
but
that's
that's.
That's
the
use
case
adopting
adopting
a
cluster
that
otherwise
has
sort
of
like
bringing
your
own.
B
A
Fair,
it's
a
fair
ask:
isn't
it
yeah
nothing
else.
A
The
good
work
done
by
ankita
reminded
me
of
is:
if
there's
anyone
interested
in
learning
more
about
the
end-to-end
tests,
you
know
how
we
write
them,
how
we
run
them,
how
we
debug
them
feel
free
just
to
ping
again
the
slack
or
on
here,
and
then
we
can
arrange
like
a
like
a
zoom
session
just
to
go
through
that,
if
anyone's
interested,
but
just
to
forward
you,
it
does
take
a
lot
of
time
to
run
these
things
and
it
kind
of
suck
up
a
lot
of
your
your
life
doing
it.
A
But
they're
really
really
helpful,
so
yeah
if
anyone's
interested
just
feel
free
to
ping,
it
ping
on
the
slack
and
we
can
arrange
a
zoom
in
cool.
Is
there
any
final
topics
or
anything
else?
Anyone
would
like
to
discuss.
A
Going
once
going
twice
brilliant
cool:
well,
we
can
call
the
meeting
to
a
close.
Have
a
good
day
have
a
good
evening
and
see
you
in
a
couple
of
weeks
take
care
everyone,
bye,
bye,.