►
From YouTube: KubeVirt Community Meeting 2023-04-12
Description
Meeting Notes: https://docs.google.com/document/d/1nE09vQWcCTW-9Ohe9oCldWrE0he-T_YFJ5D1xNzMtg4/
A
While
everyone
gets
the
notes
pulled
up,
is
there
anyone
new
on
the
call
or
need
the
meetings?
I
would
like
to
introduce
themselves
and
say
hello,
we'd
love
to
welcome
you.
A
This
is
an
open,
Agenda,
so
feel
free
to
make
sure
you
have
signed
in
on
your
account,
that's
enrolled
in
the
mailing
list,
Google
group,
and
that
should
give
you
edit
access
in
case.
You
need
to
add
something
to
the
agenda.
We'll
look
into.
B
B
Great
name,
cool,
which
means
next
week,
looks
like
we'll
be
creating
the
the
first
V1
aftec.
A
Are
they
just
trying
to
jinx
themselves
this
time
with
that
whole
Chill,
Vibes
name,
it
sounds
like
I'll
just
have
to
see
if
it
lives
up
to
its
name.
B
It
let
me
find
the
page
I'll
put
it
into
chat,
it's
got
a
a
wonderful
logo,
I.
B
Hugging
the
kubernetes
symbol,
it
I
mean
it
from
the
the
Sig
release.
Chatter,
on
slack,
it
does
seem.
B
Really
really
smoothly
great
Shield.
B
And
so
the
other
stuff
I
have
there.
So
hopefully
this
isn't
used
to
anyone,
but
I
didn't
bring
it
up
because
it
is
all
happening
next
week
at
kubecon.
We've
got
the
Cubit
project
meeting
now
this
actually
during
the
kubecon
week,
but
before
kubecon
starts
it's
on
Tuesday
from
10
30
to
12
30,
talking
about
Community
the
roadmap
and
use
cases
it
is
somewhere
in
the
let's
click
on
the
link.
It
tells
you
what
room
it's
in.
It's
in
the.
B
I
can't
remember
what
it's
called
Congress
Center,
that's
the
one
and
I
think
that's
on
the
third
floor,
but
best
fourth
floor.
There
we
go
fourth
floor
of
the
Congress
Center
and
then
we'll
we're
gonna
have
some
kind
of
lunch
afterwards,
on
the
same
day,
Fabian
and
Ryan
are
giving
a
talk
at
the
openshift
Commons
at
2
30.
about
convert
powering
GeForce
now
and
then
I've
got
a
list
there
of
the
other
kids
that
specific
talks
there.
B
So
I've
got
to
maintain
a
talk
which
is
talking
with
our
end
users,
a
panel
with
a
delegates
from
our
end
users,
and
then
we've
got
yep.
B
Operator
day,
two
stuff
by
Simone
having
a
230
and
then
later
in
the
afternoon,
is
Uber,
giving
a
non-root
Made
Easy
talk.
Now,
that's
all
happening
on
Thursday,
so
Tuesday
and
Thursday
are
a
big
cubic
days,
but
we
will
have
a
booth
there.
Wednesday
Thursday
and
Friday
we've
got
a
full-time
boost
at
cubert.
It
will
be
in
the
project
really
in
the
I
can't
quite
make
sense
to
the
map
because
they're
totally
disjointed,
but
we're.
D
B
Somewhere
in
the
in
the
top
Corner
top
right
corner
of
that
map
near
the
yarn
wall,
so,
if
you're
looking
for
us,
that's
where
we'll
be
yes
and
the
other
point
I
have
there
I
believe
is
the
cubert
summit
recordings.
Hopefully,
everyone
saw
the
email
they're,
all
posted
they're,
all
there.
If
you
see
any
any
glitches,
please
let
me
know
I'll
see
if
there's
a
problem
on
YouTube's
end
or
my
end.
B
Yeah,
it
was
kind
of
nice
flicking
through
them
and
actually
being
able
to.
You
know,
take
a
bit
more
information
in
so
yeah
thanks
to
everyone
who
attended
and
has
been
checking
out
the
videos
afterwards
and.
B
A
A
E
Sure,
oh
sorry,
I
just
a
little
confused
I
have
a
question.
E
Is
there
any
intended
way
to
propagate
updates
from
virtual
machine
spec
to
Virtual
Machine
instance,
I
found
that
there
are
no
ways
to
update
labels,
node
selectors
and
Affinity
rules,
for
example,
I
guess
they
can
be
updated
online
and
I
see
that
some
vendors
are
actually
doing
that,
for
example,
Harvester
they
do
update
not
selector
before
the
live
migration
to
migrate
on
specific
node
and
I
was
thinking
to
implement
the
same
thing
in
our
platform,
but
I
see
that
virtual
machine
image
spec
is
not
is
immutable,
so
I
have
a
question.
F
F
There's
some
Nuance
there
I
think
that
the
decides
into
hot
plug
it
ties
into
a
couple.
F
Other
conversations,
I,
don't
know
if
there's
a
general
rule
about
what
we
should
do
or
not
do
it's
just
generally
been
accepted
so
far
that
you
would
accept
changes
on
the
next
reboot
but,
like
you
pointed
out
there
are,
there
are
legitimate
use
cases
where
it'd
be
really
useful
to
do
it
on
the
Fly,
such
as
changing
labels,
so
that
when
you
do
a
migrate,
you
pick
up
new
label
selectors
or
you
know
so
that
you
you
can
control
what
nodes
you
migrate
to,
for
instance.
F
E
Okay,
that's
a
good
idea
anyway,
if
I
understood
it
correctly.
For
now,
there
is
no
way
to
update
such
information
like
labels
and
every
time
you
have
to
recreate
V
my
object.
F
Correct
that
is
how
it
works
today
and
we're
I
would
just
say
we
should
proceed
with
caution.
We
don't
want
to
do
it
generally
for
all
Fields
the
same
way,
because,
as
I
was
hinting
with
hot
blood,
for
instance,
there
might
be
things
where
or
like,
for
instance,
some
things
don't
make
sense
today,
as
as
of
127.
like
changing
resource
limits,
there's
just
no
way
that
can
translate
to
the
VMI,
because
kubernetes
doesn't
support
it
as
of
today.
F
Now
there's
potential
that
in
the
future
that
could
change,
but
there's
an
implied
migration
at
the
very
minimum
behind
accomplishing
the
change
and
that
by
the
way,
is
something
that
we
could
explore.
We
have
been
mentally
thinking
about
is
when
you
make
a
change
to
the
VM
next
time
you
migrate,
the
VMI,
it
picks
up
the
new
changes,
so
you're,
not
necessarily
waiting
all
the
way
until
a
full
reboot.
E
So
I,
just
thinking
even
kubernetes,
allows
you
to
update
some
fields
in
ports
back.
However,
everybody
thinks
that
it
is
immutable,
but
it
is
not
fully
immutable.
You
can
still
update
image,
for
example,
without
recreating
Parts,
not
sure
why
it
was
needed,
but
I
think
we
can
allow
to
modify
some
Fields
like
Affinity,
not
selectors
and
stuff,
like
that,
and
do
not
allow
updating
any
other
fields
which
are
not
supporting
the
live
updates.
F
So
the
Affinity
is
a
funny
one
that
you
bring
up,
because
when
you
like
that
effect
scheduling,
but
it
doesn't
affect
running
so
if
you've
got
an
existing
pod
running
somewhere,
you
change
its
Affinity.
It
doesn't
I
mean
you
can
take
effect
immediately,
but
it
doesn't
do
anything
until
you
move
the
Pod
around.
E
Yeah,
that's
true:
I
think
this
might
be
solved
by
running
live
migration
after
that
change.
D
E
A
A
All
right
looks
like
a
niche
yeah.
G
And
the
next
one
is
mine,
so
there
is
this
per
request,
so
basically
they
would
like
to
extend
Q
first
API
for
adding
local
time,
and
basically
this
will
configure
the
timing
of
the
guest
based
on
the
time
zone
of
the
host.
G
So
I'm
not
familiar
with
this
flag,
but
it
seems
to
be
a
suggested
flag
for
running
Windows
in
liver
documentation.
However,
I'm
wondering
what
happened
if
we
migrate
to
a
host
that
has
a
different
time
zone,
does
anyone
have
experience
with
this.
F
That's
curious
I
mean
usually
when
you
take
virtualization
out
of
the
loop.
Usually
the
general
standard
has
been
you.
You
leave
the
hardware
clock
on
the
machine
in
UTC
and
then
you
configure
local
time
at
the
operating
system
level
per
the
user
preferences
and
with
that
there
wouldn't
need
to
be
a
change
at
the.
F
If,
if
that
were
the
approach
taken,
you'd
need
it
at
the
the
guest
level,
which
would
mean
that
you
would
need
guest
agent
cooperation
to
make
that
work
right.
G
Yeah
I,
don't
think,
is
this:
this
option
is
really
for
a
cloud
I,
don't
know.
G
G
Reach
things:
if
you
see
a
card,
maybe
can
you
go
on
fixes?
Yeah
I
mean
that
was
this,
this
issue
up
and
did
I
reopened,
because
I
was.
G
G
G
A
All
right
then
looks
like
cfp
notice
from
Andrew,
so
anyone
is
interested
in
submitting
a
talk,
as
always
we're
here
to
help
review
and
support
you
in
writing
and
submitting
those
cfps.
F
B
To
highlight
this
on
behalf
of
Lee,
because
I
hadn't
seen
it
I
get
any
responses.
H
So
we've
done
something
similar
in
CBI
when
we
went
from
Alpha
One
to
beta1
and
the
way
I
understand
it.
As
long
as
the
new
version
of
the
object
has
the
same
Fields
plus,
maybe
some
extra
ones,
because
you
know
it's
it's
a
newer
version,
then
the
update
happens
automatically.
You
don't
have
to
do
anything
special,
so
I
was
sort
of
confused
when
I
saw
the
the
email
it
was
it.
It
should
just
magically
happen.
So
at
least
that's
my
understanding.
I
No,
so
the
if
there's
a
there's
a
kubernetes
there's
a
kubernetes
controller
that
takes
care
of
all
that.
But
basically
it
has
to.
I
Happen
right,
all
the
resources
have
to
be
I,
think
holds
and
updated.
Like
you
have
to
get
the
you
know,
version
string,
API
version
string,
updated
in
that
TV
and
that
doesn't
happen
automatically.
H
I
Control
I
think
that's
mentioned
in
this
email
somewhere.
That
does
it
like.
If
you
look
at
the
code
Alexander
that
we
did
to
do
this,
we
well
we
list
all
the
resources
and
explicitly
call
update
to
do
it.
I
That's
not
my
understanding,
but
but
there
and
that's
because
we
can't
rely
on
that,
but
there
is
a
part
of
the
kubernetes
project.
That
is
a
controller
that's
like,
so
the
other
thing
is
not
just
updating
the
resources
in
SCD.
There
is
a
a
field
in
the
crd
itself
that
has
stored
versions
and
you
have
to
delete
an
entry
in
there
and.
C
I
There
are
either
you
can
build
all
this
functionality
yourself
or
there's
like
a
controller
that
does
it
but
yeah.
So
all
the
resources
have
to
be
updated.
Have
the
API
version
strings
updated
in
that
CD
and
the
crd
itself
has
to
have
a
the
stored
versions
set
to
just
the
you
know.
Most
recent.
A
Yeah
and
well,
if
anyone
wanted
to
chime
in
on
the
mailing
list,
it
looks
like
that's
going
to
be
a
lively
conversation.
A
And
jumping
into
bugs
we
have
a
few
staged
here,
there's
anything
that
you
want
special
attention
to
feel
free
to
add
your
own
bugs
to
the
bug
review.
A
Printed
so.
H
A
A
Trying
to
are
there
like
custom
parameters
that
you
can
set
for
communities
host
health
checks.
A
A
It's
been
a
long
time
since
I've
used
other
standard
hypervisors,
but
I
feel
like
at
least
VMware
or
hyper-v.
You
might
have
some
ability
to
not
schedule
nodes
when.