►
From YouTube: SIG Cluster Lifecycle 2021 01 26
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).
B
All
right,
I
don't
need
to
clean
host
hello.
Today
is
january
26
2021
holy
cats.
This
is
the
standard
sick,
cluster
lifecycle
call.
Let
me
punch
up
the
agenda
here.
B
Have
have
we
passed
the
point
of
submitting
sessions
for
kubecon?
We
have
for
the
virtual
kubecon
eu.
B
All
right,
so
we
should
probably
do
the
psa.
Then
we
should
submit
our
sessions
for
the
kubecon.
Usually
we
do
it.
We
coordinate
via
here.
B
Do
we
have
folks
that
wanted
to
submit
because
the
the
sig
calls
there's
like,
usually
a
limited
number,
or
at
least
there
were
before
I
don't
know,
who's
coordinating
anymore,
because
I
know
nancy
has
switched
roles.
B
So
do
we
have
folks
that
wanted
to
do
a
session
for
a
kubecon.
B
D
Basically,
we
kind
of
arrived
to
the
decision
to
send
an
email
to
the
mailing
list
and
if
some
project
wants
to
send
a
session
according
to
the
latest
guideline
from
the
cncf,
they
can
that
without
the
agreement
of
the
sick,
okay,
the
su
projects
are
in
a
way
eligible
to
do
that.
D
So
what
I
didn't
do
in
that
email
is
to
at
least
ask
the
people
for
sending
a
session
to
drop
us
an
email
and
say
hey.
We
are
sending
a
session
here
just
so
you
know.
So
maybe
I
can
send
another
email
in
the
same
thread
to
kindly
ask
people
to
do
that.
C
Mini
cube
hasn't
sent
one
yet,
but
I
think
we
plan
on
it
so
I'll.
Let
you
guys
know
when
I
do
that.
E
I'm
going
to
raise
the
point
to
the
next
meeting,
so
we
coordinate
there.
D
And
I
don't
think
me
or
fabricio
will
volunteer
to
present
this.
B
Okay,
all
right,
the
next
couple
topics
are
you
then,
so
I
think
we
could
probably
move
on.
So
the
action
item
is
to
drop
a
reminder.
Midi
keep
said
they're
going
to
send
one
for
b2
said
he's
going
to
bring
it
up
during
the
next
kubecon
call
or
circuit
next
cluster
api
call
and
then
comedian
isn't
going
to
do
one.
So
I
don't
think
we
need
to
belabor
the
point
here.
We
could
just
follow
up
easternersly.
D
B
I
might
try
to
use
it
as
a
forcing
function
for
me
to
make
a
presentation.
So
let's,
let's
see
here,
what's
the
what's
the
due
date
again
for.
D
B
D
Yeah,
this
is
something
that
I
think
the
courser
api
provider
for
nested
opened
as
a
discussion
some
time
ago
that
we
are
not
very
consistent
in
the
naming
of
our
youtube
videos.
D
D
What
the
naming
of
youtube
videos
for
the
whole
kubernetes
project
should
be.
It's
pretty
much.
The
format
we
arrive
to
is
sick,
the
name
of
the
sick
and
potentially
the
name
of
the
meeting
with
a
dash
in
front
of
it.
If
it's
the
main,
seek
call,
you
don't
include
any
of
that,
and
in
the
end
there
is
the
date
which
is
sort
of
different
of
what
zoom
produces.
Zoom
doesn't
have
the
dashes
between
years
and
month
and
day.
B
I
I
don't
care,
I
think
that,
having
some
level
consistency
might
help
if
people
are
searching
for
things.
B
D
B
Yeah,
I
think
just
take
it
to
the
list
and
with
a
recommendation
and
we'll
see
what
happens,
I'm
sure
people
won't
follow
it
for
a
while
and
then
we'll
have
to
follow
up
again.
B
All
right,
you
got
the
next
topics
for
the
mirror.
D
The
gcse
gcr
repositories,
slash
packets,
we
use
for
art,
for
cr,
artifacts
and
below
you
can
see
the
changes
that
will
happen.
In
fact,
we
already
support
these
new
repositories
in
buckets,
but
we
realize
that
some
new
artifacts
are
missing
in
there.
D
I
think
it's
safe
to
do
the
gcr
change
already,
but
the
case
release
dev
one
was
problematic
and
I
think
it's
still
pending
a
certain
fix,
but
once
that
issue
is
resolved,
which
will
probably
happen
in
the
121
cycle,
you
should
just
be
ready
to
migrate.
D
Both
images-
and
you
know,
binaries
if
you
are
downloading
them
from
the
gcs.
B
D
Because
I
I
found
out
about
this
aaron
whispered
me
on
a
github
issue,
without
that
I
wouldn't
have
known
that
we
are
making
the
switch.
So
I'm
hoping
that
this
this
will
reach
cooperatives.
B
D
B
B
All
right,
well,
I
think,
follow
up
on
an
email
thread
to
the
broader
group,
but
I
don't
necessarily
know
it'd
be
nice
to
know
the
scope
of
the
potential
view
of
the
impact,
because
I
do
think
there
are
people
who
consume,
especially
people
who
do
like.
B
D
D
Yeah
so
for
kubernetes
we
had
a
planning
session
last
week,
which
was
only
been
for
british,
who
joined
the
call,
not
many
participants
in
the
kubernetes
meetings.
Recently
we
basically
discussed
the
pending
work
me
for
british
were
planning
for
121,
which
is
not
that
much,
but
next
the
next
call
we
are
going
to
try
formalizing
a
roadmap
for
cuba
dm,
which
is
going
to
be
like
a
document,
something
like
a
plot
for
the
future.
D
If
somebody
is
interested,
please
join
the
call
so
that
you
can
help
us
with
this
discussion.
B
You
might
want
to
broadcast
that
to
the
email
list.
I
think
that
would
be
a
a
good.
I
mean
you
want.
If
you're
having
an
attendance
like
fall
off,
I
think
broadcasting
to
the
email
list,
as
well
as
maybe
even
to
devel,
might
not
be
a
bad
way
of
saying
like
hey.
If
you're
interested
in
engaging
do
the
one
part
outreach
one
part
roadmap
planning.
B
All
right,
it's
funny
because,
like
after
you
get
past
the
hype
cycle
for
these
projects,
you
go.
You
almost
go
into
like
maintenance
mode
right
like
it
works
now,
and
you
have
a
lot
of
users,
so
they
kind
of
view
it
as
this
trusty
thing
that
they
use
not
realizing
that
to
maintain
this
thing
requires
effort
and.
D
D
It's
one
of
my
long-term
goals
to
try
to
solve
that
and
new
features
and
lesser
of
focus
for
me,
but
yeah.
We
have
to
fix
a
lot
of
things.
It's
not.
B
D
This
is
just
a
couple
of
updates
that
we're
working
with
six
scalability
and
sick
windows
to
either
add
new
signal
for
cube,
adm
and
scale
testing,
or
currently
we
have
some
problems
with
the
kubernetes
or
windows
tests
for
both
of
these
we're
using
gce
as
the
platform.
So
it's
a
bit
slow
to
debug
in
the
quality
infrastructure,
but
we're
still
doing
the
best.
We
can
to
get
this
signal
in
a
good
shape.
D
So
to
buy
the
static
is
a
long-term
deal
to
do
that,
but
this
is
likely
going
to
be
like
a
kubernetes
test
that
coexists
in
parallel
to
the
existing
one
for
a
while.
D
Basically,
we
want
to
get
parity
between
what
cuba
can
do
today,
which
is
your
q
q
is
like
costary
pi,
prime,
which
is
it's
a
provider.
It's
a
bootstrapper,
it's
everything
and
I
think
they
want
to
basically
remove
parts
of
that
so
that
we
can
get
kubernetes
at
least
as
the
bootstrapper
and
see
if
we
can
configure
kubernetes
in
the
same
way,
cube
up
is
being
configured
today.
You
know
it's
bash,
so
it's
very
configurable,
but
at
the
same
time
it's
difficult
to
maintain.
B
D
Yeah,
so
I'm
working
with
some
of
the
polish
developers
who
work
at
google
like
wojciech
and
some
of
his
team
to
get
this
going.
We
currently
block
on
some
testing
for
semantics,
but
this
is
going
well
thus.
D
B
Oh
sorry,
I
was
muted,
we
can
move
on
fabricio.
E
E
Then
there
is
also
pr
form
from
metal
cube
fork,
so
we
which
is
implementing
the
scaling
strategy
for
kcp-
and
this
this
will
allow
basically
to
upgrade
your
control
plane
in
a
bare
meter
scenario
where
you
don't
have
a.
E
Spare
machine
to
use
for
the
for
the
upgrade,
but
the
implementation,
of
course,
will
work
also
with
other
providers
and,
last
but
not
least,
the
the
load
balancer
proposal,
so
the
plugable
load
balancer
proposal
is
making
good
process
progress.
Thank
you
to
jason
nadir
and
your
senior
and
all
the
other
folks
which
are
working
on
on
this
effort.
B
That's
good
any
questions,
comments,
complaints,
concerns
for
forbidzio.
F
Thank
you
yes,
so
on
chaops,
the
primary
thing
that
happened,
that's
across
project
interest
is
we
found
this
or
an
issue
was
reported
against
us.
That
turns
out
to
be
an
ubuntu
issue
or
a
kernel
issue.
But
if
you're
seeing
incredibly
slow
downloads
from
gcs
or
in
general
anywhere
with
relatively
long
round
trip
time,
then
you
might
want
to
look
at
the
sort
of
thread
and
work
around.
F
B
Never
sideways
or
backwards
yeah,
so
that's
amusing.
I
did
see
like
one.
I
think
I
can
share
this
information.
B
We
did
some
profiling
internally
for
other
issues
with
regards
to
total
throughput
performance
for
certain
sets
of
operations
with
regards
to
newer
ubuntu
kernels,
and
we
did
see
appreciable
slowdown
on
sets
of
operations
within
the
newer
kernels,
which
we
thought
was
fascinating,
but
you
know
we
haven't,
there's
not
like
public-facing
information.
I
can
point
people
at
so
this
is
kind
of
another
data
point
that
resonates
with
that
information
that
we
have.
B
So
one
thing
that
we
talked
about
scalability
testing,
one
thing
that
might
be
interesting
is
that
I
don't
we
used
to.
I
in
fact
I
believe,
jay
and
I
wrote
the
test
so
long
ago
that
I
forgot,
but
I
don't
think
they
were
blocking,
but
there
were
round
trip
latency
performance
tests
that
were
created.
I
think
the
reason
why
they
were
never
turned
on
by
default
instead
of
blocking
ci
is
it's
subject
to
all
the
whims
and
follies
of
providers,
and
but
there
are
end-to-end
tests,
or
at
least
there
were.
B
I
don't
know
if
there
still
are,
that
basically
do
sort
of
low-level
latency
tests,
and
I
believe
that
you
know
back
when
I
worked
at
red
hat.
We
actually
used
that
for
blocking
signal
for
ci
on
the
downstream
product
to
make
sure
that
there
wasn't
regression
in
certain
behaviors
for
the
cni
and
kernel,
so
that
might
be
interesting
from
a
cop's
perspective,
because
you're
only
you
primarily
focus
on
aws.
But
you
know:
that's
not
the
only
provider,
but
it
might
be
interesting
to
put
that
in
the
battery.
F
Yeah,
that's
a
great
idea.
Also.
We
actually
have
our
first
chaos
specific
tests,
we've
moved
to
cube
test
two
or
are
moving
to
cube
test
two,
and
so
we
can
actually
start
to
introduce
our
own
tests
now
and
can
do
something
like
that.
The
other
thing
I
thought
of
is
there
is
a.
I
don't
know
if
it's
a
scalability
test,
but
there
is
a
test
which
at
least
produces
numbers
on
api
server.
Latency
again,
I
don't
think
it's
blocking,
but
I
don't.
F
Maybe
we
could
borrow
some
of
their
machinery
to
sort
of
track
regressions
there
yeah.
B
B
All
right
next
up
sheriff
sharif.
C
Yeah
yeah,
so
mini
cube's
main
goals.
This
quarter
are
fully
supporting
arm
on
linux
and
darwin
platforms.
We
now
release
a
an
arm64
binary
for
linux
machines
and
the
other
one
is.
We
want
to
migrate
to
having
our
default
container
runtime
inside
of
a
inside
of
minicube
be
container
d
instead
of
docker,
but
still
allow
users
to
build
docker
images
like
they
expect
to
with
a
docker
demon.
C
B
Are
there
any
other
topics
that
folks
want
to
discuss,
because
we
are
kind
of
at
the
end
of
our
agenda
going
once
twice
three
times?
I
think
there's
there's
one
thing
I've
been
thinking
about
doing
for
a
while,
and
I
think
this
might
be
beneficial
is.
It
relates
to
the
outreach.
But
what
might
be
interesting,
which
is
something
we've
never
really
done?
We've
done
surveys,
but
one
thing
I
was
thinking
about
is:
maybe
the
idea
of
setting
up
like
a
listening
session.
B
It
might
be
interesting
to
set
up
as
a
sig
level
to
have
like
a
a
listening
session
with
people
who
are
deeply
invested
or
using
our
projects
kind
of
do
an
outreach
with
the
cncf
to
see
if,
like
we
can
get
direct
feedback
from
what's
what's
working
well,
what's
not
working
if
they
had
a
laundry
list
of
things,
just
a
a
smaller
sort
of.
What
am
I
thinking
a
focus
group
that
can
talk
about
the
pitfalls,
the
issues,
the
other
things
that
they
may
like
to
see?
F
A
Yeah,
I
think
I'll,
just
echo
kind
of
what
justin
said.
I
think
it's
a
great
idea.
The
only
thing
I
wonder
is:
does
it
make
sense
to
kind
of
scope
it
down
at
some
level
to
certain
sub-projects,
because
I
it
might
get
a
little
crazy
if
we
try
to
do
all
of
the
sub
projects
at
once
and
get
a
little
noisy
and
distracting.
B
So
it'd
be
interesting
to
get
like
their
take
on
why
they
made
certain
choices
and
stuff
all
right.
Remember.
D
The
first,
the
first
time
I
encountered
something
like
that
was
during
the
lps
working
group
session,
where
they
invited
certain
european
shoe
manufacturers
that
has
gigantic
coasters
in
europe,
and
it
was
probably
one
of
the
best
zoom
calls
in
kubernetes
I
ever
joined,
and
then
they
stopped
and
I
haven't
seen
any
other
sick-
do
something
like
that
or
a
group.
D
D
Well,
yeah,
something
else
is
that
I,
like
I
agree
with
jason
that
we
probably
shouldn't
unify
a
single.
You
know
a
60-minute
zoom
call
for
a
lot
of
different
su-project
consumers,
but
we
can
do
it.
I
mean.
Is
it's
probably
going
to
end
up
like
a
session
where
people
are
just
reporting,
cops
and
kubernetes,
or
something
like
that
which
we
don't
like
to
end
up
having?
So
maybe
we
can
a
way
to
approach.
D
Such
a
meeting
would
be
to
first
sort
of
qualify
a
certain
consumer,
as
you
know,
eligible
to
join
the
call
and
present
their
use
case
and
everything.
B
Right
it
sounds
like
we
should
probably
do
a
little
more
homework
on
this,
but
why
don't?
I
start
to
draft
a
document
here
because
I've,
I
think,
that
the
community
and
the
consumers
would
benefit
from
this
sort
of
style
of
engagement
because
a
lot
of
times
there's
there's
like
an
intermediary
or
a
set
of
intermediaries
in
between
where
you
don't
get
that
direct
feedback.
B
I
do
know
of
some
key
high
level
high
profile
customers,
who
are
consumers,
who
might
be
interested
in
doing
that,
like
the
the
t-mobile
folks,
they've,
already
sort
of
engaged
a
number
of
fronts,
there's
a
t,
there's
walmart
walmart
labs
so
and
there's
of
course
apple
now.
So
I
think
that
that
might
be
a
good
way
for
us
to
start
to
get
that
that
feedback,
loop,
cranking.
B
Right
any
other
last
topics,
good
months
twice
three
times
take
care.
Everybody.