►
From YouTube: KubeVirt Community Meeting 2023-07-26
Description
Meeting Notes: https://docs.google.com/document/d/1nE09vQWcCTW-9Ohe9oCldWrE0he-T_YFJ5D1xNzMtg4/
A
A
And
with
that
welcome
everybody
to
today's
cubert
community
meeting.
It
is
the
26th
of
July,
which
seems
totally
insane,
but
here
we
are
oh.
If
you
could
please
put
down
your
name
in
the
attendees
list
at
the
top
there,
as
I
said,
I've
put
the
agenda
into
chat
and
yeah
we'll
get
going.
A
Do
we
have
any
new
members
this
week
or
people
that
have
maybe
lurked
before
in
this
meeting
and
haven't
introduced
themselves
and
I'd
like
to
take
that
opportunity?
Now.
A
Next
on,
our
list
is
oh
yeah
and
yeah.
If
you've
got
anything
we'd
like
to
talk
about
anything
you'd
like
to
highlight,
or
any
questions
like
to
ask
about
the
agenda
and
the
open
floor,
companies
are
the
one
and
likewise
pull
requests
and
bugs
so
I'm
in
there
we
will
get
to
them.
First
up.
We
have
the
schedule.
Check-In
click
on
this.
A
Ta
say
that
it
was.
This
must
be
this
week,
so
we're
starting
the
1.28
provider,
which
I
saw
an
email
about.
Thank
you
Brian,
oh
next
week
was
that
three
weeks
time
we'll
have
the
kubernetes
release.
A
All
things
going:
well
all
the
variety
another
thing
I'm
just
going
to
jump
ahead
a
little
bit,
because
this
is
another
kind
of
a
check-in
and
we've
got
some
cvfps
which
are
open
and
some
of
them
close
really
soon
got
four
K
CDs,
which
are
coming
kubernetes,
Community,
Days,
they're,
quite
small
city-centric
Regional
events
I've
never
been
to
one
but
I
hear
they're
a
lot
of
fun.
A
I've
got
Austria,
Sri,
Lanka,
Texas
and
Denmark,
and
we've
also
got
the
open
source
Summit
in
Japan,
which
has
just
opened
up,
and
you
can
see
the
the
submission
dates
there
and
all
the
links
there,
if
you're,
not
interested
if
anyone's
going
to
flock
Ryan
is
giving
a
talk
there.
A
August,
second,
to
the
fourth
close
that
all
right
I
saw.
So
it's
kind
of
like
a
mailing
list
thing
is,
is
Leon
the
call
at
all.
A
Fair
enough,
he
sent
an
email
I
think
it
was
the
end
of
last
week
asking
about
the
populating
the
the
sigs
that
we
created.
A
Basically,
it
is
a
mechanism
so
that
we
have
kind
of
component
and
and
Sig
specific
reviewers
and
approvers,
so
that
was
a
change
that
was
made,
but
we're
now,
as
Lee
White
rightly
points
out,
need
to
populate
those
and
so
here's
creative
thread
which
you
can
read
over
here
and
he's
created
a
meeting
next
week,
which
I'll
just
click
on
here,
because
I
can't
remember
which
day
it
is
also
Thursday
at
what
is
that
BST,
oh
UK,.
B
A
I
think
just
wanted
to
raise
awareness
of
that.
A
A
We
could
try.
I
did
try
on
the
the
one
from
read.
Ai
I
did
try
and
denialist
that
domain,
but
it
doesn't
seem
to
have
worked
at
all.
B
A
Why
don't
we
move
the
open
floor
and
I
can
see
if
I
can
figure
out
how
to
dump
people
joao?
Are
you
here
yeah.
C
So
hello,
everyone,
so
basically
I
wanted
to
raise
a
discussion
about
the
recent
alert
that
I'm
I
opened
up
here
to
add
to
Cubit
and
it's
related
to
some
CRC
efforts
that
you
are
finding
on
the
cluster
nodes
when
the
the
users
are
basically
creating
virtual
machines
using
a
walk
volume
for
the
safe
provisioner
that
is
not
using
RX
phones
so
and
Igor
is
also
present
in
the
call.
It
is
the
the
question
if
we
should
actually
add
this
opinionated
alert
right,
because
this
is
very
specific
to
a
storage
provisioner.
C
My
idea
is
that
if
we
are
aware
of
the
problem-
and
we
can
do
something
to
at
least
raise
awareness
for
the
users
that
this
problem
exists,
it
might
cause
them
issues
with
their
VMS
and
even
severely
integrated
performance
with
the
cluster.
We
should
alert
them
even
if
it
really
doesn't
totally
fit
what
the
Cubit
is
supposed
to
do.
D
Yeah
sorry
for
the
delay,
can
you
hear
me
yeah
great,
so
yeah
I
would
like
to
share
my
kind
of
concern
regarding
this
and
the
alert
we
are
speaking
about
related
to
a
specific
storage
provision.
Okay,
let's
say
it's
a
specific
CSI
among
other
csis
rights
that
we
theoretically
support.
Now
we
don't
now
the
there
is
a
code
overhead
introduced
here
in
order
to
to
implement
this
current
alert.
D
To
be
more
specific,
we
need
to
add
the
persistent
volume
shared
indexer
into
the
build
controller,
just
for
the
sake
of
watching
the
virtual
machines
and
those
who
have
this
specific
data,
consistent
volume
claim
with
the
set
provisional
professional
and
we
have
some
incorrect
attributes
inside
the
physical,
the
persistent
volume.
Only
then
the.
E
D
Risk
or
the
CRC
for
the
Iowa
or
CRC
errors.
Now,
from
my
perspective
well,
this
is
fine
to
have,
for
example,
alerts
that
are
related
to
operator
observability,
okay
for
operator,
health
for
control,
plane
and
some
common
alerts
related
to
workloads
in
Cuba.
That
can
indicate
some
states
some
information
about
the
virtual
machine,
but
this
is
but
this
looks
to
me
as
a
coronary
case
and
I'm.
Also
I'm
I
feel
that
we
don't
have
a
quorum
regarding
whether
we
would
accept
this
alert
into
our
code
or
not
so.
D
Hear
some
opinion
of
others
other
open
source
projects,
let's
say
let's
say,
for
example,
for
each
CD:
okay,
when
it's
ecd,
for
example,
it
doesn't,
let's
I,
don't
know,
imagine
a
theoretical
scenario
where
it's
in
the
offer.
It
doesn't
play
nice,
some
providers.
Okay,
so
should
we
should?
Will
it
CD
accept
an
opinionated
alert
regarding
a
use
case
where
hcd
runs
on
top
of
specific
provider
and
or
whether
that
alert
should
be
outside
of
its
CD
core
Ripple
and
to
have,
among
some
other
reports
related
to
observability.
C
Yeah
and
I
want
to
talk
that.
Maybe
we
are
looking
also
at
this
pneumatic
and
the
overhead
is
just
for
the
alert,
but
actually
personally
think
that
this
actually
might
be
useful,
for
example,
because
after
this
Edition
we'll
also
still
have
information
about
which
volumes
previous
VMS
were
using.
So
you
can
have
more
historical
and
statistical
information
about
the
volumes
that
VMS
were
using,
so
that
might
also
be
useful
for
users
in
some
other
contexts,
so
that
actually
might
be
also
pointing
in
favor
of
this
Edition.
C
But
yeah
I
understand
your
concerns
and,
as
I
said
in
the
beginning,
it's
not
really
related
to
car
keyword
functionality,
but
then
again,
I
think
this
is
important
for
users.
This
has
been
a
serious
issue
for
many
of
them
and,
although
it's
a
corner
case,
maybe
be,
it
makes
sense
to
to
try
to
prevent
them
for
facing
this
sort
of
serious
problems.
F
E
G
E
We're
designing
a
way
to
avoid
these
CRT
errors
in
the
first
place.
I
yeah
I
mean
I,
definitely
agree
that
it
would
be
nice
to
report
them,
but
I.
H
C
But
right
now,
what
will
be
the
alternative
left
to
users
and
one
because
any
any
other
option-
and
we
thought
about
maybe,
for
example,
this
to
CVI
right,
but
the
the
question
would
be
the
same
cdis
for,
for
example,
but
the
volumes
it's
not
really
mixed.
It
doesn't
really
make
sense
to
add
it
third,
so
we
don't
have
like
a
good
place
to
have
this
and.
C
F
E
Is
better
but
it's
still
not
ideal,
but
I
think
it
is
better.
A
A
Yeah
I
I
I'm,
not
I,
didn't
realize
that
you
probably
can't
see
me.
I
Hey
how's,
it
going
welcome
yeah
good
thanks.
How
are.
A
You
and
very
well
you've
got
the
next
point
if
you'd
like
to
take
it
from
here.
I
Yes,
here
is
a
problem
when
we,
when
I
trying
to
build
the
Fedora
with
test
tooling
image
in
the
processor
I
got
the
foreign
error
message
in
case.
Maybe
someone
don't
know
why
we
do
this
because
Fedora
with
tests,
the
touring
image
is
used
by
some
e2e
tests.
I
So
we
try
to
build
multi-arc
Fedora
with
testature
image
to
enable
more
if
we
test
on
64
platform
and
we
have
written
a
script
for
each
and
in
this
script
we
use
word
install
to
install
some
RPM
package
in
the
virtual
machine
and
then
we
use
vertices
play
Q
generator
Q
equal
to
image
and
when
it's
wrong
word
says
pray,
it
goes
the
following
Arrow,
it
shows
it's
unable
to
mount
a
PTR
FS
file
system
volume.
I
H
Yeah,
hey
Howard,
how
you
doing
hi
Brian
here
I
was
looking
at
this
today
and
yeah.
I'd
agree
with
you.
The
the
worker
nodes
are
missing:
support
for
the
ptrs
s
file
system,
but
that's
due
to
the
worker
work,
worker
workloads,
coaster
being
a
openshift
cluster
which
basically
Rel
and,
by
extension,
Red
Hat
core
OS,
have
dropped,
support
for
btrfs,
so
I'm
not
sure
what
direction
we
can
take
with
this
we've
kind
of
struggled
to
keep
up
to
date.
H
With
these
these
test,
VM
images
as
well
so
I
believe
this
is
based
the
test.
Tooling,
one
is
based
on
Fedora
35,
which
is
a
long
time
out
of
days
and
the
one
the
real-time
one
is
based
off
of
fedora
32.,
so
yeah
I
I'm
not
sure
what
the
way
forward
for
this
would
be
I
either.
H
I
can
provision
these
locally
from
my
machine
here
and
get
them
uploaded
to
Quay
and
then
have
them
there,
or
we
may
have
to
look
at
moving
these
via
VM
images
to
maybe
a
center
stream
option
where
we
wouldn't
have
to
update
them
as
regularly
and
I
believe
there
wouldn't
be
a
requirement
for
ptrfs
there
either.
I
H
G
H
Don't
think
we
can
easily
add
that
there
and
from
what
I
saw
today-
and
it
wouldn't
be
a
simple
thing
to
fix
so
I
I
can
see
two
paths
forward
since
we've
kind
of
struggled
to
maintain
the
version,
the
floor
versions
and
these
VMS
either
move
to
kind
of
a
Centos
stream
VM
with
test
tooling
or
basically,
it
would
come
down
to
someone
on
the
info
team,
like
myself
or
Daniel.
H
H
Yeah
that
that
is
one
route
yeah
but,
as
as
you
see
here
as
well,
that's
based
off
of
Fedora
35,
which
is
already
I,
think
three
versions
back
which
should
already
nine
around
the
corner.
So
we
might
have
to
talk
about
moving
to
the
center
stream
testing
VM
at
some
stage.
Plus
for
now,
I
can
talk
with
you
offline
and
we
can
get
these
published
and
we
can
move
on
from
here
then
from
that.
A
Are
there
other
benefits
to
having
a
center
stream
VM
for
this
kind
of
stuff.
H
Yes,
it
will
just
yeah,
it
would
just
be
this.
We
wouldn't
have
to
the
updated
cycle,
wouldn't
be
as
strenuous.
So
basically
it
wouldn't
be.
Every
I
think
it's
six
months
for
Fedora,
but
like
the
real
time,
the
real-time
VM
is
back
on
Fedora
32,
so
they
haven't
been
updated
in
quite
a
while,
so
yeah
it's
just
might
be
better
for
just
if
we
go
to
Center
stream
nine,
then
we
don't
have
to
update
for
quite
a
while.
A
A
F
H
G
A
B
Yes,
I
just
wanted
to
to
update.
There
was
a
little
bit
of
progress
in
this
area
and
the
one
that
is
is
looking
for
feedback.
B
Is
there
is
a
concrete
example
of
how
to
do
a
a
network
binding
using
a
plugin,
at
least
in
theory,
I
would
say
with
a
concrete
example
what
needs
to
be
done
and
what
happened
at
this
stage,
and
if,
if
you
want
to
look
at
it,
maybe
you'll
see
something
that
you
think
is
missing
can
can
dig
in,
and
there
is
another
one
that
is
going
to
be
finished,
I
hope
tomorrow,
which
is
about
slip,
something
that
no
one
uses.
But
it
is
a
good
example
of
something
that
is
a
bit
different.
B
That's
it
probably
will
add
more
examples
for
each
binding
and
to
complete
the,
and
then
it
would
be
from
an
example
to
a
concrete,
so
something
this
is
what
we
want
to
do
eventually
feel
free
to
feedback.
Thank
you.
B
B
A
I
did
I
did
see
on
that
thread.
Someone
I
can't
remember
who
it
was
I
did
ask
what
the
outcome
of
the
previous
meeting
was.
B
A
Fine,
so
yeah,
if
you,
if
you
are
interested
in
that
look
out
for
Eddie's,
invite
for
what
sounds
like
next
week.
G
Oh
yeah
hi,
so
it's
a
pretty
small
PR
I.
It's
already
reviewed
and
I'll
get
them
it's
there.
I
just
need
an
approval.
It's
for
exposing
a
metrics
from
literature.
This
is
the
only
one
which
is
missing,
I,
think
and
yeah.
We
will
definitely
it
will
be
beneficial
for
us
to
have
it,
as
you
can
see,
I
think
I'm
looking
for
approval
like
for
a
week
yeah.
So
if
anyone
can
spend
like
five
minutes
on
it,
it's
very
very
simple
and
a
short
one.
C
D
I
sorry
I
will
see
myself
on
on
that
one.
A
Alrighty
we've
got
two
design
proposals,
I'll
just
kind
of
bring
these
up.
You
can
read
them
in
your
own
time.
I
just
wanted
to
yeah
throw
some
item.
One
is
from
Lee
about
common
instance
types
and
the
other
one
is
also
about
a
lot
collecting
I
subscribe.
A
From
Cuba
planets
yep,
so
those
therefore
the
links
have
a
look
at
them.
Provide
feedback
all
that
good
stuff.
This
one
is
well
I,
don't
know
if
you
saw
Miguel's
email
and
blog
posts
celebrating,
they
can
now
write
Network
policies
for
secondary
network
interfaces,
so
there's
a
blog
post
down
here
and
there's
also
a
certain
dogs
that
is
included.
A
A
So
that's
interesting,
there's
a
it's
a
big
read,
but
it's
mostly
configs.
Thank
you
very
much
for
that
Miguel,
as
we
saw
earlier,
we've
got.
1.28
provider
is
now
available.
Again,
that's
just
a
highlight.
Thank
you!
Brian
and
a
good
conversation.
That's
happening
about
that.
Daniel
raised
I,
think
Daniel's
on
holiday
at
the
moment,
or
at
least
out
of
office,
about
what
could
have
been
a
unit
test
rather
than
end-to-end
test
yeah.
A
A
Where
is
it?
Writing
better
tests
and
scrapping
older
tests
when
they
come
on
non-functional
and
yeah,
looking
after
the
resources
that
we'll
rely
upon
when
it
comes
to
testing.
So
please
have
a
read
of
that
and
we
have.
F
Yes,
Shelley
just
posted
this
to
the
kid
for
death
mailing
list.
F
Because
it's
essentially
integrating
with
volume
populators
in
the
PM
templates
and
sorry
in
the
VM
definition.
Instead
of
like
data
volume
templates.
A
Awesome.
Thank
you
that
looks
great.
It's
nice
to
see
all
these
design
professors
come
through
lots
of
like
where
to
dig
down.
A
One
bug
to
look
at
as
far
as
I
could
tell
if
you
do
have
a
bug
that
you
would
like
attention
on.
Please
add
it
here.
Otherwise,
we've
got
this
one
that
was
rated
by
Vasily
I
think
today,
and
he
was
hoping
to
get
a
second
brain
to
have
a
look
at
it.
F
A
Okay,
okay,
she's
also
ccued
all
right
well.
If
anyone
else
is
interested
just
to
have
a
little
look
there,
it
is
at
the
bottom
of
the
page,
and
that
brings
us
to
the
end
of
our
current
agenda.
Let
me
just
check
to
see
that
nothing's
been
added
and
we
did
get
rid
of
those
two
butts.
Thank
you
to
whoever
slipped
into
the
cncf
cupert
account
and
done
to
them.
E
Stu
I
uncooked
yeah
I
couldn't
log
back
into
my
normal
account.
A
Oh
snap,
all
right
well,
yeah,
thanks
for
doing
that,
and
yeah
I'll
have
to
look
into
a
more
long-term
solution
to
that,
because
I
suspect
next
week
the
same
thing
is
going
to
happen
and.
A
Yeah
and
what
I
found
about
those
things
is
because
we
have
a
public
meeting
that
has
no
like
what
are
they
called
waiting
rooms
or
anything
like
that.
Basically,
anyone
that
has
that
enabled
on
their
side
that
says
that
they
want
to
join
this
meeting
that
bot
automatically
joins.
A
Those
meetings
seems
to
be
a
yeah,
a
potentially
growing
headache
for
the
future,
but
yeah
if
they
don't
have
anything
at
the
last
minute,
they'd
like
to
to
raise
any
Kudos
I'd
like
to
throw
around
or
anything
they'd
like
to
just
pop
in
before
we
all
take
off.
A
Oh,
okay,
that's
good
to
know.
I
had
a
look
around
and
I
couldn't
see
anything
because
it
doesn't
actually
list
who's
in
our
meetings
we
just
get
the
recording
and
some
like.
We
don't
get
a
lot
of
analytics
from
it,
but
I
will
I'll.
Take
another
look
yeah.
Let's,
maybe
sync
up
offline
sounds
good.
Thank
you.