►
From YouTube: Kubernetes SIG Multicluster 20180619
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
B
B
A
Okay,
so,
as
I
said,
there
is
nothing
specific
on
today's
event
also
last
week
Nicola.
Actually
we
had
some
some
updates
with
respect
to
each
of
the
projects,
like
the
three
sub
sub
projects
that
we
are
running
and
he
missed
you
with
respect
to
Q
Gong
Zi
update,
so
it
was
benefit
for
new
developers.
I,
don't
see
as
Buddha
presents
today.
B
As
an
update
for
Cuban
CIE,
you
cently
hide
point
Foley's,
where
we
added
a
bunch
of
features.
It's
some
improvements
to.
You
can
remove
clusters
from
an
existing
building
else,
train
whistle,
fingers
added
a
few
flags
and
improved
some
documentation
and
added
some
examples
as
well,
and
in
case
you
saw,
there
was
also
a
big
block
host
on
GCP
that
it's
now
cloud
supported
product.
So
that
is
awesome
as
we
are
having
them.
A
B
B
So,
as
we
said
during
our
cube
con
demo
as
well
like,
we
have
a
prototype
working,
there's
still
some
design
challenges
we're
working
through.
Yes,
we
are
exploring
that
and
see
how
that
goes,
but
in
general,
like
the
module
of
building
controllers
with
less
religiously,
is
exciting
and
like
multi-gesture
English
as
an
example,
but
we'd
love
to
see
lot
more
and
I
know
like
Lindsey
was
also
working
on
one
example
for
exporting
cube
conflict
for
all
the
clusters
that
was
exciting
and
even
in
cube
con.
B
D
B
B
A
I
know
that
she
was
exploring
so
this
ingress
problem.
We
want
to
be
fun
to
explore
solving
the
Angus
problem
in
foundation
v2
also,
so
such
he
was
flying
text
or
figuring
out
the
genetic
solution,
but
so
far
he
has
not
been
able
to
converge
to
a
solution
which
can
work
for
all
the
providers,
because
mainly
because
there
are
specific
features
which
are
specific
to
talk
providers
and
all
you.
A
Possible
one
of
the
possibilities
was
that
having
an
active
controller
which
can
use
or
leverage
cube,
emceeing,
either
as
a
library
or
an
active
runtime
which,
which
can
just
be
the
most
of
the
meat
of
the
controller
and
the
controller,
just
leaves
exactly
someday,
but
that
basically
works
only
for
TCP.
So
so
we
are,
we
are
I
mean
with
respect
to
invest
there
isn't
anything
concluded
the
in
traditional
v2
it'd
be
great.
If
you
can,
you
know
if
some
sedition,
yeah.
B
Yes,
there's
definitely
interest
in
multi
class,
dangerous
outside
GCP,
as
well
as
a
first
implementation.
We
did
it
with
GCC
because,
like
GCP
has
that
capability,
global
anycast,
IP
and
other
cloud
providers.
Don't
have
that.
So
that's
why
it's
possible
with
GCB,
and
we
want
to
extend
it
to
others
as
well,
and
we
see
as
we
get
like
customer
demand
and
as
other
clouds
also
support
that
use
case,
we'd
love
to
add
it
and
the
controller
that
we
are
exploring
like
it
about
witness
registry.
B
So
it
can
be
packaged
as
part
of
the
Federation
as
well.
It
can
be
an
independent
controller
that
anyone
can
adjust
that
controller
location.
So,
yes,
you
want
to
solve
that
use
case,
and
that
controller
should
do
that
and
we
are
also
talking
with
signet
work,
buoy
and
him
making.
Sure
lady
today
is
restrictive,
that
you
cannot
like
it's
a
single
cluster
English
again,
there's
no
way
you
can
specify
multiple
clusters,
they
do
cluster
selection
or
percentage
traffic.
Splitting
there's
a
lot
of
fancy
things
you
can
do
with
particular
standing
this
and
the
API.
B
D
A
D
So
I
don't
believe
there
was
been
that
much
activity
in
the
last
two
weeks,
I'm
trying
to
remember
if
anything
happened
after
the
last
sick
meeting,
I
think
there
were
some
small
changes
but
I
think
at
this
point
mostly
will
I
made
some
small
changes
to
the
API
and
then
letting
them
sit
for
a
little
while,
hopefully,
within
the
next
few
weeks,
if
they're
no
complaints,
I
will
propose
a
push
up,
PR
to
move
the
API
to
beta
other
than
that
I.
Don't
think
there
are
any
significant
project
updates.
D
The
move
to
CRT
seems
to
have
gone
reasonably
well,
but
I
haven't
heard
any
integrators
complaining
or
people
saying
that
it
is
and
difficult.
I
know
that
on
the
Federation
side
there
has
been
some
work
being
done
to
work
on
this
with
the
CRT.
It's
the
V
API
server,
which
I
believe
has
taken
some
time,
but
I'm
not
familiar
with
that
work.
C
I'm
still
in
process
of
trying
to
transition
to
cube
builder
I'm
having
a
little
bit
of
trouble,
because
the
CRD
support
for
sub
resources
is
pretty
immature
and
110b
a
little
bit
better
in
in
111,
but
you're
trying
to
figure
out
how
to
how
to
debug
problems.
They're,
probably
very
few
people
have
seen
because
nobody's
really
using
it.
C
Yet
I
have
mostly
working
it's
just
that
we
do
use
status,
which
is
a
sub
resource
for
a
recording
cluster
state,
which
is
pretty
much
foundational
to
everything
we
do
and
currently
I
actually
can't
create
c-d-c
RDS
with
the
sub
resource
with
the
CR
DS.
The
queue
builder
is
generating
and
I'm
trying
to
plug
Y
and.
C
D
C
A
Make
sense
yeah
with
respect
to
the
high
level
controllers.
I
did
finish
work
on
the
scheduler
types
that
was
merged,
also
I
guess
shushies
sachets
PR
is
waiting
for
review
on,
probably
maybe
an
NG
TM
from
iPhone
is
needed.
A
105,
either
Ivan
or
a
ball
hat
have
to
have
a
look
once
that
is
true.
Then
we
will
have
the
service
discovery
and
and
the
scheduler
types
would
be
working
I'm,
currently
working
on
sort
of
sporting
HPA
functionality
from
v1
into
one
of
the
high
level
cities.
A
The
types
here
so
I
should
ideally
be
able
to
do
it
in
a
week's
time.
They
have
more
details.
We
can
talk
in
tomorrow's
tomorrow's
partition,
rock
group,
sync
yeah
that
is
more
or
less
from
my
side
yeah
and
for
the
usage
of
folks
who
do
not
attend
the
petition
work
group
tomorrow
sure
she
might
be
giving
out
a
demo
on
the
cross
cluster
service
discovery
in
Federation
v2
and
how
we
have
achieved
that
so
anybody
who's
interested
can
join.
That
meeting
tomorrow
same
time.
A
So
one
of
the
targets
we
kept
for
for
v2
was
also
feature
parity
with
respect
to
whatever
features
we
were
able
to
achieve
in
v1.
So
this
one
is
targeted
towards
that.
But
the
usage
is
a
little
different
like
we
have
a
separate
DNS
API,
which
needs
to
means
that
needs
to
be
created,
a
resource
which
is
called
multi
cluster
DNS,
which
needs
to
be
created
by
certain
information
compiled
a
user
so
that
DNS
records
can
be
created
in
to
an
external
DNS
server.
The
rest
of
the
stuff
is
more
or
less
similar
to
p1.