►
From YouTube: 20230223 SIG Architecture Community Meeting
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,
hello,
everybody
today
is
February
23rd
2023,
this
kubernetes
architecture,
meeting
and
I
believe
we
have
a
pretty
short
agenda
today.
A
Okay,
here
we
are
hippie
hacker.
Take
it
away.
B
C
A
while
back,
we
decided,
we
went
into
the
seg
what
the
shape
of
kubernetes
was,
and
it
looked
like
this
and
we
have
reached
it.
We're.
C
D
C
C
But
when
you
first
reach
that
it
does
what
we
normally
do,
it
shows
you
every
endpoint
in
kubernetes.
So
if
you
want
to
bring
up
the
next
page
of
you
know,
just
go
to
API
snoop.cncf.io,
what's
fun
about,
that
is,
it
shows
the
whole
shape
of
kubernetes
and
all
of
the
endpoints.
C
If
you
want
to
Mouse
over
them
and
whatnot
all
of
our
history
for
our
different
releases
and
what
we
you
know,
you
can
click
on
alpha
or
beta
and
see
what's
up
and
coming
it'd
be
nice
somewhere
to
get
the
velocity
of
where
they
are.
In
the
you
know,
the
development
promotion
process,
but
if
we
scroll
down
just
a
little
bit
to
some
check
boxes
there,
the
first
check
box
there
excludes
conformance
ineligible.
A
C
Means
obviously,
to
be
for
conformance
to
be
an
eligible,
it's
got
to
be
alpha
or
or
stable,
and
then
it
has
to
be
approved
by
this
group.
So
this
is
what
it
looks
like
currently.
If
we
want
to
see
our
success,
we
also
exclude
the
endpoints
who
who's
in
whose
eligibility
is
pending
that
second
checkbox.
This
is
the
100
as
of
today.
I
want
us
to
stop
celebrate.
We
did
it,
there's
a
nice
dance.
We
should
do.
D
A
Yes,
fantastic
great
work
like
you
guys
have
been
working
for
years
and
it's
been
a
long
route
really
appreciate
it.
That's
fantastic!
Absolutely.
C
C
But
right
now
the
main
thing
is
is
is
what
we
really
think
should
still
be
part
of
it
that
we
put
off.
So
this
celebration
is
just
a
slight
small
dance
and
if
you
uncheck
that
second
check
box,
this
is
the
the
objective
and
key
result
I'll
find
in
one
like
one.
One:
click.
A
C
Right
and,
and
so
if
we
go
back
to
our
report
or
wait,
let's
go
celebrate.
The
other
thing
go
to
the
the
go
to
the
top,
the
history
right.
So
if
you
click
on
conformance
progress,
it's
the
at
the
very
in
the
green
at
the
very,
very
top
of
the
whole
page,
yeah
yeah,
that
so
far,
we've
had
gray
stuff,
that
is,
technical
debt
and
we
have
completed
100.
The
technical
debt
as
of
today
and
the
red
stuff
is
where
we've
colored
in
that
by
adding
tests.
C
The
the
turquoise
stuff
is
where
the
community
has
done
a
great
job
at
bringing
things
forward.
You'll
notice
there's
not
been
a
whole
lot
of
new
endpoints
since
Kobe
either
we
have
reached
maturity.
A
C
B
A
C
Right
I'm,
aware
of
we'll
have
to
go.
We
can
we'll,
let's
put
that
on
the
to
check,
Creon
to
make
sure
but
I'm
I'm
I,
just
you
know
top
of
the
memory
that
top
graph
is
going
to
change.
If
we
approve
this
in
this
meeting-
or
you
know,
whenever
the
consensus
is
made,
one
color
that's
missing.
Is
these
new
newly
yet
eligible
endpoints?
C
This
graph
is
based
off
of
PR
being
merged
that
we
haven't
merged
yet
and
that's
rion's
part.
That's
next!
That
defines
the
body
of
work
for
the
likely
the
rest
of
the
year
be
sure
that
we're
happy
with
this
work
so
that
I
can
put
it
on
the
objective
and
key
results
for
the
cncf
for
II,
for
for
this
year
I'll
hand
that,
over
to
urian
I
took
may
I
took
some
of
the
steam
out
of
it,
but
this
this
is
the
these.
D
B
B
When
I
asked
for
an
illegible
endpoint
to
be
moved
onto
the
list
last
year,
I
got
some
very
good
food
feedback
from
Jordan
really
gets
saying,
but
some
of
these
things
should
not
be
on
the
list
anymore,
which
we
knew
all
along
with.
We
basically
said
in
several
meetings
that
we
we
will
review
when
we
get
to
the
end.
B
So
now
we
did
the
review
I
created
an
Excel
list,
a
lot
of
people
actually
in
front
of
writing
list
looked
at
it,
the
identified
endpoints
that
should
be
actually
moved
off
the
ineligible
list
and
be
tested.
So
there
were
some
comments
on
that.
So
I
basically
made
this
PR
to
move
endpoints
from
the
ineligible
endpoints.
Let's
do
a
new
list
which
is
endpoints
that
will
be
tested.
B
Let
them
run
stable,
like
we
always
did
for
the
two
weeks,
then
promote
that
to
conformance
and
as
the
performance
promotion
get
merged.
We
will
make
another
PR
to
actually
remove
the
bending
in
the
eligible
endpoints
off
this
list,
so
this
list,
the
idea-
is
the
next
12
to
18
months
at
the
most.
This
list
actually
needs
to
be
empty.
What
we,
what
I
would
recommend?
B
We
retain
the
list
that
there
is
endpoints
that
come
in
that
show
up
as
technical
debt,
but
they're
in
beta
there's,
there's
some
funny
nuances
in
the
way
opposed
to
pick
things
up
when
they
they
actually
get
introduced.
So
this
is
a
bit
of
a
storage
space
where
we
can
keep
that
so
worth
monitoring
apis
new
with
monitoring
con
performance,
but
one
of
the
other
things
that
we
will
do
is
look
for
endpoints
that
come
in
that's
not
tested.
Yet
what
should
be
on
this
list?
B
A
You
said:
there's
a
bunch
of
discussion:
is
there?
Can
you
put
a
pointer
to
that
discussion
here
like
a
bunch
of
these,
are
are
like?
How
do
we
test
delete
core
V1
node
in
in
our
test
framework?
Are
we
gonna?
Is
that
something
we
can
test
in
our
test
framework,
or
is
that
why
it
was
on
the
list
before
and
now
we're
gonna
modify
the
test
framework
then
to
be
able
to
test?
That
is
that
the
idea.
C
The
discussion
so
far
have
happened
in
slack
and
in
surrounding
the
message
to
the
Sig
architecture.
Mailing
list.
That
says
here
is
the
idea
and
the
spreadsheet.
So
there
was
a
there
was
the
overall
concept
that
we
couldn't
put
forward
and
then
the
on
Excel.
What's
it
called
the
sheet
with
all
the
Google
Sheets,
with
all
the
endpoints
and
talking
about
why
they
were
there,
and
there
were
people
that
came
back
on
specific
once,
and
this
is
the.
What
do
you
call
it?
C
D
A
I
mean
I'm
all
for
adding
more
right,
like
like
none
of
assuming
none
of
them
cause
problems
for,
like
you
know
the
kind
of
reasons
we
consider
conformance
tests,
I
I,
just
I
I-
want
to
know
that
you're
you're,
confident
you
have
solutions
for
the
issues
that,
like
oh
there's,
a
ton
of
storage
ones
on
here
and
I
think
that
they
were
listed.
C
Are
there
reasons
they
should
be
removed
and
will
come
up
with
solutions
for
them?
Because
that's
that's!
What
we're
here
for
is
to
help
create
the
solutions.
We're
we're
not
here
to
say
this
can't
be
done
because
there's
not
a
a
solution
to
video
up.
We
get
pushback
when
they
say:
there's
not
a
way
to
test
this
at
all
right,
and
so
those
discussions
have
kind
of
occurred
and
I.
C
Think
at
this
point
it's
discussions
around
just
getting
more
eyes
on
it,
and
it
can
change
over
time
that,
but
we
just
wanted
that
definition
directly
in
in
in
a
file
controlled
by
this
sig
in
that
owner's
file,
so
that
the
authority
we're
trying
to
put
it
right
in
front
of
the
correct
Authority.
A
C
These
are
the
nuttier
ones,
they're
harder.
That's
why
they
that's
why
we
said
no
just
don't
for
now,
because
we
wanted
to
keep
velocity
going
in
Focus.
Now
we
have
picked
all
the
low-hanging
fruit.
That
tree
has
got
a
line
about
Stephen
and
reach
tall.
Now
we
got
to
get
the
ladder
out
and
get
the
rest
of
the
apples.
Okay,.
B
There
should
be
discussion
on
going
forward,
so
one
of
the
last
points
on
my
agenda
is
because
we
have
the
two
yaml
files
now
and
it's
probably
anybody's
allowed
to
make
a
PR
and
say
or
sickness.
D
B
So
so
is
it
really
just
making
the
work
public
and
encouraging
as
I'm
the
person
that
tried
to
get
the
list
knocked
down
as
fast
as
possible,
so
I'm
all
for
br's
taking
things
off
again,
but
then
we'll
have
a
history
of
why
things
are
on
our
off
the
list?
Oh
yeah,
so
it's
yeah.
A
Yeah,
you
know
going
back
to
this
like
what
we
did
early
on
part
of
the
ineligible
list
came
from.
Hey
is:
does
this
affect
user
work,
user
workload?
Portability
right
like
registering
nodes
and
unregistering
nodes,
is
sort
of
like
cluster
construction.
That
I
can't
see
from
a
conformance
point
of
view.
I
guess
you
could
have
like
like.
If
you
want
to
say,
if
the
control
plane,
you
want
to
be
able
to
create
your
own
cubelet
like
virtual
cubelet,
then
maybe
you
want
to
know
that
and
you
want
to
use
that
with
any
kubernetes.
A
You
want
to
know
that
the
way
that
you
register
and
and
deregister
works
across
them,
but
that's
those
are
very,
very
unique
use
cases,
which
is
why
we
could
afford
to
make
them
not.
You
know,
they're,
not
part
of
the
lying
fruit.
They're,
not
not.
A
lot
of
users
are
going
to
be
worrying
about
that
and
then
there's
permissions
issues
around
the
test.
Frameworks
and
things
like
that.
So
anyway,
I'm
open
for
for
putting
more
on
there.
A
We
have
to
make
sure
that
when
we
do
like
how
do
you,
how
do
our,
if
it's
a
hosted,
kubernetes
right
like
gke
or
right?
Yes,
how
does
that
test?
Suite
doesn't
enable?
Does
the
cloud
provider
or
whoever
the
service
provider
hosting
that
cluster
allow
a
user
process
to
execute
those
yeah
those
apis
and
if
not,
then
we
it
could
be
pushing
somebody
out
of
conformance
for
a
reason.
That's
not
really
the
spirit
of
the
program.
C
I've
thought
about
this
as
well.
There's
two
two
parts
here:
one
is:
we
need
to
update
the
conformance
submissions
a
bit
to
because
we
need
to
declare
that
the
running
the
conformance
suite
is
destructive
because
it
needs
to
be
right.
C
Then
we
go
ahead
and
have
tests
that
are
conformance
eligible,
but
not
conformance
required
that
so
they
they
are
part
of
our
con.
Whenever
anybody
makes
a
submission
to
you
know,
cncf
slash,
Kate's
conformance,
it
includes
the
results
of
these
tests
and
these
tests
are
ones
like
delete
node,
and
so
when
we
see
people
we
just
get
a
list
and
that
next
conformance
release
of
things
that
we
know
that
we
want
that
will
force
failures
in
the
next
release.
I.
C
Would
be
a
good
idea
and
the
other
one
we've
gone
back
and
forth
in
is:
is
our
back
I
I
I
have
never
used
a
cluster
in
my
entire
career.
That
does
not
use
our
back,
and
so
that
particular
Edge
case
feels
like
one
that
we
need
to
Corner
the
the
people
that
would
fail.
Like
add
that,
to
the
thing
start,
the
discussion
you
didn't
pass,
R
back.
Can
you
tell
us
why
you
know
and
is
that
what
what
percentage
of
people
that's
going
to
upset.
E
C
E
E
It's
not
clear
to
me.
What
conformance
for
our
back
would
mean.
A
Well,
let's,
let's
yeah,
okay,
so
so
stepping
back
a
second
conformance
for
our
back
would
just
mean
that
you
can
the
conf
the
the
rbac
endpoints
work
and
are,
and
when
you
add,
say
an
enforcement
works,
it
works
right.
The
the
rules
do
what
what
you
expect
them
to
do
now,
the
conformance
program
in
general
and
it's
a
psych.
So
the
fundamental
thing
that
very
frequently
gets
mixed
up
clusters
themselves
can't
be
defined
as
conforming
or
not
just
like
it's
it's
implementation.
A
It's
the
distribution
has
to
be
able
to
produce
a
conforming
cluster
so
like.
If
you
go
and
create
a
GK
cluster
straight
out
of
the
box,
we
may
well
have
things
configured
in
such
a
way
that
we
wouldn't
pass
conformance
tests.
But
if
you
set
I'm,
not
sure
if
that's
true,
but
if
you
set
the
right
options,
I
think
it
is.
When
you
create
the
gke
cluster,
you
can
create
a
conforming
cluster.
Any
user
can
do
that.
So
that's
the
that's
the
criteria
to
be
able
to
pass
conformance
and
get
your
little
badge.
A
E
Sort
of
but
again
like
going
back
to
the,
how
do
you
actually
make
sure
that
it
does
what
it?
What
you
expect
it
to
do
is
actually
is,
is
really
difficult
because
you
have
to
be
able
to
send.
You
have
a
logically
have
to
send
every
possible
request
right,
because
you
can
imagine
someone
who
wires
up
a
deny
authorizer
to
deny
someone
to
create
I,
don't
know
a
namespace
named
openshift.
E
You
would
never
be
able
to
catch
that
in
in
conformance
like.
If
so,
if
that's
your
goal,
you're
not
going
to
be
able
to
do
that.
C
C
It's
one
and
I'm
just
saying
that
as
a
silly
example,
but
there's
got
to
be
some
way
to
test
just
the
general
functionality
of
our
back
as
a
whole
across
the
endpoints
we
that
are
that
are
high
priority
or
academic.
Some
type
of
of
we
can't
test
like
you're
saying
for
every
possible
input
to
it.
We
can
test
for
the
general
functionality
and
expectation
that
the
most
common
scenarios
are
do
work
with
our
back
on
a
cluster
that
allows
the
use
that
is
configured
to
allow
the
usage
of
our
back.
A
Yeah
I
mean
we
can
we'll
we'll
discuss
it
more
over
time.
I
think
that
you
also
mentioned
optional
features
and
that's
a
a
mountain.
We've
tried
to
climb
a
couple
times
and
we've
fallen
off
each
time
and,
and
there
doesn't
seem
to
be
an
appetite
for
it,
but
I.
It's
not
something.
I
I
think
there's
value
in
it,
and
and
maybe
now,
as
the
Project's
more
mature,
there's
more.
C
Our
back's,
the
only
one
that
I
thought
we
should
bring
inside
of
the
optional
into
the
into
the
you
need
to
support
this
somehow,
as
it's
an
expectation,
the
optional.
A
Things
I
don't
think
we
should
make
an
expectation
like
like
I,
don't
think
we
should
think
you
can't
say
your
kubernetes,
if
you,
if
you
can't
enable
our
back
I,
mean
imagine
I'm
imagining
things
like
like
a
a
small
Edge
cluster
or
something
where
they
just
wanna.
They
only
have
one
user,
because
there's
just
one
user
managing
like
there's
one
automation
that
manages
it
and
doesn't
care
right
like
they
may
handle
authorization
for
creating
resources
externally
to
the
cluster
itself.
A
C
It
we
don't
have
time
in
this
release,
but
maybe
for
the
next
one.
Let's
just
add
some
simple
string
up
the
simplest
R
back
thing
as
the
as
a
probe
and
see
in
the
submissions
for
that
release.
How
many
fail
just
so,
we
can
actually
know
which
distributions
are
currently
submitting
that
wouldn't
pass.
It
sure.
C
Yeah
the
provisional
thing
that
way
it
allows
us
to
have
a
really
structured
conversation
around
probing
for
the
for
the
optional
things
and
using
this,
as
the
first
use
case,.
A
Okay,
all
right,
thank
you!
That's
awesome,
great
progress,
and,
and
now
now
we're
gonna
I
guess
put
a
more
big
bunch,
more
work
on
your
plate.
So
that's
you.
You
achieved
your
goal
and
your
PR.
Your
reward
is
more
work.
D
B
If
we
can
get
some
eyes
in
that
PR
and
just
merge
it
in
and
we
will
happily
go
sort
out
the
rest
of
the
work
thanks
for
all
the
support
there
was
really
I
know.
I've
been
nagging
all
around
to
everybody
to
review
things
to
get.
D
A
A
A
Okay,
thank
you,
everybody
and
have
a
great
weekend
we're
almost
we're
almost
to
the
weekend.
I
guess
there
in
New,
Zealand,
you're,
probably
already
on
the
weekend
right.
It.