►
From YouTube: 20190614 scl cluster api grooming
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
A
A
A
C
I
think
it's
resolved.
It.
B
D
I
think
that
we
could
make
a
comment
that
this
would
be
an
excellent
instance
of
an
infrastructure
provider.
I'll
follow
the
bootstrapping
proposal,
but
it
doesn't
need
to
be
tracked
within
this
repository
and
then
my
thoughts
about
using
terraform
is
that
it's
not
the
greatest
programmatic
way
to
solve
this
problem.
But
it
is
super
fast
and
super
happy
and
it
works
sometimes
so
no
reason
to
discourage
people
all.
D
A
A
A
C
D
A
C
B
D
A
A
A
So
another
another
thing
while
raising
them
on
here,
if
you
want
to
sway
it
haven't,
listened
the
project.
The
more
commits
you
have,
the
more
work
you
do,
the
more
windy
chop,
the
more
water
you
carry,
the
better
your
influence
capabilities
aren't
round.
So,
if,
if
you
don't
chop
the
wood,
if
you
don't
carry
the
water,
don't
expect
to
have
much
sway
in
some
of
the
decision
making
team.
F
A
B
A
Maybe
we
should
have
a
link
in
the
main
doc.
Somebody
want
to
open
an
issue
to
have
a
new
kinship
like
a
new
contributor
doc
to
cluster
API
before
and
somebody
open
an
issue,
please
do
yeah.
Do
we
have
the
communities
org
as
part
of
that
list?
Yeah
I'll
double
check.
Well.
Well.
Well,
he's
checking!
You
need
to
go
to
the
Pyrenees
or
Pablo
and
hit
new
issue,
and
then
there's
organization
and
membership
requests
you
need
to
organize.
You
need
to
get
a
membership
to
karate
sings
as
the
order.
A
A
A
B
A
A
A
C
A
But
this
is
part
of
cluster
coddled
it
again
because
there's
there's
nothing,
there's
no
constraints,
including
heum.
That
say
your
add-ons
have
to
be
in
this
order.
Is
it?
Is
it
blocking
its
cluster,
cutting
uploading
the
creation,
or
is
there
something
inside
of
the
orchestration?
Because
it's
this?
This
might
be.
A
C
D
I,
my
I
have
a
different
opinion.
I
feel
like
this
is
clearly
a
cluster
cuddle
question.
The
clustered
one
of
the
reasons
that
cluster
cuddle
is
annoying
to
me
is
this
exact
thing.
The
fact
that
we
have
three
different
manifests
I
mean
the
only
thing
that
cluster
cuddle
does.
For
me,
the
COO
control
doesn't
is
the
pivot
and
I
feel
like
passing.
Cluster?
Could
cuddle
a
list
of
manifest
in
a
single
file
or
a
directory
with
different
files
should
be
sufficient?
We
don't
need
three
arguments
for
that.
Yeah.
B
B
B
A
C
The
problem
is,
is
this
is
specifically
to
the
actuator
interface
and
for
at
least
the
machines
so
far.
The
proposal
that
we're
looking
at
eliminates
the
actuator
interface
for
machines
and
we've
also
talked
about
potentially
even
eliminating
the
actuator
interface
for
the
cluster
as
well.
Potentially
41
alpha-2.
B
Pablo
had
expressed
some
interest
and
trying
to
put
together
a
proposal
that
would
get
rid
of
the
cluster
actuator
and
follow
the
model
that
we're
trying
to
do
with
separate
controllers
for
machines.
But
do
it
for
clusters
whether
it
gets
into
v1l
I,
don't
know,
but
if
they
can
that'd
be
awesome.
Do.
B
B
B
C
Builder,
handles
ok.
B
E
B
A
B
Note,
looking
at
the
release,
notes
for
the
beta
for
controller
runtime
automatic
certificate
generation
for
Web
books
has
been
removed
and
web
hooks
will
no
longer
self
register,
use
controller
tools
to
generate
a
web
hook,
config
and
use
something
like
certain
manager
to
generate
your
cert.
If
you
don't
already
have
a
solution,
the
next
release
of
queue
builder
will
support
this
out-of-the-box.
Oh
yeah
yeah.
A
B
A
You
can't
reconcile
T's
it's
a
one-time
shot.
That
was
even
like
a
security
problem
right
and
we
have
the
ability
to
do
initial
deployment
through
all
the
plumbing
of
Covidien.
So
you
can
do
everything
you
need
to
do
today
through
Covidien,
but
I.
Don't
know
if
it's
plumb
through
properly
in
a
way
that
people
like
through
custard
guy
so.
A
Can
do
whatever
you
want
to
do
so
like
it's
got
the
plumbing
for
everything.
So
if
you
want
to
annotate
a
node
you
can
you
can
specify
whatever
you
want.
So
it's
got
that
the
plumbing
is
very
wrong
them
and
we
even
do
that
as
part
of
initial
bootstrap,
or
we
put
in
a
CRI
information
as
an
annotation
into
the
nose
on
the
initial
construct.
B
So
the
comments
I
added
to
this
earlier
today
were
one:
do
we
want
to
come
up
with
a
way
that
any
like,
regardless
of
bootstrap
tool,
regardless
of
infrastructure,
you
can
specify
in
one
place
in
the
machine
back?
These
are
the
labels.
These
are
the
annotations
that
I
want
from
my
node
and
I
can
be
confident
that
they
will
get
set,
whether
I'm
using
cube,
ATM
or
not,
and
so
that's
one
question
the
other
is:
do
we
reconcile
beyond
the
initial
setting
of
these
values
the.
D
B
B
Yes,
I
mean
the
bootstrap
config
is
100%
provider
dependent
and
provider
specified.
So
what
I'm,
suggesting
as
a
possibility
for
discussion
and
not
necessarily
in
favor
of,
is
adding
some
controller
logic
to
cluster
API
itself
to
go
and
modify
do
a
one-time
modification
of
a
node
as
soon
as
it
comes
up
to
set
variables
and
annotations.
F
Guys
both
trying
to
recall
because
I
am
assured
that
something
that
this
is
dawning
in
gardener
project,
it
sounds
for
a
presentation
in
the
community
I
think
it
was
Holly
who
was
working
on
this.
Probably
I
can
ping
him
and
because
I
remember
something
about
updating
the
labels
in
changing
in
nodes
and
the
problem
that
you
can
add.
But
later
do
you
have?
The
probably
could
you
remove
I
mean
it's
like
a
programmatic,
it's
made,
so
somebody
3ds
equation
of
reconciliation.
So
probably
okay,
that's
him!
Well,.
B
A
Well,
you
you
could
it's
not
it's
not
a
so
you
have
to
be
super
careful
in
that
it's.
You
should
initially
set
the
values,
especially
for
teens
and
labels.
An
administrator,
high-level
administrative
access
can
do
that
too,
as
well,
but
building
it
into
into
automation,
especially
rectification
for
automation,
is
a
huge
security
constraint
right.
A
So
the
reason
why
we
never
we
were
not
allowed
to
incur
ATM
was
because
there's
big
security
holes
that
need
that
couldn't
be
plugged
right,
so
that
they're
only
allowed
to
be
modified
by
an
administrator
or
by
initial
bootstrap
concerns
using
about
tanks
both
paints
and
labels,
because
you
can
do
special
things
with
taste
and
littles
I
mean
you.
Can
you.
B
B
B
A
So
this
seems
very
tailored.
Do
folks
have
like
the
thing?
That's
nice,
a
little
bit
weird
about
this.
Is
a
lot
of
people
manage
their
machines
through
machine
deployments?
If
you
have
specific
IP
addresses,
it
means
you're
explicitly
managing
your
machines
separately,
which
is
a
bad
bad
and
you
could
plug
through
the
configuration
through
kopitiam
to
do
this.
A
C
B
A
A
E
C
C
B
F
B
A
B
A
B
Do
we
still
want
to
have
the
taints
represented
on
the
machine
speck
I
realize
this
is
a
different
question,
but
it's
very
similar
to
the
labels
and
annotations.
We
were
just
talking
about
like
it's
it's
up
to
some
external
provider
to
use
them.
So
maybe
we
just
make
this
part
of
the
cube
idiom
bootstrap
provider
and
take
the
taints
off
the
machine.
Spec
I,
don't
know
I.
B
Cuz
that
I
mean
you
know
you,
you
don't
know,
looking
at
a
given
machine,
if
the
infrastructure
provider
you've
chosen,
which
currently
the
infrastructure
and
bootstrapping
provider
be
1,
alpha
1,
you
have
no
idea.
If
those
tant's
are
gonna
be
applied
or
not,
it's
not
all
providers.
Look
at
them
do.
B
A
F
A
E
A
C
A
B
E
B
C
A
C
E
B
B
Know
I
mean
it's
the
same
to
me:
it's
like
half
of
the
equation
for
853,
which
was
the
one
right
before
this
about
managing
different
kubernetes
distributions.
So
we
could
open
one
issue
that
says
we
want
bootstrap
and
infrastructure
providers
to
be
separate
and
then
just
link
all
of
these
to
that
and
close
close
these
or
we
could
just
keep
all
of
these
open
until
the
proposal
is
accepted
and
we
get
stuff
coded
and
D
1
alpha
2
is
closer
than
ping,
ready,
whatever
makes
sense,
I.
A
F
Be
but
the
time
we
get,
this
disco
show
having
a
link
to
this.
This
kind
of
umbrella
issue
you
mentioned
makes
sense,
because
later
they
can
can
be
like
East
cases.
We
can
be
from
say
why
we
need
that
separations
because
of
these,
and
these
use
cases
so
I
think
it
could
available
to
leave
them
open
and
but
have
an
umbrella.
A
dimension
suggested.
A
A
A
A
A
Do
we
want
to
start
this
work?
This
is
a
good
thing
for
other
people
who
want
to
get
engaged
does
help
define
what
the
API
Samantha
should
be
like,
and
you
know
get
the
verification
in
place
that
this
is
a
great
thing
for
new
people
to
get
involved
in
working
from
the
outside
in
this
is
a
good
idea.
I,
don't
know
if
I
consider
a
backlog,
as
there
is
long
term.
A
B
A
A
C
B
C
A
A
A
B
A
We
should
we
call
it
for
today,
yeah.
B
C
A
Right
we'll
call
for
today,
I
will
rip
through
all
the
cluster
cuddle
stuff
and
get
that
done
before
the
next
one.
Maybe
what's
a
good
time
for
folks
to
I
folks
want
to
reconvene,
should
do
one
more
week
on
Friday
or
just
to
involve
broader
people
and
do
more
PSAs.
This
time,
I'll
send
out
the
meeting
invite
properly.