►
From YouTube: KubeVirt Community Meeting 2022-07-06
Description
Meeting Notes: https://docs.google.com/document/d/1kyhpWlEPzZtQJSjJlAqhPcn3t0Mt_o0amhpuNPGs1Ls/
A
Hopefully,
you
can
see
my
screen.
Yes,
we
do
yeah
cool
cool,
in
which
case
welcome
everybody
good
afternoon
good
morning,
good
evening,
wherever
you
are,
this
is
the
cubert
community
meeting.
I
will
post
the
agenda
in
the
chat
and
you
can
please
add
your
name
to
the
attendees
list,
chat.
A
This
is
going
to
be
a
little
a
little
shaky
there.
We
go
all
right.
Do
we
have
anyone
new
on
the
call
that
would
like
to
that?
This
is
their
first
keyboard
community
meeting
and
they
would
like
to
introduce
themselves
with
a
few
short
words.
B
I
I'm
alay
I'll
be
working
with
cuboid
for
some
time
now
I
I've
been
attending
these
calls,
silently
just
to
listen
in
but
yeah
from
now
on
I'll,
be
a
little
more
active,
so
just
wanted
to
say,
hi.
B
C
A
Going
once
all
right,
I've
got
a
couple
of
really
quick
things.
One
is
that
everyone
who's
attended
any
meeting
through
the
month
of
june
will
know
that
we've
been
attempting
to
get
rid
of
the
duplicate
cubot
calendar
entries
they
should
now
all
have
been
removed.
A
That
is
all
another
thing,
and
I
realized
as
I'm
starting
this
up,
that
I
should
send
this
out
as
an
email,
so
I
will
and
give
the
duplication,
but
in
short,
a
few
of
us
met
with
the
cncf
technical
documentation
or
technical
writing
special
interest
group
who
have
a
monthly
meeting
to
help
out.
Whoever
turns
up
answer
the
technical
writing
questions
and
we
talked
about,
lastly,
restructuring
or
potentially
restructuring
a
user
guide
to
help
it
sort.
A
We
were
fortunate
enough
to
have
a
new
user
who'd
just
gone
through
the
experience
of
of
trying
to
get
involved
in
qbert
and
using
the
user
guide
and
having
a
lot
of
struggles
with
that,
and
so
we
were
able
to
have
a
good
chat
with
him
and
a
technical
writing
team,
and
so
they,
they
kind
of
generally
recommend
that
quick
start
material
is
the
best
thing
to
focus
on
to
start
with,
and
so
that's
something
I've
created
an
issue,
but
I
think
I've
posted
in
the
engineering
to
tackle
some
of
this
and
in
doing
so
discovered
that
the
and
so
on,
that
raised
an
issue
the
same
day
that
the
catacota
website
is
no
longer
public.
A
We
have
some
cuber
labs
that
were
in
catacoda
that
are
no
longer
accessible,
which
means
you
go
to
a
website.
You
click
on
one
of
the
try
buttons
and
it
goes
down
if
anyone's
here,
that's
familiar
with
our
website.
If
you
could,
please
take
a
look
at
that,
my
recommendation
would
be.
A
Maybe
we
still
need
to
look
into
potentially
migrating
to
an
alternative,
such
as
kilakoda
was
suggested
and
then
using
the
catacota
repo
issue
or,
if
you
wanna,
do
something
else,
so
maybe
not
totally
remove
it,
but
if
we
could
take
it
off
the
website
that
would,
I
guess,
prevent
people
from
from
going
to
a
dead
link,
yeah
happy
to
expand
on
any
of
that,
apparently
at
once,
but
also
happy
to
move
on
to
the
next
point.
D
Okay,
well,
I
can
take
on
the
responsibility
of
trying
to
clean
up
the
katakota
stuff
this.
This
is
chandler
by
the
way,
oh
hey,
so
I
I'd
most
recently
kind
of
taken
over
ownership
of
the
catacotta
scenarios.
D
If
it's
an
easy
one-to-one
transition
to
killer
coda,
then
maybe
we
can
skip
the
step
of
removing
it
from
the
website,
but
I
think
to
you
know:
if
it
takes
more
than
a
week,
then
it
probably
should
be
off
the
website.
In
the
meantime,.
A
Right
on
the
the
the
person
that
raised
the
issue
that
I've
got
here,
issue
34
was
kind
enough
to
link
to
a
like
how
to
migrate,
and
it
seems
to
be
just
relatively
simple.
Just
a
couple
of
steps,
all
right
I'll
go
check
that
out
next.
A
Man,
which
case
alay
you've
got
something
in
the
agenda.
Would
you
like
to
please.
B
Sure
so
I
was
running
into
a
bug.
I.
D
B
I
I'm
not
sure
if
this
is
a
bug,
but
I
was
running
into
a
scenario
where,
if
I,
if
I
try
to
create
a
bmi
with
bad
spec,
dot
affinity
yaml
through
cube
cuddle,
it
complains
about
not
being
a
valid
ammo.
But
if
it
happens
through
ansible
kate's
module,
then
it
does
not
complain.
It
just
allows
it
to
get
through.
I
learned
that
ansible
kids
module
don't
do
client-side
validation,
that
is
the
open
api,
crd,
spec
validation
on
the
client
side.
B
So
because
this
and
I
found
out
that
it
was
not
implemented
on
this
website.
So
I
was
just
wondering
if
this
is
a
bug-
or
this
is
intentional.
I
was
wondering
if
I
can
create
an
issue
to
you,
know,
try
and
fix
this,
but
wanted
to
get
the
community's
opinion
on
this.
D
B
So
I
know
that
deployment
does
not
go
through
with
the
template
same
spec.template.f
value
and
in
fact,
when
I
did
cube
cuddle
create
minus
minus
raw.
That
is,
it
does
explicit
post
request
against
the
api
server.
It
did
not
fail
the
validation
so
it
allowed
because
it
was
not
going
through
the
client
side
validation.
B
It
went
right,
so
I
am
unsure.
If
this
is
a
ansible
case
thing,
I
think,
by
design
and
spokane
skips
client-side
validation,
because
they
don't
want
to
import
all
the
python
client
data
structures
and
maintain
it
over
time.
So
they're
using
dynamic
client
which
skips
over
the
client
side,
validation,.
D
Okay,
that
makes
sense,
and
so
it
sounds
more
like
and
the
ansible
philosophy
in
general
is
you
should
vet
your
yaml
some
other
way
before
you
go
into
ansible?
Yes,
and
I.
B
B
So
what
would
be
the
path
forward
from
here
on?
Should
I
create
an
issue
and
try
to
follow
up
with
some
kind
of
server-side
proposal,
server-side
validation
proposal
for
this
I
have
walked
through
the
deployment,
validator
and
potentially
cubot
vmi
validator
can
also
use
a
similar,
similar
functionality.
D
A
Okay,
creating
an
issue
as
a
starting
point
for
creating
a
large
proposal
or
a
pr,
I
think,
is
the
right
way
to
go,
and
then
people
can
put
their
thoughts
on
before
getting
into
any
deeper
work.
As
to
you
know
whether
it's
a
good
starting
point,
that's
my
understanding
anyway.
A
Thank
you
unless
anyone
has
anything
else
to
add,
we'll
go
to
underrate's
point
of
that.
Gpu
live
migration.
Thank
you.
F
F
Since
the
nvidia
already
released
them
kernel
drivers,
we
are
already
investigating
how
to
develop
the
code
on
top
of
converge.
To
do
gpu
live
migration.
Any
help
is
welcome.
We
are
in
the
investigation
phase
right
now.
I
already
talked
with
ryan
from
nvidia.
That
is
part
of
this
community
here,
and
he
say
nvidia
is
doing
nothing
at
all
today.
To
do
that,
give
me
some
ideas,
but
any
ideas
are
very
welcome.
Now
we
are
investigating
how
to
do
that
in
a
proper
way.
F
G
F
Because,
on
our
case,
all
the
servers
has
the
same
hardware
like
four
nics
four
gpus
each
and
then
we
plan
to
migrate
the
users
and
attach
to
when
you
create
the
other
pod.
The
same
way
you
when
you
have
different
mac
address,
you
do
something
regarding
re-attach
the
the
network
you
we
need
to.
We
would
like
to
reattach
the
gpus
the
same
way
you
do
on
on
network
cards.
F
G
Yeah
that
sounds
reasonable,
but
how
the
virtual
machine
can
live
without
the
gpu
actually.
F
This
is
something
that
we
are
discussing
with
ryan
from
from
nvidia.
He
say
this
is
the
the
way
that
hyper-v
and
vmware
do's.
That
is
something
like
hibernate
the
vm
and
you
he
he.
This
is
reattached,
something
like
that.
Okay,.
G
So
this
is
more
likely
not
online,
not
live
migration,
but
like
offline
migration,
so
you
need
to
hibernate.
It
then
migrate,
I'm
not
sure
if
cubiert
is
supporting
it
right
now
am.
F
I
right
that's
why
we
are
looking
for
some
directions
and
since
I'm
talking
to
them
on
the
open
floor,
also
about,
can
I
talk
about
that.
F
We
are
trying
to
install
windows
11
on
top
of
of
convert
and
it
requires
tpn
version
2..
I
know
that
kvm
is
capable
to
do
tpm,
but
I
didn't
find
any
instructions
how
to
do
that.
Any
ideas
from
the
community
how
anyone
already
tried
to
install
windows
11
on
convert.
F
Anyone
has
tried
to
install
windows
11
on
top
of
convert
because
it
requires
tpm
version.
2
and
security.
Boot
is
already
there
enable,
but
then
so
far
so
on,
I
was
not
able
to
to
to
make
it
happen
over.
H
We
we
do
support
tpm
version
2
for
this
exact
purpose,
and
it
is
in
the
latest
cube
vert,
at
least
like
versions
beyond
release
53,
so
the
main
branch
does
have
it.
Do
you
know
specifically
why
it's
not
working?
Do
you
have
an
error
that
you're,
seeing
yes.
F
I'm
using
for
53
for
sure
53.1,
I
think
and
hope
when
I
enable
the
tpm
something
is
wrong.
Perhaps
you
can
give
me
the
the
ammo
to
to
to
make
it
happen.
D
H
Our
default
templates
should
make
that
work
out
of
the
box.
So
if
you're
running
into
issues
we
we
would
be
very
interested
in
finding
out
more
about
that.
F
H
Well,
we'll
follow
up
sorry,
I
had
missed
that
discussion,
so
we'll
go
check
to
make
sure
you're
getting
the
help
you
need.
G
I
F
Missed
I
talked
to
you
and
we
talked
about
gpu
live
migration
and
we
was
thinking
to
start
to
develop
it
since
nvidia
is
not
doing
that,
and
I
I
opened
the
floor
here
to
have
anyone
given
an
idea
how
to
to
make
it
happen
behind
the
scenes,
because
now
we
are
investigating
only
how
to
do
it.
F
I
Yeah
I
mean
I
I
think
like
like
what
I
was
saying
news
that
for
at
least
for
where,
where
I'm
at
right
now,
it's
like.
I
think
it's
that
I
wouldn't
say:
that's
it
not
so
we're
not
interested
it's
more
like.
We
have
a
bunch
of
things
that,
like
we
have
sort
of
in.
G
I
In
our
focus
right
now-
and
it's
like
this-
is
something
that
I
think
is
very
interesting
like
and
even
previously
I've
spoken
about.
Like
one
other
topic,
that's
interesting
to
us.
That's
sort
of
in
this
area
like
which
is
like
offline
virtual
machine
migration
with
pass-through
gpus,
which
is
something
that
was
something
that
we've
taken.
I
Some
interest
in
we've
worked
on
a
little
bit
in
the
community
and
had
some
discussions
and
there's
there's
already
been
some
design
around
this,
like
mike
hendrickson
and
a
few
folks
have
done
some
work
around
doing,
save
and
restore,
and
that's
something
like
that
was
that's
along
these
lines,
but
it's
it's
mostly
with
pass
pass-through
that
you
know
that
we
were
looking
at
to
to
do
this
with
some,
but
I
mean
I
think,
like
there's,
I
guess
the
point
is
like.
I
I
think
it
would
be
good
to
have
a
discussion
on
this.
I
just
need
to
get.
I
just
need
to
find
some
time
to
like
prioritize
some
of
this
stuff,
and
I
think
it's
valuable.
So
I
did
you.
Did
you
start
a?
I
think
you
started
something
on
the
mailing
list.
I
can
maybe
follow
you
there
on
just
what
my
thoughts
are
and
kind
of
what
we
can,
how
we
can
progress
this.
F
I
Sure,
okay,
yeah,
we'll
follow
up
with
you
on
the
mailing
list.
We'll
start
a
discussion
there:
okay,
wonderful.
H
Hey
andre
looking
at
the
mailing
list,
I'm
I'm
not
seeing
anything
any
queries
from
you
regarding
windows,
11
on
there,
I'm
only
seeing
four
different
emails,
one
title
press
any
key
to
boot
from
cd
keyword,
rambus,
deduplication,
yeah,
I'm.
F
Missing
one,
let
me
see
if
it's
on
this
on
the
convert,
maybe
list
or
I
put
on
kvm
daily
list.
I
don't
remember
exactly
if
I
don't
put
here,
I
I
gonna
open
a
new
one
right
now:
okay,
okay,
thank
you.
A
All
right
so
so
I.
A
There's
the
tpm
2.0
as
well,
which
takes
us
down
to
the
ad
proposal,
live
vibrations
abridged.
What
network.
G
Actually,
I
was
telling
about
this
feature
before
now.
I
just
prepared
the
proposal
and
I
want
to
take
your
attention
to
look
into
it
if
anybody
interesting
in
life,
migration
for
the
bridget
support
network,
please
take
a
look.
G
The
only
problem
I
faced
is
the
integration
test,
which
is
running
it's
like
operator
test
for
updating,
cube,
weird.
If
I
understood
it,
if
I
understood
it
well
and
right
now,
there
are
no
special
condition
for
non-migratable
virtual
machines,
I'm
trying
to
use
cgrom
and
the
virtual
machine
with
the
read-only
cd-rom,
which
is
shouldn't,
be
able
to
live
migrate.
G
A
Jordan,
sweet
eskegon
but
open
floor
about
separate
keyboard,
keyboard
storage.
C
Yeah
hi,
so
there
is
this
initiative
of
separating
the
storage
code
to
storage
package,
similar
to
what
the
network
team
did
with
the
network
package.
So
alicia,
and
I
were
talking
about
it-
how
how
we
can
do
it?
What
should
be
done?
C
What
functionalities
should
be
moved
like
what
would
be
the
best
way
to
to
review
the
storage
code,
because
it's
pretty
much
everywhere
from
controllers
to
client
commands
to
launcher
the
hot
plug
is
all
over
the
place
so
like
I
started
with
moving
the
snapshot
and
restore
and
some
utils
for
pvcs
and
data
volumes
in
the
pr
I
attached
here,
but
we
were
wondering
if
you
have
any
opinions
about
the
best
way
to
do
it.
C
What
should
be
moved?
What
shouldn't
be
moved
stuff
like
that.
I
One
question
for
you:
so
you
so
you're
trying
to
move
all
the
storage
code
out
of
the
keyboard
keyword
repo!
Is
that
what
you're
trying
to
do.
I
See,
okay,
I
yeah
I
have
to
okay.
I
don't
I'm
not
really
an
expert
on
this.
I
mean
this.
I
don't
know
if
mike's
around,
but
it
would
be
good
to
talk
to
mike
hendrickson,
see.
Maybe
he's
got
an
opinion
on
this
or
alexander,
maybe.
C
Like
I'm
not
sure,
regarding,
like
the
whole
albert
launcher,
stop
stuff,
I
think
alicia
and
I
have
agreed
that
for
that,
like,
for
example,
that
should
remain
where
it
is,
but,
like
all
all
the
different
disks,
are
they
considered
storage
or
not,
because
I'm
I'm
not
sure
if
they
are
like
they're,
not
coded.
Our
storage
team
wrote
exclusively,
for
example,.
J
So,
just
to
generalize
a
bit
more
the
the
question:
what
could
make
the
life
easier
for
the
community?
J
I
don't
know
it
happens
that
there
are
some
storage
related
bugs
and
maybe
it
would
be
easier
if
we
can
group,
I
don't
know,
even
feature
under
the
same
package
to
be
able
to
identify
better
the
reviewers.
I
don't
know,
maybe
if
something
is
changed
under
that
package,
maybe
just
being
people
like
charlie,
for
example,.
J
J
J
I
don't
know,
try
to
group
storage
functionality
and
but
we
also
need
to
define
exactly
what
are
those
yeah.
E
I
can
understand
what
you
mean
about
taking
responsibility
for
more
than
stories
than
you
have
before,
like,
for
example,
the
container
disk,
whether
you
should
put
it
up
to
storage
more,
like
primitive
storage
features
like
if
you
should
put
it
under
storage.
E
I
don't
have
a
solid
opinion
about
whether
it
will
be
like
more
easier
for
me
if
this
will
be
grouped
under
storage
or
not,
but
I
can
see
one
clear
advantage
is
that
we
can
reduce
the
boiler
plate
of
the
ci
jobs.
If,
for
example,
everything
will
be
consolidated
at
the
wrong
place
in
storage,
then
we
can
go
around
if
changed,
and
then
the
lane
would
need
to
be
run
like
for
every
each
and
every
change.
A
I
I
don't
know
that
I
suspect
that
some
of
the
more
opinionated
people
on
this
subject
are
not
on
this
call
at
the
moment.
Is
this
worthwhile
putting
to
the
cuba
dead
mailing
list
for
toronto
as
an
opinions,
fair.
C
Yeah
I
talked
with
michael
personally.
He
is
not
so
strong
about
this
initiative,
it's
something
that
duncansburg
is
really
into
so
we're
trying
to
look
if
it's
something
that
we
really
want
to
do
or
what's
the
opinion
of
the.
J
I
mean
the
other
alternative
would
be
that
to
do
what
shall
you
already
did
in
the
pr
and
then
maybe
for
a
future
feature?
Then
we
can
try
to
to
put
it
there,
because
I
think
one
one
of
the
other
part
that
cubert
does,
that
could
be
considered.
Storage
are
all
the
disk
type,
but
again
not
sure
if
moving
those
in
the
subdirectory
trillion.
E
Maybe
let's
have
a
more
technical
discussion
to
schedule.
It.
A
One
thing
we
could
do
as
well
is:
we
could
set
aside
some
time
specifically
that
in
next
week's
community
meeting
and
if
we
send
an
email
out
now,
the
the
people
that
have
opinions-
or
this
may
impact
in
the
broader
community
can
then
attend
that
meeting.
You
know
with
advanced
notice
just
about
it.
B
Hi
this
is
again
I
am
trying
to
get
started
to
run,
make
commands
like
make
cluster
up
and
make
clusters
inc
with
doc
d
ind,
but
I'm
using
a
podman
environment
in
in
os
x.
I
was
wondering
if
anyone
has
anyone
here
has
got
that
working
I'm
gone.
I
I
hit
a
couple
snags
and
then,
after
that,
I'm
hitting
an
exec
error
for
the
arch
64
environment.
G
To
be
honest,
I
was
trying
to
run
this
make
command
on
my
os6
with
the
docker
on
for
desktop
and
m1
processor,
and
it
wasn't
working
to
me
I'm
not
sure
if
podman
will
work
as
well.
B
Okay,
that
is
helpful
data
point.
Thank
you.
A
Right
that
brings
us
to
the
end
of
the
generator
floor
parts,
real,
quick
scrub.
A
A
A
Next
week,
one
thing
that
occurred
to
me
with
this
kind
of
I
think,
would
it
be
helpful
for
future
community
meetings
if,
before
the
meeting
or
myself
went
through
these
pull
requests
and
lined
them
up
here
like
a
couple
hours
beforehand.
Rather
than
doing
this
in
the
meeting.
E
Yeah,
I
think
it's
better
to
prepare
it
before
ahead
of
time.
H
Windows
11
install
question.
You
know,
I
see
the
email
now.
Thank
you
for
sending
that
now
that
I
see
what
you're
asking.
Are
you
asking
how
to
enable
the
tpm?
Because
we
do
have
that.
A
Looks
like
everyone
else
has
replies
yeah
a
lot
of
moving
on
threads
this
week.
I've
noticed.
A
A
And
I'll
be
in
my
clean
up
in
the
low
generator
tool
rather
than
trying
to
address
these,
because
I
don't
really
know
much
about
anything
I'll,
just
kind
of
like
bring
them
up
here,
read
out
the
heading
and
then,
if
someone,
if
that
kind
of
like
interests,
anyone
by
all
means
can
continue
reading
it
or
answer
and
respond.
A
A
A
B
Mate
which
the
handle
vmi
cleanup-
oh
right,
I.
B
Okay,
no
problem
yeah,
I'm
working
on
trying
to
test
that
draft
pr
and
that's
where
I'm
hitting
issues
with
trying
to
make
trying
to
get
the
make
commands
working.
So
once
I
have
that
working,
I
will
convert
that
into
pr.
A
Thank
you.
That
sounds
great.
I
totally
lost
it
yeah
and
I
hope
that's
okay,
it's
something
in
your
draft.
I
presume
that
means
that
we
don't
need
to
necessarily
focus
on.
A
A
Basically
brings
us
to
the
end:
is
there
anything
else
that
anyone
would
like
to
to
bring
up
or
or
return
to
in
the
the
bug,
scrub
mailing
list
review
or
pull
request?
Any
discussion
that
they
could
dig
deeper
into?
Rather
than
ending?
The
meeting
now
clicks
hollow.
A
All
righty,
in
which
case
I
get
back
18
minutes
of
your
life
back
one
final
thing
we
go.
I
am
trying
to
put
together
some
stuff
for
kubecon
north
america,
so
if
there
is
anyone
on
the
call
that
is
highly
likely
to
attend,
please
let
me
know
it
would
be
very
helpful
to
help
coordinate.
A
Apart
from
that,
I
hope
everyone
has
a
lovely
day
and
we'll
see
when,
when
is
it
it?
It
is
in
detroit
and
it
is
in
stop
my
head.
It
is
the
end
of
october.
I
think
it's
I
should
know
this.
I
think
it's
the
22nd
to
the
27th
of
october.
F
Okay,
before
you
understand,
I
think
we
are
one
of
the
largest
implementation
of
convert
modern
hunter
cluster
with
1250
nodes
per
cluster.
Before
you
understand
that.
A
A
Right,
sorry,
who
are
you
representing
sorry.
F
A
Fantastic
and
that's
that's
yourself
and
others
or
yeah.
A
F
A
Excellent
well
I'll
I'll.
Keep
that
in
mind
I'll
keep
in
the
loop
for
for
our
presence
and
keep
it
at
kubecon.
A
Right
on
and
we'll,
I
think
we'll
leave
it
at
that.
Thank
you
for
everyone
for
attending,
see
you
next
week
and
have
a
lovely
day.