►
From YouTube: SIG Cluster Lifecycle 2022-01-25
A
A
So
the
first
group
topic
that
we
added
is
it's
about
the
annual
reports
that
special
interest
groups
have
to
do
every
year.
It
was
a
change
that
was
introduced
last
year.
We
did
a
report
and
the
basically
the
theodr
of
these
of
these
types
of
reports
is
that
you
have
to
summarize
what
is
the
health
of
the
group
like?
What
is
the
health
of
the
soup
projects
like?
Do
we
update
donor
files?
A
I
have
the
original
email
from
frobridge,
so
it
has
more
details
about
the
whole
thing
and
we
established
like
a
soft
deadline
for
this,
which
is
the
15th
of
february
in
a
couple
of
weeks
from
now
roughly,
and
we
decided
that
if
we
don't
get
the
responses
in
time
from
other
subprojects,
we
are
going
to
have
to
go
and
start
asking
people
individually,
which
means
potentially
somebody
get
it
getting
a
ping
or
slack,
or
we
may
as
well
log
an
issue
in
the
respective
project
repository,
and
hopefully,
people
can
act
on
that
and
the
the
last
point
I
heard
is
that
being
fabricio
can
potentially
split
this
task
of
pinging
people.
A
But
this
is
now
a
requirement
from
the
steering
committee.
I
think
originally
it
happened
because
you
see
in
commodities
we
had
some
stale
groups
and
that's
not
great
for
the
image
of
the
project
as
a
whole.
A
But
yeah
we
could
get
back
to
this
and
also
want
to
discuss
this
point
with
fabricio
that
maybe
we
can
split
the
world.
He
has
some
topics
here
I
mean
I
know
what
these
are
about,
but
let's
see
for
which
you
join
us
later.
A
Does
everybody
have
any
topics,
other
group
topics
or
psjas.
A
You
might
have
seen
an
email
that
I
sent
to
the
mailing
list
with
a
summary
of
what
is
happening
in
124.
We
are
making
a
number
of
changing
a
number
of
changes.
Unfortunately,
I
think
this
landed
in
the
same
cycle,
which
is
not
ideal,
but
also
the
cycle
is
so
low
now
that
it's
four
four
months
like
it's
really
difficult
to
spread
the
work.
At
this
point
I
had
concerns
at
the
beginning,
but
I
guess
nobody
complained,
so
we
proceeded
with
the
all
the
changes
that
we
planned.
A
A
So
you
can
check
these
out.
I
think
we
are
going
to
repeat
them
in
this
meeting
if
we
read
the
email
again,
but
I
want
to
highlight
the
first
one,
which
one
of
these,
I
guess
is
that
the
master
two
control
plane
level
intent
migration
is
in
flight
and
we
noticed
that
this
broke
immediately.
Some
aspects
in
kind
well
can't
use
kubernetes,
because
kind
was
waiting
for
the
master
label
on
a
node.
A
This
was
one
of
the
changes
we
had
to
do
it
kind
of
actually
said
the
pr
and
the
kind
maintainers
accepted
it.
A
The
summary
of
the
changes
that
mateo
the
arab
change
is
that
we
are
no
longer
having
the
master
label
on
kubernetes
at
all,
but
for
one
release,
which
was,
I
think,
something
that
jordan
suggested
is
that
we're
going
to
have
both
the
old
taint
master
and
the
mutant
control
plane
on
the
nose.
A
This
is
to
provide
the
easy
migration
stepping
stone
for
users
and
next
release
we're
going
to
remove
the
master
date.
So
in
125
we
are
no
longer
going
to
have
any
references
of
the
master
label
maintained
in
cuba.
A
Instance,
like
the
kite
project,
did
it's
a
breaking
change
for
them?
Then
again
in
one
120,
we
already
announced
the
plan
and
everything
break
break.
Breaking
users
is
inevitable
around
this
change.
Somehow
I
anticipate
users
log
in
tickets
and
that's
why
we
have
the
cap
ready
to
with
a
very
clear
plan,
to
install
them
how
to
adapt,
and
also
I
saw
that
so.
The
second
point
here
is
that
I
saw
the
entrance
sweet
in
kubernetes
only
tolerated
the
master
taint
as
a
non-blocking.
A
So
there's
a
bit
of
a
design
problem
in
the
suite
where,
if
you
have
taints
on
some
nodes
that
are
with
no
schedule,
this
means
that
the
the
switch
will
walk
on
these
stains
until
the
no
schedule
is
removed.
Like
the
the
effect-
and
I
disagree
with
this
design
and
the
the
way
the
way
the
suite
is
dealing
with,
it
is
how
it
has
a
flag
where
you
can
specify
which
taints
you
want.
A
It
was
like
a
non-blocking
paints
or
something
like
that
is
the
name
of
the
flag.
D
A
D
Right,
I
mean
yes,
we
when
we
change
that
we
we
will
have
to
update.
I
think
we've
started
double
labeling,
as
you
have
incubated
m.
I
don't
think
we
are
doing
the
taint
migration
yet,
but
I
think
we
are
starting
to
generally
tolerate
like
wild
cards
instead
of
tolerating
individual
taints.
D
They
they
do,
you
don't
specify
the
name
you
just
specify
like
exists.
A
Key
yeah,
yes,
which
is
a
little
weird,
but
yes,
it's
very
weird
yeah.
I
know
this
yeah
yeah
just
this
is
a
basically
a
concern
about
this
change
with
the
tents
is
that
it
probably
will
need
an
action
required
in
the
release
loads.
Whenever
we
do
that,
because
the
framework,
the
the
suite
will
block
on
conformant
tests,
conformance
tests
in
user
questions
and
production
questions,
but
you
know
it's
part
of
the
rename
effort
and
it
has
to
go
away.
D
A
I
guess
I
can
also
notify
kdf
when,
when
we
are
actually
going
to
introduce
the
removal
of
the
master
teams
at
some
point,
another
side
notification
of
the
site
is
that
I
said
appear
to
improve
the
strict
validation
of
component
configs.
We
have
in
cubanium,
which
is
only
a
little
bit
of
an
improvement.
We
already
had
warnings
for
fields,
fields
that
are
duplicated
by
the
user
by
mistake
or
for
fields
that.
A
Are
unknown
in
the
conflict,
so
the
user
passes
a
field
that
is
with
a
typo
and
they
get
no
indication
of
the
mistake,
and
this
is
something
we
already
had
these
two.
But
I
read
with
this
change
about
in
a
third
one,
which
is
a
case
sensitivity,
so
it's
actually
important
to
pass
the
right
case:
sensitivity
for
a
particular
field.
A
For
instance,
in
cubarium
we
have
the
control,
plane,
endpoint
field,
which,
if
the
user
passes
the
the
first
letter
with
a
capital
c,
the
field
is
completely
unrecognized
by
by
the
json
parser,
and
this
change
is
going
to
ensure
that
they
get
a
warning.
A
It's
called
so
we
already
have
six
ktml.
This
one
is
called
the
six
case.
Json
is
actually
a
fork
of
the
standard
library
json
package,
because
it's
currently
missing
some
functionality
for
strict
and
marshalling.
A
So
sorry,
jordan
took
basically
created
a
like
a
fork
and
api
machinery
accepted
the
project
and
I'm
using
this
now,
or
rather
I'm
not
using
it
directly
because
it's
not
possible
to
use
it
directly.
I'm
using
api
machinery,
the
api
machinery
repository
that
renders
this
particular
repository
yeah,
I
mean
it's.
A
This
is
very
nice
because
I'm
doing
this
because
I've
seen
multiple
reports
by
users
who
pass
a
field-
and
they
say
yo
kuberium-
has
a
bug,
but
the
reality
is
that
they
have
a
typo
in
the
case,
sensitivity
of
a
particular
field.
A
Yeah
one
topic
about
the
audio
report
is:
can
we
like
me
and
you,
or
maybe
somebody
else
I
don't
know
like
formally
decide
how
we
want
to
ping
the
the
sub
projects
if
they
don't
submit?
I
think
my
idea
was
that
you
can
take
all
cluster
api
star
projects.
A
Okay,
maybe
also
it's
a
good
idea
for
you
to
give
us
access,
or
at
least
me
so
so
that
I
can
see
who
has
submitted
already.
I
don't
have
access
to
the
results.
Okay,
we
do,
but
thanks,
but
that's
all
this
topic.
I
think
I
already
explained
the
survey
and.
B
B
A
B
Brought
up
to
the
to
the
different
products.
B
B
A
I
wonder
what
is
going
to
output
for
projects
like
cuba
in
our
case.
A
A
Problem
one
is
that
we're
probably
going
to
get
a
lot
of
outputs,
because
we
have
many
other
files
in
different
repositories
and
I
don't
think
it's
feasible
for
the
leads
of
this
group
to
be
able
to
like
address
all
the
owner
files.
We
can
log
issues
in
the
respective
repositories,
but
you
know
I
don't
think
yeah
we
can
do
the
cleanup
ourselves.
B
This
makes
sense.
To
be
honest,
I
don't
have
idea
of
what
the
output
could
be.
I
saw
that
there
are
forks
which
are
already
doing
the
cleanup
of
our
maintainers
files.
B
B
Okay,
next
topic
is
yeah.
I
saw
a
male
in
the
in
ours
list
about
promoting
about
accepting
bringing
iran's
provider
into
the
city
repository.
There
are.
B
Two
plus
one,
but
they
are
asking
for
the
other,
leads
to
provide
a
feedback.
I
personally
don't
see
problem,
but
please
take
a
look
in
case
you,
you
don't
have
seen
they
may.
B
A
E
Yeah,
I
still
have
to
go
through
my
notification
list,
but
yeah
I'm
definitely
plus
one.
I
don't
see
a
problem
with
this,
given
it
doesn't
really
overlap.
A
D
I'm
in
favor,
I
think
it
looks
good.
I
mean
I
yeah
license
long
fixed
license
thing
which
sounds
like
we're
doing
then
yeah.
It
sounds
great.
I've.
B
Yeah
and
also
I
participated
at
the
first
office
hour
of
this
project,
and,
and
there
is
a
interest
from
red
that
as
well,
they
have
a
similar
provider
and
they
are
discussing
with
the
team
about
how
to
merge
the
two
effort
together
and
those.
So
if
I
remember
where
new
relic
force,
no,
not
new,
nearly
so
yeah.
B
Okay,
so
thank
you
also
twilio.
I
saw
people
from
twilios
showing
interesting
in
this
effort,
so
it
seems
that
is
eating
a
spot
that
that
kind
of.
A
Well,
instead
of
collecting
plus
ones
from
the
mailing
threads
mailing
list
thread,
we
can
do
it
on
a
kubernetes
or
tickets.
B
Yeah,
I
I
think
that
what
what
is
asking
is
just
go:
no
go
for
opening
the
ticket,
so.
B
A
A
E
A
A
But
I
added
a
couple
of
comments
here.
This
is
not
blocking,
they
can
change
it
later
I
mean
we
can
go
on
the
kubernetes
or
ticket.
At
this
point
I
don't
mind:
okay,.
B
And
then
my
last
point
is:
is
a
quick
check
on
about
the
cubico
maintainer
track?
If
we
have
already
submitted
some
talks
or
we
if
we
have
to
go
in
the.
D
As
far
as
I
know,
we
have
not
submitted
any
talks.
I
I
have
not
checked
today
what
the
state
of
the
variance
is
or
are,
and
whether
whether
we're
going
to
be
in
person
or
not
what
the
chances
are,
it
was
looking
good.
I
heard
there
was
a
new
variance.
I
don't
know
it
would
be
good
to
do
yeah.
D
I
don't
know
we
have
an
idea
for
something
we
could
do
for
chaos
if
we
if
we
need
to,
but
obviously,
if
we're
in
person,
it
would
be
great
to
do
something
like
the
collaborative
one
we've
been
talking
about
for
a
while.
B
Okay,
I
I
I
will
raise
the
point
tomorrow
in
the
copy
office
hour,
because
it
is
a
a
good
place
to
raise
many
sub
projects
to
to
reach
menisca
project
and
see
if
someone
want
to
propose
their
own
projected
dive.
They
have
to
reach
out
to
the
leads
that
that's
the
procedure
but
yeah.
That's
basically
what
I'm
going
to
tell
that.
A
I
think
it's
going
to
be
hybrid,
at
least
what
what
I
saw
the
last
time
I
went
to
the
kubecon
website.
A
good
idea
from
us
might
have
been
to
also
collect
information
in
the
survey
for
super
projects.
Whether
a
sub
project
is
going
wants
to
present
the
cubicle
this
year
or
something
like
that.
B
A
Yeah,
that
is
true
it
in
the
past.
We
had
issues
collecting
responses
so
from
the
mailing
list,
at
least
who
is
going
to
send
or
not.
You
know
that
that
has
been
a
problem,
so
the
survey
might
have
been
a
good
collection
point
for
that,
as
well.
Maybe
for
the
next
year
survey.
B
Yeah,
in
this
case
they
made
from
sick
country
backs
or
or
cncf-
I
don't
remember,
was
pretty
clear.
The
submission
every
surprise
that
have
the
chance
to
submit
a
talk,
submission
should
be
done
by
siege,
lead
and
and
basically
torque
will
be
selected
about
in
base
of
a
viability.
A
It
changes
every
year.
Yeah,
it
just
means
the
project
can
submit
themselves
now
see
lead
has
to
submit
yes,
okay,
yeah
I
mean
if
somebody
wants
to
submit-
and
they
see
the
recording
of
this
meeting,
they
can
contact
us
and
we
can
send
it
for
them.
No
problem.
B
Okay,
quick,
quick
update
from
capitland.
We
are
slowing
down
for
getting
the
release
that
we
want
to
advance
the
release
beta
2
was
cut
last
week.
Rc
is
planned
this
week
and
yeah.
We
are
looking
forward
for
feedback
from
provider
source
users.
A
I
wanted
to
ask
somebody
asked
me:
you
know
with
my
team
at
vmware,
they
wanted
to
use
quest
api,
but
they
asked
me
the
naive
question:
is
there
going
to
be
a
v1
which
I
didn't
have
the
response
to
like?
Do
we
have
a
rough
like
timeline
for
everyone?
Maybe
two
years
one
year
or
something
in
the
future.
B
A
Yeah
I'll
give
the
simplest
possible
cube
idea.
Basically,
when
it's
ready,
it's
ready,
it's,
it
will
take
perfect.
E
D
D
I
think
we're
now
at
the
stage
where,
like
we're
finding
issues
with
lack
of
support
with
like
cloud
controller
manager
and
like
external
dns
and
things
like
that,
so
we're
sort
of
moving
through
the
ecosystem
and
driving
a
lot
of
changes,
and
I
think-
but
I
think
it's
all
it's
actually
going
pretty
well
and
so
pure
ipv6
will
hopefully
be
relatively
widely
supported
soon,
not
just
in
chaops.
D
I
think,
and
on
that
grounds
we
were
thinking
about
that
that
that
might
be
a
good
talk
to
give
us
a
kubecon
maintainer
talk,
because
it's
not
really
about
chaos.
It's
more
about,
like
you
know,
like
the
the
ecosystem
type
thing.
D
So
that
was
that's
all
I
had
on
on
chaops
we've
like
I
was
worried
that
i36
was
going
to
prove
a
bit
of
a
time
sink,
and
it
has
indeed
done
so,
but
it
is
driving
good
good
changes
in
in
everything,
we're
doing
lots
of
cleanups
in
kops
and
a
couple
of
other
projects.
A
Yeah,
I
I
think
it's
a
great
idea
for
a
talk,
because
if
you
have
this
title
in
the
talk
like
ipv6
with
aws
and
chaos,
you
are
going
to
potentially
attract
all
those
people
that
understand
and
you
want
to
use
ipv6
and
you
can
have
a
bit
of
a
conversation
like.
What
is
your
current
adoption
like?
What
is
the
infrastructure?
A
Do
you
have
like
dual
stack
single
stack
up
ipv6
and
it's
really
good
to
get
the
feedback
and
also
you
can
then
pipe
the
feedback
to
seek
network
even
and
it's
just
very
useful
for
the
kubernetes
as
a
whole.
D
Exactly
yes,
it
would
be
very
much
a
like
a
practical
talk.
I
think
one
of
the
things
that
at
least
the
people
that
are
adopting
it
in
the
chaops
community
are,
I
think,
finding
is
that
you
know
a
big
part
of
the
appeal
of
ipv6
is
pure
ipv6,
like
not
dual
stack,
because
dual
stack
means
you
still
have
to
manage
ipv4.
So
what's
the
point
that
sort
of
thing,
so
that's
that's
a
big
now
that
it's
realistic
to
do
that.
F
Okay,
so
yeah
we
haven't
been
here
in
a
while,
but
also
we
haven't
had
much
to
update
on
125
took
a
long
time
to
release,
and
then
we
had
to
hotfix
it
the
next
day
because
that's
how
life
works.
So
that's
that's
ready.
F
It
was
just
a
bunch
of
sort
of
random
odds
and
ends
from
the
end
of
last
year
that
were
that
was
released,
and
so
some
of
the
some
of
the
groundwork
was
laid
for
vm
drivers
for
arm64,
which
seems
to
be
more
and
more
of
a
thing
that
people
need
so
yeah.
So
126
is
going
to
be
a
lot
of
things.
It
should
be
an
arm.
64
compatible
vm
driver
for
kvm
for
linux
and
with
some
of
the
work
for
a
mac,
a
mac
os
arm,
64
vm
driver
as
well.
F
We
don't
have
there's.
No
so
like
hyperkit
is
never
gonna
support
arm64.
As
far
as
we
can
tell
so
we're
gonna
work
just
directly
with
qmu
and
see
see
how
that's
gonna
go.
That's
the
plan.
F
Currently,
the
the
biggest
thing
for
126
is
supporting
the
deprecation
of
docker
shim,
because
we
need
to
continue
to
have
a
docker
container
run
time
inside
of
our
clusters
for
sort
of
developer
reasons,
and
so
we're
gonna
figure
out
what
we're
gonna
do
there
either
use
one
of
the
external
maintained
ones
or
maintain
one
ourselves
and
the
other
big
thing
is
we
already
no
longer
care
about
master
labels
or
anything
like
that,
but
we
still
haven't
fixed
the
name
of
our
main
branch
to
maine
and
all
of
our
internal
jenkins
stuff
also
says
master
everywhere.
A
Yep,
the
the
cuba
migration
from
ambassador
to
bait
was
pretty
straightforward.
We
just
wanted
to
synchronize
everything
with
a
push
of
a
bottle
to
happen
at
once.
You
know
it
was
fairly
simple.
We
just
had
the
prs
ready
and
somebody
did
the
rename
of
the
repository
with
administer
administrator
writes
you
know
just
do
the
rename
then
merge
the
prs
immediately
and
we
didn't
actually
see
any
errors
in
the
enter.
F
A
B
A
Want
to
keep
docker
support
for
docker
as
a
like
a
default
option.
Honestly,
I
I
have
some
concerns
around
cri,
dr
dear
how
much
mirantis
is
going
to
invest
into
maintaining
the
project
because
we
have
the
same
concerns.
A
F
Option:
yeah
yeah,
that's
so
many
people,
so
many
like
mini
cube
users
expected
though,
and
more
and
more
people
are
using
mini
cube
as
a
replacement
for
docker
desktop
on
non-linux
machines,
and
so
for
us,
it's
really
like
container
d
as
a
default
is
would
be
challenging
in
terms
of
it.
Would
we
wouldn't
be
able
to
make
that
change
soon?
So
that's
the
that's.
That's
the
only
reason
we're
sort
of
backed
into
this
corner.
F
That's
the
biggest
thing
is
bill
is
building
images
and
we
have
our
own.
We
have
our
own
sort
of
mini
cube
image,
build
tag
whatever,
but
it's
not
fully
feature.
It
doesn't
have
any
sort
of
feature
parody
with
the
docker
yet
and
that
will
that
works
with
container
d
and
cryo.
So
eventually,
that's
probably
what
we're
gonna
have
to
point
people
to,
but
it's
not
ready
yet.
A
A
F
D
I
I
know
nerd
ctl,
that's
when
I
use
this
build
kit,
I
haven't
tried
it
does
that
work.
Is
that
what
you're
doing
or
to.
F
Do
the
builds
we
so
we've
talked
about
supporting
nerd
ctl
like
under
the
hood.
It's
not
what
we're
doing
currently,
but
it
is
something
we're
aware
of.
D
I'm
not
sure
whether
bill
kid
is
what
docker
is
docker
build,
build,
build
x
or
which
build
x,
build
yeah,
so
maybe
yeah
anyway.
The
other
question
I
had
was
you
mentioned
mac
os,
not
supporting
hyper
kit.
What
about
the
other
one?
The
virtualization
framework.
F
So
the
only
the
only
other
so
parallels
we
don't
have
any
licenses
for
parallels,
and
I
have
no
idea
if
that
supports,
arm64
or
not
parallels
in
minicube
has
always
been
like
a
best
effort
on
anybody
who
wants
to
continue
maintaining
it
can,
but
none
of
the
core
maintainers
have
ever
maintained,
parallels
and
virtual
boxes.
Virtualbox.
D
Yeah
I'll
put
a
link
in
the
chat
or
something
on
the
chat
in
the
in
the
doc
that
that's
the
one
I'm
thinking
of,
and
there
are
some
examples
out
there
where
people
have
it's
like.
It's
like
a
very,
very
high
level
one.
So
I
don't
know
I
was
just
wondering
whether
you
were
aware
of
it
and
I
would
have
been
wanting
to
try
it
out.
So
I
was
wondering
whether
you
tried
it
out,
never
saying
it
doesn't
work
or
something.
D
Yeah,
this
is
not
something
I'm
aware
of
personally:
okay:
cool
cool,
that's
good,
because
yeah
well,
yeah.
I
agree
that
we
should
try
kumu,
because
it's
very
universal,
it
might
even
work
on
windows.
But
this
one
seems
like
a
good
choice
on
mac
os
they're.
They,
I
think
you
can
use
it
from
like
the
app
store
as
well.
So
you
don't.
B
I
have
a
question
back
on
the
on
docker
versus
alternatives.
I
saw
recently
podman
also
portman,
creating
as
a
command
for
creating
a
sort
of
virtual
machine,
or
something
like
that.
Are
you
aware
of
what
is
the
status
of
this
effort?.
F
B
C
After
I
realized
that
the
80
adm
oops
thanks
yeah,
it
looks
like
justin
you,
you
added,
like
I
guess.
The
final
final
part
of
the
migration
from
the
std
manager.
D
D
It's
it's
a
wonderful
organization
with
very
very
cool
people
that
are
part
of
it,
zero
members-
I
I
remember,
I
think
it's
just
me.
D
That
is
a
kernel
that
none
of
this
has
been
updated
for
a
number
of
years.
You
will
note,
but
that
was
a
kernel
that,
like
included
the
things
which
we
needed
when
kernels
were
problematic
back
in
2019.
A
Okay,
well,
I
don't
have
any
questions
I
think,
like
eventually,
we
have
to
update
the
cube
adm
documentation
to
reference
lcd
idm.
I
will
continue.
A
A
Have
you
been
trying,
like
more
proactively,
to
find
contributors
for
the
project?
To
maybe
other
companies
like
I
know,
aws
was
interested
at
some
point
with
their
external
hcd
for
cluster
api.
C
E
C
E
D
I
think
I
think
it
goes
in
fits
and
starts
right
like
we
have.
I
certainly
personally
have
times
when
we
do
more
on
it
and
times
we
do
less
on
it
and
often
around
like
around
like
new
lcd
versions
or
trying
to
like
converge,
but
I
need
to
get
more
consistent
on
on
contributing
to
it
and
and
doing
all
the
reviews
and
things
like
that
as
well.
A
Yeah,
if
there's
a
company
that
really
wants
to
use
it
and
change
it
in
a
certain
way,
they
will
have
to
invest.
So
that's
how
it
works.
Basically,
if
it's
currently
usable
for
certain
parties,
if
somebody
wants
changes,
they
should
invest
the
human
resource,
that's
how
it
works
in
all
consoles.
Basically,.
A
Yeah
I
mean
I
don't
have
any
other
comments
any
do.
Does
anybody
have
comments
or
questions
for
this
year?
The
idiom.
A
All
right,
unless
there's
any
other
topics,
we
can
get
deadbeats
back.