►
From YouTube: SIG Cluster Lifecycle 20200922
Description
Recording of the sig-cluster-lifecycle biweekly meeting held on 20200922
A
Hello,
everybody
and
welcome
to
the
sick
cluster
life
cycle.
Biweekly
meeting
today
is
tuesday
september
22nd
2020..
I
am
your
moderator,
facilitator,
justin,
santa
barbara.
I
work
at
google
a
reminder
this
meeting
is
being
recorded
and
what
we
put
on
the
internet
and
so
therefore
to
please
be
mindful
of
our
code
of
conduct,
which
boils
down
to
being
a
good
person.
A
There
are
a
moderate
number
of
us
here
today,
so
please
try
to
use
the
hand
feature
I'm
on
the
web,
so
I
will
try
to
observe,
but
I
think
we
can
just
generally
self
police.
There
is
an
agenda.
There's
an
agenda
document.
I
will
paste
a
link
in
chat
if
nobody
has
beaten
me
to
it.
Please
do
add
your
attendee
or
please
add
your
name
to
the
attendee
section
and
any
group
topics
or
public
service
announcements
you
might
want
to.
A
We
will
go
through
any
group
topics
where
there
are
none
currently
and
then
we
will
go
through
the
various
sub-projects
and
do
call-outs
on
those
just
to
roll
up
status.
I
am
filling
time
to
see
if
anyone
adds
a
group
topic,
but
I
don't
see
anything
on
the
agenda.
If
there
is
anyone
that
would
like
to
introduce
themselves,
please
feel
free
to
do
so
now
or
if
there
are
any
other
group
topics
that
have
not
been
on
the
agenda,
please
just
chime
in
now.
Otherwise
we
will
move
to
project
call-outs.
A
B
Hi
everyone
so
on
quick
update
from
kubernetes,
we
start
we
already
started
the
work
for
120
and
the
first
things
that
we
are
taking
thanks
to
ruboni
are
a
set
of
deprecation
like,
for
instance,
dynamic,
kubernetes,
configs
and
fasting
and
customize,
which
was
replaced
by
json
patches
and
also
a
graduation.
So
we
are
going
to
graduate
kubernetes
alpha
cert
to
cover
min
cert.
B
C
A
B
Yeah,
I
can
give
also
an
update
on
cluster
api
custom
api.
Basically,
we
are
working
to
get
release
v0
3.10
already,
probably
we
will
get
a
first
rc
by
the
end
of
the
week,
the
the
most
that
there
are
many
small
changes
in
in
this
release.
The
most
relevant
one
is
is
that
we
we
found
and-
and
we
solved
the
memory
leak
in
in
in
the
cluster
source
tracker
in
a
cluster
in
a
workload
cluster
tracker,
which
is
basically
a
cache
that
is
created
for
every
every
workload
cluster.
B
The
attachment
basically
contains
a
client
for
every
workload,
cluster
and-
and
basically
the
problem
was
due
to
a
controller
on
time,
which
was
not
handling
very
well
a
client
to
get
created
and
destroyed,
created
and
destroyed,
and
and
so
on.
So
the
the
the
real
problem
was
fixed
in
in
control
around
time,
then
we
pick
it
up,
then
the
new,
the
new
version
of
controller
runtime
and
and
along
the
way
we
we
also
implemented
some
something
some
coding
program
around
how
this
cache
is
managed.
A
B
Oh,
thank
you.
Thank
you
brian,
for
remembering
me,
so
we
we
in
parallel.
We
are
starting
discussion
for
for
the
for
what
will
be
in
the
one
alpha
four
and
one.
One
of
the
topic
that
we
started
is
the
idea
of
having
a
management
cluster
operator
that
basically
will
allow
to
manage
in
a
declarative
style
or
git
or
github
style,
the
management
cluster
itself.
So
you
can
basically
apply
provider,
the
provider,
the
lead
provider
in
a
declarative
session,
and
we
had
a
first
meeting
about
this.
B
A
Wonderful,
thank
you
and
the
I
I
guess
the
appropriate
next
step
there
would
be.
It
looks
like
there's
an
issue
open,
which
is
looks.
I
think
it's
three
four
two
seven,
but
I
think
the
next
step
would
logically
be
to
attend
the
cluster
api
meeting
on
wednesdays,
for
people
that
are
interested
right.
B
Yeah
for
sure
the
the
the
next
step
agreed
on
on
the
on
on
the
meeting
was
for
me
and
warren
to
try
to
submit
to
today
to
the
community
in
a
google
doc
a
first
version
of
the
api,
and
so
we
can
start
it
iterating
fast
on
the
google
cool
document.
Basically,
we
are
starting
to
work
on
a
cap
in
incremental
fashion,
and
maybe
I
should
link
the
document
here.
B
A
A
Okay
looks
like
I
have
the
next
subject
update
for
cops
the
biggest
I
guess.
Cross-Productive
news
is
that
we
actually
managed
to
do
a
community
or
a
maintainer
release,
as
in
a
member
of
the
community
or
a
maintainer,
was
able
to
do
a
release
and
get
it
mostly
pushed
without
requiring
you
know
too
many
magic
scripts.
A
There
are
still
some
magic
scripts
and
special
permissions
that
need
to
be
run,
but
we
are
making
progress
on
getting
to
the
working
group
kate's
infra,
and
I
think
that
can
that
the
publishing
of
containers
and
releases
of
containers
is
now
fully
automated
there
and
the
remaining
pieces
relate
to
legacy
locations
by
the
locations
we
used
before
and
to
non-container
artifacts.
So
binaries
like
the
cops
binary
itself,
but
we
are
getting
much
closer.
A
C
D
D
D
A
E
Yeah
so
two
weeks
ago
we
released
one
thirteen
zero
in
tandem
with
kubernetes
119,
and
then
yesterday
we
released
one
thirteen
one
for
kubernetes
one.
Nineteen,
two
y'all
really
fast
and
release
two
like
back
to
back
so
so
we
should
fully
support
that.
The
focus
for
the
next
little
bit
here
is
gonna.
Be
stability
like
stability
of
our
like
integration,
testing,
infrastructure,
we've
been
having
issues
with
that
and
it's
sometimes
reflected
poorly
on
the
releases.
So
we're
gonna
take
a
step
back
and
make
sure
that
their
releases
are
stuff.
E
A
C
Yes,
I
don't
mind
giving
one
for
image
builder,
so
we
have
moved
to
new
office
hours
and
we
have
been
having
office
hour
calls
with
a
few
more
people
recently.
So
we
are
getting
a
few
more
people
on
board.
I've
been
on
board
a
couple
people
in
my
team
now
I
know
vmware's
also
onboarded,
quite
a
few
people,
so
that's
kind
of
the
update
from
our
side.
The
cli
is
moving
along
quite
nicely
and
then
I
noticed-
or
we
were
kind
of
nominated
for
kubecon
deep
dive.
C
I
think
in
the
last
office
hours
or
plus
the
life
cycle
meeting,
I'm
looking
at
the
at
the
meeting
notes
and
only
lumiere
and
and
tim
were
there,
so
I
think
they
nominated
us
to
do
it,
but
they're
not
on
the
call
today,
maybe
catch
them
offline.
I
just
wanted
to
confirm
that
we
were
actually
doing
it
and
it's
not
something.
That's
still
to
be
done.
A
Other
people
may
have
a
better
reflection.
My
recollection
is
that
we
were
basically
unsure
around
which
projects
would
benefit
from
from
a
session,
the
downside
being
that
as
a
virtual
kubecon
it.
It
is
a
little
different
from
a
it's
a
different
experience
from
a
physical
one.
I
that's
fair
to
say
and
whether
you
would
want
to
but
yeah.
I
agree
that,
following
up
on
slack
is
probably
the
the
right
thing
to
do.
A
My
guess
is
that
if
you
feel
like
it'd
be
valuable
for
your
project
and
you
want
to
that,
that
would
be
great.
You
do
have
to
like
pre-record
a
video
cool
but
yeah
following
up
on
slack,
just
to
confirm
that
and
make
sure
that
we
got
the
the
requests
and
would
be
good.
I
can't
remember
the
deadline.
I
think
it's
coming
out
very
soon,
though.