►
From YouTube: Kubernetes SIG Cluster Lifecycle 20171121
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
B
So
basically,
it's
time
for
us
to
like
cubed
him,
I've
turned
the
and
we
still
have
the
same.
You
know
a
clueless
file
as
we
had
a
year
ago,
but
the
actual
reviewers
and
contributors
to
the
project
has
shifted
over
time.
I
should
be,
and
now
the
proposal
is
to
add
four
new
review
s1
year
approver
and
remove
three
reviewers
from
list.
Are
there
any
objections
to
beasts?
I've
checked
with
for
the
people
and
I
think
it
should
be
okay,
but
like
this
kind
of
mass
shift
requires
just
at
least
I
mention
in
there.
A
C
D
I
would
love
to
be
removed
as
well.
I
would
ask,
are
we?
Are
we
following
the
ladder
or
is
that
sort
of
you
know
because
I
mean
there
are
people
that
we
are
like?
I
was
certainly
not
technically
qualified
and
technically
team
you're,
not
quite
a
site
and
like
I,
haven't
checked
everyone
else,
but
I
was
like
you
know.
Those
are
the
ones
that
are
easy
to
like
do.
D
That
but
I
mean
like
the
the
the
the
technical
qualifications
are.
You
have
like
made
like
30
PRS
to
the
subtree
and
have
reviewed
ten
significant
like
things
and
like
been
there
for
18
years
of
experience
in
go
and
all
this
sort
of
stuff.
You
know
it's
like
I,
don't
know,
I'm
totally
happy
with
over
doing
here.
I'm,
just
like
it's
more
of
a
meta
question,
like
particularly
tim,
is
someone
on
the
steering
committee
of
whether
that
whether
we
are
still
observing
that
strictly
and
how
strictly
we're
doing
that,
I
don't.
A
Say
I
don't
think
that's
ever
been
observed
strictly,
I
think
there's
been
a
lot
of
grandfathering,
other
things
that
have
come
into
place
there.
There
has
been
comment
in
the
steering
committee
in
to
address
roles
and
responsibilities
of
SIG's,
but
that
is
like
priority
two
at
this
point
of
the
steering
committee.
The
most
important
one
is
still
the
incubation
issue
that
seems
to
be
ongoing
and
so
I.
It's
definitely
been
it's
on
the
radar,
but
folks
have
not
prioritized
it
in
priority
order.
A
A
So
it's
a
it's
a
huge
conundrum
that
exists
in
the
project
and
I
think
the
original
manifestation
or
creation
of
those
roles
was
through
Brian
and
it
wasn't
necessarily
done
by
a
broader
group,
so
I
think
it's
it's
subject
to
change
and
I.
Think
it's
subject
to
the
roles
and
responsibilities
of
the
sig.
B
Yeah
and
also
yeah,
and
when
we
added
cubed
M,
it
was
like
impossible
to
have
this
like
18
years
of
experience
with
the
project,
so
we
we
basically
just
added
black
six
seven
people
or
so
we're
we're
testament
as
well.
Was
there
and
things
like
that?
We
don't
still,
as
you
said,
like,
follow
the
exact
dates,
but
they
aren't
necessarily
as
they
are
necessarily
required
in
this
case.
It's
more
like
the
participation
and
involvement
level
and
Tim
has
Indian.
We
would
at
least
ten
significance
be
our
last
cycle.
B
A
I
know
on
the
agenda
items
the
next
one
we
have
is
unofficial,
meetup
in
Austin,
but
before
that
I
wanted
to
quickly
discuss
any
freeze
items
that
we
need
to
address
in
part,
because
freeze
is
coming
this
Wednesday.
Are
there
any
things
that
require
like
immediate
attention
or
review
within
arcing
or
if
we
can
help
somebody
else
that
requires
attention?
A
B
B
Which
of
them
should
be
mentioned
here
like,
for
example,
will
enable
certificate
rotation,
yes,
otherwise,
otherwise,
folks
API
serving
cert,
will
expire
before
they
have
a
chance
to
upgrade
to
cover
this
one
time.
This
is
kind
of
significant
and
Peter
has
been
kind
enough
to
send
a
PR
that's
going
to
be
merged.
B
One
issue,
though,
is
Sergey
yeah
one
issue
that
Sergey
absurd
was
that
a
TD
rollback
wasn't
necessarily
working
the
way
it
should
be
I'm
manually
checked
yet,
but
we
should
definitely
make
sure
this
time
that
3
3
2
3
1
back
2
3
0.
That
rollback
must
work
because
well
obvious
reasons.
If
we
claim
to
support
upgrades,
we
also
shouldn't
claim
to
support
some
rights,
and
this
time
it
involves
my
CD
version
change
as
well.
D
D
D
D
E
D
Far
as
I
send
that
via
this
would
be
more
of
a
disaster,
recovery
type
thing
that
is
more
independent
from
n
CD
and
the
core
requirement
is,
it
has
to
go
to
back
up
on
it
c2
and
restore
n
c3,
and
therefore,
presumably
it's
me
three
NS
any
to
right
other
than
that.
I.
Don't
think
it's
practical
to
do
a
lot
more,
like
I,
want
to
do
the
full
like
arc.
Tateo
zark
thing
right,
which
is
a
lot
lot
more
involved,
but
maybe
maybe
that's
the
only
maybe
that's
the
minimum
solution.
I.
A
D
A
B
So
well,
there
were
a
couple
of
related
to
qubit
and
yes,
but
I
have
none
in
tracking
so
should
be
under
control.
I!
Think,
for
example,
dynamic
qubit
configuration
landed
last
week,
just
cool
still
in
alpha,
though,
but
we
will
help
Mike
dolphins
to
push
this
to
be
that
well
by
like
having
having
Ely
tested
in
in
Q
by
damn
yes
for
background
information.
This
is
feature
that
makes
it
possible
to
configure
configure
all
the
qubits
in
your
cluster
using
a
base
config.
Not
so
you
can
basically
control.
B
You
can
make
information
of
the
state
of
the
beside
statement
just
flow
down
from
your
masters,
your
nodes,
which
is
the
thing
we
can't
now
right
now
we
have
the
bizarre
cubelet
cube.
Adm
droppin
in
a
system
becomes
file
in
the
beds
and
at
the
ends.
We
can
eventually
get
rid
of
this
as
we
store
things
in
component
version
of
API
inside
of
the
cluster
in
the
config
map
that
is
then
distributed.
So
could
a
witness,
so
Peter
has
been
a
great
help
here
and
and
recently
made
a
PR
to
do
so.
A
B
Yeah,
it's
not
going
to
be
that.
Hopefully
the
cubelets
API
include
based
on
the
Vita
I
really
hope
so,
but
that
is
also
uncertain,
but
but
the
cubed
dynamic
compression
still
has
some
tweaks
to
be
done
in
order
to
work.
Hopefully
the
cubit
API
itself
will
be
betta.
That's
the
plan,
okay,
which
I
guess
you're
thinking
about
well.
B
Exactly
so
that
is
hopefully
gonna
make
Vita,
but
not
the
Qiblah
dynamic
configuration
loading
pulling
from
the
clock
twenty
things,
but
in
in
110,
and
that
is
that
will
make
possible
to
do
things
like
like
right
now.
If
you
configure
your
service
Sybil
to
something
else,
instead
of
the
default,
you
will
have
to
go
to
each
cube.
Let's
get
into
the
system
V
file
and
change
the
value
there
for
for
the
DNS
server,
it's
it's
horrible,
so
this
is
gonna,
make
it
possible
to
the
float
that
information
down
to
all
the
heads.
A
A
B
I,
like
the
process
so
far
has
been
basically
myself
and
a
couple
of
other
guys
manually,
hacking
things
on
VMs,
trying
to
validate
things
like
the
to
three
days
before
we're
cutting
in
a
in
a
place
where
we
basically
can't
do
changes
anyway
anymore,
so
definitely
would
would
love
some
help,
but
that
needs
work
from
for
myself
and
that
is
to
take
out
the
guide.
How
to
do
this
thing?
I
mean
it's
not
hard,
but
it's.
It
needs
documentation
right.
This.
Nobody
knows
knows
it's
possible
unless
it
is
yeah.
A
I
think
maybe
what
I
can
do
that
seems
reasonable.
I,
don't
signing
myself
for
things
for
a
night.
It
seems
reasonable
to
have
a
little
like
hack
Guide
to
beta
testing,
and
it
can
outline
how
you
can
mock
your
system
to
unify
us
with
a
per
build
per
tag,
build,
and
then
you
can
do
this
on
your
own
environment.
I.
Think
that's
a
reasonable
thing
to
do.
I
could
I,
don't
think
that
will
take
too
long,
and
maybe
we
can.
We
can
have
uniformity
on
how
we
do
some
of
the
pre
pre-release
testing.
E
B
F
B
A
That
Jace
last
release
cycle
did
an
early
iteration
of
the
release
notes.
Do
we
have
if
I
get
this
document
out
and
I'm
able
to
push
it
out
to
the
community?
Do
we
have
like
a
listing
of
all
the
major
feature
enhancements
somewhere
that
we
can
reference
that
says
like
hey,
we
have
an
updated
version
of
cube
ADM,
here's
here's
the
feature
list
because
folks
will
enable
or
disable
weird
stuff
they
do
all
the
time.
This.
A
B
Well,
yeah
I
should
put
something
together
soon.
I
mean
a
couple
of
notes.
I've
done
this
for
like
in
my
cubicle,
slides
and
things,
and
when
I
presented
that's
cumulative,
meaning
lost
last
week
and
things
so
I'll.
Think
me.
If
you
see
the
record
requests
somewhere
and
the
high
level
things
okay
and
then
yeah
generally,
what
what
we
need
help
with
is,
as
you
said,
testing,
but
also
documentation,
you
definitely
need
more
people.
Updating
adopts
for
retailers
have
done
a
great
work
with
mine
pages,
which
is
still
a
bit
working.
B
A
D
I
said
when
kubernetes
1/8
is
ready,
we
do
a
sort
of
curation
thing.
We
have
a
yeah,
hopefully
soon,
typically
yeah.
Typically,
this
is
later
than
we
like.
Obviously,
there
are
some
pretty
serious.
The
the
a
great
API
server
blowing
up
clusters
thing
is
giving
me
poles
we're
seeing
also
some
calico
issues
so
yeah,
but
when
I
open
a
doubt.
C
D
B
D
Yeah
you
can
like
we
end
to
end
tests
like
a
running
against
one
time
right
now,
right
so
yeah
you
can
specify
any
kubernetes
version.
You
want
it's
just
that
b18
release
of
cops
is
the
one
that
actually
yes,
so,
yes,
you
can
like
run
the
1-800
alpha,
whatever
we're
on
or
beta
what
right
and
you
can
like,
specify
Q
minus
1
9.
It's
just
that.
The
the
release
of
one
800
of
cops
signifies
that
we
believe
that
it
is
a
good
idea
to
run
kubernetes
1
8
with
cops.
A
Okay,
so
that's
pretty
much
the
highlights
for
release
and
updates.
Next
agenda
item
was
unofficial,
meetup
in
Austin,
actually.
B
B
A
Bobby,
maybe
you
at
the
contributor
meetup
we
can
discuss
this
because
there
would
be
a
contributor
meet
up
and
we
can
start
to
talk
about
like
who
could
potentially
involve
this
cycle
and
commit
to
it
and
I'll
talk
with
Joe
to
about
what
obligations
that
are
going
to
have
the
next
cycle
as
well.
So.
A
B
Yeah
I,
don't
know
it
was
just
some
something
like
that:
Craig
some
on
some
other
people
and
myself
just
started
chatting
about
if
we're
gonna
have
a
full
month.
So
this
time
we're
not
gonna
have
a
formal,
sigilyph
life
cycle
meeting,
as
we've
had
the
two
of
their
cube
cons.
We
had
one
or
two
ago
where
Robert
was
the
host
in
Seattle
and
we
had
one
in
Berlin
where
Luke
was
the
host
but
yeah,
not
not
this
time.
E
E
A
A
A
B
B
Basically,
Robert
and
myself
are
gonna.
Do
a
class
lifecycle
update
so
well,
then
we
could
think
up
as
well.
If
we
like
to
view
that's
during
cube
con,
so
it's
always
too
much
things
going
on
price
on
the
same
at
the
same
time.
So
but
anyway,
that's
that's
happening
and
I
have
a
presentation
on
the
keypad
and
things
well.
B
C
C
The
gist
of
this
is
when
you
do
creating
it
in
it.
You
get
lots
of
nice
debugging
output.
One
of
those
things
is
the
joint
command.
You
can
copy
paste
to
a
networker
node
to
join
previously
Park
aeri
to
eight
that
join
command
was
also
pretty
easy
to
just
generate
yourself,
because
the
only
thing
that
buried
was
the
token.
C
Now
that
we
added
the
key
fingerprinting
for
the
server
it's
like
you
can
still
generate
it
yourself,
but
it's
more
work.
So
we
wanted
that,
basically,
to
make
this
thing
more
usable,
add
a
new
command
to
print
that
join
command,
on-demand,
so
that,
if
you
come
back
to
your
your
cluster,
that
it
was
initialized
before
you
want
to
add
a
new
worker,
you
can
just
say
on
the
on
the
master.
Give
me
the
joint
command
and
then
copy/paste
that
I.
C
A
C
D
C
The
other
thing
is,
you
can
generate,
I
mean
you
can.
If
your
rip
building
an
installer
today,
you
can
generate
the
joint
command.
Like
you
just
you,
you
have
to
pull
it
and
sort
of
like
open
SSL
to
generate
the
key
fingerprint
if
you
have
the
server
certificate,
which
is
annoying,
but
it's
not
impossible
right
now.
This
is
just
a
usability
feature.
Is.
B
C
A
I
think
there
can
be
a
case
probably
made
for
this
to
add
this
as
an
addition.
If,
if
we
need
to
or
want
to
at
the
end
of
the
cycle,
it
doesn't
seem
like
it
would
be
a
it's
not
a
stretch
right
other
things
are
definitely
a
stretch
at
any
new,
basically
net
new
behavior.
This
is
like
expanding
upon
existing
behavior
and
making
it
more
usable,
so
I
I
think
that
falls
in
this
weird
gray
area
where
folks
are
generally
okay
with
those
type
of
communities.
The
API
changes
after
today
are
usually
frowned
upon.
I.
C
The
other
change
that
we
had
already
landed
for
this
release
is
turning
the
insecure
warning.
We
hadn't
1/8
into
an
error
so
now
in
one
in
one
nine
year,
you'll
have
to
either
pass
in
the
key
pinning
argument
which
is
annoying
to
it's
annoying,
to
generate
right
now
or
you
have
to
pass
it
in
the
flag
that
says
you're
running
in
secure
mode.
So
it's
going
to
be
sort
of
press.
The
issue
for
more
people
means
you
can't
just
ignore
that
warning
anymore.
You
have
to
actually
do
something
to
fix
it.
C
B
C
B
Also
so
a
small
implementation
detail
is
that
we
actually
can
have
multiple
tokens
right,
so
we
don't
know
exactly
which
token
we
should
print
in
a
command
kiss
well,
but
we
could
do
things
like
if
there's
only
one
token
print
that
one,
if
there's
only
one
token,
with
the
right
properties,
print
that
one
and
otherwise
ask
the
user,
which
that
we
can
do
want
to
see.
I
was.
C
So
that
would
be
like
one
use
case
where
you're
building
us
an
installer.
You
want
to
generate
a
new
token
to
provision
a
set
of
nodes.
You
just
generate
the
token.
The
other
I
also
propose,
maybe
adding
a
token
get
command
so
token
get
and
I
took
an
ID
with
sort
of
similar
semantics,
where
you
can
tell
it.
You
just
want
the
full
token
with
the
secret
or
that
you
want
to
print
the
joint
command.
For
that
token,
ID
yeah.
D
C
There
might
be,
there
might
be
actually
a
use
case
where
in
cou
coda,
we
add
a
command
just
to
print
the
certificate
hash
or
just
just
output.
The
certificate
hash,
because
that's
kind
of
the
building
block
and
then
the
Kubb
radium
specific
thing
is,
is
like
bunging
that
together
to
make
the
joint
command.
A
A
Going
once
twice
three
times,
I
think
we
can
possibly
end
early
for
once,
give
you
back
14
minutes
and
if
we
see
folks
before
then
happy
holidays
of
some
type
for
you
I
know
the
folks
in
the
US
will
probably
be
off
on
Thursday
and
Friday.
So
take
care.