►
From YouTube: KubeVirt Community Meeting 2022-03-02
Description
Meeting Notes: https://docs.google.com/document/d/1kyhpWlEPzZtQJSjJlAqhPcn3t0Mt_o0amhpuNPGs1Ls/
A
A
Okay,
so
I
think
I'm
going
to
take
over
the
leap
for
today,
because
I
think
the
usual
host
seems
to
be
not
able
to
join.
A
B
I
can
hear
you
is
it
just
faint
for
other
people
or
I'm
not
sure,
maybe
my
volume's
just
really
high.
A
I
just
just
don't
want
to
increase
the
input
level
to
100
or
something
like
that,
because
I
guess
then
it
gets
very
noisy
for
everyone.
Okay.
So
what
I
wanted
to
say
is:
I
am
going
to
share
the
document
for
the
meeting
notes
for
everyone
so
that
everyone
please
fill
in
your
attendance
on
that.
A
And
without
further
ado,
so
I
don't
rob
you
of
more
of
your
time
than
necessary.
I
would
say:
welcome
to
the
cubert
community
meeting
2nd
of
march
2022.
C
Hi,
this
is
a
brad
solar
from
mainsail
industries,
first
time
here,
but
user
of
kubert
for
for
a
little
bit
of
time,.
B
Great
welcome
yeah
welcome.
Have
you
signed
our
just
curiosity?
Have
you
publicly
said
that
you
are
using
it
in
our
adopters
file?
B
C
Is
that
on
the
dock
you
put
in
chat
or
is
that
somewhere
on
the
site.
B
I
will
link
you
to
it
right
here:
we'd
love
to
have
you
if
you
are
willing
to
contribute
to
this
document
as
an
end
user.
If
you
are
an
end
user
and
maybe
give
some
information
about
your
use
case
there,
that's
always
alpha
for
us
with
the
cncf
as
we
go
like
graduate
from
incubating
to
ga
and
things
like
that,
and
it
gives
a
little
bit
more
traction
to
the
project
for
people
to
see
more
adopters.
E
Hi,
my
name
is
andrew
bird,
I'm
from
red
hat.
This
is
my
first
cubert
community
meeting.
It's
been
a
long
time
since
I
used
qubit,
but
I've
used
the
downstream
cnb,
now
openshift
virtualization
for
a
little
while
yeah
and
then
and
don't
mind
me
it's
just
like
one
o'clock
in
the
morning
at
the
moment.
So
I'm
a
little
bit
like
half
present.
B
E
B
Thanks
for
joining
yeah,
where,
where
are
you
located
at
one
o'clock
in
the
morning,.
E
A
Okay,
then
thanks
everyone,
I
don't
yet
see
any
agenda
items.
Probably
we'll
wait
another
minute,
probably
if
someone
wants
to
bring
something
up.
C
Sure
not
a
problem,
you
know,
we've
been
looking
at
it
for
really
edge
computing
use
cases
we're
we're
more
looking
towards.
You
know
us
dod
type
of
use
cases,
and
we
see
that
there's
a
a
lot
of
people
want
to
go
to
containers,
but
a
lot
of
their
existing
workloads
are
still
virtualized
and
what
we're
looking
at
is
you
know
how,
to
you
know,
put
together
solutions
that
can
utilize
kind
of
best
of
both
worlds,
and
I
think
cooper
kind
of
falls
definitely
into
that.
B
D
Yeah,
I
I
just
want
to
bring
up
on
this
meeting
that
I'm
actually
started
to
work
on
the
vehicle
support.
V-Host
is
another
type
of
interface
for
the
vm.
It's
a
it's
a
high
performance
interface,
something
like
srvo.
D
However,
it's
different,
it's
it's
it's!
It's
meant
for
a
local
local,
high-speed
networking,
because
it's
using
in-memory
sharing
of
data.
So
basically
the
use
case
that
I'm
having
is
basically
a
one-node
kubernetes
cluster,
and
I
I
try
to
I
can
I
can
share.
If,
if
I
can,
I
can
show
you
a
picture
of
of
the
use
case.
What
I'm
working
on,
if
sure.
A
A
Already
with,
if
you
already
have
a
pr
already
or
something
or
if
it's
not
energy
early
stages,
it
doesn't
matter.
D
It's
not
it's.
Actually,
it's
actually
a
output
request
and
it's
not
mine
for
requests.
It's
something
like
a
resurrection,
oh
okay.
So
so,
basically
there
is.
There
is
some
support
in
in
this
request
for
vhost
the
host
interface.
However,
the
problem
with
this
pull
request
is
that
it
has
no
ci
test.
So
basically
it's
unmergeable
right
now
and
I'm
not
solving
this
issue.
Actually,
however,
I
I
get
the
pull
request.
D
I
try
to
set
everything
up
using
using
the
using
the
development
setup
of
keyboard
using
using
kind.
D
Where
is
the
kubernetes
and
clemo
and
everything
I
try
to
try
to
change
this,
this
docker
environment,
this
kind
environment
for
developing
kubert
to
actually
set
up
the
the
v-host
interface
interacting
with
vpp.
D
Basically,
this
pull
request
is
try
to
connect
the
vm
via
vhost
and
and
open
vswitch,
but
I
try
to
do
it
the
same
with
with
vpp.
That's
a
virtual
switch,
that's
alternative
to
to
obs
the
openv
switch.
D
However,
from
the
cube
weird
kuber
perspective,
it's
the
same
because
the
the
the
setup
it's
done
through
a
socket.
So
basically,
it
doesn't
matter
if,
on
the
other
side
is
vpp
or
obvious
for
the
code
of
of
cubit,
I
will
probably
share
the
picture
so.
B
B
If
you
have
the
host,
pin,
add
your
you
can
make
yourself
a
host
and
then
you
can
add
co-host.
D
A
Well,
yeah,
what
what
you
can
at
least
do
is
probably
send
something
to
the
cube
or
dev
mailing
list,
for
example,
so
that
you
can
share
it
so
that
we
can
discuss
it
at
least.
So,
if
you
are
not
able
to
yeah.
D
I
can
find
it
yeah,
but
I
I
want
to
just
like
say
what
I'm
doing
there:
what
changes
I'm
doing
in
the
in
the
development,
environment
and
and
basically
my
ultimate
goal-
is
of
course
to
set
it
up
and
and
to
do
my
use
case.
However,
I
was
I
was
fighting
with
with
all
of
this
like
three
to
four
weeks
to
get
it
up
and
running,
because
I'm
also
new
to
to
kubevirt
and
some
stuff
around
it.
D
So
basically
I
wanted
to
share
the
the
setup
with
the
world.
It's
it's
it's.
When
I
finish
it,
it
will
be
a
little
bit
ugly.
So
it's
not
it's
not
in
a
state
that
I
can
merge
it
into
cube
beard
as
an
example
or
as
a
nice
example,
because
because
I,
although
I
made
a
lot
of
hex
dirty
waiting,
it
depends
on
on
the
user
space
cni
that
is
set
up
being
the
v-host,
but
the
user-space
cni
doesn't
support
vpp
in
such
a
way
that
I
want
to
have
it.
D
So
basically,
I
I
did
a
lot
of
dirty
tricks
which
make
it
very
not
nice
to
merge
it
somewhere.
However,
I
can
I
can
open
source
it
on
on
some
fork
somewhere
and
then
and
make
a
link
just
just
to
help
others
to
to
to
start
quickly
with
v-host,
because
there
is
no
no
example
for
this
to
be
used.
There
are
some.
There
are
some
documentations
in
the
pr
in
the
original
pr,
because
there
was
a
pr
then
it
was
closed.
D
Then
it
was
resurrected
and
then
it
was
closed
again
and
and
stuff
like
that.
So
basically,
the
setup
could
help
new
people
that
try
to
use
the
v-host
to
start
up
and
not
to
waste
weeks
with
some
crazy
setups
that
that's
my
immediate
immediate
goal.
However,
I
was
also
noticed
by
miguel.
I
I
think
that's
the
one
here
that
that
maybe
we
can
use
such
a
setup
as
some
kind
of
testing
or
or
something
like
that,
because
something
like
that
was
done
for
srvo.
A
I
think
maybe
this
might
be
a
reasonable
contribution
to
the
qrci
project
which
we
are
using
it
only
for
testing
our
stuff.
So
there
are
a
couple
of
other
setups.
I
am
just
going
to
share
the
link
in
the
chat.
Probably
I'm
not
sure.
I
guess
that
you
already
saw
that
right.
A
So,
okay
yeah
in
general,
how
ci
works?
I
guess
that
you
also
know
that,
but
I'm
just
to
recap
a
little
bit.
We
are
spinning
up
clusters
that
are
running
inside
virtualized,
vms
or
vms,
that
are
running
inside
docker,
which
make
up
our
cluster,
and
I
think
the
the
similar
thing
is
for
our
kind
stuff.
A
We
have
a
couple
of
kind
setups,
for
example,
for
sr
irv
and
for
vgpu
testing,
which
we
also
used
just
to
test
those
scenarios,
and
I
guess
that
your
testing
also
falls
into
that
category,
where
we
could
probably
set
up
some
kind
stuff
inside
the
kuber
ti,
so
that
you
have
your
testing
environment
ready
so
that
you
can
so
that
we
are
able
to
to
set
to
test
that
in
cube
in
kubert
as
a
separate
lane,
for
example.
A
So
I
could
probably,
if
you
want,
you
could
just
ping
me
via
slack,
for
example,
and
cue
for
death,
and
then
I
might
be
able
to
show
you
around.
If
you
want.
D
The
thing
is
that
the
priorities
right
now
for
me
are
to
to
set
it
all
up
or
make
make
the
use
case
that
I
have
because
it's
a
project
with
with
some
deadlines.
However,
in
the
end
when,
when
the
project
is
successful,
I
I
we
definitely
want
to
to
push
the
the
merge
request
and
to
help
with
the
ci
cell.
D
So
after
that,
we
definitely
want
to
to
explore
the
option
to
use
the
development
environment
with
the
ci,
but
until
then
it's
probably
not
necessary
to
to
dig
deeper
into
this,
but
I'm
definitely
want
to.
After
I
want
to
dig
into
this
option.
A
A
H
One
thing
one
thing
I'd
like
to
say
about
this
and
to
philip
like,
if
you
can
like
you,
don't
need
to
have
something
right
now
that
works,
not
that
it
does
not
work,
but
you
don't
have
to
like
push
the
entire
thing
that
works
end-to-end
like
with
tests
and
all
that,
if
you
can't
just
like,
provide
a
simple,
github
gist
that
people
can
use
and,
as
you
said,
that
will
prevent
them
from
wasting
weeks
trying
to
figure
out
a
correct
configuration
or
how
to
use
the
pieces
they
have
available
right
now
that
would
move
the
status
quo
forward
quite
a
lot
and
would
help
lots
of
people.
D
Right
now
I
have
I
have
some
dirty
working
solution
for
for
making
pinging
between
between
the
vm
and
vpp,
however,
not
using
the
dpdk
as
I
wanted,
but
using
the
linux
kernel.
So
it's
not
finished.
However,
it's
quite
dirty-
and
I
want
like
a
little
bit
a
little
bit-
make
a
little
bit
nicer
and
to
write
for,
for
example,
because
it's
really.
D
A
Yeah
thanks
for
the
work
and,
like
I
said,
every
everything
is
appreciated,
I'm
not
exactly
familiar
with
with
the
vo
scenario,
but
I
think
this
has,
of
course
value
yeah
and
let's
follow
up
with
it
and,
like
I
said
just
just
ping
me
anytime,
if
you
want
something
to
have
properly
set
up
some
qbci
stuff,
I'm
really
happy
to
help
on
that.
D
Yeah,
thank
you.
I
tried
to
find
the
pubert
dev
email
thread
and
and
post
it
in
the
chat
in
the
meantime.
So.
A
A
Miguel,
thank
you
for
the
mail
trade.
I
think
that
is
the
thing
that
philip
wanted
to
point
us
to
right.
D
It's
not
been
chat
for
everyone.
I
think
it
is.
G
A
Okay,
so
we
don't
have
any
per
request
that
needs
attention
at
the
moment,
as
I
see
at
least
from
the
list,
so
I'm
going
to
share
my
screen.
Let
me
see
where
is
it.
A
A
A
A
A
A
Cluster
so
you're
saying
we
are,
we
are
deploying
the
arm
version
or
the
amd
version
for
an
arm
note
or.
B
Well,
that's
what
this
this
code
snippet
essentially
leads
me
to
believe.
So
is
the
vert
launcher
version?
I
don't
know
what
the
handler
is
crashing.
Maybe
we
maybe
we
have
one
blanket,
I'm
sure
we
do
that
as
well,
so
we
have
one
daemon
set
and
it's
going
to
be
one
version
or
the
other
as
far
as
architecture
goes
likely.
I
don't
know
in
the
same
interestingly,.
H
Enough
yeah,
sorry
to
cut
you
off.
Interestingly
enough,
I
saw
like
on
the
upstream
slack,
a
user,
mentioning
I
mean
speaking
about
this
as
like
opening
a
feature
request
to
to
allow
this,
I'm
not
sure
where
I
could
paste
the
link
for
it.
H
But
yeah,
if
I
understand
correctly
kind
of,
wants
to
to
to
have
like
nodes
with
different
architectures
and
deploy
a
vm
indicating
like
the
architecture
of
for
it
in
the
spec,
and
that
kubert
should
schedule
it
to
the
correct
node.
B
H
I'll
tag
you
in
in
a
thread,
so
if
you
could
kind
of
chime
in
there
as
well.
B
Okay,
I'll
try
ping
me
offline
or
in
a
different
like
on
slack
or
google
chat.
If
I
don't
respond,
I'm
kind
of
overwhelmed
with
github
lately
the
pings
go.
Oh.
H
No
okay:
okay!
Don't
worry
I'll
I'll
tag
you
on
on
the
option,
slack.
C
H
B
Scenario
to
test,
but
we'd
definitely.
I
B
B
I'm
not
sure
if
I
can.
I
don't
know
what
to
think
of
this,
so
I
feel
like
halted
and
always
are
kind
of
our
default.
If
you're
going
to
be
using
verse
ctl,
it's
probably
what
should
be
the
default
for
start
and
stop
if
somebody
wants
to
set
their
own
run
strategy
on
start,
like
start
and
some
other
cli
option
to
enforce
what
run
strategy
is
set
as
a
result
of
start,
then,
okay,
that
could
make
sense.
A
So
I
see
that
there
is
also
a
pr
that
is
work
in
progress
on
that,
I'm
not
sure
if
that
would
be
somehow
adequate.
But
I
guess
david
that
you
like
the
resources
to
look
at
that
right.
B
A
A
A
B
B
A
B
B
That
we
can
point
this
is
a
really
common
question.
We
get
it
all
the
time
and
it
would
be
nice
if
we
had
just
a
ride
up.
If
we
don't
already
about
how
to
handle
it.
H
I
mean
the
way
I
know
that
you
can
have
kind
of
sticky.
Eyepiece
here
is,
if
you
use
like
static
ipam,
at
least
I'm
unaware
of
any
other
cni
that
will
like
persist
vip
across
on
a
two
new
pod.
B
H
If
exactly
right,
but
yeah
that
for
that
you,
you
need
to
have
like
dhcp,
not
not
ipam,
as
at
least
as
I
understand
ipam,
okay,.
A
Miguel,
would
you
would
you
be
able
to
to
follow
up
on
that
at
least
post
some
short
answer
on
that
what
they
might
be
able
to
do
here
or
don't
you
I.
A
A
Okay,
I
am
assuming
that
you
did.
A
A
A
A
A
A
Yeah
in
general,
one
could
of
course,
always
assume
that
that
we
would
have
some
kind
of
integration
test,
but
I
don't
think
that
this
makes
sense
here,
because
it
can
always
make
it
ever
bigger
and
it
might
still
fail,
I'm
not
sure
yeah,
but
it
could
also
be
likely
a
limitation
of
the
underlying
system.
I
don't
know
what
to
make
of
that.
To
be
honest,.
B
A
Okay,
I
just
added
some
comment
on
that
with
this
question.
I
hope
that
adequately
describes
what
you,
what
you
were
saying,
david.