►
From YouTube: Kubernetes SIG Multicluster 20180522
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
A
A
C
A
Anyone
here
from
Google
is
it's
Christian
planning
to
join
Christian
can't
make
it
today.
This
is
Ian
at
Google,
oh
yeah,
now
you
I,
don't
think.
We've
met
before
you
Christians
new
manager.
Yes,
I
heard
rumors
of
such
a
thing
good
to
meet
you
nice
to
meet
you
as
well.
Let
me
pull
up
the
notes
and
we
can
see
if
we
can
get
started.
D
Yes,
guess
so,
but
last
couple
of
meetings
we,
the
historically,
did
not
have
much
on
the
agenda.
At
least
last
four
meetings
have
been
like
that,
so
I
was
expecting
that
there
might
be
some
update
regarding
cluster
registry
today
and
because
I
saw
some
mails
and
by
Jonathan
about
one
of
the
latest
releases.
A
Okay
sounds
good:
yeah
I
was
going
to
suggest
we've.
We've
had
quite
a
few
companies
coming
to
me
in
the
past
few
weeks
wanting
to
contribute
here,
Cisco
CERN,
who
else
was
there
there
been
a
few
and
so
I
thought,
be
useful
in
the
absence
of
anything
else,
for
us
just
to
have
a
detailed
update
on
the
three
major
projects.
D
A
D
I,
remember:
I
am
putting
that
in
the
agenda
now.
I
remember
one
more
point:
I
am
sure
she
had
the
discussion
about,
so
she
was
going
to
start
analyzing,
the
implementation
of
federated
ingress
controller
or
some
mechanism
which
can
benefit
the
ingress
controllers
in
kubernetes
clusters.
B
A
F
A
D
C
A
A
D
An
update,
I
meant
if
it
is
like
an
agenda
in
advance,
it
might
let
people
there
might
be
people
who
are
just
observing
the
meeting
notes.
So
if
we
put
an
agenda
in
advance
for
say
next
meeting,
then
more
folks
might
join
in
to
get
that
thing.
Yeah
yeah!
Let's
have
project
updates.
Next
time
a
little
bit
of
yeah
I
can
I
can
right
away.
Put
I
mean
immediately
after
this
meeting
put
an
agenda
for
the
next
meeting
as
a
project
updates
on
the
three
for
the
benefit
of
contributors.
A
D
D
D
So
Federation
v2
there
are
two
I
mean
it
started
off
the
two
main
efforts.
One
was
the
lower
layer
if
you're
the
base
API
that
we
considered
the
base
API
objects,
which
ideally
should
be
sufficient
for
high-level
functions
or
any
kind
of
controllers
to
be
build
around
the
concept
of
rotation
of
editing,
multiple
clusters,
so
that
no
earlier
stuff
is
generally
has
big,
has
generally
been
taken
care
by
the
redhead,
guys,
Maru
and
I
won.
D
Currently,
the
three
resources
are
sort
of
API
wise
are
sort
of
complete
at
their
alpha
level,
post
that
the
latest
most
contribution
by
Meru
walls
federating
all
possibility
of
rotating
the
CRD
types.
Also,
so
you
can
name
it
Federation
of
CR
DS.
So
there
is
a
mechanism
that
is
now
possible
where
any
new
user
defined
CR
G's
can
also
be
federated.
D
But
for
all
of
these
types,
the
lower-level
lawyer
mechanism
would
be
sink
alone.
So
if
if
resource
is
created
in
in
Federation,
using
these
lowly,
atas
they'll
simply
be
sink
and
or
placed
in
two
different
questions,
so
the
current
state
is
up
till
here.
The
next
step,
I
guess,
is
moving
most
of
this
controller
logic
and
the
API
definitions
to
actual
CR
DS,
so
that
the
need
of
separate
API,
server
or
aggregation
of
get
case
ever
also
goes
away
about.
D
The
higher
level
API
is
I
am
into
the
causes
of
of
implementing
the
scheduling
types
which
include
the
department's
and
the
replicas.
Six
shashi
has
completed
one
round
of
or
one
level
of
implementation
of
this
service
discovery
which
includes
DNS
records
creation
include
an
external
DNS
server,
so
he
has
also.
He
also
has
an
Elfa
implementation
of
the
same,
using
some
mechanisms
which
is
sort
of
not
probably
the
final
go-to
mechanism,
but
it's
like
a
stopgap
as
of
now.
D
A
D
So
when
I
mentioned,
when
I
was
updating
about
the
latest
contributions
by
Moreau,
so
that
this
exact
implementation
of
propagation
also
has
been
done
so
earlier
before
this,
it
was
something
like
for
all
the
for
any
API
type
that
you
define
in
Federation.
For
example,
you
say
you
want
to
federate
a
secret.
You
define
three
types
of
federated
template
placement
and
override,
and
there
was
a
controller
written.
D
So
if
you
have
a
API
a
type
installed
in
the
API
server,
it
would
anyways
run
the
sync
loop
and
based
on
placement
and
overwrites
will
place
the
particular
resource
with
some
modifications
or
without
modifications
into
individual
clusters.
Right
now
he
has
implemented
another
resource
which,
as
far
as
I
remember,
the
name
of
that
is
I
even
can
correct.
Me
is
I
I,
don't
exactly
remember
the
name
of
that
right
now,
but
that
resource
can
using
that
resource.
D
C
C
C
The
federated
config
allows
you
to
basically
specify
what
type
of
resource
you're
wanting
to
federate
and
in
it
you
can
specify
whether
to
enable
or
disable
propagation
and
the
existing
Federation
v2
code
base
does
have
a
push.
Propagator,
that's
that's
implemented
as
part
of
the
repo,
but
certainly
that
could
be
expanded
to
include
other
types
of
propagators.
D
E
D
A
C
A
A
A
B
F
This
was
a
this
was
I
think
a
very
valuable
transition
effort
because
it
reduces
the
amount
of
code
in
that
repository
significantly,
which
should
make
the
maintenance
burden
going
forward
significantly
lower.
It
also
gave
an
opportunity
for
us
to
use
the
queue
builder
tool
which
hopefully,
will
make
continued
maintenance
of
the
repository
manager.
F
Regenerating
client
code
in
such
an
agenda
documentation
a
more
lightweight
process
rather
than
us
having
to
maintain
our
own
scripts
and
tools
for
doing
that.
I
also
added
meeting
spaces
into
the
cluster
registry
object,
which
was
I've,
been
decided
on
a
few
weeks
ago,
but
after
the
carotene
conversion,
I
made
the
relatively
small
change
necessary
to
add
namespaces.
At
this
point
there
are
still
a
few
smaller
tasks
outstanding
and
follow-up
item.
I.
E
F
Beta
I
think
at
this
point
there's
not
much
of
anything
blocking
it
one.
The
one
thing
that
I
might
want
to
do
is
there
has
been
a
bit
of
discussion
about
whether
we
could
add
a
simple
status.
A
simple
conditions
object
to
the
status
I
think
I
will
probably
post
a
PR
for
that
I've
been
thinking
about
thinking,
venemous
lied
about
controller
off
vs.
off
or
off
for
users,
interacting
with
the
cluster
registry
and
I,
make
the
public
a
change
to
the
API
to
try
to
better
demarcate
those
use
cases.
A
A
No,
okay,
well,
I'll,
give
a
quick
summary.
So
these
are
four
organizations
that
have
contacted
me
recently
and
wanted
to
get
involved
in
multi
cluster
in
general.
So-
and
some
of
you
probably
saw
the
presentation
at
Q
con
Europe
a
couple
of
weeks
ago,
and
they
are
very
keen
to
contribute
some
of
the
work
they've
done
to
open-source.
A
Then
Cisco
also
approached
me
recently
about
wanting
to
get
involved
and
they
have
I
gave
them
a
couple
of
options.
One
was
to
help
us
with
the
federated
ingress
stuff,
which
sounds
like
it's
pretty
overstuffed
at
the
moment,
but
independently
of
that,
the
other
option
I
suggested
to
them
was
to
build
a
multi
cluster
simulator
so
that
we
can
test
large
multi
cluster
deployments
in
various
different
guises,
which
they
have
taken
on
and
we'll
be
coming
to
this
group
soon,
I
think
with
some
designs
and
proposals.
A
The
general
idea
is,
you
know
it
should
be
easy
to
it's
pretty
difficult
at
the
moment
to,
for
example,
set
up
a
you
know,
hundred
cluster
Federation
so
having
something
similar
to
what
did
we
call
the
thing
for
scalability
testing
in
kubernetes?
It
has
a
name
that
is
eluding
me
for
now,
but
we
similarly
built
simulator
to
you
know
build
pseudo
clusters
of
5,000
nodes.
We
would
similarly
like
to
build
sort
of
pseudo
Federation's
of
maybe
a
hundred
clusters,
so
they
will
be
working
on
that
github.
A
Just
restarting
you
this
morning,
I
actually
suggested
that
they
join
us,
but
I
guess
maybe
they
didn't
make
it
and
they
also
want
to
get
involved.
They
have
some
internal
they're
actually
using
multi
cluster
and
I
believe
Federation
internally
in
github,
and
so
they
want
to
kind
of
just
get
a
little
closer
to
this
group
and
figure
out
where
they
can
help
them.
What
work
they
can
do.
I
will
follow
up
with
them
outside
of
this
and
get
some
more
details.
A
A
At
this
point,
Jonathan
correct
me:
if
I'm
wrong,
it
doesn't
seem
like
there's
a
large
amount
of
new
development
to
be
done
there
and
so
I'm
on
that
basis,
when
people
step
up
to
help
I'm
sort
of
telling
them
that
they're
probably
more
valuable
on
the
generation
v2
stuff,
which
has
a
bigger
backlog
of
development
work.
But
if
that's
not
accurate,
please
do
let
me
know
if
you
want
more
people
to
help
you
out
on
the
cluster
industry.
That
can
certainly
be
done,
and
the
same
goes
for
cube.
A
Mci
I
was
under
the
impression,
certainly
that
the
non
Federation
version
of
was
mostly
done,
and
that
doesn't
seem.
That
is
a
big
backlog
of
work
there
and
you
can
certainly
figure
out
what
to
do
with
the
federated
version
of
multi
cluster
ingress,
which
seems
like
it
has
a
lot
of
people
interested
in
working
on.
E
This
is
Ian
I.
Think
for
your.
Your
assessment
of
the
amount
of
development
required
right
now
is
correct,
probably
for
the
multi
cluster
ingress,
as
well
as
the
cluster
registry.
I
think,
for
maybe
the
new
members
would
be
good
to
understand
their
scenarios,
but
also
to
encourage
them
to
leverage
things
like
the
clusters.
Registry
makes
sense,
so
we
can
get
more
feedback
from
who
are
actually
using
it.
Yeah.
A
Yeah
I
have
definitely
been
doing
that
and
that's
been
a
common
thread
through
all
of
the
discussions.
I've
had
most
people
seem,
you
know
to
need
a
cluster
registry
and
they
want
to
use
the
one
that
Jonathan
is
built,
so
yeah,
absolutely
that's
kind
of
taken
as
a
given,
but
but
it
doesn't
seem
like
there's
a
lot
of
actual
developers.
Work
to
be
done
there.
A
A
So
in
fun,
if
you
could
just
make
sure
that
the
relevant
people
have,
if
we
do
need
to
do
more
updates
next
time,
it
seems
like
we
covered
most
of
the
ground
except
cube
mci.
So
we
can
yeah.
We
can
put
cube
mci
on
the
agenda
for
status,
update
next
time
and
then
I
also
reach
out
to
these
new
contributors
and
get
them
to
tell
us
what
they're
up
to
cool.
F
F
F
Do
not
under
a
the
problem
is
when
I
attempt
to
upload
the
videos
youtube
tells
me
this
video
cannot
be
processed
and
provides
no
more
information
as
to
what
could
be
causing
it
to
not
be
processed
I
haven't
yet
dug
into
the
source
code.
Two
attempts
to
track
down
the
problem,
I
think
I
probably
have
more
important
tasks
on
my
plate.
I
was
having
I
was
getting
some
help
from
sig
contribute
from
the
from
sort
of
the
kubernetes
admins
with
this,
but
I
don't
think
I
can
continue
to
just
ask
them
to
upload
the
videos.