►
From YouTube: Community Meeting, April 4, 2023
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
All
right
today
is
April
4th.
This
is
the
kcpe
community
meeting
welcome.
Everybody
looks
like
a
light
crowd
today.
Maybe
some
folks
will
join
I've
got
the
issue
for
the
community
meeting
created.
A
A
C
Yeah
so
a
couple
weeks
ago,
we
sort
of
opened
a
discussion
on
the
kcp
forum
for
users
that
were
trying
to
do
workspaces
and
service
type
things.
We
know
that
right
now,
a
lot
of
our
a
lot
of
the
ways
that
we
model
permissions
for
cross
workspace
concepts
are
based
on
user
identities
and
that
sort
of
breaks
down
when
you
had
a
bunch
of
service
accounts
into
the
mix.
C
So,
while
thinking
through
how
we
might
support
workspaces
as
a
service,
one
of
the
things
I
I'm
trying
to
focus
on
is
like
do
we
are
we
better
served
with
a
different
way
of
modeling
these
types
of
entitlements
across
workspaces
and
right
now
we're
kind
of
bending
over
backwards
to
make
some
of
the
our
back
we're
using
work.
C
So
I've
started
writing
a
proposal
in
the
enhancements
repo
for
how
we
might
model
entitlements
in
a
way
that
sort
of
sidesteps
a
lot
of
these
issues
by
not
tying
them
to
users
specifically
and
then
critically.
Another
part
of
this
design
that
you
know
requires
some
thought.
C
This
similar
to
quota
is
kind
of
a
A
system
that
isn't
well
served
by
putting
a
bunch
of
data
in
at
CD
and
propagating
things
throughout
the
workspace
stream
when
things
change
so
trying
to
find
a
middle
ground
of
an
implementation
that
works
for
bootstrapping,
but
also
a
plugable
interface
that
you
know,
others
can
add
more
capable
things
later
is
a
goal.
So
if
any
of
that
sounds
useful
to
you,
please
take
a
look.
Any
feedback
is
is
welcome.
C
A
B
B
A
The
agenda,
so
I
will
give
a
status
update
on
the
one
time
please
this.
B
A
Based
update
so
I
have
it
all
compiling
and
most
of
the
end-to-end
tests
do
pass
I'm
working
through
individual
ones
as
I
find
them.
So
one
that
was
a
bit
of
a
challenge
was
the
validating
and
mutating
webhook
code.
We
had
modified
in
the
fork
to
be
multi-cluster
aware
and
in
125
or
126
there's
a
new
admission,
plug-in
called
validating
admission
policy
which
does
not
use
web
hooks.
A
It's
probably
going
to
be
more
inefficient
or
I
should
say
less
efficient
because
it
is
instantiating,
validating
and
mutating
web
plugins
on
the
Fly,
but
I
think
I
probably
could
cache
some
of
that.
The
bigger
issue
is
that
each
instance
of
the
plugin
has
a
cash
for
clients
that
connects
to
external
web
hooks
or
even
service-based
web
books,
and
that
cache
is
fixed
at
200.
It's
an
lru,
so
each
time
I
created
instance
of
the
plugin.
We
get
an
lru,
that's
sized
at
200,
which
is
probably
not
super
efficient.
A
So
we
can
look
to
increase
efficiencies
later.
I
just
want
to
get
this
working
for
right
now
and
then,
if
we
need
to
make
some
changes,
Upstream
I
think
there's
some
ways
that
we
could
do
that.
So
we
can
reuse
the
client
cache,
because
it's
not
even
reused
Upstream
between
the
validate
and
mutating
plugins
David.
You
got
your
hand
up.
A
B
D
Sorry,
updating
the
comments
and
and
putting
my
mic
back
yes,
I
just
wanted
to
give
a
short
notice
about
the
fact
that
after
quite
much
work
you
are
Kim
updated
the
latest
and
the
kcp
glbc
for
them
who
those
who
didn't
know
this
kcp
side
project.
It
was
done
it's
a
month
ago
by
by
another
team,
providing
Global
Ingress
for
kcp.
D
That's
mainly
work
coming
out
from
the
quadrant
Community
project,
and
this
was
an
integration
but
based
on
very
old
kcpe
and
still
using
the
provisional
Advanced
scheduling
hacks
that
have
been
dropped
in
the
meantime
and
and
that's
very
huge
work
here
now
all
the
tests
are,
you
know,
green
and
passing
and
after
updating
to
the
latest
kcp,
and
in
fact
that
would
be
the
first
step
to
really
a
real
Global
Ingress
for
kcp,
which
would
be
based,
and
that's
the
next
step
of
the
work
on
the
coordination
controller
pattern.
A
Very
cool,
thank
you,
I'm,
assuming
this
is
supposed
to
say
a
POC.
A
Very
cool
that'll
be
exciting,
I
I
think.
If
there's
time
when
that's
updated,
it
would
be
really
cool
to
demo
it
at
the
community
meeting
and
then
also,
if
I
know,
it's
a
lot
harder
and
a
lot
more
work.
But
if
you
wanted
to
put
together
a
video
that
we
could
throw
up
on
YouTube
that'd
be
cool
too
yeah.
B
B
Okay,
well
have
a
great
week:
everybody
that's
a
short,
but
that's
okay
and
we'll
see.