►
From YouTube: Kubernetes Federation WG sync 20181017
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
B
D
Tell
me
that
I
just
post
on
of
questions
with
the
Middle
East
and
the
question
that,
because
we
are
going
to
release
edition
old
who
so
as
so
I'm
just
wondering
I,
think
that
women
need
to
finalize
some
concept
like
with
his
hope,
host
cluster,
which
is
member
Craster
and
also
with
his
Federation,
so
that
we
can
make
sure
that
people
can
talk
with
same
concepts.
So
I
hope
that
we
can
have
some
concluding
process.
D
And
after
we
get
some
agreement,
and
maybe
we
need
to
create
a
PR
to
update
the
rid
me
of
the
header
image
document
and
that
and
for
now,
I
think
that
we
I
got
some
comment
from
a
row
from
from
Paul.
I.
Think
that
we
agree
that
we
use
the
concept
upon
host
cluster,
which
is
running
the
phoenicians
controller.
And
we
also
want
to
use
member
cluster
to
add
the
two
to
just.
And
we
also
want
to
use
member
cluster
to
identify
the
clan
sir,
which
is
running
pessimal
workload
and
I.
E
D
E
Of
imagine
that
we'll
have
more
terms
than
we
want
to
put
into
right
into
the
readme,
so
it
will
be
helpful
to
just
collect
them
all
in
like
a
single
place
in
the
docs
folder
to
so
that,
once
that
list
gets
larger
than
we
want
to
like
have
present
right
in
the
readme
that
there's
a
place
for
things
to
go.
Yeah.
C
C
E
C
E
C
C
H
G
G
Is
simply
distributing
my
administrative
context
service
account
that
kind
of
thing
so
to
me,
I've
always
thought
of
the
Federation
is
the
group
of
clusters
for
which
configuration
is
applied
so
those
they
coordinated
fashion,
not
very
catchy
but
I.
Think
that's
technically
what
it
is.
It's
baseline.
B
We
want
stamp
when
we
were
walking
with
Edition
women.
One
small
definition
is
the
trust
and
gently-used
was
a
tool
or
of
a
control
train
used
to
run
an
application
into
multiple
clusters,
a
kubernetes
application
into
monthly
freshness,
so
to
me,
I
think
condition
might
still
be
similar
or
same
indeed,
and
can
be
defined
as
tools
which
can
be
used
to
running.
Ok,
this
application
into
multiple
projects.
Does
that
seem
happy?
In
fact,
if
you
search
for
hydration
docks,
there
are
traditions
of
still
existing
in
kubernetes
stocks,
kubernetes
at
I/o,
so.
B
So
if
you
go
to
those
documentation,
I
couldn't
find
a
page
which
basically
lists
that
this
is
Federation
or
definition
of
addition
and
this.
Why
Federation?
What
is
the
need
of
it
and
that
kind
of
stuff?
But
to
me
the
app
definition
is
a
tool
which
is
used
to
run
kubernetes
applications
into
multiple
clusters.
As
you
were
saying,
the
other
one
is
not
necessary
to
get
you
listing.
G
And
I'm
just
not
sure
that
we
have
to
focus
on
applications
just
as
easily.
You
know
doing
non
workload,
configurations
and
clusters,
so
tavini
configuration
is
kind
of
a
catch-all
and
configuration
includes
configuration
and
workloads,
but
maybe
there
needs
to
be
a
better
catch.
All
term
configuration
yes.
B
F
E
E
H
D
H
E
B
Okay,
moving
ahead
so
yeah
we
have
been
talking
about
this
0
dot,
0
dot,
to
release
I
think
it's
time
we
probably
cut
out
one.
So
there
is
only
one
PR
that
budgeted
jobs
or
job
scheduling
references
we
are,
which
is
outstanding,
which
was
back
to
this
milestone,
so
I
I
mean
I,
did
fix
the
review
comments
and
all
that
stuff
and
some
items
I
have
kept
for
follow-up
years
so
to
the
viewers.
B
This
is
a
shout
out
that
if
you
find
that
it's
okay
and
the
fixes
are
okay,
we
can
go
ahead
and
merge
that
and
this
week
one
of
us
can
then
cut
the
leads.
I
won't
be
available
office
anymore,
so
it
will
need
to
be
done
by
somebody
else
and
if
that
doesn't
seem
to
be
fit
for
March,
we
can
probably
remove
it
from
the
milestone
and
then
go
ahead
to
the
Reeves
I
I.
Don't
find
it
logical
to
delay
that
release
further
yeah.
E
I
I'm
glad
you
put
that
on
the
on
the
agenda
for
today,
because
I
realized
last
night
that
we
talked
about
the
intention
to
start
releasing
regularly
and
then
we
didn't
do
it
yet.
So
I
would
also
like
to
do
the
Oh
Toyota
to
release
later
this
week,
Maru
I
think
you
indicated
you
had
some
interest
in
doing
that.
Do
you
want
to
take
that
as
an
action
item?
E
E
I
One
thing
to
know
tomorrow
or
to
whoever
does
the
release
is:
if
we
use
the
github
tagging
technique,
it
does
not
create
an
annotated
tag,
so
the
only
so
the
way
you'd
get
an
annotated
tag
is,
if
you
create
a
tag
in
your
own
repo
and
then
push
upstream,
that's
something
to
keep
in
mind.
We
have
some
commands
in
the
make
file
that
create
a
version
based
on
having
annotated
tags,
not
just
simple
tags.
D
E
We
I
think
still
need
to
get
the
chart
published
into
a
repo,
so
I
think
it's
I
think
it's
okay
to
do
a
release
of
the
binaries
and
the
images,
and
then
we
can
bump
them,
probably
for
now
bump
the
version
of
the
chart
in
the
like
in
the
Federation
repository,
but
Guanyin
I
think
I.
Think
that
you
and
I
had
an
action
item
there.
We
that
we
that
we
dropped
that
we
should
take
backup
to
figure
out
where
what
chart
repository,
we're
gonna,
publish
this
yeah.
E
I
B
I
think
the
activity
of
publishing
a
chart
can
be
post-release
also.
So
when
we
cut
out
a
release,
the
image
and
the
binaries
will
be
created
that
you
friend,
binary
on
the
controller
manager,
image
and
relevant
Yama's
or
the
conflicts
will
also
be
created
which
will
be
available
in
our
current
report.
B
E
B
B
E
D
D
E
Upon
and
I
have
had
a
running
thing
for
a
while
that
I've
been
kind
of
bandwidth,
starved
to
that
that
should
be
changing
soon,
to
have
a
kubernetes
blog
post
on
Federation,
very
fun.
I'm
gonna
have
a
little
bit
more
bandwidth
for
that,
so
maybe
you
and
I
can
set
up
some
structured
time
to
get
a
next
draft
of
that
together.
J
E
Before
we
did
that,
I
wanted
to
take
another
pass
through
it
and
having
had
the
experience
recently
of
doing
a
lot
of
writing
of
prose
about
this,
like
now,
it's
a
little
easier
for
me
to
put
this
stuff
that
we
talk
about
all
the
time
into
actual
written
words
and
just
take
another
pass
through
that.
So
maybe
I'll
take
that
as
had
something
to
do
by
Friday
in
my
time,
and
we
can
share
for
a
review
for
the
wider
group
later
or
early
next
week.
Does
that
work
for
folks.
H
B
E
So
so
Han
created
a
test
repository
and
to
test
that,
like
the
the
pluralization
and
naming
was
handled,
the
way
we
wanted
to
it
looks
like
there
might
be
a
bug
in
queue
builder,
where
the
queue
builder
resource
path,
annotation,
they
that
is
supposed
to
control
the
plural
name
of
the
CD
might
busted
I
actually
before
this
meeting
was,
was
updating
my
own
deal
builder
to
to
try
to
see
if
I
could
recreate
it.
But
if
there's
a
bug,
I
think
we
ought
to
fix
it,
because
we
use
that.
E
You
know
we
use
it
in
this
project
and
you
know
we
want.
We
want
to
make
sure
that
gets
fixed,
I
guess.
One
thing
that
perhaps
we
could
do
if
that
pluralization
is
not
going
to
work
is
what
if
we
just
change
the
name
since
the
the
resource
is
already
in
an
API
group
that
identifies
it
as
multi
cluster
DNS.
B
J
E
So
I've
talked
to
Phil
with
Roc
about
adding
that
stuff
back
in
I.
Actually
have
it
partially
done?
I
just
need
to
wrap
it
up
to
add
the
API
back
in
to
fix
that
first
and
then
we
need
to
establish
for
sure
whether
there's
a
bug
or
not
so
Hans
asked
a
couple
folks
on
slack,
but
I
don't
think
anyone's
responded
yet.
G
B
B
In
the
future
for
Federation
I
think
I
just
wanted
it
that
out
to
one
of
the
number
there
and
I
think
we
were
able
to
fix
that
so
recent
coordinates
just
released.
A
study
seems
to
fixed
on
so
overall
this
feature.
If
you
want
it
work,
as
we
initially
designed
I
think
it
requires
that
particular
coordinates
release
but
I'm,
not
sure
all
we
can
limit
or
notify
the
users
like.
If
you
want
this
feature,
you
need
this
for
DNS
I'm,
not
sure
that
is
not.
It
landed
on
you,
kubernetes
release,
so.
E
That's
that's
a
really
interesting
question
that
is
probably
like
the
first
of
potentially
many
dependencies
that
will
want
to
be
like
really
crisp
about
what
the
required
versions
are.
I
have
no
good
ideas
of
how
to
do
that,
but
I
think
it
will
be
important,
so
I
would
love
if
some
courageous
individual
might
have
some
good
ideas.
They
were
willing
to
share
about
how
to
communicate
that
I.
J
B
E
Does
anybody
happen
to
know
that
that
might
be
able
to
tell
us
about
it
now
exactly
how
kubernetes
itself
identifies
the
required
version
of
dependencies?
I
mean
I,
I'm
I
think
that
there
are
probably
still
places
where
there
are
dependencies
on
specific
tools
being
on
the
hosts,
etc.
That
are
are
probably
not
fully
documented,
but
I
I
wonder
if
anybody's
familiar
with,
like
the
overall,
for
example,.
B
D
Yeah
so,
based
on
my
understanding,
I
I
think
that
currently
communities
also
don't
have
that
kind
of
document.
But
currently
it's
men
and
depending
on
the
Mini
Cooper
I,
think
recording
the
Mini
Cooper.
There
are
some
different
components
and
also
given
the
warran
of
all
those
components
and
those
components
will
also
be
updated
before
every
release
of
chronitons.
B
Yeah
going
here,
but
in
a
normal
deployment
also,
there
are
mechanisms
to
deploy
kubernetes
using
cuba
again
or
some
of
the
tools,
so
they
also
pull
up
version
runtimes
or
containers
or
dependency
tools
together.
So
they
there
should
be
some
mechanism
of
figuring
out
say
for
a
given
company
leaves
what
other
dependent
dependencies
versions
to
Holden.
D
B
B
Was
I
was
trying
to
work
out
in
the
Google
DNS
enrollment
I
felt
like
every
feature
needed
a
cname
record,
and
but
there
is
a
workaround,
but
still
some
there
are
some
pros
and
cons
and
yeah.
Maybe
we
will
document
that
in
the
user
talk
so
I
think,
like
the
kind
of
enrollments
the
feature
needs
to
work
in
it's
pretty
much.
There
are
so
many
other
dinner.
It
really
depends
on
what
then
Roman
you
need,
and
mostly
it
was
designed
to
work
anywhere,
but
there
seems
to
be.
B
It
is
untested
area,
so
we
might
discover
some
of
this.
So
currently
there
is
work
around,
but
it
is
with
some
limited
teacher.
So
that's
probably
okay.
Oh
no
I
think
yeah.
It
also
might
take
some
long
time
to
fix
I
believe
this
is
the
external
DNS
issue.
I
am
talking
about
individual
cloudiness
in
the
default
concentration.
We
might
not
be
okay,.
B
Yeah
I
think
really
discover
issues
as
in
when
this
is
tested.
This
is
also
like.
This
also
puts
in
a
thought:
do
we
need
to
I
mean
if
we,
if
the
stakeholders
are
interested
in
the
service
discovery
feature,
do
we
need
to
introduce
the
tests
or
a
tweak
in
kind
of
test
which
can
verify
this
feature,
but
that
leaves
some
set
of
games
providers?
G
G
F
E
Yeah
this
this
is
something
that
I'm
I'm
kind
of
wondering
to
myself
about.
In
terms
of
you
know
just
using
DNS
as
an
example
with
external
DNS,
you
can
integrate
with
all
these
different
provider
api's.
It's
it's
like
natural
that
that
different
features
may
work
different.
You
know
well,
two
different
degrees
on
different
DNS
providers.
B
B
E
B
I
actually
had
to
talk,
but
more
like
them,
I
miss
putting
it
over
here
and
I,
see
that
it's
time
so
so,
as
part
of
the
review
of
that
my
Pia,
which
is
about
the
job
technique,
differences,
there
was
an
item
which
came
up
that
it
needs
multiple
override
fields
and
recently
at
at
our
integration
group.
Also,
we
came
across
this
same
requirement
that
the
users
of
foundation
basically
need
to
have
multiple
override
fields.
So
some
examples
are
like
a
user
which
deploys
a
service
or
propagates.
B
B
So
I
couldn't
come
up
with
a
really
generate
way,
because
earlier
she
did
try
that
and
it
had
its
own
problems
and
it
did
not
work
out
which
was
having
the
override
also
same
as
template,
and
then,
whatever
the
specified
in
that
override
template,
you
merge
it.
The
penalty
template
what
I
am
thinking
right
and
I'm,
putting
this
out
for
validation.
What
I
am
thinking
right
now
is
that
the
way
it
is
implemented
right
now
is
in
the
type
conflict.
B
It
is
the
type
type
and
picture
it
specifies,
a
single
path
which
basically
is
mapped
into
the
override
Phebe,
and
that
is
what
is
it
into
the
target
resource
so
in
place
of
having
a
single
path,
we
can
have
a
map
of
key
and
values.
A
key
basically
specifies
the
the
field
name
and
the
value
is
the
paths,
and
that
could
be
a
list
of
I
mean
the
list
of
key
value
appears
it's
not
absolutely
generic
way,
but
most
of
the
cases
might
actually
work
out
using
this.
So.
E
B
E
One
one
thing
that
I
was
remembering
about
this-
is
that
I
we
used
that
one
of
our
motivations
for
having
the
the
overrides
resource
be
separate
is
that
we
felt
like
the
initial
design
choice
that
we
made
to
kind
of
constrain
the
problem
space
a
little
bit
and
make
things
simpler
of
just
overriding.
A
single
field
was
something
that
we
were
likely
to
come
back
and
redress
because
you're
right
or
fun
there's
a
lot
of
there
are
a
lot
of
different
places
in
the
API
where
you
would
you're.
E
You
know,
potentially
want
to
use
different
annotations
in
certain
clusters
or
otherwise
differentiate.
Student
yields
and
I
seem
to
recall
that
we
kind
of
identified
it
like
the
other
end
of
the
complexity
and
flexibility
spectrum,
from
like
treatment
of
a
single
field
with
something
like
arbitrary
JSON
or
ml
patches,
or
something
like
what
the
customized
with
the
k-project
does,
where
you
can
specify
kind
of
sparse
ml
constructs.
G
K
B
B
K
I
misunderstood
you
to
me
in
a
path
to
a
different
to
a
different
API
object
and
I
was
a
bit
concerned
about
the
performance
there.
The
other
customers
with
angle
suggestion.
So
if
somebody's
needs
to
new
to
think
it's
okay
is,
is
it
worth
just
taking
on
a
standard
template
eyes,
templating
substitution
mechanism
to
at
least
at
least
drawing
inspiration
from
it
rather
than
inventing
our
own
one,
which
you
know
I?
K
E
My
own
preference
would
be
you
avoid
any
of
that
and
like
let's,
let's
be
really
crisp
as
we
look
at
the
design
for
this
feature,
which
I
think
is
necessary
to
have
like
for
just
in
terms
of
the
things
that
people
are
going
to
want
to
do.
Well,
let's,
let's
make
sure
that
we
avoid
any
temptation
to
put
a
template
language
into
this,
like
the
I.
E
Still
remember,
the
great
template
language
Wars
of
2015
and
kubernetes
I
think
it
was
that
kind
of
wound
up
with
no
official
adopt
adopted
template
spec,
although
there
are
there,
are
places
in
the
API
where
there
are
kind
of
very
basic
substitution
mechanisms.
My
own
preference
would
be,
let's
not
have
any
of
that
for
now
and
and
I
think
that
as
it
comes
as
people
may
want
it
to
figure
out
what
to
do
or
if
there's
a
way,
they
can
use
tools.
On
top
of
Federation
to
accomplish.
K
Yeah,
my
suggestion
was
actually
something
slightly
different
than
what
you're
saying
I
think,
which
is
so
first
point.
Let's
avoid
developing
a
proprietary
templating
language
absolutely
agree
with
that
part
I'm,
not
as
in
favor
of
building
the
simplest
thing
to
start
with,
because
I
just
don't
think
we're
very
good
at
deciding
what
the
simplest
useful
thing
is,
and
we've
got
a
bunch
of
simple
things:
we've
built
that
that
have
been
too
simple
and
Mike
useful.
K
And-
and
so
I
don't
know
of
one,
but
I
would
strongly
suspect
that
there
is
a
reasonably
good,
suitable,
templating
language
for
llamó
or
whatever.
It
is
that
we
need
at
the
moment
and
that
we
can
therefore,
perhaps
so
I
would
like
us
to
support
the
simplest
possible
case,
like
the
one
we
have
today
with
a
single
value.
I
would
like
us
to
support
what
seemed
like
incrementally
the
next
simplest
one,
which
is
such
a
template
with
two
values.
K
You
know
dollar
Bell,
one
and
dollar
bail,
two
kind
of
thing,
but
that
can
equally
easily
be
extended
to
things
like
expressions
etc,
and
can
we
not
just
adopt
an
existing
thing,
not
necessarily
a
kubernetes
thing,
but
just
a
standard
generic
template
in
solution
and
say
stick
the
config
for
that
in
a
config
map,
and
we
will
apply
it
to
your
thing
and
in
the
simplest
case,
is
you
know,
key
1
equals
value.
1
key
2
equals
value
2,
but
you
can
do
all
those
other
stuff
as
well
and
and
see
where
that
breaks.
B
I'm
done
yeah,
agree
Quentin,
so
that's
what
I
am
Not
sure
if
you
did
catch
that
or
you
missed
it,
that's
what
we
were
actually
talking
about
earlier.
So
a
really
generic
solution
is
is
desirable,
but
we
might
as
well
be
implementing
so
I'll
look
into
the
possibilities
of
templating
solution.
Also,
if
I
mean
we
find
issues
that
we
did
not
find
anything
centric
than
the
one
that
we
talked
about
using
the
key
value.
Clear
is
what
I
mean.