►
From YouTube: SIG Cloud Provider 2019-11-13
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).
B
B
Okay,
on
the
GCP
side,
we've
been
concentrating
on
getting
the
network
proxy
stuff
working
for
our
managed
kubernetes.
The
good
news
there
for
anyone
who's
interested
in
using
the
network
proxy
is
that
we've
we've
confirmed
that,
in
fact,
it
works
just
fine,
unmanaged
kubernetes,
so
we're
able
to
use
it
as
a
complete
replacement
for
the
SSH
tunnel,
which
is
great
news
for
our
ability
to
actually
remove
the
cloud
provider
from
the
kubernetes
api
server,
so
that
that
is
definitely
good
news.
Anyone
on
the
IBM
side,
yeah.
D
B
B
A
A
This
was
mainly
asked
from
folks
at
Google
and
Amazon,
and
so
the
initial
prototype
essentially
just
update
the
cue
controller
manager
in
the
cloud
controller
manager
to
read
a
config
file
which
determines
which
then
specifies
a
list
of
controllers
that
you
want
to
put
in
a
shared
leader,
election
resource,
so
we're
still
designing
the
cap
and
we're
still
we
soonly
get
approval
from
api
machinery,
but
this
is
the
prototype.
So
if
you
want
to
look
at
broadly
just
want
to
get
an
idea
of
what
we're
proposing
and
what
changes
are
gonna
go
into.
A
Think
118
would
be
ideal.
Yeah
yeah,
it's
gonna
depend
on
so
I'd
like
to
get
this
kept
like
updated
as
soon
as
possible
and
then
from
there
I'll
depend
on
the
bandwidth
of
the
API.
Machinery
leads
so
yeah,
but
you're
otherwise,
I'm
not
sure.
If
they're
all
I
mean
I'm
hoping
it
won't
be
blocked
because
we're
gonna
be
try
gate
it.
We're
gonna,
also
feature
gate
it,
but
yeah
I,
don't
know,
I
know
that
the
other
election
is
kind
of
a
contentious
topic
because
of
performance
reasons.
D
B
B
B
Alright
did
anyone
have
anything
else
they
wanted
to
add
for
any
of
the
providers
and
you
providers.
We
didn't
cover
all
right
on
the
agenda.
First,
the
fun
items
we've
got
Keuka
next
week.
I
would
like
to
take
folks
out
to
dinner.
So
please
pingyuan
slack.
If
you're
available,
I
was
thinking,
Thursday
night's,
probably
gonna,
be
the
most
or
the
least
busy.
But
let
me
know
if
other
nights
work
better
and
then
just
ping
me
and
then
I'll,
try
and
book
us
somewhere
to
have
dinner
so
I'm
good.
A
A
So
last
week
we
merged
two
pr's
that
bumps
the
that
promotes
the
promotes
three
labels
that
are
added
by
the
club
provider
from
beta
to
GA.
So
the
biggest
problem
with
this
is
that
we
marked
labels
as
beta
by
actually
prefixing
the
word
beta
in
it,
and
that
requires
actually
adding
a
new
label
when
we
promote
right
so
starting
in
117,
we're
gonna
actually
add
both
the
beta
labels
for
zone
region
and
instance
type
and
the
GA
labels
for
zone
region
instance
type
2
nodes
and
then
we're
gonna.
A
We're
gonna,
keep
supporting
the
beta
labels
up
to
121
and
then
we're
gonna
remove
support
for
the
beta
labels
and
1.21.
So
this
is
just
kind
of
a
PSA
you're
gonna
start
what,
when,
when
you
have
users
on
117
you're
gonna
start
seeing
instead
of
the
failure
domain,
beta,
kubernetes,
io,
/
known
it's
gonna,
be
topologies
of
community
saw
io
/,
none
so,
and
we
should
be
encouraging
users
to
use
the
G
labels.
I,
think
that's
it.
Oh
yeah
one
other
thing
we
have
lodge,
we
added
control
loops
in
the
cubelet
and
the
cloud.
A
The
cloud
note
controller
to
also
backfill
the
G
labels
on
an
existing
node
that
gets
upgraded,
so
yeah
it'll,
it'll
it'll
be
applied
for
both
like
new
and
existing
notes,
so
it
should
be
a
pretty
clean
upgrade.
The
only
the
own.
The
downside
is
like
duplicate
labels,
but
not
much.
We
can
do
about
that.