►
From YouTube: Kuberentes SIG Release Meeting for 20230711
Description
Kuberentes SIG Release Meeting for 20230711
A
A
It
was
mine
yeah,
so
we're
already
recording,
but
I
want
to
remind
you
that
these
meetings
we'd
like
to
conduct
them
under
the
kubernetes
code
of
conduct.
So
please
be
very
nice
to
each
other
and
yeah.
So,
let's
go
ahead
with
the
recurring
topics
and
the
agenda
today
that
I
posted
on
the
on
the
chat
so
I
believe
Marco.
You
have
several
topics
today.
Oh,
but
first
thank
you
Joe
for
taking
notes
and
do
we
do.
We
have
any
new
members,
so
I
would
like
to
say
hi.
B
A
C
A
C
I
left
some
topics
about
release
engineering
at
the
research
work
that
the
team
did
so.
The
first
point
that
I
would
like
to
bring
up
is
the
July
patch
releases
that
were
originally
scheduled
for
tomorrow
are
now
delayed
to
the
next
week
to
Wednesday
July
19,
to
be
able
to
get
all
the
PRS
that
we
have
merged,
because
this
is
going
to
include
the
last
planned
124
release.
So
we
want
to
include
the
PRS
like
go
updates
and
the
general
we
also
were
asked
to
merge
some
cherry
picks
past
the
deadline.
C
We
decide,
considering
everything
to
delay
the
release,
so
we
are
going
to
do
it
next
week.
This
is
speaking
of
releases.
Also
chirping
deadline,
as
mentioned
to
be,
the
announcement
is
also
moved,
and
it
is
this
Friday
July
14th,
the
next
topic
that
I
would
like
to
bring
up
before.
Do
we
have
any
question
about
releases.
C
Okay,
so
the
the
next
topic
is
that
I'm
not
sure
if
he
mentioned
at
the
submitting
before,
but
because
I
was
not
able
to
attend
the
last
one
but
Sasha
and
I
think
Carlos,
I,
hope,
I
didn't
miss.
Anyone
are
working
on
updating
the
base
image
to
Debian
travel
to
book
form
for
all
the
images
that
we
use.
There
was
some
announcement
on
mailing
list
about
that,
just
as
an
Fai
in
Suffolk
to
pay
attention
to
this
is
still
work
in
progress.
I,
don't
know
if
Sasha
has
any
update
about
that.
One.
D
Hey
yeah,
so
we
are
working
on
updating
most
of
the
images
and
we
had
a
discussion
about
Cube
cross.
We
will
probably
stick
with
all
the
creep
costs
or
Bullseye
related
Cube
course,
images
to
all
the
release
branches
and
was
updated
to
Bookworm
only
for
128,
but
this
will
be
part
of
the
release,
notes
and
yeah
I
think
it
will
take
me
some
time
to
upload
or
update
all
of
those
images
but
yeah.
We
are
not
in
a
hurry
right.
So.
C
C
Okay-
and
the
next
topic
is
that
we
had,
we
won
28
0r4
released
last
week.
It
was
released
successfully,
but
there
were
some
minor
hiccups
that
might
be
worth
mentioning,
especially
if
we
run
into
suffix
similar
next
week.
So
the
first
one
is
that
generating
the
s-bomb
was
showing
some
sort
of
null
pointer
exception,
and
this
is
fixed
by
Adolfo,
but
I
think
we
still
need
to
include
this
change
in
Corel
generally
just
running
the
release
again
is
what's
needed
to
unstuck.
C
Another
thing
is
that
we
are
running
into
a
rate
limits
again
we're
doing
the
image
promotion.
This
is
the
unfortunate
part.
I
am
not
exactly
sure.
Why
is
that
the
case
that
might
need
some
additional
investigation,
but
this
is
something
that
might
happen
again,
let's
cross
fingers
that
it's
not
going
to
happen,
but
we
will
see-
and
the
last
part
is
that
some
user-
some
release
managers-
are
having
problems
with
sand
grid.
C
I
don't
know
if
you
should
eventually
consider
getting
rid
of
scent
grid,
but
generally
that
tends
to
be
of
the
problem,
so
not
too
bad,
and
the
last
point
for
my
side
is
that
136
to
124
release
branches
are
updated
to
Globe
on
the
20.,
thanks
to
Carlos
for
taking
care
of
that,
I
think
that
we
all
need
to
merge
some
final
PRS,
like
updating
the
publishing
bot
also
publishing
part,
is
currently
down
or
broken.
We
are
waiting
for
that
to
be
fixed
before
updating
any
rules
and
I.
A
E
A
C
Yeah,
you
see
me
again
so
I
wanted
to
provide
some
updates
regarding
how
are
we
doing
with
OBS?
It
was
the
plan
and
how
is
it
all
going
to
look
like
so
the
first
part
where
we
he
stopped
before
me.
Getting
away
was
trying
to
figure
out
like
how
to
connect
OBS
with
the
release
Pipeline
and
to
cut
to
publish
packages
to
OBS.
Then
we
are
cutting
a
release
and
that
part
is
done.
C
We
have
been
actually
running
that
in
the
pipe
in
the
release
pipeline
for
some
time
and
packages
are
getting
built
and
published
successfully
by
OBS.
Now
there
are
some
important
parts
that
needs
to
be
done
and
the
first
one
is
especially
being
testing
like
when
we
were
implementing
the
initial
part
and
the
focus
wasn't
getting
into
work
and
also
getting
its
tests.
C
So
we
need
to
step
back
and
to
implement
some
tests
for
that,
and
this
is
like
the
one
of
the
highest
priority
tasks
going
to
128
release
and,
besides
that,
the
second
high
priority
task
that
we
have
is
sorting
out
infrastructure,
and
this
is
something
that
I
have
been
working
in
the
last
on.
In
the
last
few
days
on
that,
how
is
it
going
to
work?
C
It
is
that,
because
the
OBS
platform
has
sort
of
limited
bandwidth-
and
we
don't
know
how
much
bandwidth
do
we
need
for
packages,
because
we
are
using
Google,
infra
and
they're
not
really
able
to
provide
us.
Any
stats,
I
already
checked,
with
Ben
L
there
and
with
Google
build
admins,
but
there's
no
way
to
get
any
usage
information
so
to
make
sure
that
we
don't
overload
the
OBS
platform
and
that
we
that
we
can
provide
some
guaranteed
stability
so
that
users
can
always
pull
packages.
C
We
are
going
to
take
off
some
load
load
from
OBS
and
like
to
distribute
it
between
their
mirrors
and
our
own
mirrors,
so
they're
still
going
to
build
packages
they're
going
to
host
them
they're
going
to
serve
them,
but
we
are
also
going
to
serve
them.
Obvious
is
going
to
be
one
single
fronted,
one
single
repo.
It
is
just
that
they
will
also
point
to
our
own.
C
The
plan
is
that
we
will
only
mirror
our
own
packages,
so
we
will
not
have
additional
packages
and
the
way
it
works
is
that
it
is
basically
going
to
be
S3
bucket,
nothing
too
special
and
we
will
have
some
CDN
in
front
of
the
test
free
bucket
and
the
CDN
solution
that
we
are
going
to
use
is
going
to
be
cloudfront
at
the
beginning.
If
you
get
some
additional
bandwidth,
we
are
going
to
use
fastly
because
we
are
using
fastly
for
DLK
this
IO.
C
However,
we
have
only
10
petabytes
per
month
and
the
arcade
sio
is
going
to
require
at
least
six.
So
right
now
we
can't
use
fastly,
but
we
requested
that
they
bump
us
to
20
petabytes
per
month.
If
that
happens,
we
might
be
able
to
switch
four
packages
to
use
faster.
Otherwise
we
will
continue
with
cloudfront.
Everything
is
on
ABS.
We
have
the
donation
of
three
million
dollars
of
cloud
credits.
C
We
are
close
to
million
right
now
with
CI
and
everything
so
2
million
is
about
left,
so
I
think
that
should
be
more
than
enough
to
fit
packages
and
everything
else,
especially
that
hopefully,
OBS
is
going
to
take
some
part
of
the
load
as
well
their
mirrors
and
also
we
have
mirrors.
So
we
can
sustain
all
the
load.
We
can
guarantee
that
packages
are
up
and
yeah.
That's
basically,
the
plan
I'm
going
to
share
a
PR
for
the
infra
part,
so
I
shared
it
in
chat.
C
C
E
C
I
am
not
sure
if
I
have
anything
else,
I
will
be
working
on
docs
and
on
feature
blog
posts,
I'm
going
to
propose
that
so
yeah
I'm
a
I'm
going
to
make
sure
that
the
once
we
are
ready
that
we
try
to
like
get
the
word
out
to
make
sure
that
users
migrate
I
also
had
some
discussion
with
cncf
cncf
is
going
to
help
us
as
well.
When
we
are
ready,
they
can
include
some
blog
posts
and
stuff
like
that.
C
C
A
F
Yeah
so
I
have
a
question
about
how
do
we
determine
which
projects
get
to
be
lucky
enough
to
become
Master
blocking
and
release
blocking
I'm
asking
a
question
because
I'm
in
the
middle
of
effectively
rewriting
a
lot
of
the
node
ede
jobs,
some
of
them
already
qualify
under
that
definition
and
others
don't
so?
I
was
wondering
if
there
was
a
dock
somewhere
that
talked
about
that.
F
C
A
A
Okay,
yeah
also,
these
are
the
the
general
guidelines
but
I.
Remember
that
a
few
times
when
someone
has
required
to
move
one
category
to
the
other,
either
blocking
to
informing
or
reforming
to
blocking,
there
can
be
discussions
about
it.
It's
not
like
set
in
stone
depending
on
different
variables,
so
you
guys
should
be
good.