►
From YouTube: KubeVirt Community Meeting 2022-06-29
Description
Meeting Notes: https://docs.google.com/document/d/1kyhpWlEPzZtQJSjJlAqhPcn3t0Mt_o0amhpuNPGs1Ls/
A
A
A
Final
reminder
check
your
calendars,
make
sure
that
you
are
subscribed
to
the
cncf
organizer
invite
for
the
weekly
meeting,
so
we
don't
lose
track
of
that.
We
have
some
duplicate.
Cooper
meetings
out
invites
out
there
and
we
just
want
to
simplify
everything
and
make
sure
we
clean
that
up.
So
we're
going
to
deprecate
and
remove
the
other
invites.
A
B
Hello,
everyone
yeah
totally
scared
my
mind,
but
every
month
the
cncf
technical,
docs
sig
runs
runs
their
meeting
and
part
of
the
the
I
guess,
pretty
old,
the
one
doc
we
had
from.
I
think
two
years
ago
that
I
guess
is
still
a
bit
active
talked
a
bit
about
the
what's
the
word,
I
guess
the
the
out
of
date
and
kind
of
like
semi
chaotic
nature
of
the
the
current
user
guide.
B
B
The
cncf
offers
projects
such
as
ours
a
possibility
of
like
a
six
week
or
potentially
longer
project
from
the
cncf
technical
documentation
team,
and
so
I
was
going
to
see
a
you
know
what
their
bandwidth
was
like
and
how
they
want
us
to
proceed
and
also
kind
of
see
what
it
is
they
do,
but
the
the
two
things
that
have
struck
out
at
me
from
going
through
the
the
user
guide
at
the
moment
is
the
organization.
B
If
you
look
at
our
documentation
compared
to
say
kubernetes
or
a
lot
of
other
the
graduated
incubated
projects,
they
do
kind
of
follow
a
relatively
consistent
organization,
which
seems
to
make
sense.
I
think
they've.
All
many
of
them
have
potentially
folded
on
from
the
kubernetes
changes
that
I
think
went
through
in
2019.
B
Just
like
you
know,
concepts,
tax
tasks,
reference
and
then
things
kind
of
get
subgrouped
into
those.
So
that
was
one
possible
project
for
them
that
maybe
they
could
help
us
with,
and
the
other
thing
that
struck
me
is
potentially
like
to
ease
the
burden
of
starting
with
nothing
and
getting
into
like,
like
a
useful
environment
which
is
beyond
just
installing
qbert.
B
You
know,
potentially
a
series
of
concise
tutorials
to
get
people
you
know,
have
their
networking
set
up,
have
their
storage
set
up
and
and
virtual
machines
running
with
cubot
on
kubernetes,
so
kind
of
like
introductory
tutorial
kind
of
material.
So,
if
anyone's
keen
to
to
come
along
it's
seven
o'clock,
I
can
share
the
information.
A
Okay,
all
right,
actually,
that's
a
pretty
neat,
invite
and
recommendation.
If
I
can
make
it,
I'm
gonna
try
to
cool.
A
All
right
that
does
cover
everything
on
the
agenda
note,
so
we
have
a
couple
things
on
open
floor.
D
Oh
yeah,
I
take
it,
you
can
hear
me
yes
good.
D
So
recently
there
was
an
effort
to
add
like
a
linting
job
to
prowl,
and
the
goal
is
to
like
it
currently
only
points
out
a
couple
of
places
on
the
end-to-end
tests
and
the
idea
is,
to
over
time,
add
more,
let's
say
more
code
to
the
link
to
be
linked
targets.
Now.
The
thing
is:
this
is
not
required
by
ci,
which
means
that
there
isn't
really
anything
that
will
prevent
people
for
adding
linting
errors
or
things
that
would
be
flagged
as
linting
gears
as
we
go
along.
D
So
it's
a
bit
thing
is.
If
we
I
mean,
I
think
we
should.
If
we
have
this,
we
should
make
it
required
and
because
otherwise,
there's
really
no
point
in
having
people
cleaning
stuff
up,
but
no
not
having
the
bot
making
sure
that,
like
more
stuff,
is
not
added
to
the
things
to
be
cleaned
up
afterwards.
E
Means
am,
I
am,
I
clearly
audible
perfectly
yes,
okay
thanks,
so
I
was
wondering
about
what
what
did
jafia
exactly
mean?
Is
that
the
current
minimal
linting
job,
that
just
checks
about
the
line
length
of
the
uterus
go
file?
Or
what
are
you
talking
about.
D
It's
if
you
look
at
your
well,
it's
that
plus
golden
ci.
I
think.
Well,
I
think
no,
I
I
know
so
it's
the
one
that
look
that
it's
the
one
that
looks
that
makes
sure
that
the
test
utils
does
not
grow
and
plus,
like
a
run
of
a
subset
of
the
golden
ci
linters.
E
Okay,
so
I
understood
that
edward
did
did
a
proof
of
concept
of
running
the
golang
ci
linter
with
github
actions.
If
I
understand
correctly,
but
if
I
understand
correctly,
the
other
linting
jar
that
looks
at
the
line
length
is
a
separate
one
that
is,
that
is
done
by
prowl.
So
I
am
not
exactly
sure
in
what
direction
we
want
to
go.
Do
we
want
to
change
the
whole
linting
into
this
golang
ci
lint
run
by
github
actions,
or
do
we
want
to
create
a
project
on
that
all
right.
D
I
think
let
me
just
like
add
something
to
that.
I
think
these
are
those
are
two
different
things
like
if
it
should
run
on
pro
or
if
it
should
run
on
github
actions.
He
has
like
a
prototype
for
github
actions,
but
right
now
the
golden
ci
linting
is
performed
on
prow
as
well.
So
there's
a
make
lint,
I
think,
command
on
the
make
file
that
runs
both
of
them
and
this
is
being
currently
executed
by
prowl.
E
No,
no,
I
totally
understand
I
was
just
confused
because
I
just
didn't
follow
up
completely
on
the
linting
stuff,
and
I
was
just
seeing
that
pr.
That
would
be
a
prototype
on
the
on
the
on
the
github
actions.
Okay,
so
we
are
clear.
Thank
you
yeah.
You
know,
I
think
that
lending
in
general,
of
course,
has
value
and
but
but
to
the
country.
What
I,
what
I
saw
in
recent
times
was
that
this
lending
job
was
failing
a
lot.
If
do
we
have
an
effort
on
stabilizing
all
this.
D
E
F
G
H
Here
is
andre
from
the
desk:
I'm
I'm
having
some
issues
when
creating
the
cicg
pipeline
with
convert.
H
We
are
trying
to
create
a
windows
vm
and
the
missing
part
I
didn't
find
under
the
communication
and
parts
so
far.
So
one
is
when
I
add
a
windows
iso
file
to
the
vm.
It
asks
for
press
something
to
start
the
the
the
cd
on
virtualbox.
H
To
do
the
same
thing,
it
doesn't
ask
nothing
if
the
disk
is
not
formatted
at
all,
and
this
is
something
that
anyone
find
a
way
to
have
completed,
auto
unattended,
install
of
windows,
because
I
didn't
find
a
way
to
automatically
don't
appear
that
please
press
something
to
start
the
cg
of
windows,
any
any
ideas.
A
Okay,
well,
thank
you
for
bumping
it
here
too
as
well.
We
we
need
to
make
sure
and
follow
up
on
that.
Then.
A
A
D
Yeah,
but
nevertheless,
I
think
daniel
raises
a
really
good
point
like
it
seems
we
don't
have
enough
voices,
I'm
okay,
with
moving
this
discussion
to
the
mailing
list
and
well,
I
I'll
revive
the
thread
on
lending
there.
Okay,
thanks
for
your
time,.
I
I
A
Okay,
so
we
just
we
need
a
reviewer
for
this
at
this
time,.
J
J
Leaking
between
nest
bases
like
so,
if
he,
if,
if
something
was
created
as
part
of
the
test,
it
was
left
on
the
on
the
host
or
the
node,
and
then
the
next
test
created
another
one
and
another
one
and
in
the
end
it
blow
up.
So
I
remember
it.
It
happened
with
things
that
are
related
to
storage
right
no
devices.
E
J
I
Yes,
you
mean
some
some
text,
that's
not
namespace.
We
cannot
test
in
kind
right.
J
J
That
or
or
we.
I
Not
yet
because
I
only
run
the
test
right
that
with
64
texts,
so
the
test
is
very
limited.
I
run
them
for
a
whole
day
around
the
three
tastes
for
whole
day
and
all
the
taste
passed.
I
J
I
J
J
J
E
J
I
Yeah,
okay,
I
see
so
I
want
to
move
to
next
step
is
maybe
we
can
try
to
run
testing
kind
for
a
while
to
say
if
it
is
stable
or
not,
then
maybe
we
can
do
if
it
is
stable.
We
can
do
more
thing
on
in
kind
right.
I
E
I
have
no
strong
opinion,
but
I
think
I
didn't
by
the
way,
how
sorry
for
not
picking
that
up
again,
I
just
saw
the
email
that
you
were
picking
up
your
work
again
and
that
you
posted
something.
So
I
I
owe
you
a
review
on
that.
So
be
sure
that
that
I
should
review
this
today
or
tomorrow.
A
C
No
yeah,
that's
my
pr.
I
wanted
to
show
you.
I
just
implemented
the
live
migration
for
the
pod
network
and
it
works
well
with
the
bridge
network
and
the
only
problem.
I
have
this
with
this
pr
that
I
don't
know
how
to
solve
those
tests.
In
the
end,
I
don't
know,
maybe
someone
could
help
me
with
that,
because
I'm
really
totally
stuck
with
that.
C
J
J
E
I
think
I
think
you
should
should
probably
take
the
discussion
or
raise
some
attention
on
the
prs.
If
you
need,
if
you
need
some
support
for
those
prs
to
the
mailing
list,
yeah.
C
E
C
D
One
extra
thing
I
I
think
it
would
be
very
helpful
if
you
added
to
the
pr
like
a
comment
with
the
list
of
tests
that
you
need
help
with
that
would
at
least
build
not
a
label
just
like
a
list
like
a
a
comment
with
a
list
of
tests
that
are
failing
that.
D
You
you're
sure
that
they're
broken
and
you
have
you
need
help
with
them,
because
that
would
at
least
help
whoever
will
be
helping
you
to
focus
on
on
something
instead
of
going
through
each
of
the
broken
lanes
and
figure
out
which
tests
are
the
great
ones
to
look
at
and
stuff
like
that.
I
would
at
least
appreciate
that
if
I
end
up
helping
you.
D
Welcome
and
I'm
aware
that
I
really,
I
also
owe
you
a
review
on
this
viewer.
C
A
All
right
well,
thank
you
for
that
one!
It's
promising!
Let's
see
with
that,
unless
there
are
any
other
last
minute
editions,
we
can
do
a
quick
review
of
pr's
mailing
lists
and
bug
scrub.
K
Yes,
okay,
great,
so,
as
far
as
I
know,
we
have
we're
having
trouble
migrating
gpu
vms,
because,
basically
because
it's
a
pci
card,
there
is
data
that
is
saved
within
within
this
pci,
and
this
is
basically
the
main
problem.
As
far
as
I
know,
there
is
no
progress
on
that
side.
H
G
H
They
change
a
little
bit
about
that
on
the
last
month
or
two
or
two
before
you
understand.
There
is
a
way
to
do
it
over
all
only
on
open
source
tools.
From
now
on.
G
A
H
A
Knowledge
or
contributors
for
that
we're
on
the
call
at
the
moment
as
you
progress,
feel
free
to
bring
up
other
specific
questions
to
the
group
and
I'm
sure
we
can.
We
will.
A
All
right,
I
don't
see
anything
idle
are
not
getting
attention.
That's
not
working
progress
or
draft
there.
If
I
miss
anything,
feel
free
to
drop
it
to
pr
list.
Take
a
look
at
bugs
real,
quick.
H
A
A
C
C
A
K
Maybe
I
can
just
share
what's
wrong
there
in
one
sentence,
so
basically
we
have
in
cubits
here
we
have
a
way
to
define
verbosity
for
the
different
components,
so
one
of
the
options
is
to
define
a
node
verbosity
instead
of
defining
a
verbosity
to
each
component
and
basically
the
node
velocity
does
not
affect
some
of
the
components,
especially
weird
launcher
yeah.
So
this
is
a
bug
we
either
need
to
fix
it
or
to
deprecate
the
api
or
yeah.
F
K
There
is
another
problem
actually
that,
basically,
when
we're
creating
the
grid
launcher
pod,
then
we
don't
know
which
node
it
would
be
scheduled
to.
Okay
right
now
we
determine
the
verbosity
via
an
environment
variable.
So
it's
problem
to
change
it
afterwards,
since
pods
are
immutable.
G
Yeah
well
that
can
be
solved,
but
I
think.
G
So
when
we
first
developed
it,
I
didn't
ever
had
the
word
launcher
in
mind.
When
we
talked
about
node,
not
verbosity,
I
think
it
was
just
meant
for
vert
handler
and
all
other
components
that
are
running
like
keyboard
components,
not
the
application.
I
mean
not
not
vert
launcher,
so
I'm
not
sure
how
to
go
about
it.
I
mean
I
would
I.
G
K
Yeah,
actually
it
doesn't
make
sense,
we
can
simply
document
that
it
affects
only
infrastructure,
components
and
yeah
would
be
good
enough.
I
think,
although,
for
some
reason
also
weird
api
is
not
being
updated,
but
I
think
this
is
an
easier
bug
to
solve.
G
I
think
his
jump,
maybe
from
0
43
to
054,
is,
is
a
bit
problematic
because
at
some
point
we
got
rid
of
the
of
the
config
map
and
all
of
this
configuration
of
of
the
feature
gates
moved
to
yeah.
It
moved
to
the
cr.
G
So
I'm
not
sure.
Maybe
there
is
still
some
kind
of
confusion.
Maybe
he
still
have
a
config
map
around.
F
A
A
A
If
we
did
an
iso
style
injection,
it
would
still
require
something
in
the
os
or
via
cloud
in
it
to
trigger
reading
it
and
acting
on
environment
variables.
Past.
F
And
the
environment
for
the
bash,
but
it's
a
very
vague
idea.
Sorry,
I
haven't
tried
anything.
A
A
A
F
A
A
All
right-
and
that
brings
us
closer
enough
to
time-
I
am
going
to
go
ahead
and.