►
From YouTube: KubeVirt Community Meeting 2021-12-15
Description
Meeting Notes: https://docs.google.com/document/d/1kyhpWlEPzZtQJSjJlAqhPcn3t0Mt_o0amhpuNPGs1Ls/edit#heading=h.u74oyrl72es0
A
All
right
well
looks
like
we're
filling
out
some
agenda
and
open
floor
notes.
So
thank
you
for
that.
If
everyone
is
put
their
name
in
yeah.
A
With
that,
we
don't
have
anyone
who
filled
out
anything
on
the
agenda
notes.
I
know
that
it
is
an
ongoing
theme.
We
have
conferences
and
opportunities
to
speak
and
share
coming
up.
So
if
anyone
wants
to
speak
to
or
present
on
how
they're
using
comfort
or
anything
like
that,
we
have
resources
to
help
make
sure
you
can
spin
with
up
for
those
things
easily
and.
C
A
A
See
if
we
can
go
through
any
of
the
other
things
real
quick
before
we
jump
into
issues,
we're
going
to
do
book
scrub
here
in
just
a
minute.
D
And
oh,
I
would
like
to
ask
as
we,
as
you
already
may
know,
we
have
a
client's,
a
client
in
the
server
onward
land
on
world
launcher
and
virginia.
That
is
communicating
via
grpc
calls.
So
I
would
like
to
ask
what
will
happen
when
you
change
add
a
remover
command
or
for
the
old
world
launcher
versions.
C
D
Basically,
yes,
what
I
mean
I'm
talking
about,
you
already
have
a
you
have
a
few
vms
on
your
cluster
and
you
change
the
command
one
of
the
commands.
Now
you
upgrade
the
wheel
handler
and
you
need
to
still
support
your
old
vms
before
you
upgrading
them.
So
what
will
happen?
Will
it
work?
Do
we
need
to
support
such
cases.
E
What
do
you
have
a
specific
example
like?
We
can
walk
through
a
few
to
talk
about
what
would.
D
C
So
if
you
want
to
support
hot
plugging
still
on
already
running
vms,
when
you
roll
out
this
update
like
it
was,
did
it
work
before
so?
Do
you
have
to
consider
the
case
where
vmware
started
and
you
want
to
then
suddenly
start
hot
block?
Something?
Does
it
work
right
now,
sorry,
I
think
it
was
a
little
bit
unclear
so
when
I've
started
the
vm
already
right
now
and
you
come
in
with
your
fix,
can
you
can
you
already
without
the
fix
hot
plug?
Is
there
we
or
is
this
completely
broken
at
the
moment.
C
D
I'm
I
I
started
to
work
on
if
this
is
answer,
your
question.
C
D
It's
a
it's,
I'm
changing
the
future
to
be
how,
with
the
the
correct
practices
for
covert,
for
example,
I
mean
for
the
current
implementation
is
the
the
part
where
you,
where
you
hold
plug
and
unplug
the
so
when
you
hold
back
the
srv
device
on
the
target
used
to
be
currently
is
a
one-shot
operation,
and
I
and
we
want
to
follow
our
discussion
a
while
back
ago.
We
we
want
to
to
make
it
reconcile,
so
it
means.
Basically,
it
means
that
we
move
the
logic
from
the
launcher
to
build
handler.
F
Okay,
so
you
know
what
so
sorry
can
can
I
can.
I
try
to
summarize
it
a
little
bit
more
abstract,
because
it's
just
to
make
sure
we
are
on
the
same
page
here
today.
We
have
we
have
every
added
few
months
ago.
I
don't
remember
how
much
maybe
it's
a
year
time
is
flying
here,
but
we
added
a
new
command
and
the
new
command
was
only
triggered
as
part
of
the
finalization
of
the
migration.
F
So
that
was
fine
because
it
can
it.
It
will
always
run
on
a
new
vm.
It
cannot
never
run
on
an
audio
because
it's
after
the
migration,
but
that
was
no
problem
with
backward
compatibility,
but
now
that
command
moved
instead
of
being
in
the
target
processing
of
the
migration,
it
moved
in
the
regular
reconcile
of
the
of
the
not
real,
not
relevant
to
migration.
It
just
always
checks.
F
No,
no,
the
because
now
this
is
what
happened
before,
but
now
the
the
hot
plug
itself
will
not
happen
in
the
in
the
terminal
in
the
finalization
of
the
migration
it
will
have
after
the
migration
is
considered
finalized,
and
it's
done
then,
the
the
next
reconcile
that
will
go
through
the
regular
flow,
not
the
migration
for
the
regular
flow.
Only
then
the
hot
plug
will
happen.
C
F
F
F
C
F
C
Let
me
again
try
to
rephrase
it,
in
my
words
that
we
see
that
we
talk
about
the
same
thing.
This
means
that
you
have
now
in
the
vertex
reconciliation.
Another
grpc
call,
which
is
not
there
on
old,
word
launches,
so
you
start
with
handler
and
it
would
immediately
try
to
do
it.
Your
pc
call
or
something
and
fail
there.
G
E
F
E
We
just
ignore
when
the
api
doesn't
exist
in
the
back
end.
Certainly,
there's
got
to
be
a
way
to
detect
that
so
here's
what
we
know,
we
know
that
the
old
vert
launcher
attempted
to
do
this
in
a
one
shot
way,
so
it
would
attempt
to
do
this
hot
plug
re-plug,
whatever
we
call
it
itself.
We
know
that
if
a
bert
launcher
does
not
have
that
api,
that
it
had
that
old
logic,
so
we
know
that
it
should
be
okay.
If
it's
not
implemented.
Does
that
sound,
accurate.
F
Almost
the
problem
is
that
that
logic
is
also
in
the
real
time
I
mean
the
the
the
village
lander
recognized
that
the
the
migration
almost
finished
or
finished,
and
then
it
hot
plugs.
This
is
the
old
virgin.
Now
we
wanted
to
move
it
to
the
just
to
move
that
code
to
the
records
the
regularly
console,
but
but
I
guess
we
can
do
both.
I
don't
know.
E
Vert
handler
tries
to
so
what's
the
current
logic,
with
vert
handler
than
with
respects
to
hot
plug.
F
B
F
F
C
C
I'm
not
fully
convinced
this,
but
if
you
can,
for
instance,
negotiate
an
api
version
grpc
like
you
basically
and
that
the
version
is
already
there,
you
ask
the
back
and
what
version
it
implements
and
then
you
can
choose
the
right
version
and
do
the
right
action.
For
instance,
we
don't
think
we
do
that
on
the
word
handler
launcher
interface
at
the
moment.
C
B
C
F
But
I
think
that,
because
you
said
that
I
think,
based
on
your
on
your
input,
that
if
there
is
no
on
that
on
the
server
side,
if
that
that
is
not
that
command
is
not
recognized,
then
it's
just
ignored.
I
think
if
that
is
the
case,
then
yes,
maybe
we
are
fine.
C
C
F
H
F
H
Mean
how
do
you
make
the
call
for
for
querying
for
for
asking
divert
launcher
or
anything?
I
thought
that
you
would
have
all
the
all
the
interfaces
or
the
difference
that
you
are
looking
for.
Wouldn't
you
have
it
in
the
status
already.
D
Those
are
that
unplugged,
we
don't
have
any
status,
but
we
do
this
calculation
on
the
good
luncheon
side.
F
No
all
right,
but
if
you
radic,
if
you
mean
that
we
do,
I
mean
the
code
that
all
is
now
writing
it
checks.
If
the
if
the,
if
the
status
has
less
interfacing
than
the
spec,
then
he
will
do
the
hot
plug.
If,
if
they
are
equal,
you
will
not
do
the
hot
plug.
So
it's
only
in
that
case
you
will
send
a
command
to
hot
plug.
H
Perfect,
but
wouldn't
this
mean
that
it's
a
new
word
launcher
that
already
migrated.
F
No
yes,
this
is
what
I'm
saying
you
know
on.
If
this
is
what
I
think
it's
okay,
if
we
did
the
migration
and
this
and
the
reconcile
works
there,
it
will
work
because
the
the
vm
already
migrated
it
is
for
shorter
inversion,
but
this
this
in
the
place
that
you
ask
this
question
and
do
this
query.
It
can
happen
also
on
vmis
that
are
not
migrated
like
all
like
the
existing
ones.
F
D
C
No
not
immediately
so
it
works
like
this
that
vert
handler
has
to
become
the
the
compatibility
works
like
this.
The
newer
tender
normally
has
to
be
able
to
communicate
with
old
and
new
wood
launchers
and
is
first
upgraded.
Then
your
controller
gets
upgraded
and
only
after
the
controller
is
upgraded.
New
new
launches
are
created
so
you're
right.
There
is
definitely
a
case
edward
where
you
do
a
migration,
and
then
you
still
get
old
launchers
with
a
new
handler,
because
we
have
always
zero
downtime
on
the
api
for
migrations
and
so
on.
C
F
F
C
F
C
F
D
F
No,
no,
that's
I
I
feel
like
we
took
too
much
time.
Thank
you
very
much.
A
A
G
Great
just
a
quick
heads
up
that
we
have
some
test
results
for
the
new
123
provider,
which
has
not
yet
landed
in
cupert
itself,
but
it's
already
available
in
cupid
ci.
So
there
is
a
pr
that
we
are
working
on
currently
where
the
the
new
provider
gets
added.
So
the
only
thing
is
at
the
moment
you
can't
use
it
locally,
because
I
got
the
question
already.
We
need
to
wait
until
that
vr
gets
it,
but
yeah
we
have
some.
Some
storage
and
network
results
and
compute
lane
looks
already
fine.
G
So
no
failures
on
that
and
teams
are
already
working
on
that,
which
is
great,
but
there
are
two
things.
One
thing
is
we
need
to
fix
the
operator
lane,
which
is
in
the
same.
G
Oh,
what
was
that
a
glitch
in
the
matrix?
I
guess?
Okay,
okay,
so-
and
the
last
thing
I
wanted
to
mention
is
that
we
should
make
the
check
provision
of
in
the
kubrick
ci.
G
We
should
make
it
required
for
123,
as
the
123
provider
is
already
in
there.
So
yeah.
I
think
that's
for
me
any
questions
on
that.
G
A
All
right
and
then
sorry
to
bumped
the
bug
to
last,
but
did
someone
want
to
speak
to
issue
64.98
the
goling
117
fails
to
parse?
I
I
Furthermore,
there
are
several
other
ip
addresses
inputs
in
the
vert
cattle
utility,
but
they
go
straight
to
kubernetes
as
far
as
we
know.
So
what
you
want
to
do
is,
first
of
all
understand
it
from
a
licensing
perspective.
We
can
use
a
kubernetes
fork
of
a
net
pass
ipn
or
calciter,
and
if
we
can,
we
want
to
upgrade
kubrick
to
go
away.
1.17
we
had
the
meeting
with
itamar.
He
was
the
last
one
to
do.
B
C
F
Is
a
licensed
issue,
though,
if
you
look,
if
you
look
there,
they
got
an
exception
from
the
golan
licensing,
so
there
is
an
exceptional
sign.
The
question
is:
is
it
okay?
It's
it's
it's
odd.
It's
they.
They
added
the
internal
file
that
implements
this
from
the
goal
and
the
source
sources
which
has
a
different
license,
but.
F
C
F
I
think
the
the
main
point
that
morale
once
wanted
to
to
make.
At
least
I
want
to
make
sure
we
we
raise
it
here,
so
we
get
feedback
is,
can
we
make
make
this
decision
to
to
normalize
the
the
two
api
entries
that
we
see
that
we
can
get
as
ips?
That
may
be
like
lead
with
elite
zero?
Can
we
just
normalize
them
and
use
the
all?
The
other
places
use
the
regular
parcel
of
golang
117
and
on.
C
F
C
F
Touching
the
I'm
not
touching
the
I
mean,
if
yes,
if
there
is
the
the
to
these
two
values
that
we
get
as
ips,
if
we
get
those
we
normalize
this
to
the
not
to
have
leading
zeros
and
and
that's
it
maybe
there-
we
will
use
this
sloppy
stuff,
but
in
any
other
place
that
is
that's
internal.
We
will
not,
will
not
do
it.
I
Yes,
but
still
we
we've
talked
about
data
that
is
already
there
before
a
potential
library
like
in
cache
files
or
in
the
lcd
that
contains
these
leading
zeros,
because
these
fields
are
saved
as
strings
and
not
as
a
special
go
type.
B
C
F
I'm
not
talking
about
touching
the
spec
now
the
the
manifest,
but
internally
we
interpreted
the
stem
and
that's
it.
We
need
to
do
it
once
in
the
entry
point
and
we
need
to
do
it
there
and
that's
it.
We
and
after
that
we
are
not
expecting
anyone
to
have
a
0-10
anymore.
So
it's
I
I
guess
it's
like
when
you
marcel
or
unmarked,
I
don't
remember
the
which,
which
one
of
them
is
more
confused
when
you
do
that,
you
should
get
the
correct
normalized
values.
C
F
I
A
Okay
back
covers
are
open
for
items.
I
wouldn't
actually
circle
back
to
introductions
or
raz.
Do
you
want
to
go
ahead
and
speak
out.
A
A
A
D
Yes,
this
is
mine.
I
don't
want
to
take
more
time
from
this
meeting
and
basically
there
are
two
pr's
that
I
mentioned.
If
there
is
any
power
that
can
take
a
look,
I
will
really
appreciate
it.
This
one
is
already
lgtm
and
the
second
isn't,
but
we
review
it
on
the
last
meeting.
So
I
understand,
if
it's
not
really
it,
but
this
one
is.
C
Who
is
not
active
anymore?
Yes,
approver
understand,
let's
in
such
case,
it's
also
helpful,
but
since
peter
isn't
yeah
we
should
probably
ask
peter
to
move
himself
to
the
emirates
emeratus
section
so
that
he
doesn't
get
assigned
prs
to
get
approved,
but
if
he
wants,
he
can
still
approve
it
or
assign
someone
else.
I
D
D
A
A
Roman,
do
we
want
to
get
another
signing
on
that?
One.
I
A
A
C
I
think
it's
better
to
take
other
prs
and
slack
offline,
because
it's
hard
to
discuss
them
here.
What.
K
J
C
A
Easy
enough,
we
don't
have
any
bugs
and.
B
G
K
Right
there's
been
a
few
issues
where
that
was
sim
linked
over
yeah.
I
guess
we
can
close
this
dublicated
point
into
the
yeah.
A
A
B
C
C
C
E
A
Are
they
talking
like
multiple
integration
and
stuff.
A
On
the
wrong
issue,
sorry,
oh
yeah,
6930
do
vnf
support,
covert
or
vnf's
compliant
cooperate,
I'm
guessing
virtual
network
functions.
So
would
that
be
multis
which
works
great.
A
A
A
C
A
I'm
just
handling
that
at
the
sort
of
provider
layer
and
not
a
grouper
layer,
right
yeah.
You
want
to
respond
to
that.
One.