►
From YouTube: Community Meeting, January 17, 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
Hello,
everyone
and
welcome
to
the
January
17th
2023
kcp
community
meeting.
We
have
a
few
folks
from
the
red
hat
team
who
are
at
an
on-site,
so
a
smaller
team
today
got
a
few
items
on
the
agenda.
The
first
one
was
an
issue
and
an
email
out
to
the
mail.
The
Google
group
about
the
virtual
workspaces
term
being
changed
to
views.
A
B
Okay,
so
the
issue
I
wanted
to
bring
up
was
connected
to
the
concern
of
you
know.
We
want
to
promote
the
kcp
work
upstream
and
how
do
we
do
it
and
the
problem
was
when
we
tried
to
sell
it
Upstream
the
sale
didn't
really
go
through,
they
didn't
buy
the
kind
of
well.
It
was
sold
as
as
tent
multi-tenancy
and
people
objected
that
it's
not
hard
enough
multi-tenancy
and
it's
true.
B
It's
it's
not
a
fully
hard
multi-tenancy.
There
is
some
interaction
between
the
tenants,
so
what
I
was
wondering
is?
Can
we
sell
the
concept,
a
lighter
weight
concept
of
a
higher
level
of
virtualization
right
in
and
in
fact
that
still
has
been
tried
too?
The
Upstream
people,
you
know
the
key,
the
APA
Machinery
Sig
says.
Well,
we
don't
believe
we
want
super
namespaces,
they
call
it
generally
right
and
I
was
wondering
you
know
a
two-pronged
thought.
B
You
know
we
have
some
use
cases
I
enumerated
four
in
the
agenda.
You
know
if
we
go
with
these
four
use
cases.
You
know
that
says
you
know.
If
we
just
had
this
high
level
namespace,
it
would
help
these
four
use
cases.
B
You
know
could
maybe
we
could
make
the
sale
that
way
of
of
just
this
layer
of
actually
your
virtualization,
but
the
other
question
is:
if
we
did
that
and
succeeded,
no
more
could
kcp
proceed
with
Upstream
buying
only
basically
logical
clusters,
not
workspaces
and
API
export
and
TMC
and
everything
else.
A
Yeah
I
think
that's
a
great
question
because
it
would
always
be
nice
to
reduce
the
the
number
of
things
were:
forking
from
kcp
or
I'm.
Sorry,
kubernetes,
admittedly
I'm
not
as
close
to
the
conversations
about
putting
things
Concepts
from
kcp
into
kubernetes,
but
has
has
the
idea
of
presenting
logical
clusters
already
happened
upstream
or
not
yet.
B
Well,
when
I
asked
Andy
about
this
and
I
think
those
things,
maybe
even
a
previous
community
meeting
and
I
think
I
also
had
a
separate
conversation
with
Andy.
B
My
understanding
of
what
he
was
trying
to
tell
me,
and
maybe
I
did
not
fully
understand
or
maybe
I'm,
not
remembering
fully,
but
my
current
takeaway
understanding
is
that
the
kcp
project
has
tried
to
sell
its
concept.
It's
virtue
well
has
tried
to
sell
multi-tenancy
in
the
API
server
and
the
Upstream
community
did
not
buy
it
because
it's
not
really
a
full
isolation.
They
said.
B
If
you
want
to
go,
multi-tenant
run
multiple
binaries,
don't
don't
try
to
put
the
multiple
tenants
together
in
one
binary,
because
there's
going
to
be
interference
between
them?
Sure
right
and
that's
true.
B
You
know
in
fact
we've
been
trying
to
make
you
know
with
Jamie
I've
been
working
on
making
API
priority
and
fairness
getting
that
into
the
kcp
binary
and
we
kind
of
gave
we
had
an
initial
design
that
really
did
isolate
the
tenants
and
people
told
us.
Well,
that's
too
expensive.
We
want
probably
a
design
that
gives
less
isolation
between
the
tenants.
So,
for
that
and
lots
of
other
reasons
you
know
go
just
doesn't
have
the
you
know,
true
total
isolation.
B
So
there's
this,
it
really
is
a
weak
form
of
multi-tenancy
and
I.
Think
thinking
of
it
as
an
additional
layer
of
virtualization
is
is
the
thing
that
could
make
sense,
and
that's
really
the
question
to
take
Upstream
is:
does
it
make
sense
to
add
a
higher
level
of
namespace
or
virtualization
in
the
API
server,
and
when
I've
asked
Upstream
in
the
API
Machinery
Sig,
you
know
they
said
yeah,
we
don't
believe
we
need
a
hire,
a
super
namespace
or
a
higher
level
of
namespace
and
I.
B
C
D
C
B
Point
open:
no,
we
we
know,
we
know
they
tried
to
sell
multi-tenancy
and
that
didn't
work.
Okay
and
I
can
understand
that
not
working
the
question
is-
and
maybe
this
is
a
question
for
the
API
Machinery
segment
yeah
rather
than
this
group.
You
know
why
not
high
level
namespace,
you
know
we
have
I
enumerated
in
the
issue
here.
Four
use
cases
for
higher
level
namespace.
E
Yeah
I
just
want
to
jump
in
since
representative
of
cross
plain
here.
Lots
of
what
kcp
offers
is
not
super
critical
to
us
and
I.
Think
Mike
is
pretty
spot
on
here.
That,
like
this
logical
namespaces,
is,
is
really
the
the
critical
component
that
we
care
about
and
I
think
you
know,
there's
two
different
paths.
I
could
see.
It
would
be
great
if
this
was
just
supported
Upstream
in
our
eyes.
E
Right
and
that's,
you
know
kind
of
our
goal
with
most
of
what
we're
doing
the
kind
of
like
secondary,
win.
I
guess
that
we
could
see.
Just
you
know
from
our
perspective
is
that
it
was
modular
enough
that
it
would
be
very
easy
to
build
something
that
maybe
maybe
it's
not
formally
supported
Upstream,
but
it's
very
easy
to
enable
it
right.
The
the
pieces
are
easy
to
fit
together.
E
I,
don't
have
you
know
a
design
for
what
that
looks
like,
and
the
complexity
of
that
may
be
more
challenging
than
you
know,
I'm
alluding
to
here.
But
that
is
a
that's
another
Avenue
that
we
wonder
if
that
would
be.
You
know
a
reasonable
outcome
here,
but
like
I
I
also
don't
have
the
context
of
the
previous
Upstream
conversations,
so
I'd
love
to
learn
more
about
what
happened.
There.
B
And
if
by
modularity
you
only
mean
adding
stuff
into
a
binary,
I
mean
that's
kind
of
a
given
right.
We've
already
demonstrated
that
you
can
put
together
a
binary
that
that
you
know
puts
together
stuff
and
that's
already
in
the
plan
right
they've
already
sold
the
generic
control
plane
Upstream
right.
So
if
we
could
get
not
only
the
generic
control
plane
library,
but
this
super
namespace.
You
know
my
question
is
yeah.
E
I
think
it
would
be-
and
maybe
it
actually
I
I'd
run
this
by
you
and
see
if
this
sounds
right
from
like
a
tangible
perspective,
I
would
love
for
and
I'm
sure
you
know
most
folks
in
the
k-speed
community
would
as
well.
It
would
be
great
if
the
like
Fork
of
kubernetes,
that's
currently
under
kcp,
could
be
not
a
fourth
right
like
it
could
be
as
Upstream
as
possible
and
everything
else.
We
just
added
in
yeah.
B
A
Yeah
I
think
from
my
perspective,
I
I,
don't
know
enough
about
those
particular
components
to
make
an
educated
statement.
I
I
know
this
is
something
that
Andy
and
Stefan
are
talking
about
this
week.
A
C
In
the
meantime,
we
can
take
the
time
to
maybe
elaborate
more
about
the
possible
use
case
that
can
benefit
from
that
and
that
way
we
can
discuss
those
use
case
and
those
examples
like
the
course
plan
and
all
of
that
next
week.
Right
and
maybe
you
know,
we
can
come
to
some
some
uniform
conclusion
right.
A
Yep
sounds
good,
so
I'm,
sorry
I
did
not
paste
this
into
the
Google
meet
chat,
but
if
anyone
else
has
an
agenda
item
before
we
move
on
to
the
incoming
issues,
you
can
raise
your
hand,
and
we
can
add
that
on.
A
If
not,
we
can
go
through
a
triage
really
quickly.
A
Okay-
let's
see
here
a
number
of
these
already
have
signee
so
I'm
just
going
to
do
those
really
quickly.
A
All
right
bugs
and
Flakes
I'm
going
to
just
put
into
backlog.
A
Okay
and
Steve
is
looking
for
a
way
to
separate.
A
A
Okay,
this
looks
like
a
technical
debt
kind
of
thing,
so
I
think
at
this
point
everything
is
backlog.
A
I
will
give
folks
one
more
opportunity
to
bring
up
any
other
topics.
Otherwise
we
can
get
45
minutes
back.
E
I
had
a
quick
question,
and
this
might
be
something
that
we
just
need
to
defer.
I
can
jump
in
slack
and
ask
as
well,
but
I
know
there
was
some
work
I
believe
Steve
was
leading
it
around
ultimate
alternate
storage
interfaces
specifically
focused
on
cockroach
TV
feel
free
to
just
direct
me
to
do
some
reading
instead
of
taking
up
time
here,
but
is
there
any
updates
or
status
on
that
effort?
Right
now,.
A
No
updates
yet
I
know
I,
think
he's
still
in
the
Prototype
stage.
I
know.
The
next
step
we
would
want
to
do
is
get
maybe
more
fleshed
out
prototype
with
a
single
node
going
and
then
see
what
we
need
to
do
to
get
that
working
across
shards
I
know
he
was
saying
this
morning
that,
even
if
we
would
stick
with
ETD,
that
is
a
non-zero
amount
of
work
to
get
that
working.
The
way
we
want,
but
as
of
now
I,
do
not
know
about
storage
options.
C
Yeah
we
actually
were
working
with
Steven
we
in
in
I
mean
we
developed
the
POC,
not
PRC.
You
can
actually
see
pull
request
on
the
kubernetes
fall
coffee
of
kcp.
That
is
actually
working,
we've
tested
it.
Everything
is
working
and
we
even
I
even
created
the
dummy
PR
in
the
kcp,
a
repository
to
make
sure
we
are
passing
all
the
tests
and
all
of
that,
so
everything
is
working
and
I.
Think
a
week
ago,
notified
me
that
he
is
moving
that
to
draft,
because
there
is
no
way
to
say
Do,
not
merge
or
anything.
C
However,
the
problem
now
is
more
of
I,
wouldn't
say
political,
but
currently
we
we
believe
that
the
merging
this,
as
is,
will
cause
the
Gap.
E
C
Related
to
the
previous
discussion,
it
will
cause
the
it
will
increase
the
gap
between
kubernetes
and
kcp,
and
we
don't
want
to
do
that.
So
I
think
Steve
pointed
me
to
some
work
that
was
done
initially
in
kubernetes
was
stopped
in
the
middle.
It
was
done
for
a
different
reason
to
help
with
testing
and
so
on
on
the
databases,
but
he
believed
that
if
we
will
help
Upstream
that
work
from
some
guy
from
Google,
it
will
make
it
easier
to
integrate
cockroach
and
I
want
to
mention
also
that
you
know
to
be.
C
To
be
honest,
there
are
some
gaps
even
in
the
current
solution,
in
the
cohorts
right,
both
technical
features,
wise
and
so
on
that
are
still
not
not
solved.
The
reason
I
wanted
to
push
it
is,
is
you
know
fast
as
possible?
Is
that
we
will
now,
of
course,
need
to
rebase
again
do
everything
again,
because
the
kubernetes
for
keep
changing
right
so,
but
let's
see,
we
all
believe
and
also
last
comment
is
I.
C
Also
heard
Steve
told
me
this
directly
also
that
he
believes
that
when
we
go
to
military
regime-
and
all
of
that
you
know-
because
we
are
interested
also
in
use
case
in
which
you
have
huge
amount
of
cluster
in
a
single
age.
Okay,
but
for
multi-region
Steve
thinks
that
we
cannot
go
forward
with
it
is
today
it
will
just
you
know:
it's.
C
C
I
didn't
know
that
you
are
interested,
so
I
will
keep
you
in
the
loop,
that's
good
to
know
and
well
yeah
yeah.
We
have
it.
We
have
no,
we
have
a
demo,
but
we
yeah.
We
have
a
demo,
yes,
I
I
can
send
you
a
video
of
the
demo
and
we
also
have
some
performance
results
of
the
latency.
The
the
you
know,
watch
cash.
Latency
is
all
of
that
and
performance
of
memories.
Beyond
of
that
compare
it
between
cocoa
geneticity.
A
Okay,
thank
you
for
that.
Ezra
I
was
not
aware
that
you
had
that
fork
running.
That's
awesome
and
you
mentioned
we've
had
discussions
on
the
on
the
kcp
fork
and
rebasing
and
everything
I
know.
Steve
has
been
evaluating
for
a
bump
to
I,
think
125.,
so
yeah
well,
I
think
that's
coming
soon
as
well.
I
again
I,
don't
know
if
it's
going
to
be
zero
11.
that
that
makes
it
in
but
yeah
the
the
fork
I.
B
Think
the
finer
point
on
that
yeah,
what
I
recall
from
a
recent
meeting
was
the
plan
was
first.
There
was
this
other
big
change
in
the
way
logical
clusters
are
done.
That's
coming
in
a
series
of
PRS.
The
plan
was
to
get
that
done
and
then
do
the
rebase.
A
Yeah
we're
pretty
close
on
the
sharding
I
know,
there's
a
bunch
of
edes
that
need
to
be
fixed
still
so
yeah,
that
is
your
of
operations.
You're,
correct.
A
Thanks
everybody,
if
nobody
has
anything
else,
I'm
getting
the
meeting
notes
for
next
week
started
since
we
have
a
bunch
of
deferred
topics
and
I'll
drop
that
in
the
slack
channel.
If
folks
want
to
add
anything,
yes,
fish,
new,
okay,.
D
Hi
everyone
I'm
new
to
this
community,
so
I've
been
trying
out
kcp
on
my
openshift
local
CRC
cluster,
so
I
had
an
issue
with
the
Thinker.
It's
able
to
you
know,
identify
I,
mean
sync
other
resources
from
my
local
cluster
to
a
kcp
workspace.
D
So
when
I
tail
for
the
podlocks
I
see
it
tries
to
perform
I
get
operation
on
a
private
IP
and
it
fails
due
to
you
know,
connection
I
mean
due
to
some
authentication
problem,
so
I
have
raised
the
issue
for
that
I
see
it's
more
due
to
the
backlog,
I
mean
so
is
there
I
mean?
So
what
can
I
expect
on
that?
Like?
Is
there
any
other
way?
I
can
make
Sinker
working
so
that
I
mean
just
to
try
out
kcp
on
my
CRC
cluster
yeah.
A
Let
me
look
for
that
issue
and
ping
either
David
or.
D
Here
is
the
issue
I'll
just
post
it
here.
So
this
is
what
I
tried.
I
have
added
steps
as
well:
okay,
great
and
now
it's
like
from
similar
issues
to
more
which
are
low,
which
is
which
are
the
same
excuse
so
I
guess
there
are
others
also
who
are
impacted
by
this
got.
A
It
I'm
counting
contacting
David,
because
I
have
not
touched
this
part
of
the
code
very
much
I'm
working
a
lot
on
the
API
exports,
unfortunately,
is
in
one
of
the
meetings.
Today.
That's
pulled
other
people
away,
so
I
will
I've
pinged
him
in
the
issue
and
I'll
contact
him
off
separately
to
let
him
know
it's
there.
Okay,.
D
A
Yeah
I
don't
know
that
we'll
go
into
detailed
troubleshooting,
but
you
can
definitely
raise
them.
Just
like
you
did.
Okay.
A
All
right
well
have
a
great
week,
everyone,
and
we
will
see
you
next
week.
Thank
you.