►
From YouTube: 20191216 - Cluster API Provider AWS 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
A
To
get
started
with
it
looks
like
we
have
a
PSA
that
version
0.4
dot
7
was
released,
it
adds
support
for
reconciling
the
tags
for
the
ELB
and
as
heads-up
it
does
require
new
I
am
permission,
so
anybody
who
is
adopting
it
make
sure
that
you
update
the
IAM
permissions
in
addition
to
the
images
and
binaries
all
right.
Next
up,
we
got
our
first
demo
here
so
Andy.
Take
it
away
with
the
tilt
workflow
thanks.
B
Jason,
hopefully
this
works
we'll
see
how
it
goes
so.
I
have
a
actually
before
I
start
with
what
I
have
if
you're
not
familiar
with
tilt.
It
is
a
way
to
do
more.
Rapid
iterative
development
with
kubernetes
based
applications,
and
so
I
have
a
couple
of
pull
requests
that
I've
created
one
for
cluster
API
and
a
corresponding
one
for
Kappa
that
try
to
make
it
easier
to
do
rapid
development.
B
B
Fingered
something
alrighty
so
now
it's
gonna
go
ahead
and
load
my
tilt
file,
which
will
attempt
to
pre
pull
and
preload
into
kind
the
cert
manager
images.
This
is
an
optimization
so
that,
if
you
are
frequently
deleting
your
kind
cluster
and
recreating
it,
you
don't
have
to
rely
on
the
kind
cluster
to
pull
the
images
over
and
over
again.
B
You
and
I
can
walk
through
the
tilt
file
and
some
of
the
stuff
that
it's
kind
of
nice
and
magic
in
here.
Once
we
get
this
thing
up
and
running.
Ok,
so
there's
certain
manager
is
good,
and
now
it's
found
that
I
want
to
have
a
Cappy
controller
manager,
a
capo
controller
manager.
These
are
all
the
the
amyl
pieces
and
you'll
see
that
nothing's
happened
with
them.
Yet
what
it's
actually
doing
right
now
is
running
a
command
on.
B
B
This
is
the
cluster
API
container
and
eventually
it
will
get
down
here
and
we
have
the
cube,
ATM,
control
playing
controller
and
so
on
and
so
forth.
Kappa
very
similar,
exact,
same
docker
file
and
it
copies
in
the
till
built
manager
and
all
of
this
stuff
is
up
and
running.
So
now
we
have
this
up
and
running.
I
could
create
a
cluster
if
I
wanted
to,
but
I'm
not
going
to
what
I
am
going
to
show.
B
B
I
am
going
to
do
our
long
dot
info
hello,
capital
of
meeting
and
I'm.
Just
gonna
save
this
and
flip
back
over
to
my
tilt.
Output
and
you'll
see
that
it
noticed
that
the
cluster
controller
changed,
and
so
it's
rebuilding
this
core
manager
resource
and
it's
it's
literally
just
doing-
go
build
to
get
the
whole
thing.
Rebuilt
I,
don't
know
if
there's
any
way
to
make
this
go
any
faster.
I
would
hope
that
it
wouldn't
take
45
seconds
to
a
minute
to
rebuild
just
the
binary,
but
at
least
for
right
now.
B
That's
what
it
does
on
my
laptop.
But
the
nice
thing
is
once
this
finishes:
building
you'll
see
that
it
just
does
a
copy
and
issues
a
restart,
and
the
only
thing
that
changes
is
that
file
gets
put
in
place
and
the
container
stays
up
and
running
and
there's
no,
oh,
that's
the
sub.
Oh
I,
think
I've
yeah,
sorry
for
the
the
demo,
not
working
here,
yeah
a
demo
gods.
I
was
reworking
some
stuff,
but
in
general
this
will
not
say
that
it's
falling
back
to
the
next
update
method
and
we'll
do
a
live.
B
B
So
if
you
go
to
the
master
set
of
documentation
and
are
sorry,
this
isn't
a
master.
Yet,
if
you
go
to
do,
I
have
met
Lathon,
not
yet
all
right.
Let
me
show
you
the
documentation
that
I
have,
since
it's
not
showing
up.
So
if
we
go
into
Doc's
source
developer,
tilt
I
have
a
whole
document
in
here
that
will
show
you
how
to
work
with
this.
So
once
you
have
docker
kind
and
Tilt
installed,
and
you
clone
cluster
API
as
well
as
any
other
providers
that
you
want
to
use.
B
B
So
we
didn't
want
to
put
any
hard
coding
into
the
tilt
file
in
cluster
API
to
say:
here's
Kappa,
here's
cat
B,
here's
cat
Z,
here's
cap
whatever,
and
so
we
took
a
federated
approach
where
you
essentially
list
out
where,
on
disk
you
have
cloned
the
other
providers
that
you
want
to
work
with.
So
for
Kappa,
for
example,
it
is
a
sibling
on
on
my
check
out
at
least
I
have
one
directory.
B
That's
got
all
these
different
cluster
API
repositories
and
so
I
just
say:
go
up
a
directory
and
here's
my
cluster
API
provider
AWS
and
then
I
tell
it
which
providers
I'd
like
enabled,
and
so
the
docker
provider
lives
in
the
cluster
API
repository.
So
this
is
always
an
option.
If
you
want
to
use
it,
and
you
can
put
this
string
in
place
and
then
I'll,
you
know,
if
you've
got
other
ones
like
Kappa,
you
would
just
give
it
the
name
of
the
provider
as
defined
in
the
providers
till
provider
JSON
file.
B
B
Tilta
and
I
have
documentation
around
what
the
the
fields
in
tilt
settings
JSON
mean
how
to
run
tilt,
what
the
available
providers
are
out
of
the
box
and
what
a
tilt
provider
JSON
might
look
like
and
then
what
those
config
fields
are
for
and
then,
if
you
are
interested,
you
can
additionally
customize
tilt
inside
of
your
cluster
API
directory.
You
can
create
a
tilt
directory
and
put
as
many
tilt
files
as
you
want
in
there
and
it'll
just
load
everything
that's
in
there.
B
So
if
you
wanted
to
deploy
something
additionally
with
some
extra
customize
or
do
some
other
tilt
directives,
you
could
put
them
in
there
and
then
at
the
bottom.
I
just
talked
about
what
it's
actually
doing
and
how
live
updates
work.
I
will,
of
course,
fix
the
bug
that
I
have
it's
broken,
live
updating
and.
B
Excuse
me:
that's
what
I
have
for
a
demo
too
bad
that
update
didn't
work,
but
these
are
pull
requests
right
now,
it's
1894
in
copy
14
22
in
Kappa
I
need
to
fix
a
couple
things
before
we
could
actually
merge
them
like
the
cap.
Opr
breaks,
the
quick
start,
for
example,
and
I
need
to
fix
that.
But
I
did
want
to
just
share
this,
so
y'all
could
see
this
tilt
flow
and
hopefully,
once
the
stuff
merges
you
can
take
advantage
of
it.
B
D
B
So
that
was
one
thing
that
I
struggled
with,
because
I
was
I
was
cheating
off
of
what
Chuck
had
pioneered,
which
was
taking
the
the
Builder
image
which
we
have
in
our
primary
docker
files
and
basically
installing
that
in
the
kind
cluster
and
that
had
the
full
code
tool
chain,
all
the
modules
downloaded
and
the
ability
to
compile
in
the
image.
But
the
problem
is
that
image
is
about
two
gigabytes
and
so
I.
B
Don't
know
what
the
deal
is
on
my
2017,
MacBook
Pro,
but
I
mean
I
could
go,
make
a
cup
of
coffee
while
waiting
for
it
just
to
load
the
docker
image
into
the
kind
cluster
and
then
once
it
was
loaded.
The
reloads
like
rebuilding
and
reloading
went
quickly.
But
this
way
the
Debian
stretch
image
with
the
manager
binary
in
there
is
about
hundred
Meg's
and
I.
Imagine
we
could
go
smaller
if
we
went
with
another
like
a
smaller
image,
something
smaller
than
the
Debian
slim
but
yeah.
E
One
of
the
things
that
I
noticed
at
least
a
couple
months
ago
with
tilt
was
one
thing
to
watch
out
for
is
like
rapidly
saving
a
bunch
of
different
files,
because
the
up,
because
of
the
build
reload,
is
like
queued
so
like.
Sometimes
you
can
get
into
a
scenario
where,
like
you,
save,
may
not
go
and
then
controller
not
go
and
then
like
a
test
file
go
and
it
will
actually
build
three
times.
Sequentially
I,
don't
know
if
they've
fixed
that,
but
that
can
that
can
slow
down
the
build
loop.
Sometimes
you.
B
B
Imagine
if
you're
rapidly
making
changes
and
it
still
queues
up
things,
that's
because
of
the
auto
trigger,
and
so
we
can
make
it
so
that
the
compilation
of
the
manager
binary
is
a
manual
trigger,
and
you
I'm
just
checking
the
docs.
Here
you
may
be
able
to
make
the
d
amal
b,
not
auto
done,
although
I'm
not
seeing
that
I
think
the
biggest
biggest
set
of
changes
that
people
would
be
making
will
be
code
changes
in
code
files.
B
So
if
we
switch
the
local
resource
for
building
the
manager
binaries
to
a
manual
trigger
or
at
least
allow
it
to
be
customizable,
then
that
would
probably
help
out
with
that
situation.
The
thing
that
I
really
really
want
to
do
is
make
it
so
that,
as
developers,
we
all
have
total
flexibility
of
customizing
things
like
that
without
having
to
make
a
change
to
a
tilt
file
and
then
exploit
decide.
Oh
I
need
to
not
include
that
in
my
git
commit
because
it's
a
local
change
and
I
don't
want
to
share
that
with
everybody.
B
So
that
was
the
reason
for
having
the
Tildy
directory
so
that
you
can
go
make
changes.
The
only
downside
with
the
the
tilt
D
inclusions
is
that
they
don't
have
access
to
any
of
the
global
variables
that
are
defined
in
the
tilt
file
like
the
providers
map.
So
you
basically
can
say,
deploy
some
extra
stuff.
But
if
you
want
to
customize
a
setting
in
a
global
variable
or
a
global
map,
you
can't
do
that,
but
we
certainly
could
add
some
fields
that
say
like
for
my
provider.
D
B
A
All
right,
so,
next
up,
we
start
the
group
topics.
I
didn't
want
to
let
everybody
know
that
the
cube,
idiom
control,
plane,
create
workflow,
has
been
merged
and
I
know
some
folks
have
expressed
interest
in
helping
out
with
the
rest
of
the
control
plane
implementation.
So
if
you
fall
under
that
group,
please
go
ahead
and
click
on
the
tracking
issue.
We're
going
to
use
that
to
coordinate
all
further
work.
So
it's
all
up
to
date
and
we're
going
to
start
divvying
out
and
self
assigning
work
off.
A
A
A
We
still
have
to
implement
the
control,
plane,
initialize
logic,
to
only
work
if
the
control
plane
ref
is
not
set,
and
that's
mainly
because
the
health
check
is
not
implemented
yet
so
we're
not
setting
that
through
the
cube
idiom
control
plane
right
now,
so
we
need
to
fall
back
to
that
for
functionality
and
right
now
we
have
create
workflow.
We
still
need
the
delete,
workflow
scale
up
scale
down.
A
The
upgrade
is
likely
to
be
one
of
the
last
things
that
we
can
work
on,
because
it's
going
to
touch
a
little
bit
of
everything
and
the
health
check.
You
know,
there's
some
other
documentation,
issues
that
we
need
to
sort
out.
Nadir
started
on
some
of
that
with
this
issue
1727
and
an
associated
PR
there.
So
if
you're
interested
in
helping
out
jump
on
this
issue-
and
let
us
know
where
you'd
like
to
assist.
E
A
C
A
E
Yeah
I
just
was
curious
if
someone
could
take
a
look
at
it
just
because
it's
been
open
for
a
while
I
did
the
back
port
first.
If
that's
a
problem,
I'd
be
happy
to
do
the
master
one
as
well,
but
this
is
just
getting
a
root
volume
created
with
the
instance
to
have
the
same
tags
as
the
instance
itself.
E
A
B
The
flu
that's,
the
right
share.
Y'all
can
see
the
Cap'n's
right,
trucha,
okay,
I'm
gonna
start
at
the
bottom.
We
have
setting
a
nat
gateway,
ID
on
a
subnet
causes,
seg
faults
and
I.
Think
actually,
no
I
thought
this
was
one
where
it
was
a,
maybe
wasn't
reproducible,
but
that
must
have
been
something
else
so
I
know
Jason
you
and
Vince
had
commented
on
this.
Is
this
something
that
I
guess
we
probably
just
need
to
have
somebody
look
into
it
right.
B
B
B
E
B
C
E
B
E
B
I
forget
who
reported
this,
but
basically
we
have
a
some
code
that
checks
to
see
if
an
error
that
comes
back
from
described
instances
is
a
not
found
error
and
it's
not
including
invalid
instance
ID
not
found,
which
is
what
you
get.
If
it's
truly
not
there
like
post
terminated,
it's
been
cleared
from
the
API,
so
we
just
need
to
make
a
change.
To
is
invalid,
not
found
error
minimally
to
include
that
this
code
as
well,
and
we
should
probably
do
it
in
both
0.4
and
master.
B
B
Okay,
machine
stuck
in
provisioned
after
NAT
Gateway
reconciliation-
oh
this
was
one
where
they
were
deleting
the
NAT
gateway
and
getting
it
reprovision
and
the
route
tables
weren't.
Getting
updated
and
Vince
has
a
PR,
so
double
check
which
brand
she's
doing
this
against
master.
Alright,
so
we'll
stick
this
in
and
give
it
soon.
B
C
E
B
B
A
B
It's
pretty
big
like
you
would
you
would
need
to
stick
it
in
and
gist
or
upload
it
to
a
file
site
or
something
like
I
think
I,
don't
know
that
tilt
is
necessarily
the
right
thing
like
if
we
had
a
tool
that
could
just
easily
go
and
grab
yeah
mole
and
the
logs
from
the
right
controllers
and
redact
any
secrets
that
are
in
there.
That
might
be
nice
to
have,
but
I
think
in
the
short
term.
Just
asking
for
logs
is
probably
what
we're
gonna
have
to
do.
Okay,.
F
B
D
Was
this?
Was
this
getting
logs
out
of
tilt?
Is
that
yeah
there's
a
snapshot
feature
that
I
think
it's
open
now
and
you
can
save
a
snapshot
of
your
entire
tilt
environment
and
then
paste
it
into
like
github
and
people
can
see
like
the
entire
history
of
whatever
your
current
tilt
environment?
Was
it's
pretty
useful,
assuming
you're,
not
logging,
anything
sensitive.
G
B
All
right,
I
opened
up
an
issue.
This
relates
to
the
cluster
API
issue
that
I
had
created
last
week
or
so
about
trying
to
add
a
product
project
roadmap
document,
and
so
I
created
a
rough
draft
with
some
seed
ideas.
Nothing
is
set
in
stone
here
and
I
think
nadir.
When
we
were
talking
about
this
the
other
day,
there
was
something
about
cloud
watch
in
here.
B
Maybe
that
I
don't
remember
the
details,
so
if
you
all
have
ideas
or
what
we
can
do
in
the
upcoming
versions
and
any
sort
of
backlog
features,
I
would
definitely
recommend
trying
to
put
stuff
in
here,
add
comments
and
questions
if
you
have
them
and
once
things
kind
of
settle
a
little
bit,
I'd
like
to
turn
this
into
a
markdown
document
that
we
store
in
the
repository
and
keep
up
to
date.
There
also
is
a
corresponding
cluster
API
draft
roadmap.
Let
me
grab
it
real,
quick.
Just
to
show
you
all
it
is.
B
I
can
probably
just
stick
a
link
in
here
so
draft
cluster
API
a
roadmap.
So
this
one
is
a
bit
bigger
because
it's
more
around
cluster
API
and
not
capo
specifically,
but
I
would
encourage
you
all
to
look
at
this
one
too.
And
if
you
have
questions
comments,
ideas,
please
take
a
look
and
I
did
also
post
to
the
SiC
cluster
lifecycle,
Google
Group
and
the
discuss
forum
for
the
cluster
API
one.
B
This
one
I
did
not
post
for
the
Capital
One,
because
I
wanted
to
have
a
talk
here,
first,
before
figuring
out
where
the
best
place
is
where
to
post.
But
if
you
all
are
in
agreement,
I
would
like
to
try
to
get
these
or
to
get
a
roadmap
document
again.
Rough
draft
will
be
updated
or
I
should
say
living
document
and
get
a
markdown
file
in
that
repository
as
part
of
this
release
cycle,
so
I'm
gonna
go
ahead
and
put
this
in
a
milestone
and
I
don't
have
any
idea
for
timing.