►
From YouTube: 2020-05-04 - 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
Today
is
May
4th
2020:
this
is
the
cluster
API
AWS
provider
office
hours,
cluster
API
provider.
Afraid
of
us
is
a
sub-project,
that's
a
cluster
lifecycle
so
and
this
video
is
being
recorded
and
will
be
posted
to
youtube
when
the
recording
is
done.
Please
add
your
name
to
the
attending
list
and
let's
go
through
the
agenda
so
Jason
over
to
you
for
the
release
announcement.
A
C
A
Jason
all
right,
we
have
some
group
topics
here.
So
let's
go
with
the
first
one
here
on
mounting
additional
data
disks
yeah,
so
hey
guys,
I'm
Matt
may
have
seen
me
around.
There's
our
friends
over
and
cap
Z
we're
implementing
one
of
the
one
of
the
features
that
we're
also
looking
at
external
Etsy
D
volumes
and
they
submitted
a
proposal
to
modify
or
add
additional
fields
to
the
to
the
Kubat
Kubb
admin
config
for
file
system
config.
A
So
there's
a
fork
that
they're
working
against
on
copy
and
kind
of
what
I
the
goal
that
each
of
the
cloud
providers
will
implement
their
specific
volume
mounting
in
that
regard.
So
it's
kind
of
all
work
its
work
in
progress
at
this
point,
but
there's
a
there's:
a
PR
out
there
that
I'm
working
against
and
I'm
kind
of
working
on
the
details
in
this
proposal
with
cap
Z
folks.
A
So
if
anybody
wants
to
take
a
look
and
just
see
what
the
work
kind
of
bounce
ideas,
the
direction
that
we
want
to
go,
but
you
can
kind
of
see
the
direction
that
is
currently
being
tested
in
regards
to
setting
up
these
volumes.
Ma'aming
that
in
the
user
data
section
and
things
like
that,
besides
that
just
there's
a
PR,
that's
ready
to
merge
the
custom
image,
look
up
PR.
It
looks
like
the
there's
some
issues
with
the
testing,
but
I
just
retested
it
and
it
worked
fine.
A
A
On
slack
about
some
possible
difficulties
having
an
infrastructure
provider
tell
the
bootstrap
provider
that
you
need
to
run
some
commands
to
go
mount
some
things.
Did
you
work
through
the
you
know
the
fact
that
they
kind
of
need
to
talk
to
each
other
yeah?
It
looks
like
there's
a
couple
of
different
possibilities
in
the
proposal.
A
There's,
okay,
so,
like
a
high
level,
there's
the
ability
to
like
embed
baked
those
into
the
image,
with
image,
builder
and
kind
of
abstract,
the
logic
of
actually
mounting
and
detecting
what
type
of
infrastructure
the
image
and
what
type
of
volumes
it's
using.
Underneath,
there's
also
the
option
of
including
yeah
the
bootstrap
provider
that
additional
field
that
has
the
KU
beta
baby
on
three
steps.
A
Yeah
you
got
it
yeah
I
know
we.
This
was
something
that
we
knew
was
going
to
be
a
challenge
when
we
split
the
cube,
EDM,
bootstrap
or
out
and
made
it
its
own
thing.
So
I
think
this
is
like
we're.
Just
gonna
have
to
see
what
we
can
come
up
with
and
it
may
take
a
few
iterations
to
get
to
a
place
where
it's
not
so
annoying
to
do
it
yeah.
A
A
C
A
A
C
A
A
How
so
I've
fiddled
around
with
disks
and
volumes
in
AWS
and
noticed
that,
like
you
said,
the
naming
like
the
device
that
is
exposed
isn't
super
consistent.
All
the
time
and
I
know
you
have
control
over
it
from
an
AWS
perspective
as
well.
So
is
there
some
way
to
make
this
less
error-prone
for
the
user
or
I
found?
Was
it
if
the
underlying
AWS
instance
type
supports
nvme,
which
I
believe
a
lot
of
them
have
moved
into
that
direction?
A
A
B
Now
the
the
pro
I
just
finally
got
through
the
upstream
proposal
and
I
think
that's
probably
a
good
path
for
the
initial
go
through
and
I'm
sure
we'll
figure
out
some
rough
edges
as
we
go
and
continue
to
be
up
with
tweak
it.
Okay,.
E
Additional
security
groups
and
additional
tags
will
actually
get
changed
in
AWS
if
you
modify
them,
but
there
is
no
way
to
change
those
through
a
machine
deployment,
because
you're
gonna
get
all
new
machines
and
I'm
not
bringing
this
up
because
I
think
it's
something
that
like
needs
to
be
solved
through
machine
deployments.
It's
just
a
thing
that
happened
and
is
funny
so
I
wanted
to
bring
it
up
on
the
call.
E
I
hope
not
I
mean
so
the
fix
right
is
is
really
just
like
go
patch
all
your
AWS
machines
and
then
also
patch.
The
template
like
there
is
a
removing
of
a
window
cook
in
order
to
make
that
happen,
but,
like
it's
not
so
bad,
it's
just
I,
don't
I
would
be
curious
to
find,
if,
like
anyone
is
really
like
making
nodes
that
aren't
machine
deployments.
So
like
is
anyone
really
using
those
mutable
fields?
A
I
mean
certainly,
as
we've
pushed
I,
don't
know
how
publicly
or
how
broadly
we've
pushed
this
advice,
but
the
same
way
that
with
kubernetes,
if
you're
doing
just
a
one-off
pod,
you're
doing
it
wrong.
You
know:
we've
tried
to
say
the
same
thing
with
machines
versus
machine
deployments
and
kcp,
so
yeah
I
don't
know
I.
F
C
I
was
gonna,
add
like
from
a
copy
point
of
view,
like
it's
gonna,
get
tricky
to
add
like
mutability
functionality
right
but
like
from
Katipunan
view
like,
as
you
said
like
you,
could
have
like
a
different
controller.
Potentially
that,
like
a
record,
sounds
a
template,
but
that
means
the
template
will
have
to
be
mutable
which
breaks
the
hashing
assumption
there
like
we
make
upstream
it's
a
I
guess.
It's
like
a
bouncy
act.
C
I
wouldn't
be
opposed
to
that
like
per
se,
but
that's
only
because,
like
I'm,
very
comfortable
with
the
SS
mutable
fields
like
but
I,
don't
know
about
other
providers,
so
like
I
guess
like
if
you
you
need
this
like
I,
would
actually
really
make
it
in
kappa.
Specifically
so
they're
like
we
can
keep
it
contained
and
then
learn
and
rinse
and
repeat
like
upstream
when
we
get
there
we'll
need
to
figure
out
the
whole
hashing
situation.
It
would
be
great
if
we
hang
out
at
each
field
to
say
like
hey.
C
F
A
Great,
thank
you
all
right,
I
added
just
an
FYI
here,
there's
a
pull
request
to
deregister
a
machine,
a
control
plane
machine
from
the
ELB
when
the
machine
is
deleted
to
try
and
reduce
the
amount
of
time
where
the
control
plane
machine
would
still
be
a
member
of
the
ELB
but
not
be
able
to
service
requests,
and
so
the
health
check
I.
Don't
remember
the
the
time
out
on
that,
but
this
tries
to
reduce
that
window
where
the
machine
still
looks
like
it's
unavailable
back-end.
A
But
it's
really
not
I
did
want
to
point
this
out
because
there
is
one
new
role
or
permission.
I
should
say
that's
required
on
the
controllers
policy.
So
assuming
this
PR
does
merge.
This
is
just
a
heads
up
that
when
you
go
to
the
newer
version
of
Kappa
that
has
this,
you
will
need
this
new
premiss
permissions
change.
A
And
I
know
so:
we've
had
some
internal
requests
from
some
of
our
teams
that
the
addition
of
new
I
am
permissions
is
disruptive
potentially
to
some
of
our
clients,
and
it
would
be
nice
if
we
could
find
ways
to
make
some
of
these
changes
optional
at
or
at
least
have
some
sort
of
graceful
fallback
if
they're
not
available.
So
this
is
just
an
example
of
a
PR
that
makes
a
change
and
I
don't
know
if
there's
anything
that
we
would
want
to
do
here.
C
Yeah
I'll
definitely
yeah
I
would
add
a
note
like
to
the
PR
like
to
add
a
like.
A
grease
will
fall
back.
It
gets
like
it.
This
is
it's
like
a
nice
to
have
for
sure,
but
yeah
like
could
use
this
like
I
might
not
need
that.
If
they
don't
want
to
it's
completely
fine,
it
will
be
a
president,
but
I
think
it's
a
great
one
to
have
actually,
rather
than
like
I.
Don't
know
that
the
full
story,
though,
about
like
making
some
things
up,
then
some
others,
don't
that's
kind
of
yeah.
A
I
mean
it
really,
it's
so
I
would
say
that
it
probably
affects
two
scenarios:
one
you're
upgrading
from
an
existing
management
cluster
to
a
newer
version
of
Kappa.
The
other
is
maybe
there's
some
aspects
of
Cathar
that
somebody
decides
I,
don't
want
them
doing
that
which
I
find
that
probably
less
likely
or
not
likely
like
we're
building
a
system-
and
these
are
this-
is
what
it
does.
So
you
need
to
give
these
permissions
yeah.
A
A
C
F
So
just
this,
if
you
have
one
exactly
one
subnet,
it
will
tell
you
that
you
need
a
public
one.
Otherwise
everything
seems
like
it
should
work
fine.
So
this
is
more
about
like
a
fixing
that
one
minor
bug
and
then
sort
of
setting
an
expectation,
updating
the
docs
that
I've
made
Scott
changed
to
cover
this
edge
case.
A
Yeah,
so
I
was
looking
at
this
one
and
I
had
a
honest,
genuine
question
here.
So
I
was
looking
at
the
difference
between
public
and
private
subnets
with
AWS,
and
if
I
read
it
correctly
from
the
AWS
documentation,
a
public
subnet
doesn't
have
to
get
matted
to
talk
to
the
other
hosts
on
the
internet
and
a
private
subnet
can
get
out
to
the
internet.
Even
though
it's
private
it
would
be
natin,
but
you
can
enable
or
disable
that,
based
on
your
your
networking
configuration
is
that
accurate
and.
F
D
Yes,
so
if
you're
using
Direct,
Connect
I
guess
this
is
one
of
the
these
cases
that
so
you're
actually
sending
your
internet
bound,
traffic
or
extra.
You
can
see
traffic
back
through
your
data
center
through
some
transit
through
physical
connection,
then
you
won't
have
internet
gateway,
even
in
your
EPC
yeah.
C
Yeah,
when
it
everything
is
private,
they
really,
we
didn't,
have
a
support.
I.
Think
like
we
added
that
private
ELP
at
some
point,
then
we
didn't
change
theirs
because
we're
still
creating
Nats
and
Internet
gateway
by
default,
which
we
should
so
I.
Guess
like
the
question
is
like
we
are
like
add
a
new
field
to
disable
those
or
or
like
we
add
a
field
to
enable
them
and
default
them
to
true
or.