►
From YouTube: KubeVirt Community Meeting 2020-12-16
Description
Recording of the KubeVirt Community meeting held on Wednesday 2020-12-16
A
Welcome
everybody:
this
is
the
cube
vert
community
meeting
december
16th
2020.
first
topic
in
the
agenda
is
ryan.
Hollisie
has
more
to
follow
up
on
on
the
safe
and
restore
proposal.
B
Okay,
so
one
more
topic
so
I'll
do
a
little
bit
of
review
of
some
changes.
I
made
from
last
time
on
the
api
side
of
things
to
kind
of
close
the
book
on
that
discussion
and
then
last.
The
the
last
topic
which
I
have
for
now
is
the
storing
ram
in
previously
okay.
B
So
last
time
we
talked
a
little
bit
about
the
api,
I
kind
of
tried
to
summarize
it
a
little
bit
in
this
table
and
wanted
to
as
a
way
to
kind
of
communicate
where
I
think,
where
I
think
things
are
so
kind
of
the
the
the
high
level
look
at
this,
so
offline
virtual
machine
snapshot.
B
B
I'd
describe
that
as
online
vm
snapshot
and
offline
vmi
snapshot
so
being
snapshots
of
the
vm
that
has
a
vmi
and
the
snapshots
of
the
vmi
and
then
some
a
fourth
potential
stage
here,
which
would
be
like
something
I'm
I'm
not
proposing
here.
But
something
could
do
which
is
having
a
snapshot
of
a
running
vmi
by
itself
and
that
vmi
being
running
after
so
that's
kind
of
the
the
summary
of
like
what
I
took
away
from
last
time.
But
if
folks,
are
there
any
comments
on
that?
B
Yeah,
so
it
would
be
that
when
you
do
the
save
operation
so
basically
similar
to
to
it's
it's
the
same
path
as
pm,
online
vm
snapshot,
but
in
both
these
cases,
what
I'm
saying
is:
is
that
the
at
the
conclusion
of
it,
the
virtual
machine
object,
no
longer
exists.
B
Okay,
there
any
other
questions
kind
of
about
that.
I
know
like
some
of
the
terms
on
this,
like
a
little
tough
like
online
and
offline,
are
a
little
weird
when
you
kind
of
talk
about
vmi,
as
opposed
to
when
you
talk
about
vm,
but
at
least
that's
how
I'm
trying
to
communicate
for
now.
I
can
always
change
that
all
right.
Let's
go
to
the
top
topic
I
do
want
to
cover
so
I've
been.
I
tried
to
put
together
a
little
a
few
tables
for
each
of
these
subjects.
B
I,
rather
for
each
of
these
options
for
storing
ram
so
just
kind
of
a
high
level
layer,
we're
storing
vmi
ram
how's
it
gonna
get
stored.
We
ends
up
in
a
pvc
somewhere.
So
how
does
that
pvc
get
there?
You
know
what
what's
managing
it
and
so
on
and
so
forth.
I
put
together
three
options
here
and
then
there's
those
corresponding
tables
for
pros
and
cons
for
each
of
them.
So
the
first
one.
B
Well,
let's
list
them
all
use,
it
use
the
already
attached
storage
so,
like
the
virtual
machine
instance,
has
a
pvc
attached
to
it.
That's
maybe
how
it
got
its
disk
presented
to
it,
use
that
pvc
to
store
its
ram,
since
the
assumption
is
that
the
disk
would
is
already
gonna
like
I
said,
disks
are
going
to
be
there
I'll,
just
add
the
ram
to
it,
alongside
of
it
second,
using
a
dynamic
provisioner
to
create
a
new
pvc.
B
So
in
this
case
something
would
have
to
call
out
and
say,
create
a
new
pvc
and
and
then
the
ram
would
have
to
end
up
in
that
location.
This
would
have
to
be
something
sort
of
like.
Maybe
the
launcher
could
be,
maybe
even
the
snapshot
controller
could
do
it,
I'm
not
sure
exactly
but
sort
of,
depending
on
something
to
to
provision
it
kind
of
during
the
operation
and
the
third
one
being
use
a
large
pvc.
B
That's
already
been
attached
to
the
vert
handler
to
save
vmram,
so
I
have
pros
and
cons
for
each
of
these.
I
don't
know
if
some
of
them
are.
I
don't
know
it's
hard
to
find
some
of
the
to
fill
out
for
for
each,
but
here's
some
of
the
ideas
so
like
using
our
attached
storage.
One
of
the
things
I
thought
it
was
really
nice.
Is
it
it's
actually
a
pretty
good
way
to
group
this
as
part
of
cloning
like?
B
If
you
do
cloning
later
on,
we're
gonna
have
disc
that'd,
be
nice,
I
say
no
upfront
cost,
because
the
assumption
is
the
disk
is
already
attached
as
part
of
the
pvc,
so
we're
not
adding
anything
we're
just
using
what's
there.
This
was
a
comment.
I
added
roman
made
a
comment,
but
this
I
don't
know
if
you
want
to
explain
further
on
it,
but
one
of
the
cons
I
put
out
is
file
based
pvcs
only
as
a
con.
B
I
don't
know
if
roman
here
I'm
here
for
that
nothing
to
it
right
now.
Okay,
another
con,
no
guaranteed
space,
so
I
mean
this
could
be
full.
So
what
do
you
do
if
you
save
the
disc
full?
Well,
you
can't
really
save
them
or
it
fails.
I
don't
know
some
some
other
options.
I
mean
you
could
make
the
volumes
expand.
So
it's
not
necessary
that
that's
not
necessarily
like
that.
This
is
impossible
to
overcome,
but
there
is
something
to
think
about.
B
B
C
B
Yeah
sure,
okay,
so
one
thinking
is
that
that
when
the
save
operation
comes
in
so
we
we
go
through
the
snapshot
api,
we
perform
the
snapshot,
essentially
the
save
as
being
happening
under
the
hood
on
the
on
the
vmi.
So
at
some
point
before
that
save
can
occur,
there
needs
to
be
a
create
request
for
a
new
pvc.
B
I'm
not
quite
sure
what
would
do
that,
yet
it
could
be
for
handover
that
does
it
or
it
could
be.
Maybe
the
snapchat
controller,
the
two
things
that
I
was
kind
of
between
as
options
that
go
out
and
get
that
pvc,
the
snapchat
controller,
was
interesting
because
I
think
it
does
that
already.
D
B
Don't
know
I
could
be
wrong
about
that,
and
and
then
it
would
we'd
end
up
with
a
pvc.
We
would
let
the
launcher
know
where
to
store
it,
and
then
it
would
land
the
ram
inside
of
that
pvc.
B
Okay,
so
could
do
the
hot
disc,
they're
gonna
plug
the
disk.
I
think
they're.
I
think
I
saw
that
this
is
something
that
I
don't.
I
I
got
kind
of
from
this
idea
here
for
using
discot
plug
I'm
not
familiar
with
it,
but
that's
what
I
think
you'd
have
to
do.
B
C
Yeah
that
that
would
make
sense
yeah,
you
could
probably
reuse
some
of
the
stuff
alexander
has
done
with
the
hot
plug.
I
guess
the
difference
here
would
be
that
you
wouldn't
actually
want
to
attach
that
storage
or
anything
to
the
virtual
machine
guess
so
it
would
use
all
the
same
logic
up
to
a
certain
point
and
then
it
would
then
it
would
stop.
I
think
that's
achievable.
C
B
What
do
you
think
about?
So
what
do
you
think
about
in
relationship
to
the
other
options
like?
What
do
you
think
like?
What
do
you
like
about
dynamic
pvc?
What's
maybe
of
interest
to
you,
I.
C
Where
do
you
put
the
ram
like
which
one
does
it
belong
on?
So
coupling
those
two
things
can
be
difficult
to
separate
afterwards.
I
like
the
fact
that,
with
the
dynamic
provisioning
you're
separated
the
ram
from
the
disks
as
well.
Okay,
I
think
that's
the
way
I
would
go.
I
would
use
dynamic,
it's
just
going
to
be
a
little.
C
B
So
what
do
you
think,
like
you
mentioned,
so
separating
the
brand
from
the
disc?
What
do
you?
What
do
you?
What
do
you
think
that's
good
wait
like
what
yeah?
How
come
you
mentioned
that
I'm
just
curious.
I.
C
Have
you
ever
wanted
to
start
a
clean
virtual
machine
from
that
disc
or
something
I
I'm
not
I'm
not
sure.
If
any
of
that
makes
sense,
I
just
am
hesitant
to
bundle
both
the
disc
and
the
ram
on
the
same
storage,
and
maybe
that's
not
even
going
to
be
a
problem.
It's
just
something
that
I'd
have
to
think
through
a
little
bit
more
I'd
have
to
think
about
the
use
cases
in
the
end
to
understand,
there's
ever
a
situation
where
we
would
actually
need
to
separate
those
things.
B
E
Yeah,
I
think
dynamic
pvc
makes
a
lot
of
sense.
I
think
we
just
you
know,
need
to
find
a
way
to
configure
what
storage
class
do
we
allocate
this
thing
on
and
are
there
any
rules
about
that
or
you
know,
what's
the
best
way
to
configure
that
taking
a
step
back,
though,
like
I
think
for
a
lot
of
cases,
or
at
least
what
I've
heard
from
some
overt
people
is
that
you
know
for
a
lot
of
cases
doing
these
memory
dumps
aren't
particularly
helpful.
E
So
I
just
want
to
make
sure
that
you
know
our
api
is
like
you
know.
This
is
an
optional
thing
for
for
the
snapshot.
B
So
you're
saying
the
ram
being
optional.
B
E
B
Yeah,
okay,
yeah,
that's
kind
of
what
I
was
thinking
as
well.
What
are
you
about?
Okay,
I'll,
make
those
comments
afterwards.
Okay,
that
makes
sense.
Okay.
So
let's
see
how
about
how
about
the
last
option
I
mean
I
I
just
had
one
con
here
and
I
don't
know
this
might
just
sink
this
ship
right
here
I
mean,
I
think
roman
was
the
one
who
mentioned
this,
and
I
do
concur
that
like
say,
vm
brand
will
be
readable
by
anyone
with
access
to
this
pvc-
probably
not
something
we
want.
B
So
I
don't
know
I
does
anyone
have
any
other
opinions
on
this
on
the
third
option.
E
Unless
you
could,
you
know
we'd
have
to
securely
store,
hashes
or
information
about.
You
know
the
memory
somewhere
else
to
make
sure
that
it's,
I
guess,
not
been
tampered
with
or
something.
If
that's
the
concern.
F
E
F
B
B
We
can
maybe
be
a
little
more
reliable
having
the
dynamic
pvc
allocation.
I
had
mentioned
here
that
we
could
extend
volumes
as
well.
I
mean
I
I
saw
I
mean
I
kind
of
stole
that
from
your
comment,
michael,
like
what
does
like
this
look
like
if
we
were
to
extend
volumes.
E
E
So
you
know
we'd
have
to
have
the
administrator
be
like.
Yes,
you
can
definitely
extend
this.
It's
it's
not
something
that
we
can
just
try
to
do,
because
it
may
not
work
a
lot
of
the
time,
but
you
can
you
just
edit
the
pvc
definition
and
you
get
a
bigger
pvc.
E
E
B
E
So
I
suspect
that
once
these
memory,
once
the
memory
is
stored
in
a
pvc,
it's
not
ever
going
to
be
touched
again.
So
when,
if
we
want
to
restore
from
there-
we'll
probably
you
know
we
could
use
cdi
cloning
or
something
that
would
basically
copy
it,
either
via
a
snapshot
or
just
dumb
copy
to
another
pvc
that
will
be
mounted
at
some
point
but
yeah
and
then
eventually
qmu
obviously
has
to
be
configured
to
read
that
memory
when
it
starts
up
so
maybe
we'll.
E
You
know
that
pvc
will
be
mounted
the
first
time
the
vm
started
and
then
maybe
not
anymore,
and
then
it
can
be
deleted,
but
yeah
there
will
be
some
bookkeeping
there,
but
I
do
imagine
that
you
know
once
memory's
been
dumped
somewhere.
You
know.
B
Yeah,
that's
that's
what
I
was
thinking
as
well,
because
you
yeah
you
may
want
to
do
this
multiple
times
like
you.
May
you
have
that
that
same
snapchat?
You
want
to
restore
it
five
times
for
whatever
reason
so
yeah
like
having
the
ability
to
to
copy
it
and
to
do
that
also
gives
you
the
ability
to
maybe
make
it
a
little
more
mobile
as
well
so
yeah.
I
agree
michael.
C
B
B
Previously
we
get
we'll
get
the
advantage
of
being
able
to
rely
on
their
being
or
we'll
be
able
to
rely
on
guaranteed
space
being
available,
assuming
that
we
can
get
it
from
the
provisioner
yeah
and
then,
while
attached
storage
seems
like
it
could
work,
it's
really.
I
guess
it
depends
on
how
we
want
to
allow
for
those
volumes
to
be
expanded.
B
I
guess
do
we
see,
could
there
be
a
possibility
that
both
like
attached
and
dynamic
could
be
an
option
because,
like
one
of
the
things
I
mentioned
here
is
like
requires
a
dynamic
commissioner,
I
don't
know
what
like
the,
how,
like
all
the
how
folks
out
there
like,
if
everyone's
using
a
giant
giant
commercial
and
it's
just
something
that
people
does-
and
this
is
maybe
not
a
con
at
all
like
is
this.
Is
this
realistic,
like.
B
Okay,
I
mean-
I
think
in
that
case,
then,
if
we
can
assume
that,
if
someone's
using
a
dynamic
commissioner,
it
seems
reasonable
to
me
that
the
neck
pvc
would
be
the
way
to
go
as
long
as
the
yeah
as
long
as
the
bookkeeping,
I
guess
maybe
it
could
be
the
challenge
in
it,
but
I
I
don't
think
it's
too,
I
don't
think
it'll
get
too
out
of
hand.
I
think
at
all
yeah
okay,
so
I
I
don't
know
I
so
it
would
have
folks.
B
C
If
you
start
with
artery
already
attached
storage,
I
think
you
should
look
into
a
way
of
representing
that
a
volume
is
being
attached
to
a
vmi
for
the
purposes
of
snapshot.
So,
okay,
not
necessarily
coupling
a
disk
image
with
the
ram,
but
just
having
a
reserved
pvc.
That's
attached
to
the
dmi
in
the
event
that
you
are
going
to
snapshot.
C
E
I
agree
question
I
don't
know
much
about
the
how
qmu
implements
the
memory
dumps,
but
is
there
a
way
to
do
it?
Can
you
do
like,
like
a
memory
dump
delta
so
like
first
time
you
do
a
full
dump
and
then
other
times
it's
like
a
diff,
probably
not,
because
that
would
that
would
make
a
good
use
case
for
using
it
like
a
single
shared
disk.
E
If,
if
you
could
somehow
store
deltas,
but
I
assume
it
doesn't
do
that.
B
Okay,
yeah,
I
don't
know,
that's
a
good
question.
Okay,
I
I
I
like
where
you
actually
gave
it
yeah.
I
think
yeah,
that
that
would
be
a
good
assumption
with
aria
test
storage
yeah,
if,
if
you
do
make
the,
if
you
sort
of
make
that
if
you
have
that
expectation
that
this
isn't
just
going
to
be
used
for
disk
yeah,
okay,
so
I
I
think
I
mean
I
think
both
these
could
work.
I
think
it's
guess
what
it
comes
down
to
here.
What
would
be
easier
to
like
start
with?
B
I
don't
know
I
mean
to
me.
It
seems
like,
like
I,
I
like
the
idea
of
like
having
the
ability
to
do
both
of
these.
I
don't
know,
I
think
I'm
trying
to
think
of
the
work
like
what
a
good
one
to
start
with.
Maybe
aria
touch
story
seems
to
be
a
good
one.
To
start
with,
I
don't
know.
Does
the
folks
disagree?
Should
I
not
go
that
route.
B
E
C
That's
that's
the
hard
part,
so
the
api
couples
for
hot
plug.
If
you
want
to
hot
plug
a
disk
to
a
vmi,
you
have
to
go
through
a
sub-resource
and
that
sub-resource
is
very
strict.
It
requires
a
volume
and
a
disc
to
be
coupled
together
and
it
always
expects
those
two
things
to
exist
together.
So
there
would
be
some
api
work
that
would
have
to
be
done
to
enable
something
like
this
I'd
have
to
think
about
it.
B
E
C
C
This
isn't
trivial
I'll.
Just
that's
my
conclusion,
just
trying
to
think
through
it
I
mean
it's
all.
The
technical
bits
are
there
of
how
to
actually
get
pvc
exposed
into
the
vmi
pod.
Where
you
can
do
this
thing,
it's
all
the
wiring
of
how
to
do
that
in
a
way
that's
different
than
how
hot
plug
is
meant
to
actually
attach
a
disc
to
the
to
the
vmi.
We
have
to
come
up
with
some
kind
of
different
mechanisms
to
represent
a
volume.
That's
meant
to
be
used
for
a
different
purpose
than
a
disk.
E
Case
we
basically
just
would
have
a
new
type
of
volume
that
can
reference,
maybe
a
that
will
just
mount
a
pvc
when
the
vm
has
started,
and
we
can
come
up
with
our
own,
like
you
know,
has
to
be
a
file
based
pvc,
of
course,
and
just
come
up
with
our
own
directory
structure
for
dumping
images
on
there.
B
Okay,
yeah,
I
think
okay,
so
let
me
so
let
me
try
to
kind
of
get
somewhere.
So
this
is
I'm
gonna.
Add
I'll,
add
some
more
comments
based
on
the
discussion
here.
So
like
the
dynamic
pvc
sounds
like
this,
so
that
we
could
do
it.
It
sounds
like
given
the
amount
of
things,
maybe
the
moving
parts
in
this
with
that.
If
there
comes
a
time
when
there's
an
another
use
case,
we
should
add
this
as
well,
but
I
think
maybe
like
giving
some
more
time
to.
B
I
guess
some
of
the
pieces
to
maybe
I
guess
mature.
I
don't
know
if
that
would
be
the
right
word,
but
just
give
it
some
more
time
for
some
of
the
pieces
to
to
expand
a
little
bit
and
then
this
can
come
into
play
later.
You
know,
if
there's
a
need,
so
maybe
starting
with
already
attached
storage,
with
the
assumption
that
the
volume
being
mounted
can
be
used
for
something
other
than
disk
and
then,
like
you,
said,
michael
having
some
sort
of
directory
structure
there
that
respects
that.
E
E
Yep,
what
someone
somewhere
is
basically
going
to
have
to
be
watching
for
when
you
know
snapchats
are
deleted
to
clean
up.
You
know
it
doesn't
have
to
be
like
immediate
thing,
but
you
know,
maybe
eventually
we
can.
You
know
we
have
to
clean
up
the
old
snapshots.
B
Okay,
so
that
would
be
so
what
you're
saying
is
like
if
they,
if
I
have
a
snapshot
well,
actually,
can
you
go
through
an
example,
so.
E
B
E
Are
mapped
to
snapshots
somewhere
and
and
then
you
want
to
delete
a
snapshot
because,
because
say
maybe
you're
running
out
of
space
on
that
pvc
and
you've
got
some
old
ones
that
you
want
to
free
up
or
just
some
ones
that
you
want.
You
want
to
free
up
some
space
in
that
disk
to
do
more
snapshots.
Maybe
you
have
some
policy
or
something
so
you're
gonna
delete
a
snapshot
which
I
think
the
user
would
assume
that
that
is
deleting
that
memory
dump
on
disk
and
just
the
orchestration
of
doing
that.
C
E
C
Not
exactly
how
I
thought
it
would
work.
I
thought
that
this
ram
pvc
would
have
a
one-to-one
relationship
with
a
snapshot
like
there
would
only
ever
be
one
ram
snapshot
on
that
pvc
and
then
it
would
be
cloned
and
then
you'd
have
two
pvcs,
so
you
wouldn't
have
to
necessarily
keep
the
previous
one
and
then,
if
you'd
snapchat
again,
I
guess
it
would
just
overwrite
the
the
previous
snapshot
on
the
one
that's
actually
attached
to
that
pod
and
then
the
clone
operation
would
happen
again
on
that
after
it's
been
overwritten.
E
E
So
that
would
mean
that
okay,
so
then
this
is
just
the
pvc
is
just
kind
of
a
dumping
ground
and
then
eventually
copy
or
you
could
snapshot
that
actually.
C
C
I
wouldn't
treat
it
any
different
than
I
would
a
desk.
I
guess
is
what
I'm
getting
at.
B
So
I'm
trying
to
understand
that
would
be
like
so
every
time
you
do
a
the
example
that
you
gave
dude
looks
like
every
time.
You
do
a
snapshot,
so
maybe
some
time
goes
by.
You
do
another
snapshot.
B
You
mentioned
that
that
say
that
the
ram
has
changed
and
the
expectation
would
be
then
now
that
the
the
new
snapshot,
the
next
snapshot
you
did,
would
overwrite
the
the
old
snapshot
objects
right,
so
they
wouldn't
create
a
new
one.
E
So
what
you
you
say
so
time:
zero!
You,
you
dump
the
memory,
there's
nothing
on
the
disk!
Then
you
snap
it
and
you
associate
that
snapshot
with
the
vm
snapshot
or
vmi
snapshot.
Okay,
then
you
know
you
do
another,
one
that
that
disk
is
overwritten
again
with
memory.
You
take
another
snapshot
that
becomes
associated
with
that
vm
snapshot.
Then,
when
you
want
to
delete
something,
you
just
delete
that
snapshot
resource
okay.
So
when
you.
B
E
It's
basically
just
this,
you
know
a
scratch
space
and
we'll
create
snapshots
from
that.
B
And
then,
and
then
what
you
were
saying
earlier,
michael
with
the
cleanup
when
you
delete
this
snapshot,
you
don't
expect
that
the
clone
pvc
is
deleted
or
do
you?
What
was
that
I
want
to
see
if
I
can.
E
C
That's
handled
with
sorry,
I
think
I
cut
you
off,
but
that's
that's
handled
with
owner
references.
So
it's
just
the
whole
cascade.
C
C
B
Yeah,
I
don't
really
have
a
counter,
I'm
just
trying
to
think
about
it
because
yeah
I
mean
because
I
guess
it's
our
source
of
truth
right
like
it's
like.
If
you
have
the
snapshot,
everything
that
comes
from
it
like
there
should
nothing.
If
that
snapshot
doesn't
exist
right
there
shouldn't
be
just
a
bunch
of
lying
around
pvc
clones.
B
Yeah
sorry
go
ahead:
yeah.
D
Sorry
for
coming
back
to
this
over
and
over
again
I'm
just
trying
to
understand,
I
mean
the
the
use.
The
original
use
case
is
from
what
I
remember
was
that
we
will
do
a
snapshot
and
then
the
user
will
be
able
to
send
all
of
that
to
forensic
analysis,
and
then
I
don't
know
what
I
mean
and
I'm
just
not
sure
where,
where
are
we
going
with?
All
of
this.
B
So
the
idea
would
be
that
after
the
snapshot's
been
created,
we
can
use
cloning
to
create
a
copy
of
that
of
that
that
that
vm
and
then
restore
it
in
another
location
to
do
the
forensics
so
like
another
namespace,
for
example,.
B
Oh
so
I
I
added,
I
talked
about
this
a
little
bit
last
time.
I
I
think
when
I
created
a
non-goal
for
this
athletic
because,
like
I
call
this
import
export,
like
I
think,
that's
probably
the
best
way
to
easily
just
to
kind
of
separate
it
out,
so
the
idea
would
be
to
keep
it
in
the
cluster,
so
import
export
is
what
I
call
taking
it
outside.
B
D
B
I
I
don't
quite
understand
so
like
the
idea
would
be
that
the
some
other
user
would
come
along.
It
would
store
the
vm
state
and
you
know
when
whatever
the
and
whatever
name
space
they
have,
you
know
that
namespace
could
have
different
policies,
restrictions.
You
can
use
network
policies
to
restrict
the
network,
for
example,
and
then
do
their
analysis.
There.
B
Yeah,
well,
I
don't
know
yeah
black.
Why
don't
we
can
talk
offline?
I'm
definitely
interested
in
hearing
your
thoughts
on
this.
I
can
I
can
message
you
after
and
we
can.
We
can
get
together
and
talk
about.
I
feel
like.
A
Actually,
I
wonder
this
if
I
could
break
in
here,
I'm
sorry.
This
topic
is
quite
large
and
involved.
I
wonder
if
we
need
a
separate
breakout
session,
as
pep
had
mentioned
a
few
minutes
ago
to
to
really
give
this
justice.
B
Yeah
I
mean,
I
think,
that's
fine,
I
think
for
me
from
my
perspective,
I
don't
know
how
everyone
else
does.
I
think,
like
you
know,
from
the
api
and
the
storage
side
and
a
bunch
of
things
I
think
are
like
it
gives
me
at
least
enough
that
I
feel
comfortable
like
kind
of
going
for,
and
maybe
writing
a
little
bit
of
code
on
this.
So
but
if
we
want
to
have
like
yeah
separate
sessions,
that's
fine.
I
think
I
can
do
that
from
if
we
have
any
more
large
topics
in
here.
C
I
think
your
your
best
approach
with
this
would
be
to
begin
proof
of
concepting
what
you
have
and
your
best
understanding
of
what
you
have
so
far,
and
that
gives
us
something
to
tangible
to
kind
of
talk
about
and
also
kind
of
lets.
You
go
to
the
exercise
of
getting
your
hands
on
with
what
we
have,
and
I
think
that
might
make
things
more
clear,
as
we
like
have
further
discussions
as
well
like
what
worked
and
what
didn't
work
when
you
actually
tried
to
do
this
in
practice.
B
Sure,
yep
yeah,
that's
fine,
yeah,
that's
kind
of
the
direction
I'm
leaning.
Now
I
think
at
least
now
I
don't
see
something
that's
like,
like
I
at
least
have
the
direction
that
we
want
to
go
on
a
number
of
these
things.
So
that's
good,
yeah,
okay,
I
think
yeah.
I
think
so.
I
have
a
few
notes
I'll
make
in
here
and
then
yeah.
Maybe
we
can
have
some
more
breakouts
and
be
further
as
I
get
a
little
further
on
this.
So
okay.
E
Let's
definitely
keep
the
conversation
open.
This
is
something
that
me
and
well
we're
gonna
be
working
on
for
the
next
major
release
so
and
some
other
cdi
folks
will
be
joining
on
as
well.
So.
B
Okay
yeah,
so
I
can
so
I'll
do
some
sort
of
like
dave
mitch.
They
can
do
some
sort
of
poc.
I
don't
know
how
long
to
take
but
I'll
work
toward
that.
Just
given
a
general
outline
of
what
we
discussed
so
far
and
then
yeah
we
can.
We
can
do
like
some,
maybe
some
breakout
sessions
with
some
stakeholders
here
and
kind
of
evaluate
and
if
we
need
like
very
like
some
like
larger
forms.
B
Maybe
we
can
come
back
here
and
we
can
discuss
some
enough,
maybe
change
of
direction
or
whatever
it
is
that
ends
up
happening.
A
A
Okay,
moving
on
to
the
open
floor,
I
sent
an
email
yesterday
got
a
little
bit
of
participation
on
the
documents
themselves,
but
no
real
discussion.
The
goal
is
to
try
and
diversify
the
maintainers
that
we
have.
You
know
for
the
long-term
health
and
relevance
of
the
project.
It's
it's
a
very
good
idea
to
to
get
more
people
involved
at
a
at
a
fundamental
level.
A
A
I
see
a
few
names
and
faces
on
the
documented
process
link,
but
not
sure
if
everybody's
able,
all
right,
okay-
and
so
just
you
know
hoping
that
we
can
get
some
eyes
on
this
and
make
sure
it
makes
sense.
But
the
you
know
to
emphasize
the
goal
here
is
to
get
more
involvement
and
you
know
have
other
people,
you
know
as
maintainers
or
you
know,
reviewers
or
approvers
of
code
and
any
thoughts
or
comments
on
that.
I
A
Thank
you.
I
think
that
there
is
one
more
section,
just
a
heads
up
that
I
think
we
do
need
to
add
and
that's
something
about.
You
know
a
tier
above
approvers,
which
is
you
know,
project
owners
or
something
you
know,
because
I
think
that
having
a
large
committee
of
people
you
know
doing,
the
approval
process
might
be
just
unwieldy.
A
So
I
there
may
be
some
changes
to
the
overall
structure,
but
this
is
kind
of
the
intent
of
where
we'd
like
to
go
and
past
that
there
are
a
few
things
that
we
are
looking
at
in
terms
of
areas
where
people
can
increase
our
participation.
If
there's
any
interest
so
moving
on
to
the
next
topic,
there
we
have
a
about
130.
A
I
think
I
looked
yesterday
issues
135
on
github
and
you
know
more
coming
in
all
the
time.
It
might
make
sense
that
the
idea
was
floated
at
one
point
recently
of
having
a
public
bug
scrub
where
we
would
go
or
an
issue
scrub.
That
is
where
we
would
go
through
and
look
at
those
and
triage
them
is
that
something
people
think
is
in
general
is
a
good
idea
of
doing.
Is
it
worth
our
time.
I
I
don't
want
to
sound
too
positive
today,
but
yes
again
with
this,
I
I
think
it
makes
you
know
we,
even
though
we
don't
pay
enough
attention
to
a
lot
of
these
issues
that
were
reported
or
rfs
opened
on
comfort,
we're
still
getting
new
ones.
A
Up
until
now,
we
haven't
really
had
you
know
a
clear
agenda
and
we
have
to
admit
in
a
lot
of
cases.
We
don't
have
an
agenda
created
before
the
meeting
begins,
and
I
think
that
hurts
us,
because
without
a
clear
direction
or
intent
of
what
this
meeting
is
for,
we
may
not
be
getting
as
much
attendance
as
we
could,
because
people
just
aren't
interested
so,
as
you
know,
is,
is
a
portion
of
this
time
this
hour
that
we
have
each
week
useful
as
a
bug
scrub.
I
I
was
looking
at
the
bug
scrub
recordings
from
one
of
the
kubernetes
six
and
the
first
couple
of
them,
like
they
spent
an
hour
just
going
through
all
the
issues
that
weren't
scrapped
before
so
I
think
it
has
a
potential
of
killing
this
call,
but
if
we
dose
it
properly
and
say
that
we
spend
only
20
or
30
minutes
back
scrubbing
and
the
rest
would
be
open,
floored
other
agenda,
then
it
may
work.
I'm
not
sure
how
many
people
would
be
able
to
join
multiple
times
a
week
to
participate
in
the
community.
C
I
like
the
idea
of
setting
a
time
limit,
so
maybe
20
minutes
for
for
bug
scrubs
and
we
would
prioritize
community
members
who
had
something
to
talk
about
there.
So,
in
the
absence
of
somebody
who
specifically
wants
to
talk
about
a
specific
bug,
we
would
begin
going
to
the
list
and
see
how
far
we
can
get
through.
A
Fantastic-
and
so
I
guess
you
know,
part
of
this-
is
we're
looking
to
try
and
formally
assign
roles.
We
don't
have
to
do
it
here
in
this
meeting,
but
you
know
we're
these.
Are
things
we're
thinking
about?
Is
we
should
have
somebody
able
to
help?
Keep
us
tracked.
You
know
on
track
or
guide
this.
You
know
and
in
the
spirit
of
what
we're
talking
about
it
shouldn't
necessarily
be
one
of
the
people.
A
You
know
we're
the
ones
that
are
familiar
with
this,
but
ideally
eventually
we
would
have
more
involvement,
so
you
know
we'll
probably
be
looking
for
volunteers,
for
you
know
both
the
public
issue,
scrub
or
you
know
the
building
the
agenda
for
this
meeting.
I
Could
we
make
it
so
I
mean
we
said
that
there
is
a
host
for
scrubbing,
let's
say
for
the
next
four
weeks
and
then
they
can
give
up
or-
and
somebody
else
can
volunteer
if
they
want
to.
You
know
to
make
this
like
available
for
everybody,
and
it's
it's
not
that
big
of
a
commitment
for
whoever
hosts
it
just
a
random
idea.
It
makes
sense.
I'm.
I
I'm
willing
to
start,
and
if
I
can
give
up
on
that
one
after
a
couple
of
weeks,
okay.
A
Moderating,
okay
and
yeah.
Another
thing
that
we've
noticed
is:
we
do
have
good
presence
on
slack,
but
it
tends
to
be
the
the
same
people
answering
questions
so
if,
if
you're
interested
or
if
you're
able,
if
you
could
please
spend
a
little
time
on
the
channel
and
just
see,
if
there's
you
know
any
way
to
contribute
there
because
it's
some
of
us
are
quiet,
and
I
am
definitely
guilty
of
that
as
well.
A
Okay,
no
worries:
we
will
probably
follow
up
again
more
on
email
in
terms
of
you
know,
finding
if
there's
a
fit
for
these
roles,
and
so
with
that,
if
there's
no
further
thoughts,
I
will
turn
it
over
to
pep
to
talk
about
the
kubert
summit.
H
Yeah,
just
one
thing
about
you
know
quick
clarification
about
qb1,
it's
not
about
you,
know
someone
responsible
for
the
delivering
keyword,
d1,
but
rather
about
you,
know,
kind
of
leading
the
the
effort
of
coordinating
and
the
roadmap,
etc
right
anyway,
computer
summit,
so
we
are
finally
starting
to
plan
what
will
become
the
first
online
event
for
cubert,
fully
focused
on
cube
birth,
so
some
background
on
this.
Typically,
we
we
used
to
have
face-to-face
meetings,
piggybacking
on
major
events
like
kubecon,
for
example.
H
You
know
this
year.
Everything
has
switched.
You
know
all
the
major
all
those
major
events
have
switched
to
a
virtual
format,
so
we
cannot
easily
piggyback
on
a
cubit-centric
small
meetup.
So
what
we're
doing
instead
is
a
fully
dedicated
two
birth
event
in
the
spirit
of
those
meetups.
This
will
that
the
goal
is
to
address
both
contributors
and
users.
H
This
virtual
format
provides
more
a
bit
more
flexibility,
meaning
that
we
will
have
like
kind
of
two
tracks
and
we
can
can
well
run
them
both,
and
the
only
thing
that
we
have
more
or
less
settled
at
the
moment
is
a
confirmation
that
we
have
a
platform
to
run.
This
an
online
event
platform
to
run
this
and
updates
availability
dates
of
that
platform.
So
the
proposal
is
to
have
this
online
event
on
the
9th
and
10th
of
february
running
for
half
day
each
day.
H
So,
basically,
the
goal
is
to
allow
some
overlap
between
american
and
european
time
zones.
It's
impossible
to
overlap
the
whole.
You
know
world
time
zones,
unfortunately,
but
at
least
this
this
proposed
time
we
expect
that
will
allow
most
people
interested
people
to
join.
So
beyond
that,
we
we
still
don't
have
much
more.
Yet
specifically,
we
don't
have
content.
Basically,
we
are
starting
to
plan,
and
this
is
a
call
for
you
know
anyone
interested
in
contributing
in
this
planning
effort
beyond
you
know
contributing
in
the
event
itself.
H
A
H
So
no,
we
don't
and
that's
part
of
the
preparation.
So
we
don't.
We
don't
even
know
if
we
want
to
run
a
formal
course
for
proposals.
That's
a
bit
open.
You
know
open
for
for
discussion
in
the
sense
that
you
know
a
formal
call
for
proposals
involves
getting
together.
You
know
assembling
a
committee
to
decide
and
opening
a
a
fixed.
H
You
know
period
for
well
the
whole
formal
process
of
a
cfp
due
to
the
I
mean
keyboard
is
a
relatively
big
project,
but
it's
also
relatively
small
enough.
Let's
say
that
we
believe
we
can
do
without
a
formal
cfp,
but
that's
just
the
initial
thought
again.
If,
if
anyone
believes
we,
we
must
run
through
a
formal
cfp,
basically
that
if
we,
if
we
had,
you
know
too
much
content
or
if
the
proposals
were
not
clear
enough,
etc,
we
might
have
to
well
set
one
set:
one
define
one
process.
H
A
Okay,
well
with
two
minutes
to
spare,
are
there
any
other
topics
that
we
need
to
raise
today.
A
C
I
would
imagine
actually
really
light
we'll
see
so
for
this
summit
like
the
next
announcements
and
things
like
that,
would
it
make
sense
for
us
to
go
ahead
and
postpone
making
any,
I
mean
sure,
announce
it
now,
but
as
far
as
like
coming
up
with
the
process
for
receiving
presentations
and
things
like
that,
should
we
wait
until
after
the
new
year
to
do
that
or
what
do
you
know
what's
people's
thoughts
about
the
timing
of.
H
C
My
concern
is:
I'm
thinking,
if
there's
any
sort
of
feedback
that
we
need
to
be
providing
people
who
are
submitting
proposals
or
anything
like
that.
Right
availability
is
going
to
be
difficult
until
beginning
of
the
year.
For
a
lot
of
us,
that's
something
we
could
take
offline.
It
was
just
a
thought.
B
C
On
how
we
take
in
proposals
and
how
we
want
to
proceed
with
like
getting
presentations
scheduled
and
everything.
H
H
C
A
Okay,
well,
I
believe
that
would
be
a
full
agenda
today,
so
without
further
ado
thanks
everybody
and
see
you
soon.
Okay,
thank
you.
Thank
you.
Thank
you.