►
From YouTube: 20190821 - 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
Hello
today
is
Wednesday
August
21st
2019.
This
is
the
office
hours
for
cluster
API,
a
sub-project
of
sooo,
good
cluster
lifecycle.
I
am
currently
sharing
the
meeting
minutes
and
we
do
have
some
etiquette
that
we're
using
right
now.
So,
if
you
do
have
something
to
say,
please
use
the
raise
hand
feature
in
zoom
and
now
that
I
got
signed
out.
Let
me
make
sure
I'm
in
the
right
account
on.
A
As
always,
please
add
items
to
the
agenda.
If
you
have
them,
please
add
yourself
to
the
attendee
list
and
we
will
get
started
so
the
first
thing
on
the
agenda,
I
added,
is
an
update
on
the
face-to-face
meeting
taking
place
next
month
in
September.
I
have
links
here.
The
first
one
is
an
RSVP.
Please
do
that
xxx
if
you
are
able
to
attend
and
then
the
second
link
is
for
the
proposed
agenda
and
it's
definitely
open
for
editing
and
comments.
So
please
feel
free
any
comments
or
questions
on
the
face
to
face.
B
C
So
we
pretty
much
finished
the
the
base
work
here.
I
think
that
at
this
point
we
just
need
to
test
it
and
get
bugs
opened
and
fixed
on
that,
but
there's
nothing.
That's
blocking
really
from
nothing
blocking
the
v1
up
to
release.
There
were
a
couple
of
issues
that
we
like
to
close.
One
of
them,
in
particular,
is
the
secrets
migrating
to
one
secret
purse,
one
certificate
for
secret
instead
of
having
all
the
certificates
in
secret,
there's
a
PR
out
for
that.
C
A
C
So
this
has
been
tested
a
little
bit
more,
we've
been
getting
some
bugs
and
some
fixes
for
actually
spinning
up
clusters,
which
is
pretty
cool.
So
we
just
need
more
of
that
at
this
point,
but
there's
really
nothing
major
on
this
on
this
project
to
to
finish
up
for
supporting
doing
enough
to
some
that's.
A
D
Kappa
we
of
the
queue
builder
v2
that
has
been
done
check
has
worked
on
that
right
now,
we're
working
towards
automating
the
example,
something
that
we
have
done
before
they
went
out
for
one
and
there's
a
PR
open.
There
actually
does
that.
It's
like
a
really
simple
example
like
how
to
spin
up
one
cluster
with
a
simple
control
plane
and
that's
it
for
now
we're
just
going
to
start
small
but
I,
like
others
from
the
community
have
tested
like
to
spin
up
clusters,
would
kappa
n
de
cuba
di
bishop
provider
and
has
been
working
great.
E
Yeah
I
just
wanted
to
add
that
we've
been
stress-
testing
kappa
for
the
past
couple
of
days
with
a
variety
of
bootstrap
configurations
and
infra
configurations
and
all
that
good
stuff,
and
you
know,
we've
been
opening
issues
as
we
find
him,
but
so
far
everything's
been
working
really
well,
I
just
wanted
to
add
them.
Thank
you.
G
So
like
way
too
often,
there's
one
here
open
to
more
or
less
to
most
of
stuff
except
migrate,
cubed
in
bootstrap
and
as
soon
as
that's
reviewed
and
merged
I
will
start
migrating
bootstrap
and
see
if
that
works.
But
ER
looks
good
right
now.
I
was
able
to
yeah.
Mostly
the
same
stuff
is
a
double
yes,
so
that
was
good.
Look
at
yeah.
A
A
I
D
Let's
apply
machine
lists
to
cluster
Cairo,
to
spin
up
resources,
but
now
in
accept
like
one
big
gamma
file
and
understands
like
which
operates
a
switch
and
the
one
that
it
cannot
understand
because
they're
not
in
cluster
yeah,
it
will
just
assume
they're
unstructured
so
that
we
you
can
supply,
for
example,
infrastructure
configuration
or
bootstrap
configuration
and
so
on.
This
should
be
done
by
the
end
of
the
week.
A
D
I'm
ready
so
the
main
repository
I
switch,
the
queue
build
up.
So
if
you
have
a
local
builder
installed,
please
do
upgrade
and
we're
using
our
seed,
zero,
I,
think
and
also
like
control
over
anteron
tools
have
been
upgraded
in
the
in
the
upgrade
that
we
did.
The
series
now
have
category
said:
didn't
know
that
this
existed.
So
thank
you
so
much
for
to
the
new
relic
folks
and
a
big
shout
out
to
for
the
BRZ.
They
have
fun
like
across
all
the
cluster
API
of
our
stories
for
providers.
D
There
is
one
PSA
when
you're
patching
in
we
wanna
be
one
also
to
please
use
the
patch
helper
that
Jason
contributed
to
cluster
API.
This
helps
like
to
just
use
one
way
to
actually
patch
objects,
because
we
have
seen
somewhere
issue,
would
control
runtime
in
Patras
and
losing
changes
between
one
person
and
another.
D
Currently
me
and
Jason
have
been
like
kind
of
planning
and
working
towards
having
closure
API
to
build
and
push
the
container
images
after
each
PR
gets
merged
on
master.
We
have
been
working
with
the
Kate's
infer
group
this
morning
and
I
have
a
peer
open
to
add
myself
as
an
admin,
so
that
I
can
do
some
play
around
with
github
actions.
For
that
great
thanks,
Vince.
H
C
So
this
is
an
idea:
I
had
I
just
wanted
to
run
it
across
the
group
for
a
dev
tool
that
I
found
really
useful
when
I
was
building
capti
for
v1
alpha
one
that
had
the
ability
to
generate
swords
of
llamó
needed
to
spin
up
a
Cappy
cluster,
and
it
was
really
handy
to
just
be
able
to
say.
Oh
I
want
three
control
planes
and
three
machines
and
just
have
a
spit
out
the
result.
It
gets
a
little
more
complicated
with
cluster
API
v1
l2,
because
of
course
you
have
dynamic.
C
C
So
it's
not
a
trivial
piece
of
code,
but
I
think
this
would
be
really
helpful
for
testing
Pappy
in
general
and
creating
and
kind
of
getting
out
of
the
situation
that
I
found
myself
in
when
I
was
working
on,
for
instance,
Kappa
when
we
had
the
provider
components
and
we
had
originally
just
one
cluster
setup.
So
there
was
one
control
plane
and
two
workers
I,
think
something
like
that,
and
that
was
the
work.
C
That
was
the
workflow
that
I
almost
always
used
because
it
existed,
and
it
was
really
easy,
which
means
that
I
didn't
really
end
up.
Testing
AJ
as
much
as
I
would
probably
should
have
until
it
bugs
propped
up
and
I
think
that
if
you
could
generate
something
say
like
I
need,
I
mean
three
control
planes
and
two
workers
and
be
able
to
just
generate
all
the
apples
captain.
C
So
the
the
bootstrap
provided
the
comedian
bootstrap
provider
and
the
doctor
cluster,
and
then
it
just
you
need
so
I
read
up
a
little
document
about
what
I,
what
I
was
thinking
about
the
tool
and
sort
of
how
to
use
it.
If
anybody
wants
to
work
on
this
with
me,
I
would
I
would
love
help.
Probably
I
probably
don't
have
too
much
time
to
commit
to
this.
But
it's
kind
of
an
idea
I
want
to
put
out
there
I
would
look.
I
would
love
to
get
something
like
this
for
the
community.
F
C
I
the
cluster
idiom
is
explicitly
well.
No,
it's
it's.
The
development
workflow
is
neither
gold
nor
an
on
goal
of
cluster
AWS,
ATM
or
clustered
idiom,
whatever
its
called,
and
this
is
explicitly
like
for
development.
I,
don't
know
if
this
would
be
useful
for
anybody
actually
spinning
of
clusters,
but
I
would
like
this
focus
to
be
on
depth.
Cooling
is.
I
C
D
Would
be
really
nice
I
think
the
biggest
kind
of
it
there
was
this
like
if
we
are
going
towards
cluster
ad
MSD
tool,
we
should
make
like
a
big,
no
saying
like
hey.
This
is
gonna
go
away
or
like
even
hey.
This
is
a
big
bash
script
or
you
know
like
like
a
heck
together,
go
file.
That's
completely
fine!
Yeah
I
wouldn't
know
one
like
for
users
to
start
like
relying
on
this
and
and
then
it
makes
it
look
harder
to
move
to
closer
idea
right,
yeah,
Australian,
yeah,.
C
C
Probably
yeah
I
mean
there's.
There
are
a
couple
of
ideas
down
in
the
implementation
ideas
section
of
the
doc
I
have
one
of
them
would
be,
you
know,
don't
have
it
know
about
anything,
but
have
an
existing
management
cluster
with
with
the
the
providers
that
you
want
to
use
and
have
this
thing
generate
yanil
from
the
CR
these
that
are
installed
in
the
provided
cluster,
something
like
that
you
could
kind
of
get
away
from
packaging
it
all.
C
C
Yeah,
this
is
about
a
go
really
search.
So
I
picked
up
an
action
item
in
the
state
cluster
lifecycle
meeting
to
talk
about
go
releaser
and
I.
Think
that
as
we're
starting
to
get
into
the
world
of
deploying
binaries
go,
releaser
is
a
really
nice
tool
for
releasing
binaries.
I
would
not
recommend
it
for
deploying
images
for
various
reasons,
but
mostly
because
we
have
an
image.
Releasing
pipeline
setup
indicates
infrastructure
world,
and
this
is
pretty
much
what
I'll
be
covering
in
this
in
cluster
lifecycle
meeting.
C
F
Well,
yeah
I
I
did
try
it
out.
I
did
I
had
some
snacks
I
think
we
should
probably
try
to
adopt
it.
I
imagine
we
will
likely
end
up
owning
it
or
forking
it
or
breaking
into
modular
pieces.
My
concern
with
there
was
more
that,
like
it
was
sort
of
all
or
nothing
I
know
the
hypocrisy
of
the
cops
person
saying
this,
but
the
OL
or
Nothing
type
like
workflow,
didn't
work
for
my
particular
use
case.
Fair
enough,
but
I
think
if
it's
a
new
project
that
can
adopt
it
I
think
it's
seems
reasonable.
Yeah.
C
D
C
I,
don't
think
it
particularly
matters
I
think
that
if
we
adopted
before
then
it
would
be
easier
to
get
a
release
going
if
we
adopt
it
after
then,
the
first
release
would
be
harder.
Yeah,
there's
gonna,
be
some
other
automation.
Just
right
now,
I
know
that
in
Kappa
I
wrote
some
horrific
go
script.
That
does
a
bunch
of
this
already
and
I.
Don't
like
that.
Just
needs
to
go
away.
D
C
D
Images
is
probably
more
important
given
that
today,
I
cannot
test
any
provider
or
like
cluster
API
unless
I
build
it
myself,
yeah
and
yeah.
So
that's
but
I
mean
if
this
is
not
much
effort.
Even
if
it's
like
a
really
simple
go
to
this
file,
that's
completely
okay!
We
just
need
to
think
about
it
a
little
bit
more
sure.
E
So
as
of
right
now,
the
comedian
boostrap
provider
generates
a
secret
containing
acute
config.
There
were
some
concerns
about
that
happening
as
in
that
portion
of
the
process
because
of
cap
V,
and
somebody
in
CAD
V
will
have
to
speak
up
and
keep
me
honest,
but
it's
something
about
the
predictability
of
the
load.
Balancer
name,
it's
just
at
a
different
time
in
cat
V,
so
had
like
being
able
to
generate
the
cute
config
in
cluster
API
before
the
cluster
is
marked
ready,
would
would
enable
cat
V
in
that
way.
D
D
It
just
says
like
a
disser,
where
we
put
the
secrets-
and
this
is
where
we
do
expect
the
booster
provider
to
generate
those
secrets,
so
we
can
actually
think
about
it
and
probably
not
gonna
go
in
for
like
the
first
movie,
so
he
went
out
for
two
and
I'm,
not
sure.
If
we
were
considered
there's
a
breaking
change
could
move
it
from
Cuba
diem
to
closer
API,
so
he
could
go
into
v1.
Also
too
I
think.
J
L
L
A
Yeah,
so
for
those
who
aren't
familiar
with
what
measures
are
talking
about,
there's
multiple
places
in
various
cluster
API
repositories,
where
we
take
the
cluster
name
and
either
prepend
or
append
something
else
to
it
and
use
that
for
a
secret
or
a
config
map
or
whatever,
as
its
name.
And
if
the
name
of
the
cluster
is
the
maximum
length
of
allowed
for
a
kubernetes
name,
then
we've
run
out
of
room
and
we
can't
append
or
prepend
anything
to
it
and
so
trying
to
create
a
secret
or
any
other
object.
A
F
D
D
A
D
D
A
E
A
K
L
D
I
A
D
D
D
D
F
I
think
I
have
a
way
I.
Do
it.
I
have
these
two
decomposed
tools?
I
have
ship
bot
which,
like
uploads,
to
get
up
and
I,
have
rel
notes
which,
like
walks
through
a
list
of
commits,
which
we
generate
the
git
log
and
then
extracts
out
like
the
author,
we
could
extend
that
to
parse
the
either
the
commits
or
the
github
commit
messages
for
like
looking
for
the
standard
criminals
release
blocks.
F
A
F
A
C
It's
also
it's
also
I
was
just
looking.
It's
got
a
pretty
interesting
feature
where
you
could
just
pass
in
some
custom
release,
notes
which
we
could
all
we
could
override.
You
know
whatever
it
does
by
default,
then
actually
write
something
that
pulls
out.
The
P
like
use
like
you
use,
kubernetes
tooling,
whatever
is
using
the
release,
notes,
tool
and
you
can
pull
that
out
and
generate
those
release.
Notes
use
that
by
default.
If
we
wanted.
D
So
one
clarification
is
like
not
that
I
I
am
Pro
release
notes
per
se
I
just
like
have
like
better
release.
Note,
if
guru
me
suppose
all
the
commits-
and
we
have
nicer
commit
messages
that
would
be
a
great
start,
I
think.
Maybe
we
can
just
remove
the
release,
notes
from
the
template
as
well.
If
we're
not
planning
to
use
it
a
given,
it
just
takes
up
space
at
this
point.
D
A
All
right
so
I've
just
got
a
filter
here
for
all
Costa
Rica
issues
without
a
milestone.
So
let's
go
bottom
to
top
here
we
have
an
able
optional
use
of
local
kubernetes
api
connection
in
node
ref
controller.
So
this
is
a
request
to
use
the
in
cluster
credentials
for
certain
setups,
where
your
nodes
and
machines
and
clusters
are
all
one
in
the
same
and
that
there's
not
a
management
cluster
involved.
So
I.
A
I'm
looking
for
anybody's
comments
on
this
one
I
would
say:
potentially
we
could
maybe
add
it
after
V
1
alpha
2
comes
out
the
door,
but
I
don't
know
that
we
have
bandwidth
to
get
it
in
in
the
next
week
or
so
I.
D
O
A
I
had
talked
with
Derek
a
few
weeks
ago
and
he
seemed
to
be
okay
with
requiring
the
cluster
and
I
know
that
that's
you
know
whether
it's
require
or
not
is
not
a
done
deal
there's
a
lot
of
discussion
that
needs
to
happen,
but
we
did
have
a
chat
and
you
seemed
to
be
okay
with
it,
and
it's
something
that
I
imagine.
We
will
probably
talk
a
lot
about
at
the
face-to-face.
O
A
You
know
if
somebody
wanted
to
try
and
code
that
up
but
I'm
tempted
to
say
that
this
is
probably
like
a
p1
and
in
terms
of
milestone.
I'd
be
okay,
putting
it
for
v1
alpha,
but
it's
the
sort
of
thing
that,
if
it
doesn't
make
it
by
the
time
we
need
to
come
out
with
the
release,
then
it
could
come
out.
You
know
patch
release
afterwards
that
sound
reasonable.
A
A
A
Thank
you
all
right,
so
those
are
the
ones
for
cluster
API
for
the
cubed
iam
bootstrapper.
We
have
several
here
without
a
milestone.
So
andrew
is
asking
for
writing
out.
Etsy
hosts
and
Etsy
host
name
Jason
looks
like
you
invent
sets
them
back
and
forth
here.
Can
one
of
you
summarize
this
long
conversation.
J
A
A
J
A
A
A
O
A
O
J
This
has
been
a
specific
request
from
folks
to
support
other
os's
that
may
not
consume
cloud
in
it.
Like
you
know,
core
OS,
for
example,
with
the
ignition
format
they
don't
want
to
have
to
build
and
maintain
a
completely
different
bootstrap
provider.
That's
gonna
replicate
the
majority
of
what
we're
building
with
the
cubed
en
bootstrapper,
just
to
have
a
different
output
format,
so
instead
just
be
able
to
select
which
type
output
format
you
want
from
the
cube,
ADM,
bootstrapper
and
weed
they
can
consume
that.
A
E
Selecting
I
would
expect
like
if,
if
they're
gonna
be
two
output
formats,
I
would
expect
to
be
able
to
use
the
same
exact
input
and
just
toggle.
You
know
my
format
and
get
working
output
and
I.
Think
that
works
here,
because
cue
medium
is,
is
the
actual
bootstrapper
like.
So
all
the
inputs
for
creating
valid
QAM
config
remain
constant,
but
you
know
it's
just
wiring
that
into
a
you
know
into
a
cloud
in
it
style,
format
or
wiring
it
into
an
ignition
JSON.
E
If,
if
somebody
wanted
another
bootstrapper
like
that
would
be
you
know,
I
don't
want
to
use
cube
ADM
to
join
my
notes
to
the
cluster
like
I
want
to
use
I,
don't
I,
don't
know,
I've,
never
used
anything.
That's
not
give
idiom
to
be
honest,
but
you
know
something
else
like
I
see
that
as
the
dividing
line
for
a
new
controller,
there's
Ranger
that
there
is.
A
F
A
A
A
A
A
Okay,
I'm
gonna
I'll
put
this
in
the
milestone
so
that
we
at
least
know
to
come
back
and
look
at
it
when
we're
doing
burn
down
or
final
burn
down,
and
if
you're
will
either
close
it
or
fix
it.
Depending
on
how
things
unfold
over
the
next
few
days,
all
right
and
spec
cloud
Rider
in
Cuba,
diem,
config,
Vince.
D
Yeah,
this
was
mostly
a
suggestion
to
improve
the
user
experience.
If
you
see
the
example
in
the
first,
you
can
see
cloud
providers
repeated
like
multiple
times
and
it
could
be
like
just
like
outside.
In
the
cube
idiom
config
say
flood
provider,
yes
and
then
the
controller
would
just
like
add
those
in
for
you.
A
Jason
was
interesting
like
we
might
move
to
the
external
provider,
and
this
should
not
be
configurable.
D
E
Yeah,
just
real
quick
on
this
kind
of
issue,
something
that
we
kind
of
like
Boston
v
1
alpha
2,
is
that
the
bootstrap
config
that
gets
generated
can
no
longer
have
sensible
defaults
based
on
the
infrastructure
provider.
So
it
may
be
worth
documenting
if
we're
pushing
this
to
long
term
that,
like
hey,
you
know,
if
you
have
a
bunch
of
machine
Yam
while
somewhere
they've
been
using.
You
have
to
now
specify,
like
all
of
this
cloud
provider,
stuff
and
cubed
x,
charge
because
the
bootstrap
provider
doesn't
know
how
to
do
that.
A
A
J
A
I
think
we
can
probably
brainstorm
on
that
in
the
issue
or
another
meeting
beyond
today,
because
we're
almost
out
of
time
so
I
did
put
this
one
in
the
milestone,
but
long
term,
we'll
see
where
we
go
with
it
and
then
the
last
one
is
enabling
static
check
for
this
repo,
which
fund
that
it
has
a
priority
of
backlog.
I'm
gonna
put
it
on
next
and
if
we
get
around
to
it
sooner
great.