►
From YouTube: Community Meeting, June 21, 2022
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
Hey
everybody
today
is
june
21st,
and
this
is
the
kcp
community
meeting.
I
will
be
moderating
at
least
the
first
half
of
this,
but
at
half
past
the
hour
I
will
need
to
drop
and
we're
still
talking
then
stephon
will
take
over
the
community
meeting
issue.
1320
is
up
so
if
you
have
any
topics
that
you'd
like
to
chat
about,
please
feel
free
to
add
them
as
comments,
and
I
think
we
will
start
with
sergius's
question
so
sergius
I'll
turn
it
over
to
you.
A
B
Okay,
thank
you,
so
I'm
working
on
a
little
admission,
plugin
that
is
supposed
to
lock
down
internal
kcp
labels,
annotations,
which
we
only
want
to
sort
of
like
be
modifiable
by
our
own
internal
controllers.
B
Obviously,
we
want
to
have
an
allow
list
that
is
a
list
of
labels
and
annotations
which
are
actually
supposed
to
be
sent
by
end
users
or
other
controllers
or
other
operators
which
are
not
in
the
system.
Masters
group
and
my
question
is,
if
you
know
of
any
effective
way
how
to
come
up
with
that
list,
stefan
suggested
to
iterate
over
the
api
package.
I
just
gave
it
a
little
sweep
just
before
the
community
call.
However,
I
was
apparently
not
smart
enough
to
filter
these
out
efficiently.
B
So
if
you
have
any
tips
or
suggestions
that
would
be
yeah,
I
would
be
highly
thankful
and
the
issue
is
one
one.
Three
eight
I
forgot
to
edit
in
the
comments
here
that
I'm
working
on
so
maybe
a
good
place.
If
you
don't
know
them
right
off
the
bat
here
or
you
have
any
suggestions
which
should
be
a
low
listed,
just
add
them
here
as
commands.
B
B
B
A
All
right,
I
saw
a
comment
in
the
chat
from
paul
about
david's
work
in
progress
demo
on
user
home
workspaces.
I
think
it's
about
20
minutes
long.
I
made
it
through
the
first
couple
of
minutes
before
going
on
vacation,
but
it
looked
very
promising
from
what
I
was
starting
to
see.
So
I'm
excited.
E
Yeah,
that's
right,
yeah!
It
was
just
a
mistake.
The
20
minutes.
In
fact
it's
four
minutes
long,
but
but
the
recording.
So
you
had
just
16
minutes
black.
So.
A
E
Probably
one
of
the
first
to
to
look
at
it
before
I
update
that
and
on
the
uk,
by
the
way,
I'm
still
working
on
that
this
was
so
prototypeish,
as
I
mentioned,
because
of
course
we
have
to
yeah.
Thank
you
suggest
we
have
to
to
tackle
and
define
more
precisely
what
regards
permissions
and
also
some
some
mechanism
to
gate
the
fact
that
a
user
can
see
his
homework
space
automatically
created
because
in
in
a
number
of
cases
it's
very
interesting.
E
E
There
might
be
cases
where
we
want
a
privileged
service
or
component
or
something
like
that:
external
based
on
kcp
to
gate
user
and
just
authorize
user
to
have
them
home
their
homework
space
created.
So
I
have
to
tackle
those
type
of
things,
which
is
what
I'm
doing
this
week.
So
I
mean
more
serious
and
complete
variant
of
that
should
come.
A
Very
cool,
so
we
don't
have
anything
else
on
the
agenda
if
anybody's
got
anything,
feel
free
to
raise
your
hand
or
speak
up,
otherwise
I'll,
probably
move
into
the
six
or
so
incoming
issues
and
the
milestone
epics
just
to
review
those.
A
A
So,
basically
just
as
a
reminder,
what
we're
trying
to
do
here
is
not
necessarily
go
into
detail
in
anything
in
here.
It's
more
about
trying
to
decide
what
milestone
these
issues
belong
to
and
if
we
think
something
is
important,
but
not
super
high
priority,
we
can
put
it
in
tbd
if
we
think
that
it
is
so
high
priority
that
it's
got
to
be
done
in
the
next
nine
days,
then
we'll
put
it
in
zero
six
zero.
A
I
think
we're
not
going
to
put
anything
into
zero
seven
until
we
come
back
and
do
zero
seven
planning.
So
this
one
looks
like
they
tried
to
create
and
enter
a
workspace
and
got
an
error
so.
A
I
don't
think
that
this
happens
often
enough
or
that
it
well,
if
it's
a
reproducible
100,
we
need
to
certainly
cover
this
sooner,
but.
A
D
F
D
A
Okay,
next
up,
stefan
things
not
being
fully
qualified.
A
A
D
D
It's
that
we
have
to
change
and
the
other
thing
so
I
I
I
tried
to
find
out.
Let
me
think
I
tried
to
find
a
cluster
workspace
which
is
removed
and
then
finalize
that
I
couldn't
get
the
information
easily.
So
that's
why
it's
always
easier
to
fully
qualify
them,
and
if
we
do
that,
we
have
precedence
now
that
we
use
the
workspace
path,
colon
and
an
object
name
so
like
a
file
path,
basically
to
file
it
into.
A
Enhanced
audit
logging-
I
think
this
is
definitely
not
a
0.6,
so
tbd
and
then
we'll
come
back
and
put
it
in
either
zero,
seven
or
not.
When
we
do
zero,
seven
planing.
B
A
All
right
another
one
from
you,
showed
you
ca
rotation
in
front
proxy.
That's
a
small
one!
I
mean
it
shouldn't
be
too
hard
to
be
honest,
yeah,
I'm
still
going
to
put
it
in
tbe,
because
just
to
reiterate,
like
we're,
we'll
do
planning
for
0.7
eventually,
but
we're
not
going
to
aggressively
put
stuff
into
zero
seven
until
we
do
planning.
A
A
Okay,
is
this.
A
Thanks,
stefan
all
right,
I
I'm
wondering
since
joachim
has
signed
this
in
terms
of
load
joakim.
Is
this
something
like
what
is
on
your
plate
right
now
and.
G
A
It's
totally
fine,
all
right,
I'm
gonna
put
tbe.
Thank
you
and
I
created
this
one,
which
is
definitely
tbd.
So
this
is
around
not
homework
spaces
that
david's
working
on,
but
like
inside,
of
an
org
hierarchy
being
able
to
share
a
workspace
with
other
users
which
which
you
can
do
today,
but
you
need
to
ask
an
org
admin
or
a
root
admin
to
do
it.
E
But
then
I
mean
it's:
how
does
it
relate
really?
I
mean
what
does
really
self-service
mean
as
soon
as
we
don't?
We
would
not
probably
enable,
by
default
any
user
to
create
workspaces
in
the
in
a
top
level.
Org
I
mean
that's.
That
would
not
be
the
the
approach
in
the
future,
which
is
why
we
have
all
workspaces.
D
A
Yeah,
I
think
at
this
point
it's
tbd.
I
think
that
if
someone
has
appropriate
permissions
inside
of
a
like
a
my
app
workspace.
H
A
Yeah,
please
go
ahead,
guy,
we're
happily
taking
folks
with
topics.
So
if
you
want
to
chat
about
storage
pvs,
let's
do
it.
I
I
So
the
pv
is
more
of
a
system,
administrative
component,
of
course,
but
the
question
that
we
have
about
the
pv
itself:
that
if,
if
we
let
the
user
create
the
pvc
with
deployments
or
whatever
is
the
workload
in
kcp
and
then
we
sync
it
to
a
cluster,
then
that
cluster
will
generate
a
pvc,
a
pv
for
that
right.
So
there's
the
provider
there
and
we
need
to
capture
some
information
back
into
kcp.
I
For
that
and
the
reason
for
that,
for
that
is
we
want
to
handle
the
case
where
that
cluster
goes
off
and
we
can
recover
by
reconnecting
to
that
storage
and
we
are
not
really
sure
we're
trying
to
go.
You
know
in
several
directions
there,
whether
we
we
need
to
you
know
so
pick
up
the
information
from
the
pv
and
just
put
it
into
the
pvc
or
yeah.
Stefan
sorry,.
D
Yeah
this
sounds
familiar
to
a
discussion
I
had
with
andy
earlier
today.
We
also
want
pots
for
visibility,
probably,
but
we
don't
need
them
updated
as
in
cube
like
not
every
second
or
something
like
that,
so
maybe
not
with
all
information.
So
maybe
we
have
to
think
about
mirroring
objects
back
into
kcp,
but
projected
so
it
without
information.
We
don't
need,
but
also
maybe
enhances
some
information
we
want
to
to
carry
in
kcp.
I
Yeah,
I
guess
I
mean
in
the
case
of
stores
the
reason
for
it
is
not
just
visibilities
also
for
recoverability,
because
that
that
information
will
will
serve
us.
As
the
you
know,
the
connection
information
right,
if
we
it's
a
logical
view
of
the
downstream
object.
Basically,
yes,
yes,
but
so
there's
cases
where
it
doesn't
make
sense
to
just
take
this
information,
as
is,
we
might
need
some
something
on
the
cluster
itself
to
to
provide
us.
I
You
know
like
a
plug-in
or
something
that
will
decide
what
information
makes
sense
to
push
up
right,
but
in
other
cases
it
does
make
sense
just
to
have
the
information
from
the
pv
reflected
back,
but
there
is
some
transformation
needed.
I
think,
like
you
mentioned,.
I
And
yeah
the
reason
for
that
transformations
might
be
that
it
it
might
relate
to
locations
right.
So
if
I
so,
if
I
have
a
location,
if
I
want
to
sync
that
pvc
now
to
a
different
location,
I
really
have
to
do
something
different.
I
I
need
to
be
able
to
copy
that
volume
between
locations
before
I
can
really
connect
to
that
pv
from
another
cluster
in
another
location
right,
so
so
there's
more
process
around
it,
but
I
think
that
the
initial
thing
is
just
that
we
want
the
informations
captured
in
in
kcp
for
for
syncing
back
to
another
cluster.
D
That's
that's
clear
that
we
need
that
how
we
implement
that.
So
what
is
coming?
We
don't
have
to
discuss
it
in
our
gp
here,
but
transformations
will
come
eventually.
So
there's
a
pr
from
david
and
maybe
it's
worse
to
to
meet
and
go
into
detail
of
your
use
case
and
see
how
that
matches
whether
it
fits
as
a
solution.
D
I
I
E
E
Many
transformations
I
mean
the
the
the
transformation
that
are
coming,
allow
modifying
upstream
or
downstream,
I
mean
in
in
both
directions.
Obviously
the
use
cases
we
have
now
are
more.
You
know
modifying
the
upstream
object,
as
it
should
be
viewed
by
the
downstream
sinker,
but
it
can
also
modify
some
objects.
That
are
you
know,
for
example,
it
also
modifies
the
statues
that
he
that
is
brought
back
from
from
downstream
to
upstream
so
I
mean
technically,
it
can
also
it
can
work,
both
those
directions.
D
E
That,
I
guess
that's
a
bit
some,
that's
something
we
already
do
somehow
when
you
preserve
the
statues
of
the
of
the
sinker
related
view,
because
you
don't
want
to
override
the
statues
on
the
main
object.
So
in
some
cases
you
you
get
the
information
from
the
upstream
object
for
what
is
from
the
spec,
but
you
preserve
the
status
that
initially
came
from
downstream,
so
I
mean
we
can
implement
whatever
logic
we
want
here
to
mix
both.
E
I
Have
one
question
I
still
want
to:
you
know
get
your
take
about,
so
I
really
want
to
just
get
that
flow.
You
know,
even
if
not
completely
final
design
for
it,
but
just
getting
that
flow
working
for
for
a
quick.
You
know
showcase,
and
I
I
was
wondering
I
I
do
need
to
run
some
collection
so
something
similar
to
pushing
up
information,
and
I
was
thinking
about
using
some
pushing
that
in
an
annotation,
but
then
on
the
pvc
right
basically
on
on
the
user,
the
user
entity.
I
D
E
Yeah,
but
that
would
be
as
as
stefan
said,
that
would
be
interesting
to
have
a
brainstorming,
maybe
meeting
or
session
about
transformations,
and
I
mean
the
upcoming
transformations
on
that,
because
the
the
underlying
idea
is
to
maintain
a
view
of
of
the
object,
as
it
is
viewed
and
managed
by
each
sinker.
E
A
E
Mean
I
think
at
least
what
you're
saying
is:
is
crossing
a
bit.
What
is
what
has
been
done
in
this
work,
and
it
should
be
interesting
to
discuss
that
in
more
detail.
I
think.
I
Okay,
okay,
thanks
I'll
set
up,
meaning
to
follow
up.
A
Great,
thank
you
all
right,
so
I'm
gonna
move
on
to
milestone
epics
and.
A
D
A
Good
multi-workspace
controller
development.
I
need
to
check
in
with
fabian
and
nick
and
barsha
on
some
of
this
and
see
if
we
can
get
some
things
moving
forward.
I
know
that
there's
some
open
pr's
and
things
that
need
review,
so
the
controller
runtime
fork
is
still
functional.
A
A
If
we
don't
get
all
of
the
listers
clients
and
informers
done,
I
don't
think
it's
the
end
of
the
world
like
we're
still
in
the
same
state
that
we've
been
in,
but
I
would
like
to
try
and
refocus
some
of
my
time
to
helping
get
those
closed
out
sooner
rather
than
later.
A
A
I
don't
have
quota
for
cluster
scoped
resources
and
we
haven't
really
solved
quota
for
pods
like
compute
limits
and
requests
on
cpu
and
memory,
but
we've
got
some
ideas
for
how
to
deal
with
that,
so
I'm
gonna
try
and
get
where
I
am
today:
cleaned
up
and
merged
after
it
gets
reviewed
one
more
time
and
then
go
on
to
tackling
cluster
scoped
and
pods.
A
At
this
point,
given
that
I
need
to
drop
in
a
couple
minutes,
stefan
can
I
turn
this
over
to
you
to
finish
this
out.
So
I
can
head
off
to
this
other
meeting
that
I've
got
to
run
to.
D
Looking
at,
I
think
next,
one
david.
E
Yes,
so
this
one,
let
me
look
at
the
items
so
yeah,
mainly
the
the
the
various
items
mentioned
here.
The
first
ones
are
either
already
done
or
in
process
in
progress,
mainly,
you
know
validated
in
terms
of
how
to
do
it.
Apart
from
the
the
the
points
I
mentioned
just
previously
permissions
and
the
gating
of
of
the
automatic
creation
of
home
workspaces,
and
then
obviously,
I
have
to
also
the
clean
plugin
support
has
been.
E
You
know,
export
and
and
study
implement
it
in
parallel
to
validate
the
rest
and
now
mainly
after
what
I've
been
speaking
already.
What
is
remaining
is
tests
and
end-to-end
tests.
F
We,
the
last
bits
that
are
left
here,
is
accessing
the
content
of
the
workspace,
while
it's
initializing
through
the
proxy-
I
imagine
that'll,
be
today
or
tomorrow.
After
that,
I
think
we
should
have
the
api
binding
initializer
added
as
a
service
that
should
be
fairly
straightforward.
F
D
H
The
acceptance
of
permission
claims
by
bindings,
as
well
as
adding
permission,
claims
to
the
export
and
this
so
it
should
be
taken
care
of
12,
21
and
12
22,
at
least
the
initial
versions
of
these.
Obviously,
when
you
come
back
and
make
it
more
granular,
but
the
initial
pr,
I
think,
is
working
and
is
doing
the
filtering
based
on
labels
and
has
the
controller
that
we
talked
about.
So
I
think
I
need
some
reviews
on
it
and
I
think
it
was
failing
some
one
of
the
tests.
H
So
if
I
could
get
a
couple
reviews
and
then
if
I
could,
I
can
get
into
the
test
and
if
somebody
reviewed
it-
and
I
haven't
looked
at
it
yet-
I
also
haven't
made
my
way
over
to
the
pr
to
look
at
it.
Yet
so
I
apologize
if
everything's
there.
So
I
think
this
should
be.
I
think
the
initial
ask
for
0.6
should
be
doable
within
this
week.
H
D
Good,
thank
you
next
one
on
the
list.
Sharding,
I'm
not
sure
what
you
put
on
the
list
here.
The
pr's
up,
there's
even
a
ci
drop
part
of
that
it's
not
green,
but
it
was
green
locally.
So
I
think
we
are
not
too
far,
and
you
have
seen
the
storm
of
prs,
which
I
moved
out
of
that
one
most
have
merged.
A
couple
are
still
open,
so
this
one
will
become
much
much
smaller.
Hopefully
it's
in
the
next
days
and
polynomial
the
brush
is
here.
I
don't
think
so.
Yes
watch
cash,
mostly
working.
D
D
D
D
D
D
That's
it.
I
think
all
our
epics
is
there
anything
so
there's
a
topic
where
you
think
you
are
blocked.
Please
speak
up.
I
think
we
have
a
bit
more
than
a
week
now
until
next
week
middle
of
next
week
or
something
when
the
deadline
is
so
please
speak
up
and
if
you
need
reviews
ask
for
it
actively.
So
it's
totally
fine
to
ask.
Even
if
people
ignore
it,
everybody's
busy,
so
keep
asking
keep
nagging.
That's
all
all
cool
and
I
think
then
we
are
finished
with
the
topic.
So
we
still
have
some
minutes.