►
From YouTube: SIG Cluster Lifecycle - CAPV Office Hours - 2023-08-17
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
So
yeah
hi
and
welcome
to
the
office
hours
today
for
cluster
API
provider,
vsphere
yeah.
If
you
follow
the
code
of
conduct
and
yeah
I,
think
everyone
know
in
here
knows
that
so
yeah,
let's
maybe
kick
off,
feel
free
to
add
to
attendees.
Thank
you
nadir
for
adding
everyone
and
I
See
No
Agenda
items.
A
B
Yep
Yeah,
so
basically
I
should
probably
flip
it
around
one.
Second,
so
I
think
the
first
one's,
probably
hey.
We
released
clap
B18
last
week,
so
that's
done.
B
I
think
the
major
features
are
I
mean
yeah.
It's
based
on
copy
one.
Five
soft
shutdown
is
implemented,
not
default,
but
it's
implemented,
so
you
can
configure
via
power
off
mode.
We
improve
session
handling
a
bit.
What
else
did
I
put
into?
It
is
not
I.
Think
that's
like
a
major
thing.
We
added
a
bunch
of
bug,
fixes
we
have
a
lot
of
new
automation,
but
for
users
I
think
that's
just
most
of
it
otherwise
feel
free
to
look
at
the
release.
Notes,
yep,
I
think
that's
all
about
the
release.
B
And
I
think
I'm
just
going
to
leave
next
point,
which
is
basically
that
we're
currently
working
on
128
support
and
128
support
will
just
come
down.
I'm,
pretty
sure
it
just
comes
down
to
adding
test
coverage
for
1.28,
so
the
current
status
that
gong
built,
ovas
and
they're
uploaded
and
already
like.
So
we
have
PR
open,
which
updates
to
read
me
and
links
to
those
obas
somewhere
in
Google
Cloud
and
in
the
same
PR.
We
also
modify
our
end-to-end
test
storage
actually
test
with
128..
B
There
is
one
open
thing
that
we
depend
on,
which
is
basically
the
cloud
provider
vsphere
with
128,
but
I
was
writing
with
Luke
Bryan
about
it
and
it
seems
like
there's
some
sort
of
circular
dependency.
No,
so
basically,
no
not
really
circular,
but
it's
like
republished
ovas
128.
He
likes
to
test
the
cloud
provider
128
with
our
ovas
and
then
releases
the
cloud
provider,
and
then
we
can
use
it.
But
here.
C
B
B
Yeah
I
assume
it's
fine,
it's
just
that
he
meant.
Usually
he
made
it
with
his
release
for
our
obas.
We
couldn't
because
he
wanted
to
verify
it
and
stuff
if
it
works
together.
F
Yeah
I'm
doing
some
restructuring
of
the
content
test.
Folder.
F
All
the
yamos
we
just
jumped
into
one
single
folder,
so
in
copy
we
have
main
1.5,
1.4,
1.3
and
so
on,
and
then
that
allows
us
to
Cluster
control,
upgrade
from
X
version
to
main
or
between
two
versions,
so
pretty
much
just
yeah
restructuring
stuff,
so
that
works
ultimate
Cafe.
F
So
we
can
do
because
you've
got
an
upgrade
that
may
or
may
not
be
really
important
because
of
trying
to
deprecate
and
remove
the
v10f3
and
v104
types
and
a
big
part
of
that
is
testing
how
all
references
and
other
references
are
upgraded
during
reconciles.
So
I've
got
another
PR
open
for
that.
At
the
moment,
it
looks
to
me
like
happy-
is
pretty
good
on
that
so
far,
but
because
we
don't
have
an
end-to-end
test
for
failure,
domains,
deployment
zones
or
cluster
identity.
I
haven't
tested
those
properly
yet
but
yeah.
B
Okay,
thank
you
very
much
yeah.
That's
definitely
also
one
thing:
I'm,
not
sure
if
you
mentioned
it
before
in
office
hours,
but
we
open
an
issue
to
get
rid
of
you
and
I
for
another.
Four
final
I
think
we
know
it
eight
or
ninth,
the
ninth
minor
release
already,
which
was
used
in
beta1.
B
So
it's
time
to
get
rid
of
a
free,
another
form
and
I
think
it
we
actually
have
to,
because
copy16
will
drop
even
Alpha
free,
which
means
that
our
code
won't
compile
anymore
because
we're
embedding
we
won't
have
free
types
and
or
we
will
have
three
times
Sonia.
B
F
Yeah
I
don't
know
when,
because
I
didn't
do
the
archeology,
but
we
haven't
been
running
the
upgrade
tests
for
being
happy
with
three
one
alpha
four
or
three
one
of
three
for
a
long
time.
I
don't
know
when
the
last
time
we
actually
tested,
even
after
three
version
was
but
it's
probably
quite
a
while
ago,
so
I
think
it's
fine
to
get
rid
of
it,
because
it's
effectively
not
supported.
E
B
Yep
I
think
that
should
cover
all
that
stuff,
maybe
a
note
on
the
CI
environment
or
try
to
keep
it
hold
up.
You
know
sorry
try
to
keep
it
high
level,
so,
basically
I'm
in
the
progress
of
getting
an
OU
or
our
vmc
environment
I
have
somewhat
control.
E
B
Drawing
up
first
pictures
and
then
I
don't
know
we
can
figure
out
what
you
want
to
do
going
forward.
B
I
mean
the
environment
roughly
stays
the
same,
but
there
is
some
discussion
to
be
had
with
sick
testing
just
about
they
don't
like
it
a
lot
if
there
is
if
there
are
antennas
running
via
Pro,
which
are
not
under
their
control,
but
I'm
I
will
take
a
look
at
that
of
an
hour
and
we'll
see
what
it
goes.
But
yeah
I
mean
there's
definitely
like.
Let's
say
one
effort,
just
look
at
the
current
events
that
we
have
and
clarify
your
six
testing
under
which
circumstances
or
in
which
way.
B
That's
that's
fine
and
I,
see
that
as
a
base
is
potentially
extender
and
turn
testing
to
more
stuff.
So
I
think
we
had
hump
lands,
probably
for
years
to
as
a
test
supervisor
to
also
test
GPU,
stuff.
I
think
those
are
definitely
two
big
ones
and
I
think
there
are
also
plans
which
are
more
already
possible
or
no,
let's
say
if
you
just
use
a
current
BMC
and
make
some
additions
or
some
changes,
then
we
can
also
could
also
cover
multi-ac
tests.
For
a
guest,
multiple
compute
classes,
probably
I,
have
to
talk
to
Sagar.
B
He
told
me
has
some
issue
somewhere
dying
around,
but
my
contact
is
so
multi.
Ac
should
be
possible
with
vmc,
assuming
that
we
make
some
changes
to
the
environment
and
also
node
ipam
should
be
possible.
We
just
have
to
implement
something
for
it,
but
yeah,
currently
just
getting
an
overview
on
that.
It's
also
a
question
of
priorities.
Understand
if
you
have
to
bandwidth
to
to
do
something.
A
I
think
that's
it
for
news
and
updates.
If
nobody
has
another
topic,
I
would
like
to
talk
about
one
issue.
This
opened.
A
Which
is
migrating
the
markdown
lyncher
to
something
else
which
we
are
yeah?
It's
coming
from
this
issue
actually
created
some
weeks
ago.
The
current
state
is,
we
use
a
self-built
markdown
lint
image,
which
is
hosted,
which
is
a
very
old
B,
only
works
for
amd64,
so
everyone
running
our
Max
currently
cannot
run
the
Magna
blunter
yeah,
and
this
image
is
hosted
in
our
own
gcp
project.
I.
We
tried
to
do
some
or
we
collect
some
information
in
this
issue.
A
There's
one
solution
which
would
be
some
linter
coming
from
Fox
foxy
project
from
Google.
A
A
A
E
B
A
That
that's
basically
getting
it's
not
installable
using
go
install
because
that's
just
a
huge
record
and
but
yeah
the
snippet
is
enough
to
get
the
linter
downloaded
and
installed.
B
E
B
Verify
I
think
maybe
you.
B
F
I
guess
the
alternative
that
I
don't
see
mentioned
here,
but
I
didn't
check.
The
issue
is
that
we
could
just
drop
this
linter
if
it's
difficult
to
find
a
good
replacement
and
we're
not
happy
with
the
current
situation
and
then
what
do
we
lose?
I?
Don't
think
we
run.
F
Drunk
or
cluster
API,
which
is
a
problem
I,
think
because
tables
and
stuff
can
get
unformatted
like
we
do
lose
stuff,
but
I
don't
know
if
this
lender
actually
takes
care
of
that
right
now
in
capfi,
I
don't
need
to
that'd
be,
but.
A
I
I
could
give
it
another
look
and
actually
just
provide
the
list
of
what
it
would
say
or
wants
to
get
fixed,
and
then
we
could
take
it
async.
So.
A
Be
fair,
I
give
it
a
try,
take
a
look
and
see
how
many
findings
you
would
have
with
that
linter
here
and
maybe
if
there
are
still
rules
in
there,
but
you
would
care
about
maybe.
B
F
F
So
we
wouldn't
be
able
to
use
it
for
Kathy's
book
markdowns,
at
least
if
we,
because
those
use
a
lot
of
HTML
tags
because
of
the
way
the
markdown
book
works.
A
Well,
what's
possible
with
the
the
other
linter
is
you
could,
for
example,
add
configuration
in
the
GitHub
directory
and
exclude
certain
rules
from
being
run
there
but
yeah
question
if
it's
worth
it
yeah.
E
F
Also,
the
the
license
on
the
future
tool
is
bring
the
learn
bells
up
to
the
hash
course
up
this
week,.
E
A
Yeah
it
just
annoyed
me
that
I
wasn't
able
to
run
it
locally
yeah
but
yeah.
Let's
go
to
the
next
topic.
Stefan.
B
Yeah
I
just
want
to
bring
up
this
one,
so
in
my
opinion,
we
should
try
to
get
this
one
merged.
I.
Think
the
pr
which
like
introduce
so
I,
don't
know
how
the
behavior
was
before
the
pr
which
introduced
some
of
the
code
that
we're
modifying
here.
But
it
feels
like
what
we
currently
have
is
definitely
broken.
B
B
D
B
No,
not
me
as
well,
it's
just
like
would
be
like
just
good
to
have
like
a
second
well-informed
opinion,
nothing
against
Christian
and
Korean,
and.
C
I'm
happy
also
to
to
to
to
explain
that
later.
If
someone
wants
to
page
me,
but
that's
nothing.
What
wasn't
was
broken
on
the
previous
PR,
but
we
never
actually
had
this
work
in
with
multiple
clusters,
and
we
we
just
figured
that
that
out
when
we
had
like
a
simple
storage
policy
and
failure
domains
and
since
they
started
not
being
to
be
able
to
close.
C
Yeah,
the
the
other
one
I
think
was
just
fixing
when
you
select
data
store
cluster,
which
is
like
a
meth,
a
meta
object
in
in
the
sphere
right
so
because
we
never
filtered
when
you
ask
for
a
data
started.
What's
returning
the
data
store
and
the
data
start
cluster,
so
it
was
failing
this
one.
C
It's
it's
not
related
to
that
one,
but
it's
kind
of
an
improvement,
because
we
are
not
just
relying
on
the
data
star
cluster
anymore
or
like
getting
the
data
start
cluster,
but
instead
we
are
just
getting
the
data
stores
from
a
cluster
that
we
select
and
checking
if
they
are
available
for
us
or
not
right.
So
that's
that's
kind
of
yeah.
B
See
that
oh
yeah
yeah,
that
would
be
good,
but
yeah
no
rush.
B
A
Okay,
I
think
that's
it
for
the
topics
no
action
items
listed
so
and
the
way
is
we
usually
go
over
the
pull,
requests
and
issues
again
or.
B
B
B
A
E
B
Yeah
I
think
the
plan
last
thing
we
talked
about
it
was
that
nadir
wanted
to
reactivate
some
stuff,
that
you
want
to
write
for
core
copy
and
then
put
something
on
this
PR
that
we
don't
want
to
do
it
if
I'm
not
mixing
this
up,
not
here
yeah,
yeah,
okay,.
A
D
A
Also
context
did
not
change
here
in
that
case,
compared
to
last
meeting
Arsenal
movement
here,
oh
yeah,
that's
the
proposal
topic
right.
B
B
D
Yeah
well
so
it
Microsoft
is
adding
support
for
the
semi
standard.
Yes,
metadata
standard
that
Andrew
Cuts
developed
and
that
will
upper
this
basically.
E
D
E
A
Yeah
she
said
one
was
the
one
we
talked
about.
It's
bonus,
work
in
progress.
B
I
guess
just
maybe
maybe,
while
you're
here,
that's
the
sound,
find
in
general.
The
idea
so
I
think
the
idea
was
in
addition
to
or
use
an
impossible
authentication
to
support
saml
and
if
I
got
it
right
from
LeBron
on
the
issue
or
something
they
already
have
it
working
in
club
provider.
Vision
and
you
want
to
implement
the
same
here.
But
I
didn't
look
at
it.
So
just
if
someone
has
like
high
level
opinions
or
something.
D
I
don't
but
yeah.
Let
me
ask
some
people
in
there
at
least
some.
A
Yeah
I
mean
the
change
looked
pretty
forward.
Instead
of
password
it's
using
certificates
for
creating
the
the
session.
We.
E
E
B
B
Just
more
or
less
needs
I
don't
know,
but
did
I
put
a
hold
on
for
CI.
You
cannot
cancel
yeah.
B
A
Yeah
we
we
came
from
the
discussion
when
we
were
taking
a
look
at
the
CI
environment
where
we
did
see.
Dhcp
leases
were
hold
there
for
24
hours,
which
is
not
the
case
anymore,
but
the
tri-soft
method
would
help
freeing
the
DHCP
releases
thesis
earlier
than
waiting
for
the
TTL.
B
So
I'm
not
changing
the
API
default
here,
because
that
would
be
still
a
breaking
change
to
the
API
version.
So
it's
just
in
the
cluster
templates,
which
feels
okay
to
me.
B
Right
yeah,
one
topic
that
we
should
talk
about
soon
later
is
probably
so
I'm
going
to
build
the
new
images
for
128
and
we
uploaded
Ubuntu,
20
and
22.
B
to
True,
vmc,
environment
I,
don't
know
at
which
point
or
depending
on
what
we
wanted
to
switch
to
you
continue
to
instead
of
20.
but
yeah
we'll
get
there
right.
Yeah.
E
A
That's
the
yeah
okay
for
the
kubernetes
version,
which
is
basically
updating
the
CI
variables
here
and
there
we
see
open
to
24
and.
B
B
D
Testing
will
push
budget
the
2204,
oh
just
get
a
decent
kernel.
D
B
Might
think
I
think
at
some
point
we
have
to
if
I
got
it
right.
So
if
you
want
to
use
SQL
Suite
2
in
a
proper
way,
we
should
use
Ubuntu,
22
right
and
Upstream
communities,
testing
I'm,
not
sure
what
the
deadline
was
exactly
but
I
think
they're.
Definitely
the
point
yeah.
D
Contain
a
day
will
drop
secret
three
one.
At
the
end
of
the
year
calendar
year,
yeah.
E
E
F
B
E
B
A
B
B
Okay,
yeah
I
think
it
would
provide
some
context
there.
Just
so
people
know
why
we're
doing
this
and
potentially
know
how
they
can
profit
from
it.
So
essentially,
someone
from
redhead
brought
it
up
in
core
copy
that
in
our
API
packages
we
should
avoid
dependencies
on
controller
runtime,
which
basically
would
mean
that
someone
is
able
to
consume
core
copy
apis
without
also
getting
certain
controller
runtime
dependencies,
which
also
means
they
can
use
whatever
control
runtime
version
they
want
as
soon
as
we
don't
have
dependencies
there
anymore.
E
A
B
A
Yeah
there
was
the
session
is
active
method,
which
requires
some
permissions.
Some
accounts
may
not
have,
and
basically
this
Pro
request
implements
a
fallback
to
retrieving
the
data
center.
Instead,
when
the
arrow
comes.
B
E
B
Yeah
I
started
and
I
was
thinking
like.
Oh
it's
she
but
I
mean
tiger
cheese
is
a
product
right.
So
we're
not
talking
about
internals
here
yeah.
It's.
B
Think
yeah
dishes
are
a
bit
better
I
think
the
issue
doesn't
even
mention
ticketing
anyway.
It
talks
more
like
about
yeah
what
you're
saying
an
account
of
not
that
much
promotions.
Etc.
A
E
F
E
G
So
I
don't
work
mainly
for
the
cafe
I.
So
my
background
is
that
I
am
currently
working
on
the
upgrades
controller,
that's
Downstream
for
tkg,
so
the
idea
is
to
contribute
somewhere
and
just
wanted
to
ask
if
there
is
a
good
first
issue,
because
I
saw
that
after
some
time,
the
label
good
first
issue
was
not
used
anymore.
B
Yep
I
would
do
the
following:
I'll
go
over
the
issues
tomorrow,
so
I
recently
created
a
bunch
of
issues.
I
would
basically
go
over
the
reflect
them
accordingly
and
then
continue
afterwards.
If
that's
fine,
for
you,
yeah
perfect,
just
gives
me
a
bit
more
time
to
think
about
which
ones
are
really
there
and
also
Fleck
a
bunch
of
them.
Thanks.
B
F
A
This
at
least
yeah
looks
a
bit
like
that.
I
mean
some
parts
like
kind
of
regular
clusters
will
be
or
is
available
using
Cube,
State,
metrics
and
core
copy
configuration.
Kind
of
posts
depends
on
many
objects,
but
for
the
core
copy
objects
we
have
metrics
there
too.
A
A
And
unreachable
clusters
depends
how
this
is
sort
of
faced
so,
but
what
I
know
is
there's
a
condition
which
says
at
the
vsphere
class
level,
I
think
which
says:
okay,
I
can
reach
vcenter,
at
least
so
that's
a
condition
which
could
get
exposed
here
and
I
know.
We
have
an
issue
to
create
a
configuration
for
cubesat
metrics
for
cap
V,
so
I
think
some
of
that
stuff
may
get
solved
by
that.
That
may.
B
Be
just
maybe,
let's
basically
do
the
do
the
other
issue
and
then
take
a
closer
look
at
this
one
and
see
what
the
gaps
are
and
also,
if
that
should
actually
be
done
in
cap
V
or
copy.
Instead,
there's
also
a
copy
issue
linked
directly
below
the
issue,
top
left
description.
A
F
B
This
point
I'm
not
sure
if
I
would
want
to
do
this
or
not
I
mean
I,
definitely
don't
see
a
benefit
of
putting
all
the
copy
tests
into
a
single
file.
That
sounds
like
not
a
great
idea.
What
could
be
a
maybe
good
ideas
to
put
them
in
in
one
package
and
the
copy,
specifically
the
other
one,
but
I
also
don't
see
a
big
benefit.
I
mean
how
many
tests
do
we
have
eight
ten
something
we
have
50
or
100.
E
A
And
and
depending
on
which
files
are
made
here,
lots
of
our
generated
like
the
templates,
are
all
generated,
so
they
should
be
fine.
B
C
F
A
B
B
B
A
E
E
E
B
A
So
so
what
what
I
could
think
of
is
that
our?
A
F
B
F
A
F
Yeah
I
think
the
way
I'm
running
and
dentists
locally
is
in
general
by
overwriting,
with
the
clustercado
dot
end
file.
We
get
from
our
preparation
from
our
devtools,
so
I'm
overriding
all
of
those
based
environment
and
variables,
because
you
have
to
overwrite
like
five
or
six
of
them
anyway.
F
So
yeah
I
think
we
need
to
come
up
with
a
better
slow
for
that.
Maybe.
D
We
had
asked
for
a
support,
bundle,
oh
very
center,
and
they
were
unable
to
give
us
one.
So
I
don't
know
if
it's
so
might
have
been
just
an
environmental
problem.
A
B
Okay,
but
I
mean
that
comes
down
to
splitting
the
situations,
One
issues
like
maybe
doing
another
verification
of
hey.
If
use
and
not
existing
search
policy,
that
does
it
actually
fade,
which
would
mean
okay,
can't
reproduce
this
one
and
the
other
one
is
basically
something
like,
maybe
documented
better
on
how
to
run
our
advantages
on
some
environment,
which
is
not
our
environment.
E
B
E
A
A
B
E
A
A
Yeah
I
think
the
point
here
is
currently
only
the
password
thing
works
if
I
get
it
right,
but
we
don't
have.
A
E
A
Yeah,
basically,
there
are
two
options
here:
PCA
password
which
I
think
works
already
and
V
GPU,
which
needs
implementation,
possibly.
D
E
E
A
B
D
A
Add
unit
tests
for
gets.gov
yeah
unit
tests
are
always
good.
Remember
correctly,
that
there
is
a.
B
A
look
at
unit
tests
or
can
you
see
well
I,
don't
know
how
to
pronounce
Berkeley
songs.
E
B
That
should
be
enough.
Yeah
yeah,
just
I
think
he
just
you
know,
write
just
FYI
I,
don't
know
if
people
have
time,
but
there
might
be,
there
might
be
some
people
who
will
have
time
to
just
improve
our
test
coverage
in
general,
so
that.
B
A
F
E
D
E
D
I
guess,
if
you
just
change
that
to
Dollar
question
mark
equals,
then
that
will
fix
it.
B
D
E
B
Oh
yeah
sure,
definitely
one
if
you
want
just
take
a
look:
okay,.
B
E
B
D
B
B
A
Yeah
basically,
you've
been
validate
session
is
missing
that
fails,
and
the
question
is
if
this
is
already
fixed
or
not
at
the
end.
E
B
Yeah
once
we
yeah.
A
B
Yeah,
maybe
maybe
assign
it
to
me
just
that
I
have
it
on
the
radar
somewhere
I'm,
just
assuming
that
at
some
point,
I'm
able
to
create
a
you
know
one
account
for
RMC
environment
and
then
I
know
which
permissions
and
then
I
can
just
drop
this
permission
and
see
if
you're
around
trust
problem
or
not
and
close
it.
That's
good.