►
Description
No description was provided for this meeting.
If this is YOUR meeting, an easy way to fix this is to add a description to your video, wherever mtngs.io found it (probably YouTube).
A
Thanks
everyone,
my
name
is
Daniel
Hiller
I'm,
a
software
engineer
from
Red
Hat
I
am
mostly
I'm
concerned
with
maintaining
the
cube
with
CI
infrastructure
so
where
our
ete
and
tests
run
and
not
directly
related
to
the
development
of
cupert
just
for
the
automation
part.
So
if
you
have
any
questions
later
on
that,
you
don't
want,
don't
need
to
or
that
you
that
you
don't
remember
now,
you
can
post
them
later.
If
you
want,
you
could
probably
ping
me
at
Twitter
or
look
at
my
GitHub
account
somehow.
A
So
what
is
this
about?
This
is
about
kubernetes
and
about
virtual
machines.
A
Who
of
you
is
using
kubernetes
already:
okay,
a
lot
of
people,
although
it's
a
little
bit
brightening
here
so
and
who
is
also
using
virtual
machine
somehow
a
couple
of
people?
Okay,
so
who
has
already
connected
virtual
machines
to
kubernetes
containers?
A
Okay,
so
actually
that's
what
what
this
talk
is
about.
So
as
I,
don't
need
to
explain,
kubernetes
I'm,
just
going
to
spare
you
that
so
the
the
basic
thing
is,
of
course,
that
it's,
as
you
all
know,
it's
a
software
for
running
containers.
Workloads
virtual
machines,
in
contrast,
need
some
kind
of
emulation
and
a
hypervisor
to
run,
which
is
not
present
in
the
kubernetes
cluster.
B
A
What
would
be
if
you
could
just
put
your
old
machine
that
is
somewhere
in
the
closet
and
just
take
it
and
put
it
inside
your
cluster,
where
it
would
then
probably
just
could
could
be
able
to
turn
it
off
when
you
need
it
or
to
turn
it
on
when
you
need
it
and
turn
it
off,
when
you
don't
need
it
or
also
what
would
be
or
what
what
benefit
will
you
get
when
you
would
put
your
virtual
machine
inside
a
kubernetes
cluster
and
could
let
it
talk
directly
to
the
containerized
workloads.
A
So,
most
of
the
time,
it's
like
you
have
something
like
a
a
workload
or
or
a
simple
application
that
runs
somewhere
in
a
in
a
virtual
machine,
probably
or
even
on
an
application
on
an
application
server
somewhere,
and
you
have
the
problem
of
connecting
it
to
an
existing
set
of
workloads
that
are
running
somewhere
else
and
that's
exactly
what
qubit
is
about
to
solve.
A
So
all
the
people
that
have
already
heard
something
about
Qbert
here,
a
couple
of
hands
I,
see:
okay,
so
cube
root
in
general,
addresses
the
needs
of
development
teams
that
process
existing
virtual
machine
workloads.
That
cannot
be
easily
containerized,
for
whatever
reason
I'd
say,
for
example,
you
haven't
had
yet
the
time
to
containerize
it.
A
Maybe
it's
so
critical
that
you
just
need
to
continue
running
it
like
that,
but
you
still
most
of
the
times
at
some
point,
want
to
move
somewhere
else
and
want
to
probably
get
rid
of
that
virtual
machine
stuff
that
you're
running
in
a
different
set
of
environment,
and
you
want
to
consolidate
all
this
into
one
system.
So
you
can
get
rid
of
those
split
Paradigm
of
several
environments
that
you
need
to
manage.
A
So
what
Cooper
does
is
it
provides
a
unified
development
platform
that
you
can
run
application
containers,
as
well
as
virtual
machines
inside
a
common
and
shared
environment.
In
short,
Cupid
makes
it
possible
to
manage
VMS
in
the
same
environment
as
containers
and
also
just
to
manage
VMS
as
kubernetes
objects.
A
So
I
don't
know
if
you
can
see
that,
but
that's
a
virtual
machine
definition
for
cube
word
I,
just
Sperry
the
details
of
that.
It's
just
like,
like
any
other
kubernetes
object.
Definition,
it's
a
yaml
with
a
spec
and
with
some
attachments
of
price,
for
example,
and
also
with
network
definitions
and
also
with
volumes
that
you
can
then
just
attach
to
the
virtual
machine
yeah.
A
So
much
for
Theory
my
story
when
I
tried
preparing
or
tried
imagining
what
I
wanted
to
show
here
was
that
I
wanted
to
show
probably
how
easy
it
would
be
to
migrate
a
virtual
machine
into
a
kubernetes
cluster,
so
I
thought
yeah.
The
the
way
would
obviously
be
prepare
the
virtual
machine
for
import
somehow
so
that
it
could
run
in
a
in
a
different
environment
than
probably
convert
this
disk
image
that
you
have
into
an
importable
format.
That
would
work
with
qubit,
then
import
the
image
and
create
the
VM
and,
of
course,
success.
A
So
I
had
an
old
virtual
machine
that
I
had
set
up
years
ago
for
my
kids,
because
we
don't
want
to
run
any
Windows
laptops
in
our
home
or
at
that
time
I
didn't
even
have
a
Windows
laptop,
but
my
kids
would
have
some
learning
program
that
they
would
use
for
school.
So
I
had
to
set
up
one
virtual
machine
where
that
learning
program
would
run.
Apologies
already
for
that,
it's
just
in
German
I.
If
you
have
question
on
that,
if
you
needed
some
explanations,
I'd
be
happy
to
to
help
you
on
that.
A
So,
first
of
all,
the
preparation
for
the
virtual
machine
image
was
a
real
hassle.
I
worked
a
couple
of
months
on
that,
so
just
to
find
out
how
to
probably
convert
the
virtual
machine
definition
inside
the
inside
the
image
to
align
to
what
is
currently
supported
by
cupert.
So,
for
example,
this
image
was
an
old
image
that
had
configured
an
IDE
bus,
a
storage
device
and
IDE
is
just
not
supported
anymore,
of
course,
by
by
Cuba,
because
it's
so
old
that
it
that
it
just
doesn't
support
it.
A
Also
Windows
XP
is
an
end-of-life
retrieve
operating
system,
of
course,
and
for
example,
when
you
want
to
run
it
in
any
virtualized
environment,
for
example
like
in
a
in
the
virtual
machine
manager
from
that
is
using
KVM
you,
you
just
even
need
to
select
this
this,
this
checkbox,
where
you
want
to
select
your
end
of
life
operating
systems.
A
So
normally
what
wouldn't
you
show
up
in
that
in
that
list
of
a
supported
operating
systems
so
yeah
The,
Verge,
vertio
drivers,
I
tried
for
using
storage,
but
I
didn't
find
any
driver
that
would
work
with
my
old
Windows
XP.
So
I
was
totally
annoyed
and
I
I
nearly
tried
to
give
it
up
and
just
to
cancel
the
talk,
because
I
didn't
I
didn't
manage
it.
A
That
was
that
was
very,
very
frustrating,
but
in
the
end,
I
had
a
solution
just
to
convert
this
storage
device
inside
the
virtualbox
itself
into
a
SATA
device,
and
then
it
worked,
but
still
a
lot
of
blue
screens
when
starting
up
the
virtual
machine.
A
But
in
the
end
it's
somehow
worked
so
I
think
what,
for
me
worked.
The
best
I
still
haven't
figured
out
exactly
what
qimo
operating
environment
does
to
that
virtual
machine
image,
but
somehow
it
magically
worked.
I,
don't
know
why
I
need
to
figure
that
out.
I
must
have
a
bit
sorry
for
that.
A
So
the
rest
of
the
things
was
quite
easy.
Just
to
import
the
image
into
cupboard
using
the
CDI
upload
proxy.
That
I've,
already
present
from
the
cupid
product
was
relatively
easy
so
that
it
would
convert
the
existing
qca
image
into
the
target.
I
think
it's
raw
format
that
is
used
by
keyboard
and
then
create
the
VM.
A
The
creation
of
the
VM
was
also
not
that
easy,
because
cupboard
as
the
processor
model
that
the
that
the
virtual
machine
would
want
to
have
or
wasn't
provided
by
default
on
on
keyboard,
so
I
had
to
patch
the
machine
types
that
are
present
there,
that
you
can
use
and
then
finally,
it
was
a
success.
I
had
a
running
virtual
machine
inside
Cube
word
with
the
caveat
the
sound
didn't
work.
I
need
to
work
on
that,
so,
okay,
I'm
now
going
to
start
the
demo.
A
The
demo
environment
I'm
going
to
use
is
our
standard,
Cupid,
CI
environment
that
we're
using
in
the
Clusters
to
run
integration
tests
or
e2e
tests.
Even
it
has
enabled
these
are
a
couple
of
dockerized
cluster
nodes
inside
running
locally.
A
A
About
the
project,
so
cupid
itself
is
an
open
source
project
that
is
in
a
cncf
incubation
State.
It's
been
adopted
by
several
vendors
and
end
users.
It
has
more
than
three
and
a
half
thousand
GitHub
3400
I'm,
sorry
230,
plus
contributors,
800,
forks
and
more
than
5000
per
requests.
So
it
has
a
monthly
release
schedule.
At
the
moment
we
are
working
on
extending
that
a
little
bit.
A
A
So
the
features
that
I'm
going
to
show
are
on
the
left
side
like
VM
image
report,
live
migration,
snapshot,
restore
monitoring
and
serial
I'm,
just
showing
the
graphical
console
access
to
be
honest,
but
it
also
has
serial
console
access.
So
the
other
things
that
I
wanted
to
mention
was
hot
plug
volumes
and
vgpu
and
media
edit
devices,
node
assignment
Cloud,
init
and
sysprep,
and
what
not
not
not
unimportant
I
would
say
is
a
zero
downtime
rolling
updates.
A
So
a
couple
of
features
the
community
is
working
on
is,
for
example,
memory,
encryption,
multi-arch
clusters,
so
that
we
have
also,
for
example,
arm
support
inside
a
cluster,
that
is
that
or
that
we
have
diverse
architectures
inside
a
kubernetes
cluster
that
we
can
then
use
for
cupert.
Several
instances
of
machines
and
another
feature
the
community
is
working
on.
Is
the
VM
export?
A
It's
pretty
easy
to
install.
You
can
just
extract.
The
latest
version
from
the
from
the
release
then
apply
a
yaml,
and
the
operator
will
do
the
rest
when
you're
installing
this.
A
Now
I
hope
the
import
is
already
done.
Yeah,
it
is
so
one
oh
yeah,
I
forgot
or
I
I
would
have
shown
you
probably
that
the
import
uses
a
scratch
image
to
then
convert
the
image
into
the
real
persistent
volume
client
that
I'm
using
now
so
I'm
just
going
to
show
you
now
the
already
created
virtual
machine,
so
this
is
still
starting
and
it's
still
in
scheduling.
A
The
next
thing
I'm
going
to
do
is
just
to
create
a
connection
with
a
VNC
viewer
to
the
virtual
machine.
A
And
finally,
have
this
virtual
machine
inside
a
kubernetes,
kubernetes
cluster
running,
to
be
honest,
I
must
admit:
I
didn't
yet
completely
figure
out
the
the
best
configuration
of
the
virtual
machine
so
to
provide
any
Hardware
so
that
it
wouldn't
thinking
would
be
thinking
that
it
would
would
have
some
discovered
some.
Some
new
hardware
I
need
to
fiddle
around
with
the
com
with
a
configuration.
A
So
that's
why
the
the
hardware
system
always
nags
me
about
some
some
Hardware,
that
it
has
found
somehow
and
to
be
honest,
the
VNC
viewer
I'm
using,
is
also
not
really
optimal.
A
A
The
metrics,
the
monitoring
so
first
of
all,
every
keyword
component
has
metrics
like
which
you
can
scrape
with
a
Prometheus
instance,
and
you
can
of
course
put
a
nice
grafana
dashboard
on
that
I've
just
put
up
this
one,
just
as
an
example.
This
is
from
the
cupid
API
component,
where
you
can
see
lots
of
stuff
that
that
makes
sense
to
observe
somehow.
A
Also,
of
course,
every
virtual
machine
exports,
some
metrics
if
I
need
to
reload
here,
I
guess
and
make
it
a
little
bit
bigger,
probably
or
decrease
the
time
span,
so
that
you
can,
for
example,
see
the
relevant
measures
that
that
you
need
to
look
at
when
you,
when
you
want
to
observe
your
VM
State.
Somehow
so
it
says:
CPU,
image,
usage
memory,
usage,
disk
usage
and
network
traffic,
that's
just
an
example.
You
could
probably
just
create
a
much
more
sophisticated
dashboard,
because
all
these.
A
A
So,
just
to
keep
you
a
little
bit
in
the
loop
of
what
I'm
doing
here.
The
snapshot
object
is
so
similar
or
also
just
a
kubernetes
object
that
you're
going
to
create
on
the
cluster,
and
it
then
just
stores
the
current
state
of
the
virtual
machine.
While
it's
running
in
a
snapshot
so
now
I'm
going
to
do
something
like
a
destructive
operation.
A
Of
course
it
worked
in
the
in
the
preparation
of
the
demo,
but
now
it
doesn't
work
anymore
sure.
So
let
me
see
I'm
just
going
to
remove
some
icons
from
the
from
the
menu.
Let
me
see
some
if
I
find
it
this
one
then
remove
it,
delete
it
yeah
of
course,
I
want
that
I
would
normally
clean
the
the
trash
can
just
to
show
you
that
it
would
be
there
already
after
the
after
I
restore
that,
but
for
now
this
should
be
enough
just
to
just
to
show
you
the
point
of
that.
A
A
A
A
So,
okay,
then
I'm,
just
oh
yeah
one
other
thing
I
wanted
to
mention-
is
of
course,
that
the
VNC
connection
to
the
running
virtual
machine
is,
of
course,
going
away
when
you're
going
to
stop
the
virtual
machine.
A
So
that's
what
you
saw
here
that
the
websocket
was
closed.
Somehow.
A
A
And
then
see,
for
example,
here
you
see
that
the
operation
has
already
completed
so
the
snapshot
has
been
restored.
A
A
So,
as
you
can
see,
always
the
status
of
the
virtual
machine
is
always
reflected
in
the
respective
kubernetes
objects
that
you're
seeing
on
the
right
side,
like
the
virtual
machine,
has
its
own
status,
and
the
virtual
machine
instance
also
has
its
own
status.
If
it's
then
scheduled,
then
you
can
see,
for
example,
on
which
node
it
will
run,
which
will
be
interesting
later
when
we
are
going
to
live
migration.
A
B
A
So,
okay,
you
know
we
have
something
running
here
and
that
will
be
already
the
last
part
of
the
demo.
So
now
you
see
this
is
running.
The
program
is
running
and
now
I'm
going
to
live
migrate,
the
virtual
machine
does,
and
everyone
know
what
a
live
migration
is
I
see
some
nodding
faces.
Okay,
I
should
have
put
it
probably
the
other
way
around
who
does
not
know
what
a
live.
Migration
is.
A
Okay
so
Cube
cuddle
word
migrate.
Windows,
XP,
I
could
also,
of
course,
create
a
virtual
machine.
Migration
object:
oh
okay,
but
I'm
just
doing
it
with
the
command
line
tool.
This
this
is
by
the
way.
This
is
an
integration
so
that
it
works
with
Cube
cuddle
together
as
a
plugin.
There
is
also
a
separate
binary
that
you
can
just
use
in
download,
so
you
this
is
just
another
way
to
run
this.
So
please
look
at
the
node
name.
You
will
at
some
point
in
time.
A
A
So
that
is
already
my
my
live
demo.
A
So
let
me
get
back
to
the
to
the
presentation
so
yeah
my
findings,
I
cut
it
a
little
bit
short,
but
in
general,
most
the
most
work
from
my
opinion,
because
I'm,
not
an
expert
in
that
matter.
A
A
For
me,
QE
email
system,
x86,
the
the
the
binary
that
runs
virtual
machines
inside
Linux,
has
done
wonders
to
just
being
able
to
start
it
inside
cupboard
and
to
be
honest,
I
still
need
to
find
out
what
that
hap.
What
happened
there
and
yeah
one
last
thing
is
I
want
to
get
the
sound
working
I.
There
is
some
option
that
you
can
enable
so
that
you
even
have
sound,
but
if
I
remember
correctly,
VNC
does
not
support
it
by
its
protocol,
but
I
think
there
is
some
Fork.
A
A
C
We
have
a
little
bit
of
time.
We
have
I,
think
eight
minutes
or
so
to
get
through
these.
So
let's,
let's
get
right
to
it.
Will
it
work
with
VMS
that
have
multiple
network
interfaces
I'm.
A
I'm,
not
an
expert
in
networking
I
must
admit,
but
I
have
seen
that
their
maltas
is
supported,
for
example,
so
I
think
it
should
work.
Okay,.
C
C
If
I
understood
correctly
you're
using
a
raw
disk
image
for
your
demo
is
I'm
I'm.
Sorry
I'm
not
familiar
with
qcal
2
disk
format,
not
supported
by
Cube
verd.
A
No,
actually,
it
is,
it
is
Q
card
is
from
imported
I.
Think
there
is
also
the
support
for
virtualbox
images
is
already
also,
and
you
can
you
can
use
raw
Cuca
or
I'm,
not
exactly
sure.
What
else
is
supported
to
be
honest,
I
would
need
to
look
that
up.
C
Okay,
oh
okay
and,
of
course
that's
when
my
phone
and
stuff
off
Can
pods
from
cluster
access,
the
VM
or
any
service
which
is
running
on
it.
Is
it
any
network
visibility
between
this
VM
object
and
other
pods?
Oh.
A
Yeah,
of
course,
because
it's
the
normal
kubernetes
object
it
takes,
you
can
just
use
the
normal
Network
definitions
that
you
use
for
any
other
object.
Great.
B
A
Yeah
I
thought
I
answered
that
but
yeah
in
general.
Normally
when
you
have
some
non-virtualized
workload
or
a
virtualized
workload,
that
is
not
yet
containerized
that
you
want
to
make
it
talk
to
Containers.
So
anything
that
is
still
residing
inside
a
inside
a
virtual
machine
could
just
be
easily
transferred
to
kubernetes
class.
Then
you
would
you
would
let
it
talk
to
the
containers
very.
A
I
think
there
are
quite
a
lot
of
adopters
that
are
already
using
it
in
in
production.
So,
for
example,
openshift
virtualization
is
the
downstream
project
of
cupboard
and
to
my
knowledge,
it
should
be
ready
for
production,
okay
and
but
even
Qbert
should
be
ready
for
production.
As
of
now.
C
Okay,
how
a
node
failure
is
handled
and
are
there
any
manual
actions
necessary.
A
I
would
say
that
you
would
probably
try
to
yeah.
Node
failures
are
a
little
bit
trickier.
I'd,
say
I
think
that
for
now
you
would
probably
need
some
manual
interaction
to
migrate,
yeah,
the
node
Trader
itself.
If
the
note
goes
away,
then
I
to
be
honest,
I
don't
know
what
happens.
A
It's
gone
so
yeah.
You
would
need
to
probably
try
to
try
to
make
regular
snapshots
and
backups
and
then
just
cover
that
so
yeah.
C
Make
sense
can
I
just
drain
a
node
and
the
VM
will
be
live
migrated
to
another
node.
A
The
live
migration,
I,
don't
think
it
will
work
automatically
to
be
honest
and
I.
Think
I'm
I,
you
know
I,
don't
know,
I,
don't
know,
I
need
to
stumble.
Okay,.
C
A
Are
those
are
all
possible
right?
You
can
I
forgot
to
to
mention
I,
guess
that
the
the
node
selection
works
like
in
any
other
for
any
other
part.
For
example,
like
you,
you
just
would
would
create
a
node
selector
or
an
affinity
or
anti-affinity
rule.
Okay,.
C
And
the
last
one
from
Chris
thank
you
for
not
submitting
this
anonymously
for
once.
I
was
still
friends
with
VMware
or
other
hypervisor
vendors
question
mark.
Oh.
C
All
right
well
on
that
note:
let's
leave
it
there,
please
give
another
warm
Round
of
Applause
and
enjoy
your
lunch.