►
From YouTube: KubeVirt Community Weekly Meeting - 2018-07-18
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
I'm,
just
starting
this
and
I
will
hand
it
over
to
you.
What
is
there
I
would
like
to
start
or
actually
with
my
own
item
I,
don't
want
to
be
unfriendly,
but
it's
more
on
the
organizational
side.
So,
as
you
saw,
there
were
quite
a
few
design
documents
pushed
out
to
the
mailing
list
about
several
virtualization
storage
related
topics,
including
networking,
actually
and
they're.
Here
at
a
redhead,
we
actually
had
a
big
gathering
to
speak
about
those
designs
because
they're
important
to
us.
A
We
drive
drove
some
of
them
and
we
want
to
provide
our
feedback
on
them
back
into
the
design
documents
and
we
actually
plan
to
start
working
on
that
I
think
everybody
in
the
community
will
see
when
we
when
will
do
so,
and
so
the
first
follow-up
will
be
to
to
straighten
the
design
so
that
this
really
something
which
can
be
worked
on
okay
and
good
to
see
some
new
faces
on.
This
call
welcome
and
if
you
have
any
questions,
please
step
up
and
let
us
know
one
when
there
is
so
I'm
done
with
my
item.
B
A
C
Can
anybody
see
my
screen
as
expected,
excellent,
so,
okay,
the
website
has
had
a
redesign
here
recently
and
I
just
wanted
to
share
with
everybody
the
current
website
as
it
stands
today.
We
just
might
move
that
over
just
a
little
bit,
so
this
is
got
the
old
logo
and
the
old
kind
of
structure
that
we
had.
C
Let's
keep
that
going
and
then
resources
page
which
had
videos
and
then
you'll
also
notice,
like
in
the
footer,
where
we
had
all
of
our
social
media
information
as
well
as
how
to
participate
in
the
community
to
some
extent
and
then
the
gdpr
privacy
stuff
was
kind
of
pasted
over
here
on
the
end.
So
without
further
ado,
the
new
website
has
a
new
logo
up
on
the
top
left
for
Qbert,
which
looks
really
good
and
then
you'll
just
see.
C
Basically,
like
a
nice
visual
kind
of
aesthetic
overhaul
of
the
entire
website,
so
right
now
we
have
basically
these
two
ways
to
try
kubernetes
one
of
the
things
that
I
did
just
want
to.
Let
out
of
the
box
really
quickly
is
that
we
are
working
on
two
more
buttons
and
we're
have
a
try,
cube
Burke
on
GC
P
and
a
try
cube
vert
on
AWS.
C
The
only
thing
that's
blocking,
that
from
being
released
right
now
is
basically
making
sure
that
all
of
the
a.m.
eyes
that
we
create
are
verified
through
CI
and
CD,
so
that
that's
in
process
now-
and
you
should
see
those
up
here
shortly-
hopefully,
and
in
through
the
rest
of
the
homepage.
We
have
a
nice
overview
of
what
Qbert
is
and
some
really
good
looking
icons
on
the
webpage
and
then
also
the
footer
has
been
updated
as
well,
and
you
can
still
get
to
our
Twitter
account
github
the
Google,
Group
email
address
and
also
the
calendar.
C
C
You
also
tagged
these
objects
so
that
they
can
be
organized
over
on
the
left-hand
navigation
screen
and
the
next
thing
that
we
have
are
the
docs
I'm,
not
exactly
sure
how
this
is
going
to
end
up,
but
right
now
it
basically
redirects
to
the
old
websites
user
guide,
but
I
think
that
we
do
want
to
keep
all
of
the
documentation
embedded
into
this
current
kind
of
framework
here.
So
we're
still
working
on
that,
but
it's
making
some
progress
and
then
last
but
not
least,
is
the
community
page.
C
We
want
to
make
sure
that
it's
really
easy
for
people
to
get
involved
in
this.
So
we
have
a
quick
way
to
star
the
project
to
fork.
The
project
through
github
follow
us
on
Twitter
and
then
join
the
mailing
list,
and
let
me
see
I
think
that's
pretty
much
it
for
right
now.
Please
go
out
and
take
a
look
at
the
PR
and
provide
any
feedback
to
the
website
that
you
think
needs
to
be
implemented
and
we'd
be
happy
to
take
a
look
at
it
and
I.
Think
that
does
it.
B
A
I
just
want
to
say
thank
you
for
the
revamp,
I
think
it's
step
forward.
There
were
some
small
glitches
to
me
that
acceptable
I,
don't
see
it
as
a
blocking
item.
I
think
I
would
be
loved
to
find
the
issues
and
others
please
provide
your
feedback.
If
you
have
flames,
take
care
to
not
burn
your
computer,
you
won't
burn
the
Internet
No.
Please
provide
feedback
so
that
we
can
move
forward
with
that.
I
think.
For
me,
let's
see
that
we
can
move
merchants
rather
sooner
the
maker
Thanks.
D
E
F
I,
just
I
just
wanted
to
say
from
a
design
point
of
view
being
on
the
UX
team.
I
know
that
we
discussed
you
know
the
end
user
of
the
website
and
I
think
that
we
tried
to
make
sure
it
had
more
of
a
consumer
look
and
feel
whereas
pattern
fly
is,
is
really
more
about
more
of
an
enterprise.
Admin
type
look
and
feel
so
we
wanted
to
make
sure
we
differentiated
the
community
website
from
from
that,
and
that's
why
they're
different
and
why
we
didn't
consider
easy.
F
Think
you
know
the
the
information
that
we
present
on
the
website
is
very
good
from
the
information
that
we
design
in
our
enterprise
applications
that
our
users
use.
You
know
we
want
to
make
it
very
clear,
large
buttons,
bright
colors,
something
that
a
user
is
going
to
other
than
the
blog.
You
know,
probably
just
to
get
started
and
figure
out
what
this
thing
is
about.
So
it's
really
more
informational,
whereas
our
enterprise
applications
are.
F
You
is
that
our
users
are
going
to
be
using
everyday
and
you
know
a
lot
more
dense
content,
and
so
there
are
definitely
big
differences
in
use
cases.
And
that's
that's
really
why
I
think
the
designs
are
so
are
so
different
and
if
you
look
at
a
lot
of
other
community
websites-
and
you
know
think
about
over
versus
you
know
over
toward
versus
you
know
the
over
admin
portal.
I
think
you
know,
there's
there's
a
very
strong
alignment
between
a
lot
of
the
other
communities
out
there
and
their
admin,
UI
tools,
so
I
think.
G
F
E
So
so
far
we
were
using
for
our
CI
some
internal
servers
from
reddit,
and
we
have
as
kind
of
a
soft
eyeliner
which
is
at
the
end
of
this
month
and
kind
of
a
hard
deadline
which
is
the
end
of
August.
We
have
to
move
out
of
those
between
my
of
these
service
and
your
currently
in
a
transition
phase
over
to
probably
right
now,
standard,
CI,
switchover
de
saucé
using
word
and
to
have
a
lot
of
resources.
We
have
a
perspective
and
we
may
be
able
to
use
dead.
E
Yeah
might
still
not
mean
that
we
can't
use
pro
in
front
of
that
so
they're
also
working
on
still
missing
more
into
pro
and
yeah.
It's
the
short-term
solution,
at
least
we'll
see
where
that
goes
can't
say
much
more.
We'll
try
to
keep
the
CI
service
up
as
much
as
possible,
since
it's
very
vital
for
us.
B
H
Yeah
hi,
so
basically
it's
more
the
oven,
open
questions
because
it
started
quite
a
discussion
about
it
in
the
issue
in
the
notes
limitation
started
from
we
use
of
identifiers,
so
I
mean
labels
using
the
BM
names
which
opened
that
issue
from
tossed
out
us.
But
then
we
started
discussing
about
if
we
really
do
want
to
limit
the
VM
name
according
some
thoughts
or
about
limiting
to
DNS
labels
conventions
in
VMware,
for
example.
H
H
E
I,
don't
see
a
reason
to
limit
the
VM
name
since
also
the
inverse
seems
to
differentiate
between
the
actual
hostname
and
the
VM
name,
and
can
I
mean
I
seen.
We
should
just
allow
as
well
as
names
as
pods
long
names
and
we're
sanitizing
them
anyway,
and
because
I
have
a
host
name
fields
where,
when
you
import
for
being
where
you
can
set
the
VM
name,
oh
yeah,
I.
I
I
We
were
sanitizing
it
Asura
No.
So
which
brings
me
to
my
next.
My
next
thing
right
so
that
the
sanitizing
it
I'm
not
sure.
That's
necessarily.
Is
that
really
such
a
great
idea
as
well,
because
the
the
reason
that
that
bothers
me
is,
if
you,
if
you
specify
and
you
say,
I
want,
you
know
the
name,
63
characters
plus
five
right
and
you
sanitize
it
by
truncating
the
last
five
characters.
I
That's
not
really
a
declarative
system
anymore,
right,
you're,
now
you're
you're
getting
something,
but
it's
not
really
what
you
asked
for
the
other
thing,
that's
a
little
bit
its
least
it's
what
depart
the
stool!
So
when
you
enter
party
will
say
this
house
implicated.
Okay,
well,
I
mean
if
you
want
to
behave
exactly
like
the
pod
I
guess
that
that
makes
sense.
I
mean
the
the
kubernetes
documentation
does
point
out
that,
while
they
use
253
characters
is
the
limit
for
most
resources.
It's
not
a
hard
rule.
Some
are
less
yeah,
but.
E
E
B
Because
of
what
Kubrick's
mission
is,
which
is
to
provide
kind
of
this
transition
from
previous
virtual
machine
solutions
to
a
containerized
environment,
we
do
need
a
full
course,
unfortunately,
to
support
a
superset
of
of
all
solutions
out
there.
So
as
long
as
we
pick
a
limit,
that's
large
enough
to
accommodate
reasonably
expected
names
coming
from
safe,
vmware
or
other
platforms,
then
not
as
big
an
issue,
but.
D
A
The
thing
is
that
we
know
that,
for
people
a
longer
name
is
often
required,
I
mean
which
is
exceeding
the
36
characters.
So
there
are
one
or
if
we
can
take,
the
approach
like
the
identifier,
which
is
the
name
currently,
is
equal
equivalent
to
two:
what
kinetics
s
for
plot
identifiers
and
that
we
have
a
separate
field
for
representing
or
a
more
human
readable
title.
Openshift
is
doing
the
same.
A
They
have
the
identifiers
in
their
participate
type
name
for
for
an
entity,
and
that
is
is,
for
example,
for
templates,
so
to
split
that
for
the
title-
and
maybe
it
makes
sense
to
for
the
title
for
the
title
that
it
would
also
be
no
notation.
So
we
even
don't
have
the
limit
of
the
labels,
for
example,
but
it's
an
impurity
descriptive.
What
we
could
also
enforce
is
women.
Emission
controller
is
uniqueness,
because
one
thing
is
the
length
and
the
descriptiveness,
but
let's
look
at
the
features.
What
the
name
a
die.
A
I
did
fire
his
gift
was
I
mean
specifically
the
the
identifier
of
the
entity,
the
name
it
allows
us
to
query
for
it
to
address
it.
I
think
that's
important!
That's
why
I
want
to
limit,
consume
the
limits
of
communities
or
respected,
but
for
the
for
the
title
or
would
really
like
to
be
more
in
favor
of
both
virtualization
systems.
A
Expect
that
you
really
know
the
scripture
titles,
but
you
couldn't
use
it,
for
example,
for
querying
without
the
right
tools,
I
mean
you
could
look
it
up
and
then
infer
the
name
from
from
a
very
long
title,
that's
possible
one
thing
for
the
title:
I'm
unsure
about
is:
if
we
want
to
make
it
unique
as
well,
so
we
could
enforce
uniqueness
with
an
admission
control,
for
example.
So
we
could
have
it's
hard
to
put
into
plain
words.
A
So
I
think
what
I'm
trying
to
say
is
I'd
like
to
keep
the
name
as
an
identifier
of
the
entity
and
obey
all
the
communities
limits
and
restrictions.
What
they're
ever
they
are
and
have
a
separate
field
which
is
more
friendly
to
how
it's
done
like
an
enterprise
management
systems
like
over
it
waking
up
a
free-form
identifier
or
any
I'm
a
much
remote
open
stack
but
VMware.
We
squared
e
sphere,
we
can
have
longer
names
and
I
would
enforce
a
uniqueness
on
that
title
as
well.
A
B
A
Not
speaking
about
what
it
is
today,
I'm
speaking
of
where
we
want
to
go,
and
if
we
use
an
annotation,
then
there's
no
enforcement,
then
it's
that
it's
unique
across
objects,
but
we
could.
We
could
try
to
enforce
uniqueness
and
I.
Think
that's
important,
because
if
we
like
a
effective
identifier
but
which
is
not
really
the
identifier
and
we
still
want
a
foreign
force
uniqueness,
so
that
is
how
I
would
at
least
structurally
solve
it,
identify
these
identifiers
in
communities
and
then
morph
all
four
reporter
for
more
freedom.
A
D
H
So,
just
just
to
sum
it
up.
Currently,
what
we
do
have
is
the
name
which
is
limited.
Actually,
the
DNS
label
limitation
in
kubernetes
is
decree
characters
and
we
also
have
an
annotation
for
UID.
So
if
I'm
getting,
what
you
guys
say
is
adding
another
annotation
for
the
user
display
name
for
the
VM,
keeping
the
name
title
as
an
identifier
as
we
do
right
now,
and
also
keeping
the
UID
annotations
for
special
or
unique
I.
Think
regarding
the
DNA
stuff,
that's
not
really
relevant
to
be
discussed.
E
A
A
Okay,
that's
good,
because
that
was
my
impression
and
the
name
should
just
be
the
name
of
the
entity.
We
should
not.
We
should
not
have
another
field
which
is
representing
the
same
information
like
I
know.
Today
we
have
Hubert
/vm
name,
which
is
refer
it
have.
It
has
the
same
informations
as
the
name
entity
itself,
I,
actually
wonder
if
that
is
needed
by
the
system
or
if
it's
just
added
for
convenience,
I'm,
not
sure
about
that.
E
Fabian
now
I
know
what
you
mean.
We
have
this
VM
a
name
label
in
our
examples,
because
when
you
create
it,
because
when
you
create
them
that
you
can
do
something
like
your
CTL
exposed
so
that
they
have
a
unique
name
label,
it's
not
really
it's
not
useful.
In
any
case,
it's
just
so
that
when
you
create
the
examples
you
don't
run
into
service
yeah,
Mac
mix,
it
is
nothing
to
do
with
the
system.
Person
could
forget
being
something
completely
creative
I
see.
A
E
B
J
J
This
is
for
getting
a
better
life
cycle
management
for
cube
fruits,
so
I'm
here
using
the
operator
framework
for
it,
which
will
make
it
easier
to
install
and
enable
it
to
upgrade
cube
fruit,
and
we
just
started
with
it.
Together
with
Michael
and
yeah,
it's
working
progress
and
looking
forward
to
see
it
working
soonish.
J
B
Thank
you.
Next
up,
we
have
open
shift
origin
device
plugins.
This
is
more
of
kind
of
a
just
a
public
service
announcement.
Earlier
this
week
there
was
a
issue
filed
that
device.
Plugins
were
just
simply
not
working
with
an
open
shift:
origin
cluster.
This
was
running
on
CentOS
on
bare
metal
using
the
OC
cluster
up
command.
K
L
L
M
So
the
previous
week,
I
I,
wasn't
in
on
the
under
gathering
that
everyone
was
at
so
I
looked
in
meantime,
I
looked
through
the
through
our
networking
API
right
up
and
to
see
what
was
missing
in
the
current
implementation.
So
there
were
some
small
things,
so
I
figured
I
could
tackle
those.
So
one
thing
is
a
field
that
can
tell
you
that
the
the
VMI
is
supposed
to
be
completely
disconnected
from
external
world.
So
this
support
for
this
field
is
already
up
for
view.
I
think
it's
pretty
much
there.
M
So
once
I
like
started
implementing
that
I
feel
well,
first
I
figure
out
that
at
least
me-
and
it
seems,
like
other
people-
may
also
have
this
I.
Don't
actually
fully
understand.
What's
like
the
semantics
should
be
for
this
field
for
at
least
for
the
pod
interface
case.
So
that
should
be
clarified
and
I.
Don't
know
like
I,
don't
think
we
should
discuss
that
right
now.
M
M
It
has
specific
open
flow
rules
that
block
the
traffic
that
that
comes
from
from
the
port
with
the
air
own
MAC
address.
So
we
gotta
figure
figure
out
what
to
do
with
that.
I've
sent
an
email
to
the
mailing
list
with
option
what
we
could
do
and
I
don't
know
like
whether
we
it
doesn't
seem
like
we
have
a
clear,
clear
answer
to
at
the
at
least.
F
M
For
this
is
the
algorithm
we
may
need
to,
but
then
it
doesn't
resolve
the
general
issue
of
well.
We
are
not
really
completely
isolated
from
the
underlying
as
the
end,
because
we
expose
the
custom
MAC
address
outside
world,
and
so
we
could,
because,
instead
of
trying
to
adopt
two
each
specific
as
the
end
that
may
have
the
same
issue,
we
may,
alternatively,
think
about
how
we
can
completely
isolate
this
custom
address
that
is
exposed
to
the
guests
but
isolated
from
the
external
world,
meaning
having
something
some
router
or
whatever
in
inside
the
report.
M
That
would
hide
this
MAC
address
custom
address
from
from
that
site
world,
which
obviously
like
is
more
complex
solution
in
terms
of
you
know
all
the
wiring
and
performance
wise
I
guess
so
that's
something
not
to
take
easily.
There
were
also
other
ideas.
Fine,
like
maybe
we
shouldn't
support
custom
IP
addresses
for
for
the
port,
an
interface
that's
all
great,
but
first
well,
it
already
works
for
kubernetes.
So
why
not
have
it
and
also.
M
The
issue
so
there's
that
issue
and
it
would
be
nice
to
have
some
kind
of
clarification
discussion
on
the
mailing
list
wherever
to
like,
at
least
what's
the.
What
are
the
paths
that
we
may
want
to
consider
because
it
seems
like
this
problem
hits
the
philosophical
grounds
of
compute,
which,
like
the
claim
was
done,
that
convert
is
like,
makes
everything
completely
compatible
with
and
transparent
to
underlying
Cooper
names,
but
with
custom
address
the
current
implementation.
M
It's
not
transparent,
it's
not
is
not
hidden,
it's
it's.
It's
exposed
and
it
easily
already
introduced
an
incompatibility
issues
with
open
sure,
and
probably
we
may.
We
may
experience
the
same
for
us.
So
what
do
we
do
with
that?
Do
we
complicate
the
import
networking
set
up
to
hide
the
MAC
address
or
something
to
think
about.
A
E
M
E
M
A
So
my
thing
is:
this
is
specifically
a
problem
which
is
most
visible
on
the
pod
Network
and
here
I
think
we
must
stay
compatible,
also
to
make
our
lives
easier.
So
my
take
is
here:
we
have
control
over
what
ports
we
want
to
expose
to
the
network
and
everything
else
is
not
our
under
our
control
on
the
pod
network
and
I.
Think
I
will
treat
it
that
way
so
that
what
does
it
mean
I
would
say.
A
A
This
is
as
good
see
generic,
but
what
network
it's
the
only
case
there
could
be
networks
where
you
don't
have
control
over
the
MAC
address
and
we
don't
have
control
of
the
IP
address,
and
we
need
to
recognize
that
for
the
pet
port
network
we
know
we
cannot
control
the
IP,
we
cannot
control
the
Mac.
Thus
we
obey
those
rules,
.,
I
think
that's
the
case
for
for
pol
networks.
We
only
have
control
over
what
we
want
to
expose
on
day
four,
so
UDP
TCP
services.
However,
for
device
plugins,
they
give
us
the
full
freedom.
A
There
could
be
device
plugins
which
allow
us
to
set
the
MAC
address,
nightly
l2
bridge
plug-in,
which
he
well
and
Peter
are
working
on,
but
they
could
be
other
plugins
like
you
by
casino
plug-in,
where
we
have
the
same
restrictions
as
for
the
port
network.
So
what
does
this
mean?
I
think
we
need
some
kind
of
metadata
for
the
network's
in
order
to
understand
what
functionality
they
provide.
The
most
easy
one
is
to
understand:
is
it
layer,
2
or
layer,
3
connectivity
they
provide?
Do
we?
Is
there
an
IP
address?
A
M
And
I'm
not
sure,
like
maybe
I,
missed
instant,
like
with
the
the
device
plugins
approach
like
what
we're
going
to
support
there,
but
I
think
we
may
still
have
similar
issues
there.
Like
some,
some
devices
see
device
plugins,
don't
really
support
what
they
custom
addresses
or
whatever
one
yeah
and
the
idea
behind
the
custom
address
as
far
as
I
understand
was
that
it's
a
front
and
property,
and
so
it
should
not
necessarily
be
related
to
the
underlying
network
type.
M
So
II
and
the
use
case
behind
the
custom
address
as
far
as
I
understand
is
that
you
may
have
some
weird
V&F
or
whatever
it
may
have
some
expectations
about
the
MAC
address
that
is
exposed
to
it.
But
it
may
not
necessarily
care
about
what
happens
after
when
the
traffic
after
it
leaves
the
VM
as
long
as
it's
properly
routed
and
maybe
on
the
on
the
root
path.
A
It's
still
a
strong
use
case
for
the
simple,
a
simple
non
kubernetes
case
right
so
for
the
l2
bridge,
plug-in
I,
think
there
it's
relevant
because
they
can
bridge
your
VM
into
the
local
network
and
there
it
makes
sense
to
be
able
to
specify
the
MAC
address,
because
you
want
to
you
want
to
set
up
you
th
CPS
over,
to
provide
the
right.
You
know
configuration
to
the
client,
so
I
think
that
is
the
values
case,
but
you're
right
for
SRO
V,
that's
the
opposite.
I
mean
that's
a
very
strict
physical
case.
A
We
eventually
cannot
control
it
and
the
other
side
is
Department
work
where
the
MAC
address
is
given
so
we're
little
bit
sitting
in
between
two
chairs,
but
I
think
it's
still
a
valid
use
case.
I
would
keep
it
because
it's
still
relevant
for
for
quite
a
range
of
use
cases
I
think
even
if
SRV
and
port
network
I
agree
that
we
can't
use
it.
But
what's
your
suggestion
just
block
the
the
field
for
the
pot
metal
I
want
to
block
it?
I
will
race
an
error.
A
If
we
see
that
the
network
is
not
so
I
mean
in
general,
I
would
do
a
check.
Do
we
know
that
the
network
is
supporting
the
MAC
address?
Then
we
allowed.
If
it's
not
supporting
it,
then
we
raise
an
error.
I
think
we
have
that
similar
with
other
stuff
in
pods
like,
for
example,
if
we
have
setting
your
UID
I
think
a
specific
UID
which
the
users
not
allowed
to
set.
Then
it
will
raise
an
error
if
the
pot
is
launched.
M
M
A
I
tell
you
why,
because
because
maybe
you
want
that
guest
to
have
that
specific
MAC
address
because
intend
to
and
I
think
in
the
environment
we
don't
have
control.
We
should
just
match
what
the
environment
provides.
We
respect
that
I
mean
that's
the
same
with
DHCP
right
I
mean
you.
If
you
know
that
the
amount
provides
DHCP,
then
you
don't
want
to
configure
your
own
IP
address,
because
you
can
risk
collisions
on
the
network
or
you
don't
have
connectivity
at
all.
A
So
I
think
in
cases
that
we
cannot
the
MAC
address
that
we
cannot
control
the
MAC
address.
Then
we
don't
want
to
override
it,
because
we
assume
that
the
environment
has
an
intended.
Why
it's
giving
us
that
MAC
address?
That
would
be
my
reasoning
and
actually
I.
Think
it's
true
I
mean
if
we
think
what
spoofing
filters
and
that
kind
of
stuff
there's
the
reason
to
do
it
and
yeah
it
doesn't
make
sense
to
set
it
if
the
macaronis
doesn't
get
out
of
the
setup
anyway.
So
but
I
I
see
a
point.
G
A
M
G
So
what
we're
working
on
both
sides
of
the
one
solution-
and
this
is
to
provide
LT
networking
one
end-
is
the
device
plug-in
that
exposes
carrots
for
the
bridge
into
the
pod
from
same
configuration
and
then
of
the
other.
On
the
other
hand,
adding
a
new
kind
of
network
that
is
called
a
resource
which
have
translates
to
a
device
plug
in
with
the
network
name.
G
We
have
a
small
NPI
that
we
can
craft
it
between
the
two
of
those
two
things,
and
this
is
based
on
an
environment
variable
that
is,
has
some
json
encoded
information
in
it
and
from
complete
perspective
any
any
any
time
it
has
a
network
of
type
resource.
Then
it
assumes
that
this
resource
should
exist
with
all
the
implications
of
scheduling
and
everything
that
comes
with
it
and
then
just
assume.
G
That's
the
information
about
what
the
actual
resource
state
is
encoded
in
this
environment
and
on
the
other
hand,
we
have
a
specific
implementation
that
assumes
exposing
the
bridge
from
the
host
into
this
kind
of
conforming
intern
Park.
Conforming
to
this
environment,
variable
that
qubit
can
use
in
the
future.
We
may
add
different
device
plugins
more
complex,
that
whatever,
as
long
as
they
are
defined
in
the
rasta
flood,
is
between
resources.
And
it's
been
exposing
information
about
the
API.
N
So
the
PR,
with
initial
support,
was
merged
and
it's
ready
to
use.
There
is
just
single
hook
point
on
the
find
domain
that
is
called
after
we
built
the
domain
X
and
now
the
user
can
use
a
hook
to
change
this
domain
XML
or
add
something
to
it,
and
basically,
for
those
who
don't
know
me,
the
hook
is
deployed
as
a
sidecar.
N
It
can
take
attributes
from
VM
annotations
or
from
the
whole
VM
object,
and
in
this
case
it
can
it
obtains
the
generated
on
domain
XML
and
can
change
it
and
send
it
back
so
right
now,
if
someone
wants
to
implement
this
kind
of
excited
current,
they
must
implement
well
server
and
socket
handling
and
all
that.
But
we
want
to
introduce
this
simple
framework
to
to
get
all
of
this
out
and
what
use
are
just
input
on
the
function
that
takes
whatever
response
from
convert
change
it
and
send
it
back
and
yeah.
N
A
O
Yeah
sorry
I
didn't
add
my
name
to
it.
Yes
of
CP
opening
the
the
latest
version
of
kubernetes
provides
a
GPO
manager,
it's
very
simple.
As
a
very
simple
functionality.
It
basically
provides
dedicated
CPUs
to
a
pod
and
then
pins
seep
in
to
those
CPUs
and
want
to
leverage
this
and
pin
the
VM
CPU.
The
VM
disappears
to
those
already
pins
abused
in
the
code.
There
are
some
challenges
with
this,
but
yeah
I'm,
just
starting
starting
to
work
on
this,
hopefully
for
the
next
week,
we'll
have
something:
what
would
you
rate
them.
B
Fair
enough
moving
on
to
the
open
floor
couple
notes,
first
up
its
call
for
proposals,
OpenStack
summit.
If
you
were
hoping
to
submit
a
proposal,
I
believe
it
is
now
too
late
that
the
window
close
to
couple
hours
ago.
There
is
still
time
for
cube
con
North,
America
I
looked
it
up
yesterday
it
was
early
August,
but
I
don't
have
the
exact
date.
I
will
fix
that
in
the
notes
here,
but
please
feel
free
submit
the
topic.
The
the
more
the
merrier
next
up,
Qbert
client
Python
to
customers
is
this
morning.
B
Q
Question
I
just
wanted
to
bring
some
attention
about
this
repository
because
I
don't
know
if
everybody
is
found
habited,
but
disabilities
have
to
generate
it
from
open,
API
specification
from
the
cupboard
and
it's
supposed
to
provide
the
SDK.
The
Python
SDK
to
you
know
the
access
today,
cooper-cooper
API.
Q
Q
One
is
about
the
testing,
because
every
time
when
this
kind
gets
generated,
we
need
to
test
it
somehow,
and
the
question
is
if
we
want
to
test
it
on
the
side
on
the
coop
website
or
on
under
the
client
Python
repository
side,
because,
as
I
said,
this
code
is
auto-generated,
so
it's
actually
being
generated
from
the
site
or
inside
of
the
repository
and
then
it's
being
directly
pushed
inside
of
the
code.
Client
Python.
So
there's
no
space
to
do
a
view
or
some
kind
of
a
think.
Q
It
will
just
land
it
there
and
then
we
might
to
see
if
it
felt
or
not.
So
this
is
one
question
which
I
would
like
to
bring
up
on
what
side
we
you
should
test
it.
Second
thing
is
versioning
because
since
we
already
have
several
branches,
like
you
know,
released
0.4
and
0.6
and
master,
there
are
some
incompatible
incompatible.
Things
like
this,
when
we
were
factoring
from
the
Flying
V
ends
to
VM
instances,
and
it
stopped
so
we
will
probably
have
to
create.
A
branch
is
also
inside
of
the
kind
Python
to
to
reflect
these.
E
E
B
Q
B
Look
how
we
have
a
github
repo
for
Cubert.
We
have
one
for
a
client
Python.
What
if
we
have
the
client
Python
testing,
which
is
part
of
the
CI,
and
you
literally
just
have
a
dedicated
project.
That's
in
github,
that
is
for
running
the
automated
tests
and
you're
actually
doing
the
automated
deployment
thing.
E
B
E
Q
Okay
and
the
second
thing
with
the
burgeoning
I,
don't
know
the
experience
itself
to
do
that,
because
not
every
you
know,
then,
when
some
pool
requests
lands
and
sorry
it
doesn't
necessarily
mean
that
it
will
actually
generate
a
new
version
of
the
SDK,
because
you
know
if
there
is
no
API
change
involved.
There
is
no
need
to
do
a
generator
decay.
E
B
E
What
we
would
probably
need
is
an
extra
suffix
or
something
then
I
mean
your
ad
yeah.
What's
different
questions,
then
how
would
we
fix
the
client?
If
it
looks
like
it
worked
for
the
release,
then
we
see
there's
an
issue,
we
fix
it.
So
we
have
well
known-
and
it's
just
part
of
the
next
release.
Then
we
would
do
an
emergency
release.
Q
E
E
They
ditched
it
and
they're
now
doing
their
own
releases
they're
in
a
more
asynchronous
way,
so
that
it's
easier
for
them
to
fit
to
bring
in
fixes,
whereas
responder
is.
If
we
need
to
do
it
here,
because
it's
almost
completely
up
because
it
is
auto-generated
completely.
That's
not
the
case
for
Glencoe
I
would
suggest
that
media
started
way,
and
maybe
we
then
see
that
we
have
two
two
independent
versioning
for
some
unexpected
reasons
it
can
happen
to
us.
I
wouldn't
say
that
it
will
happen
to
us
here,
because
we
really
generated
mostly.
Q
B
A
Yes,
I
would
actually
duck
I'll
see
that
I
engaged
them
to
join
this
meeting
as
well,
some
different
guys
because
to
be
fine,
I
think
there's
software
someone
just
a
heads
up,
I've,
been
discussing
with
them
a
little
bit
of
list,
because
there
was
Steve's,
fedora
20,
a
tree
base
and
I'm,
seeing
that
it's
going
forward
quite
nicely,
I've
been
engaging
with
the
liberal
community
and
sort
of.
Let
me
start
different.
A
That
is
all
I
wanted
to
say
so.
I
won't
deliver
to
you
more
testing
so
that
we
are
more
confident
in
consuming
them
and
they
seem
to
be
opening
up.
They
had
a
meeting
I
had
a
couple
of
meetings.
They
want
to
get
some
people
looking
after
the
quality
of
divert
to
upstream
of
some
of
the
tests,
and
so
that
Liberty
can
also
be
tested.
An
upstream
upon
releases
I
think
that's
good
news,
but
we
need
to
see
what
the
timeline
of
that
is
yep.
That's
all
I
wanted
to
say.
Oh.
E
A
A
They're,
never
second
topic
mm
the
last
one
for
me:
I
think
it's
nothing
which
requires
immediate
action.
What
we
can
think
about
it.
So
today
we
base
our
images
on
fedora
and
I.
Think
that's
good
well,
but
at
some
point
maybe
it
could
have
been
a
ploy
on
your
face
of
Qbert
I
mean
still
very
early
or
needed
a
spoiler,
but
at
some
point
we
really
want
to
make
it
robust.
I
think
this
see
I
can
I.
A
Can
I
learnt
how
important
C
is
for
us
to
provide
something
stable,
but
despite
having
the
CI,
we
want
to
get
to
really
stables
release
in
the
center.
We
want
over
the
version
1.0,
which
is
yet
another
topic,
but
our
nerd
oracle.
Are
we
confident
in
the
container
world
of
having
fedora
at
the
base
OS
or
do
we
need
to
consider
to
have
a
cent
OS
based
branch
where
we,
for
example,
base
or
or
stable
releases
on
if
asset
was
yeah,
it.
E
I
tell
you
why
I
can
I
just
move
to
you,
as
the
only
reason
why
we
would
have
the
discussion
slivered,
so
deliberate
container
over
Liberty
parts
run
in
are
the
only
ones
where
we
are
the
only
ones
where
we
actually
care
about
the
content
of
the
base
image.
It's
not
relevant
for
any
other
company
component.
E
It's
just
go
if
it
wouldn't
kind
of
double
our
downloads
bandwidth
right
now,
everywhere,
I
would
just
say
you
sent
us
for
everything
else,
because
then
we
don't
have
to
switch
and
you
don't
care
about
the
base
anyway,
so
likely
with
something
of
the
newest
stuff
for
the
other
yeah.
But
that's
just
what
I
wanted
to
hear
and
it's
mostly
because
of
Stern
the
verge
container.
E
B
Much
you're
suggesting
is
if
it
were
possible
to
ship
a
recent
version
of
liver,
because
that's
really
the
challenge,
isn't
it
isn't
it
because
the
the
goal
was
sent
to
us
or
one
of
the
goals
of
stability.
Therefore,
it's
not
necessarily
bleeding-edge.
If
we're
consuming
the
latest
features
from
libvirt
there's.
Your
challenge
is
that
we
would
need
to
ship
our
own
yeah.
A
A
Yeah,
that's
true,
actually
I
think
it's
relevant
for
yeah.
For
for
containers
will
we
use
stuff
from
the
base
of
wess
right,
which
is
not
true
for
the
classical
infrastructure
components
like
controller
hander
and
launcher,
but
for
the
device
plugins
it
will
become
relevant
and
also
for
the
EM
yeah
Philippa.
E
Forever
for
every
component,
which
does
not
do
everything
in
go,
which,
by
the
way
we
should
prefer
I,
think
we
have
tissue
and
I'm
not
sure
if
we
can
control
there
for
the
whole
product
easily,
at
least
for
the
whole
project,
easily,
that
we
have
a
common
base
image
across
different
three
puss.
Even
yes,
probably
trust
impossible.
It
might
not
hit
the
starred
in
the
coracle,
but
least
it
means
extra
effort.
If
we
want
to
synchronize
that,
then
later.
G
A
E
Maybe
I
can
add
one
more
sentence
to
laughing.
Maybe
we
should
bring
to
the
Qbert
devil
list,
an
email
regarding
to
the
base
image.
We
could
define
something
like
the
base,
Fedor
advantage
image
with
the
chasm
or
something
so
that
we
can
and
ask
if
we
want
to
align
on
such
an
image
across
the
whole
repose
wave
in
the
project
and
then
I
don't
know
bump
them
on
a
regular
schedule
or
if
someone
needs
something
us
so
that
we
can
all
say.
Now
we
go
to
the
new
one
or
something
like
that.
A
We
can
we
can.
We
can
introduce
the
Qbert
based
on
I'm
just
making
the
Torah,
because
then
I
mean
that's
easier
for
users
to
track.
You
know
I
mean
we
could
ask
them
to
audience
for
where
we
can
enjoy
simple
image,
which
is
specifying
this
awesome,
and
then
we
can
that's
a
good
idea.
Yeah
I
mean
you
know
what
I
mean
is
that
in
the
front
we
have
a
simple
topper.
E
Found
aspect
so
we
mean
we
were
10
version,
that
with
text
basically
and
you
could
say,
used
to
lay
the
stick.
I
mean
it
still
might
still
be
a
little
bit,
I'm,
not
entirely
sure.
If
we're
just
not
just
adding
it
extra
later,
have
the
same
issues
again
on
top,
then.
So
we
happy
to
look
into
details,
but,
like
resolution,
l.