►
From YouTube: 20200909 Cluster API Office Hours
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
Talk
or
comment
or
something
use
the
racehand
feature
in
zoom.
You
can
find
this
feature
under
the
participant
list.
A
A
So,
let's
start
the
first
thing:
the
first
psa
that
we
have
is
that
we
kind
of
have
a
medina
track
session
in
cubicon
north
america.
This
is
also
going
to
be
virtual.
If
anybody
would
like
to
volunteer
to
do
a
cluster
api
session,
we
usually
talk
about
like
what
have
we
ripping
up
to
and
what's
the
roadmap
for
the
future
and
what
are
the
project
goals.
A
So
this
is
like
both
user-focused
and
kind
of
like
more
people
that
know
about
cluster
api,
but
also
that
don't
and
the
goals
should
definitely
be
kind
of
like
to
put
the
word
out
there
about
cluster
api.
If
someone
wants
to
from
this
group
wants
to
run
the
session,
we
have
two
spots
available.
Some
of
us
already
went
and
passed
kubecon
so
definitely
like
if
you're
new,
if
you
like
to
volunteer
you
can
reach
out
to
me
on
slack
and
yeah,
we
can.
A
A
That
all
right,
let's
move
to
discussion
topics
again
you
have
the
external
remediation
demo.
B
Yes,
can
you
guys
hear
me.
A
B
Okay,
good
so
yeah
I've
been
preparing
the
video
about
the
external
remediation
feature
that
we
we
have
been
working
on
and
yeah.
It's
like
a
three
minutes,
video
I
I
did
it
today
and-
and
I
opened
the
pr
also
for
this
feature.
I
still
have
some
some
issues
to
fix
to
get
it
past
from
ci,
but
I
definitely
like
to
to
focus
on
community
to
see
how
the
the
actual
feature
is
working.
C
B
The
basic
idea
in
in
in
this
feature
is
that
we
want
to
add
the
external
remediation
feature
into
a
mass
in
health
check,
so
so
for,
for
example,
in
cases
in
with
the
bare
metal.
We
don't
necessarily
want
to
use
the
default
feature
like
when,
when
massing
health
check
is
fun
from
the
unhealthy
machine,
it
will
set
the
condition
false
and
the
machine
control
will
will
delete
the
machine.
So
in
our
cases
this
this
will
cause
a
the
provisioning
and
provisioning
of
the
actual
bare
metal
host.
B
So
we
pro
we
wanted
to
have
a
feature
that
we
can
can
take
kind
of
a
control
when,
when
the
sort
of
on
healthy
machines
has
been
found
and-
and
we
wanted
to
mass
inhale
check
to
generate
the
request
for
these
kind
of
machines
that
are
found
unhealthy.
B
So
I
can
put
now
my
video
on
so
so
we
started
with
the
mason
deployment
where
we
have
two
replicas,
and
this
is
running
in
my
tilt
and
and
I'm
using
capd
as
a
infrastructure
provider.
Then
we
are
providing
the
sorry,
deploying
the
machine
health
check
and,
as
you
can
see,
it
find
out
that
there's
a
two
machines
and
or
two
replicas
and
both
are
on
healthy
state.
B
Then
because
we
are
using
using
capd,
I'm
I'm.
I
want
to
stop
one
of
the
containers
to
kind
of
show,
causing
the
machine
health
checks
to
find
a
one
of
the
machines
unhealthy.
B
So
after
after
we
have
stopped
this
container,
we
need
to
wait
for
a
while
that
to
machine
health
check
to
find
out
that
there's
a
unhealthy
machine.
E
B
Unhealthy
sorry,
external
remediation
feature
on
we
will
generate
the
metal
tree
remediation
request
that
you
can
see
right
there
in
the
in
the
down
right
and
what
we
have
in
there.
We,
because
we
have
this
machine
that
is
unhealthy.
We
also
set
the
owner
reference
of
this
machine
into
our
machine
remediation,
so
you
can
find
see
it
right
there
and
what
we
also
want
to
do,
which
was
not
in
the
original
proposal.
B
We
would
also
like
to
have
remediation
reference
somewhere
stored
into
the
machine
now
in
in
this
pr
I
added
in
the
status
field
of
the
unhealthy
machine,
so
we,
I
would
definitely
like
to
see
this
in
place
or
we
can
discuss.
Maybe
some
other
way
how
how
we
handle
this,
but
it's
I.
B
I
find
it
quite
good
to
have
this
reference
to
this
remediation
request
somewhere
now
I
I
start
again
the
docker
container
and
this
this
will
in
the
this
will
so
start
the
the
container
and
and
soon
our
mess
in
health
check
find
out
that
the
all
the
machines
are
healthy
again.
B
Obviously,
if
we
are
would
be
in
the,
for
example,
in
metal
tree
project,
our
own
remediation
controller
would
do
whatever
to
get
the
machine
or
node
back
into
the
healthy
state
and
and
that
that
was
the
end
of
the
video
and
then
we
we
can
discuss
with
the
with
the
others.
I
can
stop
sharing,
probably
or
well.
I
yeah,
I
think,
that's
that's
enough
for
for
now,
do
you
have
any
questions
by
the
way.
B
D
Andy
thanks
john,
that
was
really
cool.
I
am
curious
about
the
remediation
ref
and
like
how
do
you
envision
using
that
or
you
know
how?
What
does
that
help?
You
do.
B
Well
in
in
this
pr
that
I
opened,
I
used
this
remediation
ref
to
check
out
that,
for
example,
if
if
the
machine
already
have
the
remediation
ref,
I
kind
of
avoid
to
create
the
new
remediation
for
for
the
same
machine
if
it's
still
unhealthy.
So
otherwise,
in
our
in
our
original
proposal,
we
were
thinking
about
to
set
the
same
name.
B
Then
machine
have
the
in
the
remediation
request,
but
when
I
started
to
do
the
implementation,
I
found
it
more
easy
if
I,
if
I
have
this
rem
remediation
reference
somewhere,
it's
much
easier
to
use
for
fetching
the
actual
object
and
and
especially
using
during
the
deletion
of
the
remediation
request
itself
after
after
we
have
found
the
machine
healthy
again.
B
So
this
is
obviously
not
the
only
way
to
do
it,
but
at
least
the
it
felt
felt
right
at
the
moment
when
I
was
doing
the
implementation.
A
D
B
But
yeah
the
pr
is
there
and
I
would
like
to
have
a
comments
there.
The
I
put
the
bold
links
in
into
the
notes,
so
the
first
one
is
the
original
proposal
and
the
second
one
is
the
pr
that
I
opened
for
the
implementation
and
and
but
I'm
trying
to
think.
Maybe
we
need
this
reference
somewhere
in
the
future
as
well.
I'm
I'm
not
100
sure,
but
it
as
I
said
it
felt
right
at
the
at
the
moment.
D
Would
so
I
know
that
remediation
or
an
unhealthy
machine
ideally,
is
a
something
that
happens.
Rarely
I'm
thinking
about
cut
will
move
and
if
we
happen
to
have
an
unhealthy
machine
that
was
getting
moved
from
a
bootstrap
cluster
to
the
real
management
cluster.
D
B
Yeah,
I
wasn't
sure
what
what
to
answer,
because
I'm
not
that
aware
of
cluster
cattle,
how
how
it
works
and
what
needs
to
be
done
there.
But
obviously,
when
the
pr
is
still
open,
we
we
can
add
what
whatever
features
or
or
take
all
these
need
to
be
taken
count
now
and
and
try
to
think
that
we
don't
adding
anything
that
we
probably
not
really
need.
So
I'm
not
trying
to
push
anything.
B
So
I
definitely
open
for
the
other
other
options
as
well.
A
G
So
I
was
just
going
to
say
that
on
on
the
move
front,
there
is
kind
of
a
backstop
here
if
the
remediation
ref
gets
lost
some
for
some
reason.
Like
we've
moved
it,
then
this
code
actually
creates
the
remediation
requests
with
a
fixed
name.
So
if
it
tried
to
create
a
new
one,
it
would
it
would
crash
and
find
some
conflict,
I'm
not
sure
how
that's
handled,
but
there
it
shouldn't
be
able
to
create
a
second
remediation
request
because
it
uses
a
fixed.
A
C
Thanks,
I
like
the
idea
of
having
it
I'm
having
the
remediation
ref
on
the
machine,
but
it
introduced
some
problem
like
what
is
the
source
of
truth
for
the
remediation
cr.
We
might
end
up
with
only
one
or
we
might
end
up
with
a
reference,
but
without
cr
or
with
cl
and
without
reference.
So
it
might
be
a
problem.
We
need
to
figure
out
how
to
solve
it.
B
Good
point:
I
was
trying
to
add
something
today
on
the
on
the
controller
that
if
we
check
out
off
after
we
have
add
the
at
the
reference,
if
it's
not
there,
we
also
delete
the
remedy
as
a
request
itself.
So
I
I'm
not
quite
sure
it's
the
right
thing
to
do.
But
at
least
I
was
thinking
if,
if
we
can't
set
this
reference,
then
when
we
shouldn't
have
the
request-
and
maybe
we
need
to
try
again
in
the
next
loop-
yes,
but.
C
B
Yeah,
I
agree,
agree
yeah.
We
probably
need
to
have
a
discussion
also
outside
this
meeting.
B
E
The
bridge-
and
I
was
just
wondering
about
the
external
imitation
and
move-
I
was
just
wondering
if
it
makes
sense
to
block
movies
if
the
there
is
a
reminition
in
progress,
and
this
should
be,
I
think,
sit
safer
for
the
part
operator.
B
Proposal
yeah,
I
I
think
you
have
a
point
if,
for
example,
blocking
the
the
move.
If,
if
the
remediation
is
in
progress,
then
then
one
thing
what
I
I
was
thinking
how
we
handle
the
things
that
are
remediation.
B
Probably
not
you
know
we
waiting
the
infrastructure
provider
to
get
this
remediation
done,
but
if
it's
failing
we
had
some
discussions,
but
I
can
hardly
remember
now
what
we're
supposed
to
do.
That
is.
I
think
that
is
more
like
a
thing
that
we
should
do
in
in
the
controller
side
in
the
infrastructure
provider
controller,
but
yeah
there's
a
still
couple
of
things
that
I
need
to,
and
we
need
to
think
over
to
really
get
this
get
this
solid
as
possible.
B
So
there's
a
couple
of
risks
like
near
said,
with
the
reference
and
but
yeah,
I
don't
know
just
delete
it.
Yeah
go
ahead.
A
Yeah,
so
on
on
this
it
just
this
remediation
reference.
Well,
first,
it's
on
status
like
it
won't
be
preserved
in
case
move
happens,
but
it
just
seems
like
we're
trying
to
kind
of
track
something
that
we
can
probably
just
do
a
get
on
to
see
if
the
remediation
on
for
that
machine
already
exists,
because
we
have
that
if
we
want
to
use
this
as
a
convention,
that's
like
it
has
to
have
the
same
name
as
the
machine
truthfully.
A
There
can
only
be
one
so
we
can
just
get
that
one
and
if
it's
there
then
don't
proceed
to
create
a
new
one.
I
would
rather
try
to
avoid
adding
more
stuff
to
status
if
possible.
B
Okay,
yeah,
that's
that
is
understandable.
Yeah.
G
Joel,
I'm
not
sure
where
we
got
to
on
the
proposal
with
this,
but
if,
if
a
remediation
request
has
been
created
and
then
someone
tries
to
update
the
spec
of
the
machine
health
check
to
change
the
type
of
the
remediation
external
remediation
thing,
couldn't
that
lead
to
then
some
conflicts
here
like?
If
you
have
the
remediation
ref
on
there,
it
has
the
type
of
the
one
that's
existing
in
it.
So
it's
kind
of
useful
to
have
to
make
sure
that
we
don't
create
a
new
one.
That's
of
a
different
type,
because
someone's
changed
that.
A
I
do
recall,
and
I
don't
know
if,
like
we
actually
talked
about
it,
or
I
just
thought
about
it,
that
that
should
that
me
immutable
like
so
that
when
once
you
create
it
with
a
type,
you
cannot
change
that
you
can
change
the
reference
template,
but
if
you
change
the
type
without,
what's,
let's
say
like
with
the
built-in
md1,
it's
just
gonna
stay
there
like
you,
can
change
it
to
another
thing
which
kind
of
like
makes
sense
in
terms
of
safety
and
yeah
like
it's
a
little
bit
more
immutable,
but
you
can
always
delete
it
and
create
a
new
one.
G
A
Does
that
sound
good
to
you
like?
If
we
actually
don't
allow
you
to
change,
to
update
the
type,
then
we
can
remove
the
status
remediation
reference
machine
and
rely
on
the
name
of
the
machine.
We
should
document
this
as
the
naming
convention
it
has,
if
it
hasn't
been
done
already,
but.
H
B
Yeah,
well,
that
makes
sense
yeah
we
we
can
rely
on
on
the
checking
the
name
of
the
of
the
request
and
and
yeah
I
I
can
work
on
on
that
one
as
well,
and
we,
if
we
really
want
to
not
adding
anything
in
status.
That's
that's
fine,
fine,
also
I'll
check
out
the
or
change
the
implementation.
The
way
that
we
we
just
rely
on
that
name,
but
what
about
in
in
deletion?
B
For
the
request?
Is
it
working
the
same
way?
Then
we
check
the
name
and
and
get
it
then
get
the
object
and
just
delete
it
yeah.
I
think
so.
A
Yeah,
I
think
so
I
think
you
can
just
also
call
delete
and
if
it
returns
the
api,
the
api
error
not
found
you
could
just
say.
Okay,
that's
already
has
been
deleted
and
you
know
proceed
gracefully.
A
The
only
other
addition
is
to
make
sure
that
the
web
book
rejects
updates
on
the
type
of
the
machine
health
check.
That
joel
was
talking
about
before.
Okay.
B
Okay,
yeah,
that's
good
thanks
for
the
feedback.
B
It
makes
sense
as
soon
as
as
soon
as
I
get
back
to
work
tomorrow,
so
I
will
work
on
this
one
I
I
definitely
would
like
to
this
to.
What
did
we
discuss
about
the
probably
getting
this
feature
in
yeah?
Are
we
are
we
still
having
well
3.9
came
out,
but
is
it
possible
if,
if
everything's
going
well
how
soon
we
gonna
get
this
in.
A
I
mean
if
it
merges,
we
have
310,
and
this
is
one
of
the
features
that
we
we
said
that
should
go
in
310
yeah,
like
I
think,
that's
scheduled
by
the
end
of
the
month,
but
yeah
as
always.
That's
just
like
a
ballpark.
We
don't
yeah.
That's
that's
for
sure,
but
I'm.
B
Trying
to
end
this
quick
as
possible
so
that
that
will
be
being
the
three
point.
10.
A
Sounds
good,
do
I
hit
your
comment
like
we
should
try
to
make
the
purposes
up
to
date?
Definitely
did
you
have
anything
else?
I
thought
you
rest
your
hand
before.
G
No,
I
was
just
gonna
say:
we've
we've
been
discussing
this
and
then
nier's
grabbed
out
of
slack
that
we
went
for
the
like
reference
approach
and
so
like
there's
been.
I
know,
there's
been
a
hell
of
a
lot
of
conversations
about
this,
so
let's
just
make
sure
we've
got
it
written
down
somewhere
so
that
we
can
refer
back
to
the
decisions
we've
made.
A
All
right
seems
we
have
action
items
here
and
yeah,
we'll
we'll
definitely
get
it
in
310,
make
sure
it
makes
progress
and
once
emerges.
We
can
always
tag
like
an
alpha
or
beta
tag
for
so
that
you
can
use
the
word,
go
modules
or
import
it
and
yeah.
We
can
generate
some
artifacts
as
well
yeah
any
other
question
all
right
fabrizio
for
management.
Cluster
operator.
E
E
A
A
All
right
definitely
sign
up
for
the
doodle
if
you're
interested
in
it
and
yeah,
which
you
have
the
survey
results.
D
E
So
the
during
the
last
cuba
corner
mayor
basically
proposed
an
issue,
a
survey
and
we
got
a
lot
of
very
interesting
answer.
Some
of
them
are
specific
to
cluster
apis,
some
other
more
generic
to
to
the
cluster
life
cycle
topics.
E
And
yes,
yesterday
in
the
seacaster
recycle
meeting
there
was
a
discussion
around
the
answer,
so
I
don't
want.
I
don't
know
if
we
want,
if
you
want
to
go
through
the
document
today
or
you
prefer,
to
look
at
the
registration,
but
definitely
something
interesting
and
that
we
should
keep
in
mind
defining
the
next.
A
A
Yeah
definitely
take
a
look
if
you
have
some
time
there
were
some
interesting
results
here,
but
yeah
like
yeah.
That
was
there's
a
lot
of
data
here
that
we
can
probably
scrape
and
kind
of
use
to
for
the
roadmap.
For
you
know,
the
next
cycles
of
the
year.
H
Yeah,
so
sorry
I'm
distracted
it's
like
so
we
we
chatted
about
this
at
six
plus
life
cycle
meeting
yesterday.
I
think
some
of
the
interesting
points
that
were
raised
amongst
them,
where
the
fact
that
docker
is
still
quite
heavily
used
by
a
number
of
users,
their
container
runtime,
and
how
often
people
upgrade
now
we'll
need
to
do
a
follow-up
in
further
surveys.
H
But
there's
a
num
and
num
a
lot
of
users
upgrade
every
couple
of
releases
and
we
need
to
disambiguate
between
whether
or
not
they
upgrade
just
to
the
next
release,
in
which
case
they're
falling
out
of
support
over
the
long
period.
Eventually,
they'll
reach
a
point
where
they
no
longer
in
a
supported
version
when
they
upgrade
or
if
they're,
just
jumping
between
releases
so
yeah.
So
I
think,
there's
some
interesting
stuff
around
how
how
do
we
make
sure
people
are
are
upgrading?
H
A
lot
of
people
are
saying
they're
upgrading
a
lot,
but
then,
when
you
ask
them,
what's
their
latest
kubernetes
release,
it's
actually
quite
an
old
one.
So
there
seems
to
be
a
difference
between
theory
and
praxis.
A
All
right,
then,
I
think
we
can
just
call
it
like
30
minutes
earlier.
Thank
you
all
for
coming
and
see
you
all
next
week.