►
From YouTube: 20190724 scl capi 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
Hello
today
is
Wednesday
July,
24th
2019.
This
is
the
standard
cluster
API
office
hours.
A
sub-project
is
a
cluster
lifecycle.
As
always,
we
adhere
to
go
to
clinic
policy
written
down
and
the
skin
CF,
which
basically
boils
down
to
don't
be
a
jerk.
Given
the
average
attendance
of
this
meeting,
which
is
fairly
large,
we
kind
of
request
that
folks
use
the
raise
hand
feature.
This
is
actually
denoted
in
the
top
of
the
document.
So
if
there's
following
along
at
home,
it's
all
written
down
in
gory
detail.
A
B
B
Sorry,
I'm
troubleshooting
a
CCR
bucket
issue
with
Tim
Hawkins
right
now.
The
biggest
fix
right
now
is
we
were
having
issues
with
removing
the
finalized
errs
at
times,
which
was
causing
some
reconciliation
issues.
So
now
we
make
sure
we
built
the
latest
cluster
object
before
ignition.
Those
finalized
errs.
A
C
A
C
So
the
idea
is
that,
like
X
stands
for
experimental
I
guess
like
admits
from
the
golang.org
X
kind
of
extended,
and
for
now
like
we're
experimental
yet
because
we're
not
for
but
after
API
review
like
we
might
drop
the
X
and
just
do
cluster
Duckett's
that
I
own
another
PSA
is.
The
booster
proposal
has
been
updated.
Now,
there's
a
new
field
that
it's
required
and
expected,
which
is
called
status.
C
A
D
A
C
C
Million
dollar
question:
what
should
we
do
with
cost?
Ricotta
4d
one
out
for
to
the
the
tool
is
now
providing
a
gnostic
as
in
it
won't
work,
because
the
bootstrap
and
infrastructure
resources
are
not
pivoted,
but
in
a
management
cluster
it
could
work,
I,
guess
or
like
some
alpha
phase,
it
will
probably
work
so
now,
at
least
like
we'll
have
one
tool
for
all
providers.
That
said,
like
there's
a
lot
of
work
to
be
done
to
support
B
1,
alpha,
2
and
I
know
Tim.
F
C
A
G
Don't
have
deep
dependencies
idea.
Thoughts
here,
I
have
something
very
similar
in
the
doctor
provider.
There's
a
program
I
use
called
captain
control
that
I
wrote
that
does
a
bunch
of
helpful
commands
for
management
clusters.
I,
don't
think
it
is
it's.
It's
definitely
not
agnostic,
because
it
is
like
it
uses
kind
and
it
doesn't
pivot,
but
it
does
a
lot
of
those
same
things.
It
would
be
cool
if
we
could
merchants
work
together,
but
I
don't
know
if
it's
the
same
use
case.
A
So
I
have
long-standing
opinions
that
I
would
like
to
deprecated
cluster
cuddle
to
fix
some
of
the
long-standing
UX
issues
that
exist
across
providers.
It
has
a
well
defined
specification
that
outlines
the
user
stories
that
we're
trying
to
aim
it
to
simplify
the
day.
One
scenario
in
particular
for
people
who
are
coming
to
to
cluster
API
I
am
NOT
ready
for
presenting
that
I
think
I.
A
C
So
for
we
went
off
the
in
development
like
I
personally
use
cluster
cuddle
like
in
two
cases
either,
if
like
like,
especially
for
v1
out
for
one,
but
for
if
I
want
to
like
spin
up
a
cluster
and
pivot,
it
would
kind
or
if
I
want
to
have
a
management.
Cluster
I
use
some
of
the
alpha
phases,
and
this
is
like
a
document
in
Kappa,
so
I
had
to
add
years
like
either.
C
We
support
the
bootstrap,
any
particular
resources
and
we
pivot,
and
then
we
have
to
add
tests
for
it
and
all
that
funny
things
in
cluster
karo
or
we
deprecate
all
the
pivot
part.
And
we
just
say
you
can
run
kind
of
like
you,
wanna
put
you
in
a
management
cluster
and
pretty
much
it
and
then
will
rely
on
like
the
future
plans
to
actually
support
a
more
holistic,
beautiful
users.
I
know.
A
A
No
one
really
so
maybe
we
should
open
up
an
issue
and
then
blast
the
mailing
list,
because
I'm
super
surprised
that
people
aren't
mobile
on
this
open
up
an
issue
about
the
problem
and
potentially
send
that
issue
to
the
mailing
list
to
solicit
options
here,
because
I
do
know
that
people
do
depend
upon
it
and
I
do
know
that
they
depend
upon
pivoting,
because
I
asked
that
question.
So
I
want
to
get
broader
feedback.
Pablo
your
hands
up.
H
It
will
be
command
on
the
future
plans,
but,
for
me,
are
not
clear.
I
mean
the
idea
that
we
really
understand
the
problem
obtaining
the
tool,
but
this
may
mention
a
couple
of
time
that
proceeded
in
the
future
and
plans
we
have
for
more
holistic
solution
when
they
to
experience
the
most
would
mean
said
but
I
know
totally
where
which
plans
those,
probably
indeed,
in
this
issue,
review
document
or
comment.
What
will
be
the
relative
in
the
future?
There
will
be
easier
for
the
community
to
to
consider
the
two
options.
A
That's
fair
on
I
should
probably
get
that
in
line
along
the
way.
Maybe
maybe
we
can
open
the
issue
now
to
talk
about
the
current
problem
that
we
face
with
me,
went
alpha
2
and
then
by
for
next
week,
I
should
hope.
Next
week,
I
should
have
a
more
concrete
document
and
stubbed-out
idea
in
place.
I've
talked
about
it
informally
with
other
folks
I'm,
not
ready
for
broader
presentation,
because
it
required
more
time
talking
than
it
would.
If
I
had
a
document
to
go
along
with
it,.
A
H
Should
be
comment,
perhaps
notice
that
sometimes
work
is
quite
frequent,
that
we
have
some
food
requests,
have
one
single
commit
with
hundreds
of
files
which
are
change?
For
instance
sometime.
There
are
some
changes
that
are
no
strict
to
speaking
related
to
the
object
of
the
request,
but
I
probably
need
it.
So
for
factoring
something
like
that,
and
that
make
extremely
hard
to
make
a
review
because
it's
hard
to
find
which
files
to
be
in
in
terms
of
the
poor,
because
himself
so
I
don't
know.
H
If
we
could,
it
is
policy
on
the
group
on
the
community
to
all
one
can
meet
the
food
requests
or
we
can
adopt
a
more
flexible
that
if
there
are
some
known
as
no
related
or
no
essential
changes
directed
to
the
pull
request
objective
to
create
a
separate
commit
for
multiple
could
meet.
Rendering
stuff
like
that.
That
you
know
are
very
beacon
or
impossible
to
review
there.
A
Is
a
general
policy
that
actually
exists
upstream?
That
promotes
what
makes
a
good
pull
requests
that
promotes
that
your
each
chain
should
be
isolated,
so
I
would
recommend
that
you
should
not
be
during
refactoring
in
a
vendor
pull
request
just
period
unless
it's
absolutely
necessary
to
make
that
PR
fly,
and
if
you
do
do
that,
change
that
it
should
be
a
sipper
commit
message
and
that's
typical.
A
That's
actually
documented
in
what
makes
a
good
flow
request
and
that's
in
the
main
cake,
a
repository,
and
we
should
be
following
those
best
practices,
because
it
makes
it
much
easier.
The
reviewer,
because
we've
had
these
problems
in
the
past
in
KK
of
trying
to
figure
out
what
was
the
change
you
made
versus
what
is
this
monstrosity
of
dependencies
that
exists
inside
your
peer?
A
D
Things
so
yeah.
We
generally
don't
try
to
open
giant
pull
requests
that
mishmash
vendor
changes
with
everything
else
in
a
single
commit.
If
you
do
see,
it
was
probably
accidental,
and
the
second
thing
is
that,
while
this
is
good
for
pull
request,
reviews
to
have
separate
commits,
we
currently
only
do
squash
merges
when
stuff
gets
merged
in
this
project,
which
is
which
makes
it
hard
to
look
at
things
after
the
fact.
If
you're
trying
to
look
and
get
history
so
I'm
wondering,
if
maybe
we
want
to
consider
possibly
not
squashing
when
merging,
there's.
D
Yeah
I
noticed
that
happened
to
me
because
I
had
my
default
email
with
github
set
to
my
personal
Gmail
and
I
started
looking
at
commits
of
mine
that
had
been
merged
and
noticed
that
it
was
not
my
VMware
email
address
which
surprised
me
so
I
don't
know
if
we
need
to
open
up
an
issue
and
do
lazy
consensus
or
whatnot
on
this,
but
I'm.
Definitely
in
favor
of
switching
to
not
doing
the
Squatch
merges
Jason
yeah.
B
I'll,
second,
that
as
well,
it's
also
caused
us
pain
with
doing
any
type
of
back
ports
as
well,
because
the
squash
merges
then
merged
those
back
ports.
And
then
you
don't
have
kind
of
clean,
a
clean
way
to
do
back
ports
from
that
point
on.
So
while
we
shouldn't
necessarily
be
advocating
for
back
ports,
but
it's
just
another
data
point.
A
A
I
D
Next
step-
and
yes
so
I
would
say,
this
is
probably
the
last
week
to
get
in
your
responses
for
the
phases
that
face-to-face
meeting
survey.
If
you
haven't
already,
the
link
is
in
the
meeting
agenda
on
the
screen
and
it
looks
like
either
the
week
of
September,
16th
or
September
23rd
have
the
most
people
who
are
available.
I
think
they
just
vary
by
one
at
this
point.
D
D
So
again,
please
fill
this
out
and
we
are
investigating
logistics
right
now
for
the
week
of
September
16th
in
San,
Francisco
and
I
will
have
a
follow-up,
probably
later
this
week
or
early
next
week
with
some
additional
information,
but
for
right
now
I'm
planning
on
using
the
responses
to
track
who's
planning
on
attending.
So,
if
you
haven't,
please
fill
it
out.
Thanks.
A
A
Three
times
well,
that
is
great
I
hope
to
see
everyone
there.
This
would
be
awesome
if
we
can
get
a
face-to-face
to
do
a
little
bit
more
concrete
planning,
we're
all
locked
in
the
room
and
I
will
try
my
best
to
arbitrate
without
that
folks
descending
into
violence.
I
know
that
we
want
to
get
a
lot
of
stuff
done
in
a
short
period
of
time,
so
we'll
see
we'll
do
our
best
Daniel
yeah,
this
topic,
yeah.
K
J
Are
deployed
are
deployed
once
the
cabbie
controllers
are
able
to
reach
the
workload
clusters
API
server
and
you
know,
for
my
particular
use
case-
I,
actually
want
to
be
able
to
deploy
something
that
enables
that
to
happen.
So
it
is
if
the
workload
cluster
happens
to
be
in
a
restricted,
Network
environment
and,
oh,
you
know
one
way,
one
way
to
to
the
you
know
to
to
deploy.
This
is
this.
You
know
the
solution
is
on
the
cluster
itself
as
an
add-on,
but
there's
this
chicken
and
egg
problem
haven't
filed
an
issue.
A
J
D
So
the
I'm
thinking
about
this
one,
the
the
bootstrap
provider,
is
going
to
operate
before
the
infrastructure
for
your
your
server
or
virtual
machine
is
provisioned
so
and
then
like
there's
nothing
stopping
it
from
doing
additional
work
after
but
to
the
general
purpose
of
a
big
provider.
It's
just
to
generate
the
bootstrap
data.
So
are
you
thinking
about
the
cni
or
more
than
the
CNI,
for
this
use
case
pretty
much
on
that
on
that
level,
it
could.
J
D
D
You
can
certainly
have
static
pause
that
the
bootstrap
provider
may
be
created
to
do
some
add-ons.
So
I
think
maybe
let's
a
file,
an
issue
in
Cappy
and
have
a
discussion
there
and
then
we
can
try
and
figure
out.
Does
this
make
sense
in
Cappy?
Does
it
make
sense
on
bootstrap
provider
and
just
have
that
follow-up
conversation
and
github.
F
A
D
D
It's
a
slight
change
to
that
view
and
alpha-1
behavior,
but
it'll
get
picked
up
on
an
upgrade
when
you
deploy
the
new
version,
and
this
will
help
with
cluster
deletion,
because
right
now,
if
you
look
at
at
985,
if
you
have
a
cluster
actuator,
that's
doing
a
delete
and
that
all
happens
in
the
foreground
is
part
of
the
finalization
of
the
cluster
deletion.
If
there's
dependencies
that
need
to
be
deleted.
D
First,
like
machines
for
the
AWS
provider,
you
end
up
in
this
endless
loop,
where
the
cluster
is
trying
to
delete
something
say
a
V
PC,
but
because
the
Machine
still
exists,
it
won't
it'll
fail
and
the
Machine
won't
get
deleted
as
part
of
the
deletion
flow.
Unless
we
put
logic
in
to
to
make
that
happen.
So
I
think
this
will
just
help
provide
more
stability.
Around
deletions.
D
G
Happy
to
give
an
update
there
if
necessary,
please
do
yeah
so
right
now
we
are,
we
being
I,
think
Amy
and
myself
they're
the
two
people
who
are
working
on
that
most
right
now
we
are
implementing
the
controller
logic
for
for
the
config
and,
while
that's
going,
fine
I
am
before
I
do
that
I
do
want
to
get
the
talker
provider
up
to
use
V
1,
alpha
2,
so
I
can
actually
just
what's
going
on
without
having
to
use
AWS
resources.
But
it's
it's
all
fine
like
it's
going
well.
G
If
we're
gonna
go
for
the
backlog,
though
I
think
there
is
some
interesting
stuff
in
there
there's
a
lot
of
stuff,
that's
out
of
scope
in
the
issues
list
right
now,
yeah
the
interesting.
The
interesting
part
about
this
whole
process
is
the
stuff
that
we
were
automating
before
is
no.
Some
of
the
stuff
is
no
longer
like
able
to
be
automated,
like
as
an
additional
step
in
the
user.
G
Experience
and
user
creates
a
machine
and
then
creates
a
cuvette
DM
cluster
bootstrap,
config
or
chameleon
cluster
config,
and
they
have
to
manually
enter
some
of
the
data
there
instead
of
before.
Where
the
the
cabin
provider
was
aware
of
the
whole
system
now
the
booster
provider
is
just
aware
of
a
small
subsection
of
the
systems
that
see
all
the
things
that
meaning
to
see
so
we're
working
on
figuring
out
how
to
make
that
experience
a
little
bit
nicer.
One
of
the
one
of
the
interesting
parts
is
how
to
manage
secrets,
sorry
its
certificates.
G
D
Like
to
talk
about
this,
one
Chuck,
the
cloud
in
it,
so,
okay,
the
the
summary
on
this
one,
is
that
by
chance
and
luck,
the
version
of
cloud
in
it
that
we're
using
in
the
AWS
provider
is
a
fork
of
the
cloud
in
it.
Source
code
repository
that
happens
to
have
templating
and
the
templating
is
needed
so
that
we
can
pull
data
from
the
instance
metadata
yeah
and
use
that
for
things
like
hostname
or
IP
address
yep.
So
this
works
in
Ubuntu
because
they
have
a
new
enough
version
of
cloud
in
it.
A
So
I
guess
one
of
my
asks,
maybe
for
next
time
Chuck
would
be
if
you
have
items
that
you
could
Fedder
it
out
to
the
larger
group
or
people
don't
want
to
engage,
which
I'm
sure
there
are
many
there's
30
people
to
get
some
more
granularity
in
the
backlog.
So
that
way,
you
can
federated
out
some
of
the
work
and
get
a
Help
Wanted
list,
because
I'm
sure
there's
aspects
to
this,
especially
the
testing
or
somebody
can
probably
start
engaging
earlier
and
it'd
just
be
better
for
your
sanity
to
get
those
folks
go.
E
A
A
Are
there
any
other
repositories
that
we
should
consider
for
next
time
to
walk
through?
There
are
so
many
providers
that
I
don't
know
if
it
makes
sense
or
if
you
want
to
evaluate
or
walk
through
or
potentially
do
a
triage
for
B
1
alpha
2
I'll.
Leave
that
up
to
the
group.
Maybe
talk
about
that
as
a
potential
item
that
we
want
to
discuss
like
there.
D
So
if,
if
cat
V
and
cat
Z
can
do
it
as
well,
that'd
be
create
if
there's
others
that
want
to
come
in
line
that'd
be
awesome,
and
if
there
are
some
that
aren't
from
say
the
big
3
cloud
provide,
who
maybe
need
some
guidance
with
how
to
implement
either
the
infrastructure
provider
if
they
need
a
custom,
bootstrap
provider.
Please
come
find
us
in
slack
or
set
up
a
zoom
meeting
so
that
we
can
help
you
out
I'll.
L
G
G
G
Yeah,
that's
that's
fine,
but
you
could
I,
don't
think
you
could
run
a
cluster
inside
of
a
cluster
but
I'm
not
hundred
percent
sure
on
that.
But
yes,
if
you
could
you
run
a
management
cluster
and
then
you
can
create
as
many
additional
clusters
as
you
like,
assuming
that
your
host
machine
has
enough
resources
to
run
those
clusters,
we.
I
Have
to
go
back
to
Cuba
diem
and
allow
configuring
multiple
administrative
users.
Currently
everything
is
bound
to
admin,
so
you
have
to
apply
some
modifications
to
or
multiple
questions
to
be
controlled
from
the
same
location
but
yeah.
If
you
have
the
resources
you
can
create
as
many
questions,
if
you,
if
you
as
you
like,
but
you
have
to
have
the
same
clients
that
can
modify
them.
D
L
We
like
that
I'm
happy
to
do
that.
I
think
we
should
it's
currently
in
six
EMI,
which
feels
a
bit
like
a
graveyard
for
PRS.
So
I
think
we
should
take
it
into
sick
class,
the
lifecycle,
ownership
and
just
approve
it
ourselves
and
because
I'm
sure
there'll
be
other
changes.
We
won't
make
to
the
drain
functionality
and
then
we
should.
We
should
grab
that
so
that
we
can
actually
iterate
on
it.
If
we
want
to
be
in
KK.
D
L
A
Gave
a
plus
one
I'm.
I
L
A
A
So
the
the
comment
we
made
earlier
I
think
I
was
looking
for
YouTube
as
you
I,
don't
know.
If
you
were
in
line.
The
comment
we
made
earlier
is
that
there's
a
large
refactoring
work
work
that
needs
to
be
done
for
cluster
huddle
and
the
question
we
had
was
whether
or
not
we
wanted
to
move
to
the
new
world
order
that
we've
been
that
I've
been
sketching
out
on
gyrase
force,
but
it's
not
ready
for
broader
consumption.
Yet,
but
maybe
I
can
sync
with
you
to
a
particular
offline.
A
We
can
figure
out
a
path
for
us
or
anyone
else,
who's
interested.
We
can
can
do
it
on
the
kappa
channel
and
and
try
to
get
some
traction
there,
but
the
more
the
merrier
right
now
we're
trying
to
spec
out
a
sort
of
a
different
user
experience
workflow
that
cuts
across
providers.
That's
consistent,
I
think
we
have
a
solid
workflow
for
how
we
want
this
to
work,
but
the
logistics
are
many
and
thorny.
So
more
help
would
be
definitely
welcome.
A
F
F
Like
you
know,
using
this
we
can
build
a
management
cluster
and
a
workload
cluster,
and
it
is
also
related
to
the
multi-tenancy
working
group.
I
was
getting
involved
with
I
just
wanted
to
know
like
there
any
any
further
details
on
this
like
if
I'm
trying
to
learn
it
by
doing
is
there
a
way
I
can
like
do
it
somehow.
A
Yeah,
but
from
a
high
level
perspective,
cluster
API
doesn't
take
opinions
on
sort
of
the
potential
user
stories
around
it.
We
just
enable
the
infrastructure
for
people
to
do
the
different
user
stories
so
like
is
especially
with
regards
to
multi-tenancy
and
specific
deployment
policies.
Those
are
we
kind
of
consider
those
otoscope
because
they're
its
fractal
right.
It
depends
almost
entirely
upon
your
environment
and
how
you
want
to
manage
these
things.
I'm
sure
some
people
will
have
some
best
practices
and
some
people
have
highly
opinionated
versions
that
differ
from
those
best
practices.
A
A
So
if
they
have
a
specific
set
of
best
practices
that
they
want
to
recommend
for
multi-tenancy,
you
know
then
more
power
to
them
how
they
want
to
deploy.
But
that's
like
we
don't
take
a
stance
in
that
particular
area.
Usually
the
the
scope
statement
for
a
straight
guy
is
really
bounded.
Otherwise,
there's
this
weird
problem
we
could
have
or
like:
where
does?
Where
does
it
end
right?.