►
From YouTube: 20221103 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
Foreign,
hello,
everybody:
this
is
kubernetes
architecture,
meeting
for
November,
3rd,
2022.
B
C
A
Think
we
have
I,
guess
I
the
Rihanna
you
or
Stephen
want
to
take
over
I.
Think
the
only
thing
on
the
agenda
right
now
is
conformance
update
from
that
performance
sub
project
which.
C
Yes,
it
it
is
us
we
have
hijacked
this
meeting,
it
seems
is
too
give
it
to
me.
It's
fine
I've
got
the
agenda
open
here
then
I'll
roll
with.
Let
me
move
that
over
here,
move
that
through
my
middle
screen,
all
right,
you're
awesome,
share,
share
a
screen
and
I
want
to
share
that
one.
There
we
go,
you
can
see
my
screen.
It's.
C
We're
just
about
those
the
the
126
release.
Now
we
don't
think
we're
gonna
get
anything
in
else
in
because
it's
only
the
nastiest
stickiest
things.
That's
left
great
job
there
from
Steven,
with
the
last
test
that
came
in
we've
got
97.5
tested.
C
If
we
look
at
the
pie,
chart
there's
only
those
three
small
gaps
that
still
need
to
be
full
only
in
in
points
the
little
gray
spot
at
the
top
of
the
the
program
or
the
stack
graph.
B
C
Was
Sweet
Blood,
Sweat
and
Tears,
but
it's
nice.
We,
we
really
hope
Stephen
and
I.
Are
we
tried
to
do
it
before
Detroit,
which
was
too
hard
because
of
the
the
quality
of
endpoints
left?
But
we
really,
the
idea
is
we
should
be
done
by
by
Amsterdam
and
both
Stephen
and
myself
hope
to
be
there
to
celebrate
it.
C
C
So
Stephen
is
at
the
moment
researching
quite
hard
on
the
the
Pod
one,
so
pod
binding
pod
forward
and
part
attached
I'm
doing
a
bit
of
a
Roman
writing
on
across
the
three.
Then
we
still
have
authorization,
which
is
clearly
we
haven't
done
much
of
those.
So
it's
got
to
take
quite
a
bit
of
research
to
do
that,
so
we
hope
to
have
tests
or
need
to
be
tested
ready
for
review
before
end
of
the
year.
C
So
we
can
have
an
early
early
wins
again
next
year.
Early
in
January,
then
we
have
the
sticky
ones.
That's
left
behind
is
the
API
service
endpoints.
We
already.
B
C
Think
we've
had
yeah,
we
had
a
test
ready,
ETV
test
ready,
but
it
required
all
back
if
I
remember
correctly
Stephen,
you
got
to
correct
me
here.
If
I'm
wrong,
then
we
changed
it
around
not
to
use
our
back
on
advice
from
Jordan.
We
had
a
working
e2e
test
was
doing
well
on
all
the
pre-checks
got
it
in,
and
then
it
went
terribly
flaky
and
breaking
things
for
all
four
endpoints
about
it
got
reverted.
C
One
of
the
problems
is
there
is
a
bug
specifically
on
replace
API
registration,
server
status,
endpoint
there's
an
issue
for
that,
so
we
we
will
bring
that
around
when
we
have
more
attendees
to
discuss
this,
it
might
be
that
we
have
to
put
that
one
on
to
unfold.
That's
fixed
with
that
on,
as
you
know,
eligible
because
it
breaks
things
especially
on
an
h8
cluster
and
then
the
other
thing.
C
What
I'm
considering
is
early
next
year,
go
to
API
machinery
and
ask
if
there's
anybody
there
that
will
come
alongside
us
and
have
a
look
at
these,
because
this
will
be
the
last
last
one
percent,
because
at
the
moment
every
end
point
is
worth
24.25.
So
the
these
will
be
the
last
one
percent
and
we
will
ask
them
nicely.
Please
come
along.
We've
had
some
real
good
slugs
of
this
with
no
success
yet
so.
A
Yeah
yeah
and
I
think
and
I
can
I
can
see
who
I
can
pull
in
from
that
from
Google
at
least
like
or
we
could
talk
to
David
from
Red
Hat
like
so.
C
B
A
B
C
It
a
shame,
a
shame
point
which
is
not
the
idea.
Okay,
so
that
is
for
me
for
reporting
where
we
are
so
John.
If
you
do
have
somebody
in
mind,
I
really
appreciate
if
you
can
get
some
help
on
this
I
know
everybody's
crazy,
but
it's
this
time
of
year.
Well,
maybe
a.
A
C
C
So
I
will,
if
you
don't
get
some
I'll
if
I
will
be
your
conscience
opinion
in
sec
in
a
week
or
two
if
and
just
check,
if
you're
checking,
if
you
don't
mind,
okay,.
A
C
A
Yeah
performance
tests
we
have
less
than
what
is
less
than
two
percent
less
than
three
percent
left
untested,
and
these
are
the
ones
left.
I,
don't
know
if
these
ones
need,
if
the
team
needs
help
or
with
with
these,
but
if
you're
familiar
with
these
might
be
helpful
to
get
jump
in
and
help
out
with
conformance
really.
C
A
C
A
Awesome
and.
C
Then
the
next,
the
next
point
is
I'm
gonna,
throw
it
over
to
Stephen
he's
done
quite
a
bit
of
research
on
on
binding
already,
so
he's
got
some
questions.
We
want
to
take
that
Stephen.
B
B
That
of
course
follows
the
standard
scheduler
process
and
I
was
just
wondering
whether
or
not
that
sounds
like
an
okay
idea
for
the
test
to
be
able
to
do
that
and
I
found
a
an
initial
how
to
use
a
another
scheduler,
of
course,
I'm
going
to
have
to
use
the
package
scheduler,
of
course,
for
the
go
code.
But
it's
just
trying
to
just
going
with
the
general
idea
of.
B
A
A
To
schedule
anything,
you
just
need
to
hit
the
API
endpoint
and
make
it
sure
works
yeah.
That
sounds
like
a
reasonable
thing
to
me.
My
only
question
would
be:
can
you
configure
a
scheduler
which
it
looks
like
you
can
without
mucking,
with
command
line
Flags
right?
If
you
can
do
it
through
the
API,
then
that
seems
like
a
good
approach.
B
If
you
scroll
back
to
the
top
of
the
a
document
round,
there
is
the
reference
there
about
using
the
implementation
from
the
package
scheduler.
So
what
I'm
expecting
to
do
is
to
be
able
to
go
and
look
at
that
set
of
packages
and
pull
in
that
actually
directly
into
the
test
as
part
of
the
test
in
I.
B
A
Probably
fine
but
I,
don't
know
that
you
need
to
do
something
that
heavy,
like
you
really
need
a
fake
scheduler
right.
Yes,
that
you
hit
the
endpoint
and
it
does
what
you
expect
it
to
do
like
you,
don't
need
any
sophisticated.
You
know
real
scheduling,
so
I
would
consider
just
writing
a
fake
scheduler
class
that
makes
use
of
the
endpoint
and
can
validate
that
the
endpoint
does
what
you
expect,
without
necessarily
having
to
bring
in
I
I've.
Never
looked
at
this
code,
so
I
have
no
idea
what
it
does.
Yeah.
B
A
Maybe
that
may
be
I
just
if
you
think
about
what
your
end-to-end
testing
your
end-to-end
testing
kind
of
the
ability
to
inject
a
custom.
Scheduler
you're,
not
testing
the
functioning
of
that
scheduler.
So
in
that
sense
the
end-to-end
test,
in
my
mind,
is
that
your
scheduler
gets
invoked
or
whatever
in
the
way
it's
supposed
to
and
that
it
underlying
apis
that
it
relies
on.
Do
what
they're
supposed
to
but
I,
don't
know
enough
details
about
this
area
of
the
code
to
provide
a
definitive
answer
on
that.
B
Sorry,
that's
not
going
to
be
good
enough.
We're
going
to
have
to
reject
it
a
different
way.
So,
thanks
for
the
tip
yeah
sure
thanks.
C
Right
then,
if
you
do
it,
that's
why
Stephen
and
I
discussed
about
our
back
would
Arabic
be
required
for
that,
or
can
we
do
it
without?
Because
that
opens
another
one.
A
A
Yeah,
as
far
as
I
know,
there's
no
reason
why
we,
you
can
constrain
things
with
our
back,
but
if
our
back
is
off
as
I
recall,
we
usually
do
these
conformance
tests
such
that
we
attempt
or
we
we
check
for
our
back
or
something
and.
B
A
Something
like
that,
I,
don't
know,
there's
a
few
places
where
we
do
that,
because
I
recall-
and
we
like,
if
somebody's
running
the
performance
test
against
a
cluster
that
has
our
back
enabled,
then
a
conformance
test
should
still
work
and
be
subject
to
that,
our
back
that
our
back
has
to
be
set
up
properly
for
that
conformance
test
to
work.
If
somebody's
running
a
conformance
test
against
a
cluster
that
does
not
have
our
back
enable
it
should
still
work.
A
That's
that's
what
we
mean
when
we
say
our
backs,
not
part
of
conformance
it's
that
we
should
be
indifferent
to
whether
it's
set
up,
but
if
it
is
set
up,
it
has
to
be
set
up
properly.
If
we
can't
set
it
up
properly,
like
the
context
of
where
how
our
end-to-end
tests
run
aren't
going
to
have
that
level
permission,
then
we
we
should
discuss
that
like
I
think
we
we
made
some
ineligible
like
around,
like
no
registration
or
something
like
that
because
of
this
kind
of
issue.
Yeah.
A
Okay,
I
think
that's
our
whole
agenda
today,
right.
A
So
I'm
glad
we
were
able
to
help
I
guess
unless
there's
any
other
burning
issues
somebody
wants
to
raise,
although
we
have
a
very
small
attendance
today,
so
I
would
probably
just
defer
them
anyway,
please
let
me
know
going
once
twice
three
times
we're
done.
Okay,
thanks.