►
From YouTube: 20200728 SIG Arch Conformance
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
This
is
the
kubernetes
sleek
architecture,
conformance
sub
project
meeting
for
july,
28
2020,
and
please
follow
our
code
of
conduct
and
be
respectful
and
kind
with
one
another
and
shall
I
turn
it
over
to
you
hippie
to
do
the
rest
of
the
hosting
here.
C
That
sounds
great.
I
will
share
my
screen
here
shortly
share
I
I
will
need
to
have
permission.
Oh
you
said
granting.
I
think
the
beard
looks
really
good.
C
C
I
think
it's
just
the
whole
desktop
pardon
me
for
I've
not
shared
this
way
before
I
missed
my
desktop.
C
C
And
screens
I'm
on
a
laptop
in
the
cafe,
so
I
only
have
the
one
screen.
It's
minus
trivia,
please
sign
in
thanking
ryan
for
taking
notes.
Today
I
see
the
a
lot
of
the
ii
team
has
put
their
notes.
There
put
your
name
and
discussion.
We
try
to
time
box
these.
There
are
the
first
item
on
the
open
discussion
is
a
privilege
based
profiles
which
I'm
excited
to
see
so
on
our
first
link
here.
This
is
a
pr
that
john
has
asked
us
to
take
a
look
at
and.
C
A
Sure
yeah
I
mean
we've
talked
about
profiles
for
a
long
time
and
I
think
the
previous
meetings
we
were
sort
of
in
general
agreement
that
very
least
separating
out
user
level,
things
versus
privileged
things
with
useful
one.
We
have
customers
who
actually
ask
for
that.
We
have
our
people
who
actually
asked
to
be
able
to
run
that
without
getting
privileged
account
and
two
for
vendors
who
may
want
to
have
a
more
highly
managed.
A
I
think
I
think
explains
here
openshift.
I
think
they
have
managed
clusters,
they
don't
necessarily
give
clustering
privileges
and
other
people
do
too
so
try
to
go
through
and
define
the
criteria
for
what
tests
would
be
in
these
two
profiles
of
base
and
privileged.
A
We
have
talked
a
lot
about
profiles,
but
at
least
let's
try
and
in
my
mind,
let's
try
and
do
something,
because
we
just
keep
talking
about
it,
and
this
seemed
to
be
that
what
I
was
hearing
is
the
most
agreeable
to
people
the
eventually
we
may
have
more.
You
know
half
a
dozen.
We
want
to
shoot
for
roughly
profiles
and
so,
but
to
get
it
started.
I've
defined
what
I
see.
So
what?
A
A
Some
of
these
things
require
modifications
to
the
cluster
that
would
be
sort
of
not
in
the
purview
of
an
ordinary
user,
and
so
those
tests
have
to
be
categorized
into
the
privileged
profile
as
well.
So
that's
that's
what
this
does.
A
Then
there's
a
a
subsequent
pr
that
actually
takes
this
and
goes
through
all
the
existing
conformance
tests
and
categorizes
them.
According
to
these
two
things
we
end
up
with,
it
was
62
privileged
tests
and
200
and
some
odd
based
tests.
So
I
do
see.
There's
some
reviews
there
and
yeah
open
the
floor
for
comments
and
abuse.
D
So
thanks
a
lot
john
for
doing
this.
I
think
this.
This
does
make
a
lot
of
sense.
A
couple
of
questions
I
had.
I
did
look
through
the
spreadsheet
you
had
there,
which
categorizes
the
tests
did.
You
have
like
a
a
notion
that
you
followed
with
respect
to
things
like
horsepower
host
network
host
sport.
A
It
was
considered
privileged
to
host
housed
pa
host
port
host
network,
all
of
those
things
and
then
there's
some
like
there's
this
one
empty
dirt,
one,
that's
looking
for
a
race
condition
that
probably
could
probably
be
fixed
to
not
require
questions
like
the
last
one
that,
but
you
know
this
one.
A
First
of
all,
I
don't
know
why,
looking
for
a
race
condition
as
a
conformance
test
anyway,
it's
not
a
functional
test,
so
I
would
either
say
demote
that
out
of
conformance
or
fix
it,
so
it
doesn't
need
cluster
admin,
but
for
the
first
cut
we're
past
code,
freeze,
obviously
for
119,
but
we're
not
past
test
freeze.
So
I
was
kind
of
hoping
that,
if
possible,
we
could,
if
there's
general
agreement
on
this,
we
could
implement
this
in
119
and
make
make
that
happen
now,
rather
than
waiting.
Another
cycle.
D
Cool,
I
guess
the
one
thing
I
noticed
was
that,
with
respect
to
something
like
node
port,
a
couple
of
tests
that
were
using
node
port
as
well,
which
would
require
privileges
on
the
on
the
nodes
in
the
kubernetes
cluster,
those
might
have
not
been
marked
as
privileged.
I
was
wondering,
if
that's
also
something
we
want
to
address
here,.
A
Node
or
possibly
yeah,
which
one
specifically,
we
should
check
check
it
out,
no
port
in
general.
I
wouldn't
expect
to
require
those
privileges.
There
were
things
that
required
host
networking
because
they
were
testing
no
to
pod
networking
right.
D
Right
right,
so
so
I
guess
there
was.
There
was
at
least
a
few
tests
I'll
I'll,
maybe
I'll
comment
offline
on
the
document.
A
D
E
D
Yes,
sorry,
I
guess
I
should
introduce
myself
so
I'm
a
bit
shake.
D
I
work
at
vmware,
mostly
on
project
pacific
and
if
you're
aware
of
our
supervisor
cluster,
you
know
we
run
kubernetes
pods
as
as
vms,
and
so
you
know,
part
of
the
idea
with
running
kubernetes
spots
as
vms
is
also
to
make
sure
that
we
have
this
isolation
between
the
hypervisor
and
the
pods
that
are
running
as
vms
on
those
hypervisors,
and
so
we
kind
of
categorize
all
of
our
tests,
based
on
this
notion
of
hypervisor
isolation
and
some
of
the
tests.
D
So
yeah
just
a
short
background
there.
I
guess
the
the
help
that
we
could
probably
provide
here
is
with
respect
to
all
of
these
tests
and
some
of
the
work
we've
done
around
categorizing
them
into
privileged
and
non-privileged
tests.
A
Yeah
absolutely
please
take
take
a
look
comment
anywhere
on
the
pr,
probably
probably
better,
on
the
pr
than
the
spreadsheet,
but
comment
either
place,
but
we'll
see
it
on
the
pr
more
likely
than
us.
You
know
that
that
sounds
great.
I
I
want
to
make
sure
I
mean
I
think
services
functioning
as
expected
is
pretty
core.
A
We
don't
obviously
load
balancer
services
are
not
part
of
conformance
because
they
never
have
been.
They
don't
make
sense
of
product
but
meaning
external
load
balancers.
But
I'm
not
clear
if
you're
saying
node
port
services.
A
A
D
Right
so
notebook
in
itself,
you
know
essentially
requires
you
to
open
a
port
on,
for
example,
in
our
case,
the
hypervisor
and
access
to
the
hypervisor
is
privileged,
and
so
we
don't
expect
customer
reports
or
workloads
to
actually
be
able
to
open
ports
on
the
hypervisor,
because
that
breaks
the
trust
model
there.
A
I
see
it
yeah
I
mean
personally
I
can.
I
can
understand
that
sort
of
it's
allocating
a
shared
node
resource.
A
A
Any
other
thoughts
there
from
many
of
my
compatriots
on
the
performance.
F
Project
yeah,
I
think
I
talked
to
clayton
a
little
bit
about
this
for
the
call,
and
I
I
know
he
wasn't
able
to
join
so
I'm
trying
to
provide
some
input.
The
I
think,
generally,
the
current
split
base
and
privilege
makes
sense.
The
only
thing
I'm
curious
about
is
when
you
allude
to
other
potential
profile
types,
I'm
wondering
what
the
delineations
are.
F
So
like
username
spaces
is
one
of
the
things
that
a
lot
of
people
are
interested
in,
supporting
where
I
could
see
us
wanting
to
maybe
provide
more
variation
on
privilege
versus
base,
and
so
I
was
just
curious
if
you
had
like
node
privileged
versus
what
my
application
operational
environment
is.
That's
presented
like
the
container
itself
privileges
like
can.
I
appear
to
run
as
rooted
in
my
computer.
A
I
think
those
are
somewhat
open.
I
mean
I,
I
think
that
you're
you're
absolutely
right
that
that's
that
both
things
that
are
sort
of
root,
unknown
and
root
on
cluster
are
combined
here
in
this
privileged
one
and
and
I'd
be
open
to
the
idea
of
separating
those
out
at
some
point.
The
other
way
we
don't
know
yet.
We've
had
a
lot
of
discussions
about
how
we
might
divide
these
things
up
and
there's
this
tension
between
what
whether
we
divide
them
up
based
on
sort
of
functional
things
like
okay.
A
A
But
the
main
concern
with
those
is
that
that
we
would
end
up
with
like
dozens
of
them
and
that
would
be
overwhelming
for
people
as
opposed
to
maybe
grouping
them
into
very
broad
categories
of
like
lowest
commonality
denominator
for
user
functions,
that's
base,
and
then
you
know
this
sort
of
privileged
one
and
the
maybe
or,
alternatively,
it
could
be
by
use
case
it
could
be
like
stateless
stateful,
we
talked
about
having
ones
that
were
like
quote
cloud
provider,
which
basically
was
everything
in
the
kitchen
sink,
and
I
think
those
are
all
up
for
debate
right
now.
A
The
reality
is
the
one
that
we've
gotten
kind
of
feedback
from
the
world
on
is
this
division,
which
could
be
further
refined
the
way
you're
saying
derek,
but
the
reality
is.
I
believe
that
most
most
vendors
are
gonna,
be
like
okay.
As
long
as
I
can
meet
bass,
I'm
happy
because
now
I
can
say
I'm
kubernetes
and
there's
no
stick
or
carrot
particularly
for
meeting
optional
profiles.
A
They
get
a
different
badge.
The
idea
was
they
get
a.
They
get
a
base
badge
and
a
privileged
badge,
and
I
don't
know
network
ingress,
clustering
tag
or
something
like
that
or
badge.
But.
F
F
Just
trying
to
think
through,
if
there's
a
difference,
potentially
on
like
what
the
I
don't
think,
the
right
term,
it's
hard
for
me
to
say,
but
I'm
trying
to
think
about
like
the
application
operational
environment
like
that,
which
you
see
when
you
execute
your
pod
inside
your
pod,
separate
from
the
cluster
environment,
which
is
like
that
which
you
see
that
you
can
taint
a
node
right
that
type
of
thing.
F
And
so
what
I
was
wondering
is
like,
if
you
had,
we
had
thoughts
on
maybe
subdividing
privilege
to
be
privileged
to
the
cluster
versus
privileged.
In
my
application,
operational
environment,
like
I
could
see
a
running
as
root
user
is
different
than
ability
to
taint
a
node
or
something
from
inside
the
container
and.
A
I
hear
you
like,
I
think,
with
the
last
meeting
this
last
meeting,
we
had
the
same
conversation.
I
was
like
there's
these
two
different
there's
like
the
control
plane
in
the
data
plane,
you
might
think
of
it
as
right.
There's
like
as
this
application
operating,
I
can
talk
to
the
download
api
right
as
opposed
to
so
it's
what
the
view
looks
like
from
the
inside
out
or
to
the
workloads
that
are
running
versus
what
the
view
looks
like
to
the
operator,
and
I
think
the
conclusion
there
was.
I
don't
know
if
that
matters
or
not.
F
F
Do
the
only
reason
I
bring
this
up,
and
this
is
my
own
feedback?
It's
not
clayton's!
Necessarily
it's
like
putting
my
vendor
hat
on
right
now
like
when
I
try
to
say
like
what
is
a
compatibility
guide
for
openshift
right,
like
it's
clear,
we'd,
say:
hey
your
conformance,
but
like
that's,
not
really
an
application
compatibility
guide,
so
you're
like
what
is
it
that
I
see
in
my
operational
environment
for
my
app
itself?
It's
like
the
distinction
or
the
language.
F
I
was
writing
with
my
red
hat
hat
on
when
for
materials
we
put
out
as
a
our
vendor
to
say,
like
there's
kind
of
a
difference
between
like,
like
you,
said,
user
planner
data
plan,
that's
another
way
of
phrasing
it,
but
it
was
just
I'm
fine
that
we
can
evolve
it.
I
guess.
A
A
What
to
do
with
it,
yet
I
mean
the
sort
of
in
the
limit
right.
What
what
what
you're
talking
about
becomes
to
me
a
a
set
of
profiles,
you're,
basically
a
crd
for
a
profile
that
a
cluster
can
can
advertise.
I
meet
these
profiles
for
these
operational
type
of
requirements
and
therefore,
and
then,
and
then
a
packaging
product
like
home
or
whatever
can
actually
validate,
look
in
the
package
and
say:
oh,
this
needs
these
packages.
Oh
this
cluster
supports
these
packages.
A
I
can
successfully
deploy
to
this
cluster,
that's
kind
of
in
the
limit
of
to
me,
ideally
what
this
could
evolve
into,
but
there's
no.
As
far
as
I
can
tell
real
demand
for
that
right
now,
the
demand
we're
seeing
is
much
lower,
much
lower
bar
yeah,
okay.
F
Basically,
the
classical
dividing
line
we've
had
on
red
hat
side
has
been
like.
Can
I
run
as
user
zero
in
my
docker
file
or
not,
and
that's
like
the
the
common
application
operating
environment
like
dividing
lines
of
privilege,
which
is
separate
from
me
like
do
I
have
privilege
on
the
cluster
necessarily,
and
mostly,
I
was
just
trying
to
think
about
how
to
map
this
when
we
talk
about
features
that
are
coming
up
in
sig
node
on
username
space
room
mapping,
which
has
been
shockingly
popular
lately.
So,
okay,
that's
that's
all!
A
C
Is
there
any
it's
in
generation
of
these
lists
of
ones
that
are
required
for
privilege?
Is
it
a
static
analysis?
Are
we
analysis?
Did
you
analyze
the
run
time
like
as
an
electric
defense
of
some
sort?
What
was
your
primary
driver
for
separation.
A
I
mean
it's
static
for
the
most
part,
I
I
ran
the
tests
with
some
reduced
privileges
to
see
which
ones
would
fail
and
sussed
out
a
few
that
were
surprising
to
me
with
that,
because
of
like
like
this,
like
the
one
I
said,
the
man
and
the
race
commission
check
like
looking
at
it.
You
wouldn't
think
it
needs
host,
there's
also
a
couple
others
that
need
they
use
host
court
to
force
preemption.
A
I
think
there's
one
that,
like
maybe
there's
another
way
to
do,
that,
a
lot
of
the
the
the
resource,
the
resource
ones,
use
a
fake
resource,
which
I
assume
some
of
these
are
my
assumptions.
I
assume
that
to
create
a
fake
node
resource
that
that's
requires
some
administrative
privileges.
You
can't
just
do
that
as
a
user,
so
yeah
that
that
was
my
methodology.
G
Ben
curry,
here
from
vmware
thinking
about
the
privileged
example,
we're
kind
of
in
this
example
asserting
something
through
the
absence
of
tests
since
to
some
extent
right.
We
we
say:
okay,
well,
this
this
system,
you
know,
is
denies
access
to
certain
privileged
things.
So,
therefore,
we
can
pass
the
base,
but
we
can't
pass
the
privilege.
G
Is
this
model
flexible
enough?
Do
you
think
for
us
to
have
a
mechanism
where
we
can
assert?
You
know
through
the
existence
of
tests,
rather
than
through
the
absence
of
tests
where
we
have
say
a
system
that
denies
access
to
privileged
things,
and
we
have
a
certain
profile
that
says?
Oh,
we
can
assert
that
you
actually
can't
do
all
of
these
things
as
part
of
conformance.
Does
that
make
sense.
A
You're
saying
well
the
way
they
the
profiles
has
been
discussed
in
the
past
in
a
way
that,
even
before
I
got
involved,
this
sort
of
group
decided
on
it,
and
we
cannot
make
this
something
for
discussion.
I
suppose,
but
was
that
their
additive.
So
that
means
that
that
it's
not
an
either
or
power
up
profile.
It's
a
you
can
meet
base
and
then
you
can
meet
this
one.
You
can
meet
this
one.
A
You
can
make
this
one
and
then
they're
not
so
if
you
were
to
do
that,
I
mean
that
would
be
we're
not
trying
to
say
like
like
something
could
meet
base
and
it
could
mean
half
of
the
privileges.
After
the
principle,
it
would
only
be
able
to
advertise
itself
as
supporting
base
as
being
conformed
to
base
yeah
we're
not
trying
to
afford
that
base.
Doesn't
support
privileged.
That
would
be
a
different.
G
Thing
it's
not
really
yeah.
I
mean
that
all
makes
sense
to
me.
I
guess
I
guess
I
was
you
know
thinking
that,
if,
if
you
know
we
wanted
to
add
a
bunch
of
tests,
for
example
that
assert
that
you
can't
do
a
bunch
of
things
in
an
environment
that
that
you
know,
denies
access
to
specific
things
right.
So
we're
saying
that
you
know
this.
G
This
lockdown
system
is
locked
down
in
various
useful
ways
and
we
can
use
conformance
test
to
assert
that
those
things
are
going
down
if
we
were
to
do
that
and
augment
the
tests
in
that
way.
G
A
I'm
saying
maybe
not
in
the
conformance
program
as
it
is
today.
Okay,
you
know
not
to
say
we
can't
can't
move
in
that
direction.
If
people
want
to,
but
that's
not
again,
the
intent
was
to
really
the
only
really
the
only
stick.
The
conformance
program
has
is,
is
trademark,
okay
and-
and
so
you
know,
and
the
carrot
is,
is
fancy
badges
that
you
understand
right.
B
A
It
so
like
this
kind
of
alters
the
bar
for
the
stick,
because
it
allows
okay
to
be
conforming.
Even
though
they're
there
don't
don't
can't
successfully
pass
certain
tests,
but
it
allows
those
who
do
pass
those
tests
to
advertise
a
more
functional
cluster
yeah
right.
So
that's
just
sort
of
as
far
as
it
goes.
I
think
there's
interesting
ideas
around
like
I
don't
even
know.
A
If
I
call
I
don't
if
I
call
it
conformance
but
around
like
you
know,
coming
up
with
standard
use
case
type
clusters
that
can
be
and
validating
it,
but
I
guess
in
scope
is
conformance
not
in
scope.
Is
that
it's
a
quality
cluster
you
could
you
can
garbage,
and
if
it
met
all
these
tests
it
would
be
still
certified.
We
wish
we
had
the
resources
to
do
quality,
but
we
don't
really
that's.
That's.
G
D
A
That's
awesome
and
I
would
say,
if
you're
going
to
have
and
if
people,
if
there's
no
objections
here,
given
that
this
is
just
changes
to
test
code,
basically
labeling
a
bunch
of
tests,
I
think
it's
it
can
go
in
before
test
freeze
which
is
in
a
week
or
less,
and
so,
if
you're
going
to
help
today
or
tomorrow,
would
be
really
a
good
time
to
do.
It.
G
Yeah,
what
is?
Is
there
a
good
slack
channel
to
collaborate
on
the
cage
conformance?
One
is
fine
yeah
case
conformance,
okay,
yeah.
No,
this
would
be
great.
It's
something.
We've
been
looking
at
for
quite
a
while,
and
it's
it's
actually
surprising
how
few
tests
are
an
issue
for
us,
given
the
the
the
degree
to
which
our
setup
is
is,
is
you
know,
distinct
from
from
a
lot
a
lot
of
setups,
but
right
yeah?
Absolutely
we
can
help
okay,
excellent.
C
Thanks,
john
and
and
the
vmware
folks,
from
stepping
up
help
that
aaron,
I
saw
you're
trying
to
speak
so.
B
Yeah,
I
don't
know
I
I
I
do
want
to
help
see
this
move
forward,
and
I
and
I
get
all
that
there's
a
part
of
me
that
wonders
what
the
checker
balance
is
to
make
sure
that
base
is
still
a
sufficiently
useful
profile
advocating
on
behalf
of
the
user.
Like
is
base
going
to
provide
me
meaningful
functionality,
or
is
it
something
that
vendors
with
a
really
strong
bent
on
security,
end
up
cherry
picking
things
out
of
to
the
point
that
it
is
effectively
not
really
that
useful?
B
I
don't
know
where
the
dividing
line
is
there,
but
I
just
want
to
make
sure
we
we
keep
the
user
in
mind
because
ultimate
back
to
the
point
of
like,
should
we
do
negative
testing.
I
don't
think
I
don't
know
that
as
a
user,
I
would
be
really
interested
in
looking
for
which
clusters
don't
support
a
piece
of
given
functionality,
because
I
care
about
workload
portability
first
and
foremost,
so
I
want
to
make
sure
like
which
met,
which
kubernetes
offerings
are
going
to
be
able
to
run
my
workload.
B
Given
what
you
know
features
I
use
what
behaviors
I
expect.
A
Yeah,
I
think
that
the
I
mean
the
only
check
and
balances
those
of
us
who
approve
these
right
and
the
criteria
that
we
define
for
each
profile.
So
if
you
look
at
the
the
pr
and
the
community
repo,
what
I
do
is
I
explicitly
define
that
the
only
basically
difference
between
privileged
and
base
is
that
the
privileged
ones
require
require
some
special
above
and
be
above
and
beyond.
A
User
level
of
privilege,
either
from
a
workload
perspective
or
from
a
cluster
administrator
perspective.
So
I
mean
I
think
that
there
should
be
sufficient
functionality
in
kubernetes
to
support
a
lot
of
workloads
without
those
privileges.
E
Yeah,
the
other
way
to
also
look
at
this
is
like
we
have
to
augment
the
base
to
get
to
the
point
where
we
are
testing
all
the
things
that
we
need
to
test.
Yes,
that's
been
a
constant
challenge
and
you
know
I
haven't
been
able
to
spend
time
on
it
either.
So
that's
definitely
something
we
need
to
figure
out
how
to
okay,
making
great
progress.
A
In
that
regard
and
they've
added
a
lot
of
work,
a
lot
of
stuff
on
there
so
and
they're
continuing
to
work
at
it.
So
I
think
we're
making
progress.
B
Yeah,
no
okay,
thank
you
for
like
driving
this
and
look
forward
to
help
from
vmware.
That's
really
appreciate
it.
C
Phil,
thank
you
to
the
point
that
john
was
making
earlier.
Our
next
topic
of
conversation
is
some
updates
to
the
aps:
new
conformance
progress
page.
This
is
not
merged
yet,
but
it
will
be.
There
was
some
feedback,
don't
make
it
so
red,
it
looks
like
dead,
but
the
orange
is
where
our
growth
is
accumulating
for
a
backlog
of
debt.
C
The
solid
green
is
how
many
endpoints
we
have
conformance
over
the
the
more
dark
green
this
one
or
the
lighter
green
is
endpoints
that
we
have
old
debt.
That
is
now
conformance
coverage,
and
then
this
new
one
here
in
this
release,
119
we
currently
had
41
new
endpoints
and
we
were
able
to
ensure
that
all
of
them
came
with
conformance
test.
C
This
is
a
little
bit
of
a
new
update
from
what
we
had
before,
so
that
you
can
see
really
clearly
every
release
we've
had,
let's
see,
seven
a
couple
of
slow
quarters
or
slow
releases.
We
had
ten.
I
think
we're
going
to
get
around
35.
As
my
hope,
and
with
your
help
and
looking
at
some
of
the
promotions
that
we
have
and
ensuring
they
are
super
solid.
We
we
can
have
more
endpoints
added
old
endpoint
that
removed
this
release
than
we
had
for
the
prior
four
releases,
and
that's
that's
good.
C
There
was
also
an
ask
to
show
it
over
a
percentage
release
rather
than
numbers,
so
we
can
see
we
are
going
to
be
over
halfway
there.
This
this
release
very
good.
If
you
want
to
see
where
the
debt
sits
notice,
we
didn't
add
any
new
debt
in
this
release.
Just
those
41
viewing
points.
C
One
thing
nice
now
is
if
you're
on
here.
This
is
not
on
the
front
page.
Yet
when
you
click
on
it'll,
take
you
to
the
release
and
on
that
release
it
will
list
the
new
endpoints
that
came
in
for
this
particular
release
and
at
the
you'll
notice,
all
of
the
stable
ones
have
tests.
We've
got
a
few
beta
ones
that
don't
have
any
tests
yet,
so
we
maybe
would,
as
we
progress
to
having
a
bot
when
things
get
promoted
from
alpha
to
beta
or
even
into
beta,
saying,
hey.
C
C
So,
while
we
are
looking
at
currently
the
easy
to
delineate
somewhat
easy
to
delineate
privilege
area,
I
do
feel
that
most
of
our
end
users
would
love
to
be
able
to
store
data
in
their
cloud.
Not
just
have
workload.
A
Well,
I
think
those
are
those
specifically,
I
don't
know
the
details
of
how
csi
works,
but
those
look
like
they're
they're
changes
to
drivers
configuration
would
think
that
we
could
use
a
dummy
driver
to
to
conformance
test
those
endpoints.
Not
that's
not
necessarily
just
actual
storage
yeah.
I
don't
know
if
that's
true
or
not,
but.
C
We
have
a
filter
in
place
and
we
should
probably
have
somewhere
to
note.
These
are
the
endpoints
that
we
have
decided
are
optional,
because
we're
talking
mainly
around
profiles,
but
we
don't
necessarily
have
a
definition
of
what
those
profiles
are
other
than
we're
talking
about
flagging
some
tests
as
to
why.
A
C
We
have
some
crd
deletes
that
are
still
missing,
and
these
are
probably
valid
things
that
do
need
to
get
tested
that
kind
of
snuck
in.
If
you
look
back
on
the
conformance
page,
so
that's
the
updates
for
that
page.
C
That's
it
for
the
api's
new
site
itself.
Zach
is
on
the
call
he
has
done
all
of
the
hard
labor
on
this,
the
site
itself
and
including
the
database
which
we
use
underneath
the
query,
probably
anybody's
interested
we
can
get
together
and
show
you
how
we
use
the
ede
runs
and
the
data
from
those
to
decide
what
tests
have
been
hitting
what
endpoints.
C
C
As
far
as
our
quick
reviews,
I
think
we
have
a
few
moments
left
on
our
lg
teams
for
these
two
to
decide
whether
we
want
to
try
to
get
them
into
this
milestone.
C
C
When
it's
our
test,
that's
failing
versus
when
the
infra
is
failing.
So
when
we
have
a
few,
you
know,
let's
say
four
to
ten
times
where
it's
the
infra
failing.
We
start
to
not
look
so
closely
and
we
let
a
few
things
grow.
We
actually
merged
some
e
to
e
tests
that
were
flaking
and
that
was
left
to
life.
A
Yeah
I
mean
generally
we're
having
a
big
problem
with
with
this
right
now,
so
aaron
can
probably
talk
for
the
next
two
hours
about.
C
That
I
I'll
try
not
to
take
up
this
meeting
with
all
of
that
time,
but
this
one
pr,
I
think,
looks
pretty
good
and
I
think
rihanna,
if
you
want
to
confirm
if
this
promotion
merges
with
the
rest
of
our
existing
stuff,
that's
soaking
within
the
two-week
window
for
test.
Merge
we'll
get
35
points
this
for
this
release.
Our
target
was
40,
but
that
will
definitely
get
us
up
above
50.
H
Coverage
there
is,
if
I
can
speak
quickly
to
what
we
have
on
the
list,
with
basically
three
items
that
we
would
really
like
to
get
in
in
this
cycle.
This
one
here,
the
first
one
90939
it's
ready
to
go.
I
think
it's
ran
long
enough.
It's
got
two
weeks
all
green.
I
will
appreciate
some
badges
on
that
to
get
it
through.
A
H
H
If
I
explain
it
correctly,
they
would
correct
me
if
I'm
wrong,
but
I
fixed
them,
but
to
not
let
our
test
pick
them
up,
and
since
then
it's
been
running
sweetly
green
for
a
long
time.
So,
if
we
can
get
all
three
of
these
in,
it
would
be
a
beautiful
day
that
is
on
those
three
and
then,
if
you
remember
a
few
meetings
ago,
we
discussed
the
metadata.
That's
missing
from
some
of
the
e2e
tests.
H
A
B
These
these
are
in
my
queue.
I
promise
I've
just
had
other
fires
crop
up,
but
I
will
take
a
look
at
these
very
soon.
C
Excellent.
Thank
you.
Everybody
for
your
time,
I'm
tempted
to
give
you
back
around
15
to
18
minutes
of
your
time,
but
one
quick
shout
out
to
nikita
and
some
of
the
vmware
folks
that
I'm
hoping
to
meet
with
next
week,
they're
interested
in
looking
at
some
of
the
test.
Writing
and
I'm
super
excited
to
spend
some
time
with
them.
C
C
Them
otherwise
enjoy
your
day
and
stop
by
kate's,
conformance
channel
and
looking
forward
to
sharing
within
this
release,
a
large
increase
in
coverage.