►
From YouTube: KubeVirt Community Meeting 2021-06-02
Description
Meeting Notes: https://docs.google.com/document/d/1kyhpWlEPzZtQJSjJlAqhPcn3t0Mt_o0amhpuNPGs1Ls/edit#heading=h.pulua96jjnp7
A
Hello,
everybody-
I
am
chris
caligari,
your
host
and
community
organizer
for
kubert,
and
this
is
our
weekly
meeting
where
we
talk
about
usa,
coover,
usage
issues
and
anything
you
would
like
to
talk
to
us
about.
A
We
have
a
pretty
slim
agenda
for
this
week,
so
far
feel
free
to
add
a
bullet
point,
and
I
have
got
the
first.
The
first
agenda
item,
which
is
the
most
important
thing
to
happen
to
project
cooper
in
quite
a
while.
A
This
is,
it
involves
filling
out
some
paperwork
and
and
doing
defending
our
dissertation
in
front
of
the
cncf
toc
committee.
I
don't
know
how
intense
of
a
process
this
is
gonna
be,
so
I'm
just
feeling
my
way
along.
I
have
a
github
issue
on
on
this.
If
anybody
would
like
to.
A
A
All
right
so
there's
our
issue
if
you
want
to
want
to
help
out
or
follow
along,
feel
free
to
to
tag
yourself
or
or
post
commits
to
the
poll
request,
and
we
can
discuss
it
through
the
issue
in
the
pull
request.
A
Looks
like
we
have
some
folks
editing
the
notes
so
I'll
give
you
a
minute
or
two
to
fill
in
your.
A
A
A
Oh
okay,
nothing
on
the
agenda
and
I
also
have
the
first
couple
items
in
the
open
floor.
B
A
Well,
I
saw
a
news
article
talking
about
the
nvidia
dhx
super
computer
this
morning.
I
have
nothing
more
to
say
about
this
other
than
I
I
literally
saw.
The
headline
I
was
hoping
ryan
would
be
here
this
morning
to
see
if
he
could
give
us
some
deeper
insight
to
what
they
had
going
on
here.
A
So
enough
about
that.
Moving
on
to
events,
I
had
a
meeting
with
the
red
hat,
open
source
op
program
office
yesterday
and
they're
not
happy
with
with
the
project
not
submitting
a
paper
for
kubecon
n,
a
and
kvm
forum
to
our
defense.
Kvm
forum
had
a
hard
line
on
being
an
in-person
event
up
until
about
two
weeks
ago,
so
that
didn't
give
us
much
runway
to
to
prepare
a
presentation.
A
Since
everybody
just
everybody,
I
spoke
to
put
up
their
hands
and
said
they
were
uncomfortable
with
the
in-person
meeting
or
in
personal
conference,
and
it
would
have
been
an
international
travel
since
kvm
forum
was
going
to
be
in
dublin,
ireland.
A
A
If
anybody
is
talking
to
him
today,
please
have
him
ping
me:
we
need
to
figure
out
what's
going
on,
if,
if
we
have
something
for
that
conference,.
A
We
all
have
our
hardware
and
we're
working
on
getting
that
thing
set
up.
Personally,
I
caused
a
five-hour
network
outage
in
my
house
trying
to
set
up
layer,
2
vlans.
A
My
it's
for
science.
A
The
worst
part
about
it
is
my
kid
was
knocked
out
of
virtual
schooling
all
day,
and
so
we
had
some
explaining
to
do
to
the
school.
D
You
know,
but
on
a
happy
note,
on
the
all
things
open:
it's
been
a
couple
weeks
now,
since
I
played
with
it,
but
I
did
get
cube
vert
running
on
the
raspberry
pi
and
virtual
machines
that
were
arm
based
running
on
the
cube,
for
instance
inside
are
you
are.
D
C
While
we're
at
that,
maybe
that
would
be
something
we
could
discuss.
I
don't
know
if
people
are
here
that
know
about,
but
if
we
could
get
multi-arch
images
published
for
keyboard.
D
We
will
need
to
get
multi-arch
images,
but
we're
also
going
to
need
to
get
kubert
able
to
deploy
multi-arch
images
to
a
hybrid
cluster,
because
that
doesn't
exist
right
now.
Oh.
C
But
it's
like
interest.
We
found
some
interesting
steps
when
you
take
we
for
the
first
time
we
learned
how
to
build
arm
keyword.
I
think
that
was
already
there,
but
we
didn't
know
how
to
do
it
and.
A
Yeah,
you
guys
are
way
ahead
of
me.
I'm
still
messing
around
in
the
with
the
pipes
trying
not
to
allow
the
internet
into
my
network.
A
Oh,
it
looks
like
vladik
has
posted
a
comment
to
kubecon
n,
a
and
kvm
forum,
declarative
api
to
configure
and
assign
virtual
gpus
to
vms
with
cooper.
Thank
you
very
much
for
that
flattick.
A
A
And
if,
if
you
would
like,
please
create
an
issue
in
the
in
the
community
repo,
so
we
can
track
that
that
work.
C
C
We
added
a
a
periodic
ci
job
for
sysprep,
and
the
discussion
came
up
on
on
how
to
notify
the
right
people
if
that
breaks,
because
it
runs
once
a
day
and
fabian
brought
up
that
we
should
do
an
email
list,
so
we're
looking
at
that
and
and
that
might
like
towards
chris-
maybe
that
we
could
maybe
we
need
some
documentation
for
people
on
how
to
create
keyboard,
related
emailing
lists
or
request
them.
C
Maybe
maybe
we
should
manage
them
centrally,
because
permissions
and
stuff,
and
in
general,
like
if
we,
how,
if
anybody,
has
an
idea
on
how
to
manage
those
notifications.
So
we
don't
lose
track
of
periodic
jobs
that
everybody
might
care
about.
A
Yeah,
I'm
actually
a
half
a
step
ahead
of
you
on
that.
I've
talked
with
josh
burkus
about
how
to
manage
mailing
lists,
because
we
recently
changed
the
coober
dev
mailing
list
ownership
from
the
red
hat,
provided
google
account
to
the
cncf,
provided
google
account
and
hooray.
We
didn't
lose
the
mailing
list
during
that
process,
so
that
brought
up
the
question
of
how
do
we
create
additional
mailing
lists
and
because
we
are
a
sandbox
project,
we
have
limited
resources
from
the
cncf.
A
C
And
like
fabian
at
least
explicitly
requested
that
we
do
it
in
the
mailing
list.
I
don't
know
why
that
might
be
up
for
debate
for.
C
Yeah,
I
I
I
don't
know
but
or
not
everybody's.
On
slack
I
don't
and-
and
I
to
fair
I
always
thought
the
mailing
list
to
be
more.
The
official
communication
channel
in
the
slack
channel
so
far
yeah
and
the
slack
channels
are
moderated
by
the
quinase
community
or
cncf.
What
I
know
from
kubernetes
like
starting
a
workgroup
there
guideline
is
just
created
public
google
group,
which
is
a
mailing
list
and
give
the
right
permissions
to
some
cncf
account
to
manage
it
in
case
you
get
lost,
but
yeah.
That's
that's.
C
E
Yeah,
currently
we
have
a
slack
channel
in
which
we
receive
that
periodic
notification,
when
our
periodic,
but
yeah
it
is
super,
noisy
and
and
maybe
having
some
something
specific
for
for
for
specific
jobs
and
letting
people
to
to
subscribe
to
this
made
in
this
would
make
more
sense.
This
is
what
we
talked
about
with
fabian
when
preparing
this
this
job.
We
have
created
even
one
two
weird
periodics,
mainly
in
this
under
red
hat.
It
is
not
a
cncf
one,
but
yeah.
E
Maybe
we
will
need
something
something
more
specific,
because
maybe
you
want
only
to
be
to
be
notified
of
some
specif
specific
jobs
when
it
failed
and
not
of
everyone.
If
we
have
a
single
main
list,
it
is
the
same
mostly
as
a
single
as
black
channel.
Maybe
you
are
not
interested
about
all
the
noise
of
all
the
periodics
that
are
failing
and
all
the
of
some
specific
jobs,
but
they
are
having
a
main
list
for
their
job
doesn't
make
too
much
sense.
I'm
not
sure
what
is
the
right
approach
here.
E
C
A
Don't
know
just
for
that
job
but
yeah.
Sorry,
of
course.
Oh
sorry,
if
there's
already
a
mailing
list
created,
you
can,
let
me
know,
and
then
I
can.
I
can
transfer
the
ownership
that's
much
easier
than
trying
to
create
new.
Yes,.
B
E
C
Thank
you,
but
just
like
when
you
transfer
that
it
will
change
its
ads
and
then
it
will
not
work
anymore
so
like
changing
it
will
be.
You
need
to
change
that
a
few
places
I
suppose.
A
Yeah,
oh
I'll
I'll,
look
into
it
and
any
any
impact
to
the
the
actual
address
I'll.
Let
you
guys.
C
Know
all
right
yeah!
That's!
I
just
want
to
bring
it
up,
we'll
we'll
also
push
it
forward.
On
our
end.
A
A
C
Yeah
I'll
I'll
bring
it
up
async
because
I'm
I'm
not
sure
what
what
what's
what's
going
on.
F
C
What's
gonna
happen
I'll
I'll
talk
to
to
the
evolved
people
again,
I
guess.
A
A
A
A
Okay
or
requests
worthy
of
discussion,
5038.
C
Yeah
that
that's
me
as
well,
I
just
wanted
to
to
bring
up,
bring
it
up
or
share
it
because
partially
because
it
took
a
while
and
a
few
people
might
be
interested.
We
have
the
kiamo
exec
probes
image
now,
so
you
can
now
we
in
a
vm
with
the
guest
agent
present
provider,
an
exact
probe
that
executes
some
command.
You
specify
in
the
vm
and
just
you'd,
like
you
know,
with
pod
probes
products
it
probes
and
it
should
work.
C
The
redness
probe
is
passed
on
to
to
the
kubernetes
workload
so
to
the
pod.
So
if
it
doesn't,
if
it
doesn't
succeed,
kubernetes
decides
this.
Vm
is
not
ready
and
will
not
serve
it
if
it's
targeted
buses,
it's
what's
also
already
used
by
parts
like
you
can
say.
The
part
is
not
ready
and
it
won't
receive
traffic
until
it
is
okay
and
the
same
like
there's
also
liveness
probes.
C
So
if
the
liveness
probe
fails
after
some
extent,
the
part
will
be
terminated
and
depending
on
the
setup
restarted
or
not,
and
this
way
you
can
make
sure
that
your
vm
gets
restarted
if
it
breaks,
we
already
have
that
for
network
and
tcp
probes,
and
now
this
also
supports
exit
probes.
C
Yeah
I
haven't
looked
at
that
I
can.
I
can,
if
you're
free,
to
send
me
the
right
spot,
offline
or
I'll
have.
A
A
look
sure
yeah,
it
almost
seems
like
we've,
got
duplicate,
duplicate,
information
and
and
repo
dots,
and
now
we
have
a
contention
on
what
is
what's
correct.
C
C
A
C
A
We
have
many
docs
here
and
now
we
have
to
go
through
each
like
some
of
these
docs
are
really
old
and
we
have
to
figure
out
what
is
what
is
what
what
is
what's
correct
and
unfortunately,
we
have
to
go
through
every
single
page
and
and
compare
with
what
we
have
in
user
guide
when
we
did
the
user
guide
migration
to
to
make
docs
in
the
this
past
winter,
we
discovered
the
problem
and
it
just
seemed
like
we
were
looking
up
at
a
very
mountainous
problem
and
didn't
address
it
for
the
sake
of
time
and
effort.
C
I
also
earlier
noticed
that
we
have
an
examples
folder
with
vms,
and
I
I'd
want
to
get
a
vm
with
the
radius
probe
in
there.
Probably
so
it's
another
point
that
we
have
documentation
in.
I
don't
know
how
new
it
is
if
it
is
new,
but.
A
I'm
not
looking
forward
to
going
down
that
rabbit
hole.
F
F
Now
the
problem
is
that
I'm
wondering
how
to
test
this
upstream,
because
we
use
a
cluster
which
is
very
homogeneous.
All
of
the
cpu
models
are
the
same
between
all
the
nodes.
So
basically
yeah,
I'm
wondering
how
to
test
this,
because
currently
the
migration
could
always
happen.
But
it's
it's
like
a
crippled
test.
A
Itamar,
are
you
red
hat
yeah?
Have
you
ever
worked
in
the
the
d8
the
diesel
lab?
A
I
don't
think
so
check
it's.
It's
spelled
d-a-s-l
and,
and
they
have
all
kinds
of
different
machine
types-
might
be
worthwhile
to
do
a
test
with
nested
virtualization.
G
A
A
Yes,
hi
daniel,
do
you
have
any
thoughts
on
I
mixed
cpu
architecture
in
ci.
G
E
In
in
any
case,
as
itamars
had
said
before,
we
we
run
clusters
with
with
these
super
homogeneous
machines,
and
then
maybe,
if,
if
needed,
we
would
need
to
spin
up
the
different
machines
inside
or
have
different
their
metals
with
different
cpu
models,
or
something
like
that.
If
this
is
required
so,
but
currently
they
are
all
all
the
same.
All
the
machines
are
the
same.
F
But
until
then
I
mean
at
least
we
can.
We
can
have
a
small
test
or
a
lightweight
test
to
just
test
the
the
general
functionality.
A
Okay,
I
guess
that
we'll
just
have
to
move
forward
itamar.
As
always.
Let
me
know
if
I
can
help
at
all
sure
we
might
be
able
to
get
josh
to
help
us
out
with
some
additional
cncf
resources
or
or
whatnot.
A
A
Sound
good,
okay,
thank
you!
Everybody
and
we'll
see
you
next
week.