►
Description
A Kubernetes community meeting about the Azure provider for Cluster API. Cluster API brings familiar, declarative APIs to Kubernetes cluster creation, configuration, and management.
A
Foreign
all
right,
sorry
about
that
welcome
everybody.
It
is
March
30th
2023..
This
is
the
cluster
API
Azure
weekly
meeting.
We
are
a
sub
project
of
Sig
cluster
life
cycle
of
kubernetes
umbrella
project
and
as
such,
we
try
to
follow
their
rules
of
conduct,
which
basically
boils.
B
B
A
If
you
wouldn't
mind
adding
your
name
to
this
list,
just
helps
people
Network
or
remember,
who
said
something
useful
at
the
meeting
and
Now's
the
Time,
where
we
usually
let
anyone
who's
here
for
the
first
time
or
has
something
wants
to
introduce
themselves,
do
so
so
be
quiet
for
just.
C
I
think
that
might
be
me
hi
I'm,
Jim,
Walker
I'm
from
enable
which
is
a
SAS
company
and
yeah
just
looking
to
try
and
get
started
with
contributing
to
kubernetes.
This
is
my
first
first
entry
to
this
I'll
put
my
name
on
the
sheets,
but
I'm
not
sure.
C
A
Somebody
put
it
in
chat,
oh
you're,
the
best.
Thank
you
very
much
it's
up
at
the
top
here
as
well
and
in
the
slack
list
in
the
in
the
slack
Channel.
There's
a
link
to
all
this
stuff
as
well,
but
awesome
welcome
Jim
thanks
for
coming.
A
All
right,
let's
move
on,
we
actually
have
very
little
on
the
agenda,
but
that
doesn't
necessarily
mean
we
don't
have
a
lot
to
talk
about
as
we've
learned.
So
everybody
probably
knows
maybe
note
that
the
patch
releases
are
available,
there's
a
fair
amount
in
the
patch
releases.
When
you
actually
look
at
the
release
notes,
the
only
things
that
show
up
in
the
release
notes
are
things
that
we
specifically
flagged
as
this
needs
a
user.
This
is
sort
of
user
facing
so
under
the
covers.
A
You
probably
know,
there's
quite
a
bit
more
bug
fixing
and
other
things
that
go
that
went
on
so
anyway.
These
are
important
bug
fixes.
Why
can't
I
go
back
a
page,
fine
and
thanks
everybody
here,
I
think
pitched
in
on
them.
Quite
a
bit
other
related
news.
A
We
are
working
on
integrating
with
cap
e140
again.
If
you
went
to
the
Cappy
meeting
yesterday,
we're
probably
just
repeating
ourselves,
but
this
is
the
place
to
do
that
and
I
don't
have
any
updated
status
on
that
and
Willie
is
not
Willie's
taking
a
couple
days
off
so
I,
don't
think,
there's
anything
else
to
say
there.
A
D
A
E
A
I
feel
like
Jack,
might
have
an
answer
for
us,
but
he's
on
spring
break
this
week.
F
A
F
It
I'm
just
kind
of
curious
if
there's
been
across
any
of
the
Cappy
providers,
any
kind
of
performance
like
load
testing,
for
how
many
you
know
how
many
clusters
can
be
under
management
on
a
management
cluster
I,
don't
know
if
there
has
do
you
know
Mike
has
any
anybody.
Anybody
else
done
that
kind
of
testing
Kathy.
D
I
know:
there's
I
gotta,
look
back
up
in
the
slack
history.
That's
why
I
was
going
to
ask
in
slack
later
in
general,
Cappy
had
a
recommendation
of
a
few
hundred,
but
I
forget
the
exact
number
of
clusters
inside
of
the
Cappy
side.
D
Before
you
start
doing,
you
get
to
start
really
playing
with
the
number
of
oh,
my
goodness
totaled
mind
blank
on
the
were
the
not
threads
number
of
nodes,
reconciles
reconcilers,
oh
okay,
and
all
all
that
kind
of
stuff,
because
you'll
end
up
in
a
situation
where
you
know
it,
it
just
doesn't
reconcile
some
clusters
because
it's
busy
reconciling
other
ones.
Okay
or
you
know,
you'll
get
stuck
really
behind
the
queue
and
things
will
really
back
up.
F
Yeah
I
think
I
think
we
need
to
get
to
the
source
of
that
and
what
testing
has
been
done
because
I
mean
it's
likely
going
to
be
I
mean
shouldn't
be
really
much
different
today
for
cap
Z
than
the
general
copy
provider,
as
my
okay,
my
gut
my
gut
Here
and
Now,
obviously
like
once,
and
if
we
you
know,
move
forward
on
The
Proposal
with
ASO,
that's
gonna,
definitely
change
the
game
and
I
know
that
we
have
in
our
plans,
assuming
we
do
move
forward
with
that
to
to
do
some
load
performance
testing
there,
but
but
yeah
I
think
it's
a
great
question.
F
F
Right,
yeah,
the
size
and,
what's
what's
what
are
the
like
characteristics
of
it
like
if
you're
just
using
it
as
a
management,
control,
plane
and
nothing
else
like
what
is
the
ideal,
like
you
know,
stamp
characteristics
that
you
need.
You
know
memory
CPU,
Etc,
yeah,.
A
D
Yeah
from
our
side
we're
going
to
be
using
it
just
as
a
management
cluster
yeah,
so
maybe
have
a
local
Argo
on
it,
but
that
doesn't
do
very
much
load.
F
Yeah,
it
makes
sense,
I
mean,
makes
sense,
I
know.
I,
know,
Argo
has
a
bunch
of
performance
data
in
general,
so
that
does
it
does
have
some
load
there
on
Argo
but
I
think
yeah.
We
should
have
something
published
in
Cappy
at
a
minimum
somewhere.
If
it's
not
there,
then
I
think
that's
a
good
question
for
for
Cappy
right.
A
I
see
you
got
a
couple,
people
who
showed
up
a
little
late
and
we
did
really
didn't
have
a
lot
of
agenda.
You
can
see
right
there.
We
mostly
just
talked
about
the
patch
releases
and
then
Mike
had
a
good
question
about
scaling
cap
C.
A
B
Is
there
a
way
we
can
run
Focus
between
test
and
CI,
because
I'm
able
to
do
that
locally,
but
when
I
set
up
the
variable
in
make
file
looks
like
it
is
not
obeyed
in
CI,
because
I
went
and
looked
at
test
info
and
one
of
the
jobs
that
I'm
trying
to
run
is
like
the
optional
job
and
it
has
like
jinco
Focus
involvement.
Credible
explicitly
past
is
optional,
which
eventually
runs
all
the
option
of
tests.
B
Have
you
ever
tried
doing
that
like
anyone
like
or
is
it
like
just
trying
to
understand?
Is
there
anything
that
I'm
missing
or
not
doing,
or
just
that
it's
not
possible
right
now.
A
I
haven't
tried
that
hard,
but
I've
certainly
wanted
to
do
that
before
and
I
I
thought
there
was
no
I
I
didn't
see
a
way
to
pass.
Kinko
Focus
got
it
through
yeah,
but
obviously
you
can
do
it
locally.
Oh.
B
A
A
Guess
is
my
thought
if
you
could
do
that
in
general,
because
there's
a
lot
of
times
where
we
were
sort
of
stuck
last
week
on
a
lot
of
PRS,
because
the
cluster
class
spec
in
the
optional
test
was
failing
and
it
would
be,
it
would
have
been
nice
if
we
could
have
just
skipped
that
because
some
of
them
we
knew
that
was
irrelevant,
but
no
you
get
the
whole
set
of
optional
tests.
As
far
as
I
can
tell
does.
A
B
B
Yeah
I
mean
in
my
PR
just
to
actually
make
sure
that
I
run
that
particular
test
for
my
testing,
and
it
eventually
does
not
read
from
the
environmental
value
of
the
response
so
but
I
think
yeah
once
I'm
done
with
that,
I
can
look
into
ways
how
we
can
do
that
in
a
meter
way,
probably
make
it
be.
Awesome
thanks.
Man.
G
G
A
Cool
any
other
topics.
Anybody
wants
to
talk
about.
B
I
just
want
to
sorry:
I
did
not
raise
my
hand,
but
I
just
wanted
to
tell
one
thing
that
for
the
workload
and
kpr
like
this
is
the
first
time
you
know,
I
was
able
to
like
do
a
successful
authentication
in
the
ET
test.
So
this
is
a
good
sign
and
thanks
to
sale
and
Anish
for
helping
on
that
so
yeah
right
now.
The
test
tests
are
failing
for
a
different
reason,
mostly.
B
I
suspect
I
saw
lots
of
quota
allocation
issues
in
optional
tests,
but
as
far
as
my
test
was
concerned,
the
workload
identity
test,
I
didn't
see
anything
significant,
but
the
virtual
machine,
the
operation,
the
long
running
operation
did
not
complete
eventually
and
there
were
no
cloud
in
it
blocks.
So
I
didn't
get
much
on
what
went
wrong
so
I'm
just
trying
to
run
that
particular
workload
and
identity
to
be
just
isolated
so
that
if
there
is
anything
wrong.
So
that's
the
update
on
that
part
that.
B
A
All
right,
unless
we
have
anything
else,
Let's
do
let's
call.
Let's
do
a
quick,
Milestone
review.
A
A
E
A
Thanks
yeah
I
didn't
want
to
single
you
out.
I
just
had
obviously
a
tiny
amount
of
context
on
that
one.
We
just
talked
about
workload:
identity,
I,
don't
this
one
I'm
gonna
follow
up
on
I!
Think
still.
A
F
I
think
with
that
one
I
mean
right:
we
I
think
we
we
need
to
break
it
down.
Right,
I,
think
that's
kind
of
one
of
the
things
that
is
missing,
because
I
think
we're
kind
of
waiting
right
on
ASO
proposal
to
move
forward.
If
that
moves
forward,
then
we
need
to
look
at
which
resources
we
want
to
move
to
ASO
and
which
ones
we
want
to
migrate
to
the
new
SDK
right
right.
A
F
A
This
one
kind
of
hangs
off
the
same
things
moving
forward.
That's
fine.
A
A
A
A
A
This
seems
like
it's
close
to
ready
to
go.
So,
let's
add
it
to
the
milestone,
especially
since
Willie's,
not
here
we'll
just
put
him
that's
what
happens.
A
A
Ignore
all
the
in-progress
ones,
I
know
this
has
been
sitting
here
for
a
while.
Aso
is
already
on
the
milestone.
B
Yeah
so
I've
been
like
following
this
issue,
and
you
know:
yeah
David
commented
on
this
issue
explaining
like
possible
cause
of
this
issue,
but
I.
G
B
B
Let's
see,
do
you
have
any
thoughts
or
that's
fine.
If
you
have
not
went
through
this,
but
you
know,
I'll,
just
I
would
just
need
some
help
and
guidance
on
this
issue,
because
I've
seen
that
this
occurs
a
lot
of
time-
and
there
are
other
folks
in
the
community
who
are
faced
it
and
I
can
see.
Posts
are
connected
there.
So
yeah
just
wanted
to
break
this
up
and.
B
Go
ahead,
I
can
take
this.
Take
this
issue
up
and
the
I
think
the
fix
would
be
a
small
one
so
like
if
you
can
make
it
on
1.9,
milestone
in
our
local
work
on
this
one,
but
just
want
to
understand
more
on
why
this
error
occurs.
A
C
A
A
Now
we've
already
looked
at
these,
so
let's
call
it
good
unless
there's
something
else.
F
No
I
don't
think
so,
but
I
do
I'm.
Just
taking
a
look
at
that
issue,
I
mean
it
does
look
like
it's
something
we
definitely
should
prioritize.
So
you
know
I,
think
yeah
reaching
out
to
David
would
be
would
be
good
and
yeah.
Let
you
know,
let
us
know
if
we
we
can.
You
know
sure
thanks.
A
C
You
have
something
yeah
I'm,
going
to
take
a
look
at
the
issue,
adding
missing
test
coverage
for
Azure
machine
webhook,
dot
go
and
just
you
know,
going
through
setting
up
my
environment
and
getting
everything
to
build
and
stuff
like
that.
If
I
went
into
trouble,
which
I
just
post
on
the
sick
cluster
life
cycle
stack
and
are
most
of
the
people
here
in
the
western
coast
of
the
USA.
A
There's
a
more
specific
Channel,
there's
a
cluster
API
Azure
Channel
and
we're
all
we're
pretty
much
we're
all
there
and
we're
mostly
in
the
in
the
Pacific
time
zone,
but
I'm
in
the
Mountain
Time,
Zone
John's
in
Eastern
or
Central,
and
so
we're
a
little
spread
out,
but
mostly
in
the
western
time
zone.
A
Thank
you,
yeah.
Thank
you.
If
you
need
any
help,
definitely
just
post
whatever
in
the
slack
and
somewhere
help
there.