►
From YouTube: SIG Cluster Lifecycle - Office Hours - 20230124
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
Yeah
I
I,
don't
know
if
anyone's
aware
of
this,
but
yeah
there's
there's
an
outstanding
there's,
an
outstanding
issue
right
now,
with
with
kubadium
so
I
I.
Think
from
from
what
I
understand,
anyone
that
is
installing
I
guess
yeah
running
kubadium
today,
kubadium
install
is
there
I,
mean
sorry
in
it
join
and
running
a
container
runtime
that
does
not
support
CRI
V1
API,
so
this
could
be
container
D,
for
example,
1.5
and
earlier
or
sorry
earlier
than
1.6.
B
Kubidium
will
fail
because
it
will
sorry
coup
DDM
installed
from
a
system
package
that
is
maintained
by
the
kubernetes
project,
so
a
Deb
or
an
RPM
package.
B
If
you
install
yeah
a
qubidium
version,
that's
so
yeah
lower
than
one
like
lower
than
any
1.26
release,
you're
not
required
to
have
a
container
runtime
that
that
Sports
V1,
and
so,
if
you,
if
you,
if
you're
running
with
a
container
runtime,
that
only
supports
for
like
V1
Alpha,
two
kubidium
will
fail
because
it
the
package
pulls
in
a
dependency
CRI
tools
which
is
just
given
a
a
lower
bound.
And
so
it's
effectively
to
just
get
the
latest
version.
B
Version
126
remove
support
for
earlier
for
for
anything,
but
the
v1cri
API.
So
I
don't
even
know
if
this
is
really
the
Forum,
because
this
I
guess
is
is
owned.
These
packages
are
owned
by
the
release
Sig
and
it's
in
the
release
Repository,
but
I
yeah,
I
I
think
this
is.
B
This
is
I,
there's
a
problem.
There's
a
problem
here
that
affects
all
you
know
all
such
clusters
in
the
wild.
That's
not
that
doesn't
seem
to
be
very
good.
You
know
we
haven't
I
I
haven't
seen
I,
guess
a
lot
of
a
lot
of
noise
around
this.
B
B
B
You
know
this
this,
this
sort
of
change,
breaks
them
and
I
I
think
it's
also
worth
looking
at
the
way
these
dependencies
are
defined.
Oh,
you
know
why
there's
just
a
a
lower
bound,
as
opposed
to-
let's
say
you
know,
a
pinned
version
or
or
a
range
with
an
upper
bound.
D
B
That's
correct,
that's
correct,
yeah
Livermore
commented
earlier
somewhere
on
this
issue,
so
I
think
he's
he's
aware
of
it,
but
yeah.
A
I
did
see
his
comment.
We
can
just
wait
for
I,
like
I'm
I'm,
not
sure
why
it's
bullying
CRI
tools,
what
if
this
is
I'm
I
suspected
like
most
maybe
most
people
like
won't
use
the
aptee
repositories
for
to
install
things,
but
I
also
I'm,
not
sure-
and
maybe
this
is
not
a
I
guess,
like
I,
haven't
looked
at
the
emitter
builder
for
a
while.
Like
do
you
know
if
this
is
going
to
be
a
problem
for
image
Builder,
it.
C
B
B
and
although
kubadium
122
is
perfectly
happy
right
with
something
like
containerdy
one.
Five
CRI
tools
126
would
not
be
because
it's
removed
support
for
this
V1
Alpha,
2,
CRI
API.
B
B
Does
affect
older
Cuban
men
versions
and
then
the
the
reasons
as
far
as
I
understand
that
there
is
a
dependency
at
all
is
because
kubernet
actually
exacts
out
and
and
calls
the
CRI
CTL
CRI
cuddle
binary
to
perform
some
image
related
tasks,
pulling
an
image
listing
images,
I
think.
C
A
C
A
Yeah
I,
don't
think
we're
actually
using
the
rpt
repositories
either
in
here
for
for
this
specifically
for
the
qubit
kubernetes
binaries,
but
I
might
be
wrong.
I
haven't
looked
at
this
in
forever
and
it
has
changed
a
bunch.
Oh
yeah
I
mean
there
is
it's
downloaded
at
least
this
year
or
the
cni
turbo
and
yeah
that
so
this
is
not
using,
but
this
should
be
fine.
If
you
know,
if
you
use
imageable
there,
which
not
a
lot
of
people
do.
C
Yeah
lubomir
is
the
right
person
to
comment
on
this.
One.
B
And
is
this
something
you
know
this,
this
packaging
and
the
and
the
release
repo
is
that
is
this
all
under
the
release?
Sig,
or
does
anybody
know
exactly
how
that
how
that's
organized
it.
C
Is
historically
because
they
own
it
at
all
the
machinery
and
yeah?
This
is
not
the
first
time
that
this
buys
up
us,
because
basically
I
am
not
aware
of
what
happened
there
and
sometimes
you
just
break
and
you
discover
it
all.
B
B
B
D
Got
the
next
one
sure
thank
you,
so
this
is
like
in
the
spirit
of
like
cross
project
things.
One
of
the
problems
we
have
in
chaos
is
that
we
ship
a
bunch
of
manifests
for
like
cluster
level
components
and
we
recently
discovered,
like
like
cloud
provider
gcp
and
cop
provider
AWS,
and
you
know
other
things
like
that
core
DNS
things
like
that.
D
We
recently
discovered
that
the
Manifest
that
we
were
shipping
is
not
an
exact
match
for
what
cloud
provider
gcp
is
testing
and
that
they're
actually
bugs
in
in
the
one
which
we
were
shipping
and
the
one
that
they're
testing
so
we're
trying
to
like
align
those
and
basically
I'm
trying
to
get
the
idea
that
the
the
projects
like
cloud
provider
gcp
would
create
a
sort
of
supported
manifest
and
that
they
could
then
test
that
with
chaos
or
kubadm
or
whatever
my
cluster
API,
whatever
it
might
be.
D
Obviously,
I
started
with
chaops,
because
it's
the
one
I
know
best,
but
there's
nothing
really
specific
to
chaos
there
other
than
the
idea
that
we
need
some
way
to
communicate
hey.
This
is
the
Manifest
we
want
to
test
with
so
override
injector
manifest.
D
So
if
I
don't
know,
if
kubernetes
has
a
built-in
manifold
I,
don't
think
kubernet
has
a
built-in
manifest
for
cloud
provider,
gcp
but
you'd,
say:
hey
use
this
particular
manifest
in
there's
a
script
that
brings
up
a
cluster
and
so
that
script
would
use
the
tool
whatever
the
tool
might
be,
and
then
would
inject
into
that
tool.
The
the
current
development
version
of
cloud
provider
gcp
or
whatever
it
might
be
so
I'm
hoping
this
can
this
can
be
a
sort
of
it's
another
approach.
D
Yet
another
approach
to
like
the
cluster
add-ons
problem
from,
in
addition
to
writing,
operators
or
complementary
to
writing
operators.
We
can
also
maybe
try
to
get
some
benefit
to
the
some
direct
benefit
to
the
projects
themselves,
where
they're
able
to
test
in
conjunction
with
everything
else,
with
our
well-defined
manifest
is,
is
the
thing
that
I
hope
that
they
the
outcome,
the
outcome
that
I'm
hoping
for
so
yeah.
D
If
you
want
to
have
a
look
at
that
PR
and
let
us
know,
let
me
know
if
there's
anything
too
chaos
specific
or
anything
like
that,
or
if
you
know
that
you
want
to
weigh
in
in
any
way
I
just
want
to
make
people
aware
of
that
thing.
That
I
was
working
on.
A
Thanks
Justin
any
other
comments
on
these
two
topics.
A
A
C
B
C
There
are
Forks
that
are
working
on
torques
for
cluster
pi
I'm,
probably
for
cup
Z,
as
of
as
far
as
I
know,
I
have
pink
at
above
them
a.
If
you
need
assistance
you
have
to
come
to
us
because
we
are,
they
cannot
submit
it
basically
without
RS.
So.
D
I
actually
I
just
checks
like
Lumiere,
did
submit
a
session
for
Amsterdam
introduction
to
Cluster
lifecycle.
He
added
me
as
co-presenter,
but
he
and
I
are
happy
to
swap
with
anyone
else
if
they
want
I
think
everyone's
a
little
bit
like
watching
on
what
happens
with
travel,
but
it
sounds
like
you've
been
here
to
submit
one
introduction
in
the
intro
to
say
cluster
lifecycle,
one
and
I
will
I
will
I'm
trying
to
think
of
what
the
right
thing
to
do
is
for
chaops.
D
If,
if
we
want
to
submit
one
there,
there's
also
the
developer
event
as
well
the
developer
day,
whatever
that's
called
I,
can't
remember,
contributor
event.
Thank
you.
C
Yes
to
do
so,
and
it
is,
it
should
be
a
talk
about
the
customer,
Pi
release
team.
C
C
No,
it
is,
it
is
for
the
maintenance
track.
Why?
Why
not?
It
is
something
that
we
do
as
a
maintainer.
We
are
learning.
It
is
opportunity
for
others
to
contribute.
Okay,.
C
A
Yeah,
it's
like
literally
this
week,
so
yeah
we
got
a.
But
if
one
of
one
of
you
want
to
do
the
Deep
dive,
we
can
also
probably
submit
the
Deep
dive
and
yeah
I'm
gonna
figure
it
out.
A
A
A
I
think
how
things
are
going
like
I
do
see
like
specific
what
the
cluster
API
project,
like
I've,
seen
like
lots
and
lots
of
companies
that
share
like
how
they're
using
it
like
Fleet
Management
at
this
point
right.
It's
like
it's
like
Fleet
life
cycle
management
like
across
like
a
lot
of
clusters,
and
so
it
would
be
interesting
to
think
through
and
like
maybe
just
collect
some
collective
thoughts
on
what
we
think.
A
D
That
could
be
a
good
coupon
talk
or
session.
Another
talk
listening
session
or
something.
A
Yeah
I
might
miss
this
coupon
this
year,
so
no
we'll
see,
but
you
know
if
we
can
collect
them
and
maybe
talk
to
another,
or
we
can
also
do
like
the
async
on
the
mailing
list.
I
can
push
this
out.
A
I
know,
for
you
know,
cross-collaboration
with
API
Machinery.
It's
like
in
controller
runtime
personally,
like
I'm,
looking
to
add
more
multi-cluster
support
and
control
runtime
soon
and
that's
kind
of
like
a
stepping
stone
to
say
like
okay,
it's
like
what
is
what
does
it
mean
to
do?
Fleet,
lifecycle
management.
Now
it's
like
and
kind
of
ask
these
questions.
Given
the
cluster
API
has
made
it
very
simple:
to
manage
a
single
cluster,
maybe
a
few
more
clusters.
A
What
happens
when
we
go
to
like
thousands
of
costumes
right,
it's
like
that
things
get
very
complicated,
very,
very
quickly.
I'd
say
even
20
glosses
like
I,
think
they're
getting
very
complicated
at
some
point,
especially
if
a
devops
team
is
you
know
we
don't
have
to
discuss
it
now,
it's
more
like
we
can
just
think
about
it,
share
our
thoughts.
You
know
things
like
that.
D
If
we,
we
should
try
to
have
a
lunch
or
something
or
informal
lunch
for
anyone
like
that,
does
make
it
to
Amsterdam,
although
it's
even
shorter
right,
it's
only
three
days
so
I
guess
it
was
only
three
days,
but
it
seems
like
we've
compacted
it.
So
it
feels
like
the
the
extra
day
that
was
there
is
no
longer
there.
But
yes,
we
should
try
to
do
something
like
that.
A
Cool
well
thanks
folks,
have
a
great
rest
of
the
I,
can't
believe
we're
already
at
the
end.
A
At
this
point,
but
have
a
great
rest
of
the
week
and
we'll.