►
From YouTube: KubeVirt Community Meeting 2021-11-10
Description
Meeting Notes: https://docs.google.com/document/d/1kyhpWlEPzZtQJSjJlAqhPcn3t0Mt_o0amhpuNPGs1Ls/edit#heading=h.pozqokb2ojl
A
Hello,
everyone
welcome
to
this
week's
kubert
community
meeting.
I'm
your
host
chris
caligari-
and
this
is
your
chance
to
talk
about
how
you
use
kubevert
talk
with
our
core
developers
and
just
talk
about
how
how
things
are
going
for
you.
A
Okay,
he
notes
our
post
of
the
chat.
If
you
could
please
fill
in
your
attendance,
we
appreciate
that
we
do
track
that
information
go
ahead
and
fill
out
any
agenda
items
that
you'd
like
to
speak
about
today,
and
while
everybody
is
doing
that,
do
we
have
any
any
new
folks
with
us
today
that
would
like
to
introduce
themselves.
B
Hi,
I'm
philipp,
I'm
I'm
developing
some
some
networking
for
phantom
technologies
and
I
want
to
discuss
some
stuff
from
prs
regarding
regarding
b,
hosts
support.
A
A
A
A
Okay,
I
think
andre
and
I
spoke
earlier
and
it
looks
like
he's
trying
to
use
he's
trying
to
connect
kubert
with
gloucester
using
the
csi
driver,
correct
and
he'd
he'd
like
to
know
best
practices
on
how
to
make
that
happen.
F
C
A
We
just
swamped
slideshare.net,
go
away.
C
C
C
Have
okay
88
85
of
reduction
because
we
are
doing
virtual
desktops.
C
G
A
How
come
you
chose
gluster
versus
seth.
C
Because
we
know
gluster
and
vgo
on
regular
overt
in
deployment
that
we
was
using
before
implement,
convert
behind
overt
was
working
perfectly.
We
just
introduced
instead
of
using
overt,
let's
say,
cluster
we
use
convert
instead
of
of
of
earth
cluster,
and
then
we
can
reach
1500.
A
Understood
yeah
reason
I
ask
is
if,
if
you
start
purchasing
support
through
red
hat
they're,
the
the
cluster
storage
has
gone
the
soft
direction.
A
Michael,
maybe
you
can
you
can
talk
about
what
what's
happening
with
the
gloucester
versus
stuff?
You
know,
that's
been
a
hop
hot
topic.
Yeah.
C
A
Yeah,
it's
it's
basically
rook,
oh
really,
yeah
and
then
and
then
rook
interfaces
with
the
soft
cluster.
A
Look,
I
know
okay,
but
where
erasure
coding,
sorry
so
you're
gonna
get
you're
gonna
get
space
space
savings
with
with
seth
using
erasure,
coding
and
and
deduplication.
A
So
it's
it's
space
savings
at
different
layers
of
the
of
the
stack.
A
Yeah,
would
you
mind
if
I,
if
I
followed
up
with
that
note,
those
bullet
points
after
the
community
meeting
yeah?
Thank
you
so
much
yeah,
not
a
not
a
problem.
It's
it's!
Just
gluster
is
it's.
What
17
year
old
technology
now
and
they've
tried
to
they've
tried
to
keep
it
keep
a
community.
A
C
A
C
Need
momentum,
and,
but
that's
why
we
we
are
here
with
the
community:
what
is
the
best
practice?
Don't
screw
the
project
in
the
middle
storage.
A
Cluster
storage
has
definitely
gone
the
rook
in
sap
direction,
so
it's
it
took
over.
Openstack
has
taken
over
kubernetes
and
and
just
general
cluster
storage.
A
I
can
fill
out
some
some
bullet
points
for
you
after
you
know,
maybe
that
can
help
you
move
away
from
cluster.
Hopefully
you
don't
have
a
massive
data
migration
that
would
no.
E
A
Yeah
we've
we
heard
about
petabytes
of
data
having
to
move
from
gloucester
to
staff,
and
it's
just
like
cover
your
eyes
and.
C
Forget
about
a
week
later
or
a
month
later,
yeah.
A
All
pipes
are
are
moving
data
at
100.
A
G
Three,
I'm
not
sure
if
threes
on
the
call
also
discovered
that
upgrading
from
046
to
0
47
is
broken
due
to
a
pr
change.
It's
not
clear
to
me
why
our
end-to-end
tests
did
not
catch
this,
but
we
have
a
reward
of
this
pr
open
and
as
soon
as
it
is
in,
we
will
do
another
stero
47
release
and
let
me
just
also
post
the
revert
pr.
G
A
G
G
The
issues
just
if
people
install
3047.0
now
that
we
would
fix
at
least
right
now
what
we
have
in
the
revert
would
fix
zero
the
code
and
you
can
go
from
zero,
forty
six
to
zero
to
zero
zero
forty
seven
dot
one,
but
not
from
zero.
Forty
seven
dot:
zero.
Four:
zero,
zero,
seven
that
one.
So
we
probably
have
to
improve
the
fix.
G
A
Well,
let's
make
sure
we
don't
abandon
anybody,
that's
already
gone
into
470.
A
Or
yeah,
let's
make
sure
that
we
have
a
good
escape
plan
for
them.
D
Yes,
michael
gotcha,.
D
Some
weeks
ago,
we
presented
a
idea
about
the
hypothesis
for
supporting,
and
after
that
we
collected
some
questions,
and
originally
we
are
addressing
that.
We
made
some
progress
in
the
networking
issue.
Now
we
managed
to
make
the
network
in
zen
vm
work
in
bridging
mode,
and
it
will
be
described
in
the
design
proposal.
We
can
discuss
the
detail
then,
and
another
critical
issue
is
in
globalize.
D
It's
the
resource
problem.
Now
we
haven't
figured
out
a
solution.
It
looks
a
bit
troublesome
completely
running
in
from
zero,
cannot
see
the
real
resources
that
will
be
allocated
to
vms
kubulate,
uses
c
advisor
to
check
capacity.
Theodizer
obtained
information
from
system
files,
for
example.
The
memory
data
is
from
slash,
precise,
slash
memory,
mm-info
file,
and
I
didn't
find
anyway
that
in
kuberlete
we
can
intercede
or
extend
to
change
this
mechanism.
D
Besides
that,
we
began
to
look
into
where
to
cooperate.
According
to
its
document
is
a
globalized
alternative
that
can
be
customized,
including
resources,
but
now
I'm
not
familiar
not
very
familiar
with
that.
I'm
not
sure
whether
it
can
work
with
kubovert
and
there
is
some
other
problem
as
the
cooper.
D
G
D
G
I
G
Is
in
general
used
so
that
you
can,
for
instance,
create
at
least
the
idea
behind
it
is
that
you
can,
for
instance,
create
pods
on
on
gcs
or
amazon
cloud
or
whatever,
and
they
are
not
really
parts.
It's
not
just
like
you
report
them
like
ports,
but
they
probably
just
represent
an
arbitrary
application
managed
somehow,
but.
G
Yeah,
yes,
please,
but,
but
what
what
you?
What
you
definitely
lose
in
principle,
when
you're
using
the
virtual
cubelet,
is
that
there
is
any
existing
pod
implementation
for
there
for
you
already
right
and
if
hubert
would
want
to
use
it
or
if
you
would
want
to
use
it
with
sen,
you
would
probably
also
have
to
implement
something
like
a
virtual
word
handler,
it's
a
virtual
keyboard.
G
D
That
this
could
work
together,
yeah
they
made
that
provider.
I
think
we
need
to
implement
our
own
some
something
as
the
backhand
yeah.
Okay.
D
Well,
we'll
continue
to
look
into
that
from
more
detail
and
check,
and
I
think
another
question
is
that
whether
the
copylight
can
work
with
cooperate.
G
Definitely
not
yeah,
so
no
because
because
cubert
has
the
strict
assumption
that
there
is
a
container
random
implementation
running
below
hubert,
a
bit
below
cube
below
the
cubelet
like
docker
or
cryo
or
any
other
one,
and
that
they
are
name
space
and
c
group
based
so
from
the
kernel
mechanism
perspective.
If
they're,
not
name,
space
and
z
group
based
keyword
does
not
work
on
it.
So
if
the
virtual
cube
rate
would
be
used,
our
whole
node
demand
would
have
to
know
about
this
and
have
to
work
with
this
assumption.
G
G
What,
if
I
may
ask
you
a
question
yeah,
but
I
did
not
understand
yet
this.
Are
you
looking
into
send
because
you
for
cuba,
because
you
want
to
run,
send
based
applications
for
development
and
verification
on
kubert?
Or
do
you
really
want
to
use
a
kubernetes
cluster
and
keyboard
to
run,
send
hypervised
applications
on
production
systems
yeah
in.
D
D
I
C
C
We
we
have
several
people
working
on
on
convert
code
right
now,
and
they
can.
We
can
point
them
to
also
became
approvers
in
this
community.
Okay,.
G
G
If
you
are,
we
have-
and
I
think
yeah
that
that
sorry,
that
grace
is
already
yeah.
Chris
is
already
opening
the
guides
on
how
to
become
a
member
of
the
community
and
how
to
become
an
approver
so
code.
G
If
you
anyone
can
code
with
you,
it's
just
github
right,
but
you
can
become
a
member
when
you
did
some
good
code
reviews
and
contributions
over
time
and
you
get
to
sponsors,
that's
what
chris
just
said
open
and
when
you
then
also
show
that
you
managed
to
do
more
complex
features
end
to
end
we're
able
to
help
other
people
sorting
out
how
to
design
things
properly.
G
G
But
right
now
we're
yeah
there's
really
a
lot
of
a
lot
of
things
are
going
on,
and
sometimes
you
don't
get
automatically.
They
prove
a
look
at
the
pr
when
it
looks
good
to
me,
it's
already
there
and
we're
right
now,
not
always
anymore
in
in
the
position
to
do,
looks
good
to
me
and
the
proof
in
one
go
because
it's
just
too
much.
A
Usually,
what
I
do
is
I
hunt
roman
or
david
down
directly
and
and
drag
them
over
to
github,
kicking
and
screaming
yes,
but.
G
We
have
more
yes,
but
we
have
more
approvers
than
david
and
me
just
to
name
a
few
and-
and
I
think
yeah,
maybe
maybe
good
just
because
that's
this
that's
the
point
of
true:
we
have
the
owners
file
and
I'll
just
share
the
link
with
you
chris.
Maybe
you
can
open
it
for
a
second
here
we
go.
G
So
I
think
the
top
three
entries
david
radick
and
me
we're
definitely
the
people
which
can
give
the
brothers
the
proofs
without
any
issues,
but
there
are
also
others
which
are
active
like
yeah,
not
so
much.
Stu
and
fabian
were
more
on
metal
level
and
also
the
alien
danielle
belenki,
which
we
should
move
to
emiratus.
A
G
G
A
Great
yeah
just
be
vocal
and
you'll
be
noticed.
A
A
So
I'm
always
looking
for
for
new
people
to
to
introduce
myself
to
and
say,
hi
good
work
and
have
a
t-shirt.
A
Yeah
yeah,
it's
a
it's
it's
getting
reflected
in
our
in
our
incubation
process.
We've
got
we've
added
all
of
our
our
adopters
to
our
front
page
recently,
so
we've
been
trucking
along
for
about
five
years
now
with
without
any
front
page
advertisement.
A
So
that
leads
me
into
our
our
next
topic.
Actually
arm
has
recently
allowed
us
to
advertise
on
cooper.io
and
add
them
to
the
adopters
file.
A
Thank
you
that
was
a
a
tough
process
to
get
through
and
and
since
I'm
talking
about
arm,
let
me
just
sorry
I
have
to
talk
for
about
five
ten
minutes,
guys
about
community
stuff,
roman.
I
was
happy
to
announce
four
seven
can.
A
A
Okay,
happy
to
connect
you
with
howard,
andre
and
to
work
on
arm,
64,
stuff,
sure.
A
Stu-
and
I
just
did
a
demo
for
all
things-
open
in
raleigh,
north
carolina,
demonstrating
a
coupe
running
cooper
and
virtual
machines
on
arm
64..
I
think
it
went
well
stu
you,
you
think
it
went
well.
H
A
A
This
is
running
on
a
very
large
arm,
64
server
and
that
was
donated
to
the
to
the
overall
kubernetes
community
and
and
kubert
has
allowed
a
little
bit
to
gather
a
little
bit
of
resources
to
gather
github
events
and
show
some
show
some
charts.
A
K
A
And
like
it's
impressive,
it's
got,
and
I
think
the
most
impressive
thing
is
is
how
much
how
little
power
it
consumes.
So
it's
got
massive
core
count:
massive
memory,
low
power
and
it's
cheap.
A
A
So
I
every
time
you
guys
fill
in
these
attendees
right
here.
I
take
this
information
over
to
him
and
give
and
fill
in
a
report
and
we
track
how
many
users
are
in
in
the
community
meeting
week
to
week-
and
I
see
we
have
22
people
with
us
today,
but
I
don't
see
22
entries
here.
A
A
Annual
usage
survey
got
sent
out
yesterday,
if
you're
using
kubevert
in
any
way
shape
or
form.
Please
fill
out
that
survey.
We
need
it
for
our
incubation
process.
A
Another
bullet
point
in
incubation
process
is
an
official
bug
manager
not
looking
for
anybody
to
actually
fix
all
the
bugs
just
looking
for
somebody
to
make
sure
that
they're
not
sitting
idle
they're
to
take
just
like
to
handheld
hand,
hold
the
ish
github
issues
through
the
process.
A
We
are
seeing
really
good
results
with
with
the
bug
scrub.
This
year,
issues
have
sat
for
two
weeks
in
the
past
and
now
we're
seeing
issues
get
addressed
within
four
days.
So
the
bug
scrubbing
that
we're
doing
during
the
community
meeting
has
had
a
big
effect.
A
Let's
see
arm,
has
agreed
to
to
be
interviewed.
Thank
you
for
that,
but
we
we
need
to
get
five.
A
We
need
five
and
we
have
three
so
far
as
events
I
sent
out
a
email
regarding
some
potential
upcoming
events.
If
anybody
would
like
to
present
kubevert
at
one
of
those
events,
let
me
know
I'll
get
you
an
issue
created
in
the
community
repo
and
we
can
track.
A
We
can
track
the
event
through
that.
C
A
So
we're
looking
for
official
end
users,
you
would
be
added
to
the
adopters
file
and
advertise
on
covert.io
and
the
cncf
would
contact
you
directly
for
a
30
minute
interview
over
zoom.
A
A
Oh
excellent,
thank
you.
Do
you
have
a
an
email
address
that
we
can
use
to
forward
to
the
our
sponsor.
C
A
Got
it?
Thank
you
so
much.
I
will
I'll
forward
this
on
to
the
cncf.
The
ladies
name
is
elena
and
she
is
with
apple
computers.
She's
kubert's,
sponsor
with
the
cncf
uh-huh,
so
she'll
be
in
contact
with
you
for
scheduling
and
interviewing
wonderful.
C
I
can
put
another
of
the
honors
also
to
became
five.
A
And
if,
if
you
want
to
be
added
to
the
adopters
file,
feel
free
to
to
upgrade.
A
Better
you
do
it,
then
I
do
it
because
then
we
can
use
the
get
blame.
Yeah,
yeah,
okay,
747
phillip,
has
a
question
about
a
pull
request.
B
Basically,
this
is
out
course
request,
maybe
like
half
a
year
old
and
it's
about
the
viho
support
in
in
cube
beard
and
basically
there
is
an
information
implementation
that
is,
I
don't
know
almost
done.
However,
there
is
some
networking
project
that
I
want
to
create
and
I
want
to
some
sort
of
rewrite
this
request.
B
Basically,
I
have
a
very
sim
similar
like
topology,
but
not
probably
using
the
obs
dpdk
but
replace
it
with
with
fdios
bpp,
the
vector
packet
processing,
some
sort
of
like
software
router
and
basically
my
question.
My
question
is
I
I
I
checked
the
request.
I
checked
all
the
discussions
and
and
seems
like
it's
missing
like
it's
missing
the
final
kicking
implementation,
it's
missing
the
end-to-end
testing
and
it's
missing
some
sort
of
documentation
and
and
and
user
guide.
B
So
basically,
my
question
is
better.
If
this
pieces
will
be
provided,
whether
there
is
a
like
a
high
chance
that
you
can
guys
approve
it
like
I
know
you
have,
you
have
like
a
code
review
it
I
have
to.
I
have
to
do
all
the
of
the
changes
that
that
you
say
that
you
probably
wanted
a
little
bit
different.
That's
all
right,
however.
B
The
question
is,
if,
if
I
do
all
these
steps
of
all
the
tests
or
the
documentation
and
stuff
whether
this
could
get
merged
or
whether
there
is
some
sort
of
like
like
problem
like
I
don't
know,
kwemi
doesn't
support
that
and
it's
not
somewhere
in
the
comments
or
whether,
whether
you
decided
that
you
don't
want
to
go
with
the
v-host
or
or
something
like
that.
So
I
have.
B
I
want
to
get
some
like
further
brother
confirmation
that
in
this
way
I
get
when
I,
when
I
commit
to
to
make
the
effort
and
and
process
without
these
steps,
then
that
these
things
got
merged.
B
G
To
ask,
I
think
it's
some
time
it
sometime
passed
until
we
looked
into
this
and
I
think,
for
instance,
david
did
a
code
review
on
radio
already
on
it
as
far
as
I've
seen,
but
I
think
it's
hard
to
answer
that
right
now
here,
yeah
yeah.
I
understand
that
I
I
also
know
that
some
etwin
testing
was
missing
and
that
some
packages
needed
to
be
installed
and
so
on.
G
But
a
lot
of
things
also
changed
in
the
meantime,
and
I
think
we
also
have
to
look
into
that
what
it
would
exactly
do
and
how
it
fits
into
the
privilege
privileges,
structure
of
the
project.
Can
you
can
you
bring
this
to
cuba
devil
to
the
mailing
list
and
maybe
link
the
two
or
three
existing
pr's,
and
we
can
then
look
into
this
again.
G
So
definitely
not
just
go,
as
you
already
said,
do
not
immediately
go
away
and
quote
everything
and
then
hope
that
you
accept
just,
but
just
give
us
a
small
heads
up,
no
notice.
Somebody
just
sit
here
with
a
few
links
and
we
can
look
into
this
again.
I
think
then
we
can
yeah.
That
would
be
great,
so
you're
going
to
write
to
puberty
and
then
sorry
can
you
repeat
that
so
I
you're
going
to
write
an
email
to
keep
your
dev.
G
G
G
A
G
L
Yeah
I
have
one-
I
have
one
one
comment
about
this,
so
I
want
also
to
mention
this
is
not
only
now
a
technical
issue
in
sense
of
supporting
this.
It's
also
if
there
is
real
need
for
it
in
in
a
broader
community,
because
in
general
this
was
in
previous
projects
includes
that
are
related
to
virtualization.
L
This
effort
were
dropped
like
most
of
the
effort
to
have
the
user
space
application
on
the
node
side
to
accelerate
v-host
were
actually
dropped
at
some
point,
because
the
srov
hardware
was
started
to
be
very
cheap
and
very
common,
and
any
serious
enterprise
solution
will
include
them.
So
we
need
to
also
tackle
the
actual
need
here,
because
it's
questionable,
if
this
is
this,
is
really
worth
the
effort
to
to
keep
it
and
maintain.
M
It's
just
one
thing
here
and
what
that
was
a
really
good
point
you
made,
but
I
really
think
this
is
very
useful,
especially
for
single
node
deployments.
If
you
just
have
like
a
single
node,
dpdk
would
theoretically
behave
better.
So
that's
like
from
from
my
perspective,
at
least
that's
a
positive
thing
for
this
feature.
L
But
this
is,
I
don't
know
if
a
single
you
know
this
is
like
a
common
use
case.
It's
it's
one
small
use
case
and
supporting.
G
A
This
is
starting
to
sound,
like
we
need
to
have
a
a
design
document.
L
L
L
This
was
not
the
case,
because
then
srv
was
not
that
common
and
it
was
really
expensive,
but
now
that
it
is
so
common
asking
someone
to
just
buy
the
the
correct
hardware
to
have
the
correct
to
have
this
abilities
small
comma
just
you
need
just
to
start
with
the
convincing
that
it
is
needed.
B
Okay
and
and
how
I
actually
should
I
don't
know,
persuade
you
that
it's
it's
common,
that
I
don't
know
I
can,
I
can
say
like
why
I
need
it
in
the
project
and
and
stuff
like
that
and
how
I
want
to
how
I
want
to
connect
the
things.
However,
how
can
I
say
that
it's
it's
it's
common.
I
don't
know.
G
I
would
say
just
start
the
conversation
with
your
use
case,
that's
fine
and
with
a
few
pointers
of
what
we
have,
and
there
is
already,
for
instance,
we
already
have
quite
some
documentation
in
the
old
prs
on
how
this
needs
to
be
configured,
and
we
need
to
look
into
so
one
aspect
is
definitely
like
edward
said.
If
this
will
then
be
used
at
all
by
more
than
one
person,
but
in
other
aspects
also
how
difficult
it
would
be
for
us
to
maintain
it.
What
do
we
have
to
do
to
and
test
it?
How?
B
Yeah,
I
understand
your
effort
to
maintain,
it
must
be,
must
be,
on
the
other
hand,
weights
and
by
the
by
the
people,
using
it,
of
course,
so.
G
Even
yeah,
and
even
if,
if
it
turns
out
that
it's
too
hot
or
too
hard
for
us
to
maintain
it,
we
may
still
be
able
to
look
for
maybe
improving
some
hooks
which
we
have
for
so
so
that
you
can
just
do
it
anyway,
without
having
core
support.
So
just
you
know
if
there's
a
there,
are
also
other
ways,
then,
if
necessary,
we
will
see.
G
B
Okay,
I
I
yeah.
B
And
also
try
to
try
to
try
to
write
what
does
to
need
to
be
maintained
from
your
site.
Maybe
that
works
all
right.
Yeah.
A
Seems
so,
okay,
well,
that
takes
us
to
7
58,
no
time
for
bug
scrub.
So
I'm
I'm
gonna
override
a
bug
scrub
this
week
and
say
we'll
do
it
next
week
and
I
will
return
one
minute
back
to
everybody.
A
Okay,
thank
you.
Everybody
have
a
good
week
and
we'll
see
you
next
week
same
time
same
place.