►
From YouTube: Kubernetes - AWS Provider - Meeting 20200306
Description
Recording of the AWS Provider subproject meeting held on 20200306
A
Hello,
everybody
and
welcome
to
the
provider.
Eight
of
the
best
meeting
today
is
Friday
March
6th
2020
I.
Am
your
moderator,
moderator,
facilitator,
Justin,
Santa,
Barbara
I
work
at
Google,
a
reminder.
This
meeting
is
being
recorded
and
will
be
put
on
the
internet
and
to
please
be
mindful
of
our
code
of
conduct,
which
essentially
is
to
be
a
good
person
in
bigger
groups.
We
need
to
use
the
hadn't
feature,
but
I
don't
think
we
need
to
do
that
here
today.
A
B
Sure
my
name
is
Ian
Gable
I
work
for
a
small
start-up
that
builds
a
graph
database,
we're
working
with
eks
I'm,
not
sure
if
it's
appropriate,
that
I
joined
this
meeting
but
I'm
here
to
understand
how
that
Howie
KS
is
an
evolving
and
sort
of
learn.
What
is
happening
with
it
so
I'm
here
to
learn
or
and
observe-
and
you
guys
could
tell
me
if
I
should
be
here
or
not
you're,.
A
Everyone
is
always
welcome
and
you
thank
you
for
using
yourself.
We,
we
do
definitely
talk
about
crew,
binaries
open-source
and
we
also
talked
I,
think
tangentially
about
providers.
Although
I
don't
know
next,
you
wanna
talk
about
if
there
are
any
other
specific
forums
for
eks
I'm,
just
nominating
YouTube
yeah.
C
A
C
Let
me
just
pull
it
up,
so
I
can
kind
of
remind
myself.
I
was
thinking
that
it
was
I.
Think
the
main
question
was
whether
or
not
the
error
that
was
introduced
would
fail
for
any
users
kind
of
like
any
current
use
cases
that
users
have
sorry
my
daughter's
trying
to
join
the
conversation
here
so
I
think
the
the
case
that
is
introduced
is
when
the
back
end
annotation
protocol,
the
the
back
end
protocol
annotation,
is
set
to
secure
and
we
loop
over
the
listeners
looking
for
a
secure
front,
end
listener
and
don't
find
one.
C
Then
we
then
we
fail
and
so
I
think.
The
question
is
whether
or
not
there's
ever
a
case
where
the
user
wants
a
secure
health
check
but
doesn't
set
the
back
end
protocol
annotation
to
a
secure
protocol.
So
if,
if
that
is
true
people,
if
people
actually
do
that,
then
I
would
say
this,
maybe
wouldn't
be
the
correct
behavior,
but
I
I.
Think
my
understanding
is
that
if
you
want
like
I,
think
it's
a
reasonable
assumption.
If
you
want
a
secure
health
check,
then
you
would
set
that
annotation
to
either
HTTPS
or
SSL.
A
Is
it
the
health
check
that
we're
doing
or
are
we
setting
the
back
end
protocol
and
keying
off
that
I
keep
staring
at
this
PR
I,
just
can't
figure
it
out
whether
it
looks
right
and
then
I'm,
like
oh
I,
don't
know
I
feel
like?
Should
we
ask
for
tests
of
the
like
test
just
so
we
can
like
look
at
the
permutations.
C
I
mean
yeah,
so
it's
I
mean
it
is
setting
the
like.
You
look
at
the
it's
sets,
so
it
sets
the
health
check
port
right.
So
it's
it's
selecting
it's
selecting
a
secure
port
based
on
so
first
you
have
to
be.
You
have
to
have
a
secure
back
in
protocol
to
get
into
that
block,
which
is
the
annotation
and
then
it
loops
through
the
listeners
that
were
created-
and
it
looks
at
so
that
variable
protocol
is
the
sort
of
the
front
end
there's
like
in
that
listener.
There's
like
a
protocol
and
an
instance
protocol.
C
So
it
looks
at
the
protocol,
which
is
I,
guess
I,
call
the
front
end
protocol
of
the
live
answer
now
those
were
created
elsewhere
in
the
code
and
the
build
listener
function.
So
we'd
have
to
look
at
that
for
the
full
picture,
but
but
what
it's
basically
doing
is
if
our
annotation
is
secure.
We
loop
through
the
listeners,
try
to
select,
try
to
select
ace,
secure
listener
to
set
our
health
check
port.
C
A
C
A
A
A
D
So
I'm
working
on
upgrading,
Authenticator
in
cops
and
I'm
running
into
issues
about
testing
it
zero-five-zero
introduced
a
new
back-end
mode
flag
that
allows
getting
the
mapping
of
Arne's
to
our
back
entities
getting
that
information,
not
just
from
a
file.
That's
only
mounted
but
also
a
config
map
or
a
custom
resource
and
there's
a
new
backend
mode
flag
that
allows
you
to
configure
which
back-end
you
want
to
use
and
it
can
be
an
ordered
list.
D
And
so
in
order
to
preserve
the
behavior
from
the
old
version,
you
need
to
add
the
backend
mode
flag
to
override
it
to
only
look
for
the
file
that
was
volley
mounted.
So
one
might
consider
this
a
breaking
change
and
so
I'm
wondering
if
we
can
fix
this
so
that
all
three
can
work
just
fine
without
having
to
specify
back-end
mode,
ideally
or
maybe
there's
some
other
way
that
we
can
make
this
more
backwards
compatible
with
zero
for
zero
yeah.
C
D
C
D
D
C
D
A
Okay,
just
like
a
random
thought
that,
like
yeah,
if
it's,
if
it
is
the
CD
incompatibility,
then
presumably
it
was
tested
at
some
point.
It
works
on
some
version
of
kubernetes
and
maybe
then
like
that
would
be
the
only
problem
like
fingers
crossed.
But
yes,
okay,
thank
you
for
bringing
it
up.
Ie
do
not
see
any
other
items
on
our
agenda
or
we've
reached
the
end
of
our
agenda.
I,
don't
know
if
anyone
else
has
any
other
items
that
they
would
like
to
discuss.
D
Actually,
I
have
a
one
that
just
popped
into
my
head
for
the
last
two
weeks
ago,
we
are
talking
about
the
AWS
accounts
used
for
the
end-to-end
tests.
I
know
that
we've
ran
into
bugs
well,
the
cops
project
has
ran
into
bugs
related
to
China
and
gov
cloud
partitions
and
I'm
vaguely
curious.
If
we
could
have
ete
support
in
those
partitions,
if
we
could
be
able
to
run
tests
there
and
what
would
be
involved
with
doing
that.
A
C
C
A
A
A
A
If
not,
we
will
call
the
meeting
to
a
close
going
once
going
twice
going
three
times.
Thank
you,
everyone,
and
we
will
see
everyone
in
two
weeks,
not
at
coop
Khan
and
trying
to
think
whether
that's
correct,
but
we're
going
to
really
know
much
for
weeks.
But
yes,
everyone
in
I
got
my
dates
wrong.
Don't
worry,
I'll
see
everyone
in
two
weeks,
yeah
happy
weekend
all
right.