►
From YouTube: 20190336 sig arch conformance subproject
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,
this
is
Tuesday
March,
26
2019.
This
is
the
architecture
performance
of
project.
We
have
a
brand
new
agenda.
This
is
the
first
inaugural
meeting
for
lack
of
a
better
term.
I
will
put
the
meeting
units
in
check.
If
you
could
add
your
details,
there
that'll
be
highly
beneficial
just
as
a
reminder.
We
have
a
code
of
conduct,
so
please
adhere
to
the
code
of
conduct,
because
this
video
will
be
recording
posterity
and
sharing
YouTube
forever
and
ever
for
which
is
your
grandmother.
A
Ok,
so,
first
up
administrivia,
we
currently
have
a
project
board,
but
we
have
stuff
that's
kind
of
flying
all
over
the
place
and
I.
Don't
know
whether
or
not
the
project
board
resembles
reality.
What
I
found
in
a
multiple
project
course
and
I
don't
know
if
this
is
true
for
everything
so
I'd
like
people
to
give
me
their
tactics,
is
that
you
have
to
manage
the
project
boards.
So
if
stuff
flows
in
randomly
from
the
wild
and
gets
the
labels
appropriately
set
this
stuff
from
inbound
to
the
project
board.
B
Speak
to
my
experience,
having
interacted
with
it
I'm
very
curious
who
else
uses
it,
I
will
periodically
go
to
the
add
card
section
and
use
a
query
that
ensures
label
area
conformance
is
one
of
the
criteria.
I
mean
literally
everything
over
to
the
to
triage
column.
I
then
walk
the
board
backwards
if
I'm
doing
a
review
scrum.
So
I
start
with
anything
in
the
in
Review
column
to
see
if
any
of
my
feedback
has
been
acted
upon
or
if
there's
anything
that
could
use
feedback
funding.
B
B
I
then,
generally
look
at
any
other
PRS
in
any
of
the
previous
columns
and
I
try
to
review
it
and
if
I
review
it
and
move
it
to
be
in
review
column,
I,
don't
really
touch
any
of
the
issues
or
any
of
the
other
columns
and
I
have
set
the
board
up
so
that,
if
an
issue
where
a
P
o
PR
is
closed,
it
automatically
moves
to
be
done.
Column.
C
Use
it
I
typically
look
at
the
needs,
approval,
column
and
especially
if
I
get
folks
and
I'll
do
it
urgently.
If
I
don't
get
booked
by
some
means
other
than
Gabe
notifications,
then
I
try
to
pull
it
once
a
week
or
so,
although
in
the
past
couple
of
weeks,
I
haven't
done
that
because
I
wanted
to
let
some
things
pile
up,
so
we
could
be
group
go
over
them
as
a
group
go
through
the
process.
First.
C
Yes
or
if
things
gets
in
my
way,
I
will
try
to
make
sure
that
they're
in
the
board
I
think
some
other
people
helped
to
populate
the
to
triage
board
as
well
with
a
bunch
of
issues
that
previously
were
in
spreadsheets
and
whatnot,
so
having
it
kind
of
all
in
one
place,
I
find
helpful.
I
agree
that
without
automation,
it's
it's
kind
of
tedious.
Most.
C
D
C
Anybody's
died,
please
mute
most.
The
issues
are
in
communities
kubernetes
if
there
are
tests
to
get
promoted
to
conformance
or
if
there
are
changes
to
existing
conformance
tests,
but
occasionally
there
are
some
things
in
other
repos,
especial
the
community
repo
and
it's
documentation
related
to
conformance
testing
as
an
example.
A
B
A
B
So,
hang
on
because
literally
everything
else
is
manually
applied.
So
if
it's,
if
it's
about
updating
Docs
related
to
conformance,
we
expect
people
to
add
the
area
conformance
to
their
PR.
If
it's
umbrella
or
tracking
issues,
we
expect
people
to
apply
the
label
there.
The
really
tricky
one
is
if
somebody
is
modifying
an
existing
test
case,
which
is
already
in
the
list
of
conformance
you
have
to
whoever
is
reviewing
that
has
to
notice
that
conformance
is
involved.
You
know
it's
a
conformance.
B
One
of
like
the
tricky
thing
there
is
our
our
ownership
and
labeling
mechanisms
are
very
hierarchical.
So
if
we
want
to
say
that,
like
sig
apps
is
in
charge
of
all
of
the
workload
related
tests
because
two
gaps-
arms
all
the
workload
controller
is
great,
but
then
there's
a
subset
of
those
workload
tests
that
are
also
conformance
tests.
Do
we
expect
that
to
be
split
out
into
a
brand-new
directory
called
conformance
or
you
know?
How
do
we
organize
things
to
enable
that
to
happen?
Well,.
A
A
Anyways,
because
there's
a
lot
of
changes
you
want
to
make
and
the
problem
is-
is
prioritizing
what
changes
come
first
and
what
are
the
most
important
things
to
try
and
get
a
dag
flow
that
that
ripples
down
properly,
because
some
of
the
changes
that
people
want
to
make
is
that
the
intent
test
early
if
they
don't
come
as
specific
or
you're
gonna
you're
gonna
end
up
doing
a
ton
more
work
because
of
the
problems
of
dependencies,
so
I
will
happily
go
through
it
on
it.
What's
currently
there
and
probably
have
that
by
next
time.
A
C
I
did
want
to
at
least
say
some
things
about.
I
can
actually
only
stay
for
another
ten
minutes
or
so
I.
Don't
know
if
I
have
time
to
do
the
walkthrough
this
week,
okay,
yeah
I
did
want
to
say
that
when
we're
looking
at
tests
could
probably
bring
one
up
like
the
one
that
Aaron
mentioned,
you
know
it
looks
like
there
are
a
couple
ones
related
to
that
either
in
the
in
review
or
needs
approval.
C
I
can
just
pick
one
any
test,
that's
promoting
something
to
conformance
yeah.
So
if
you
look
at
the
changes
there,
it's
some
of
these
are
quite
subtle
and
some
of
them
are
not,
but
my
general
advice
would
be
if
you're,
not
a
domain
expert.
Please
call
in
domain
experts
from
the
relevant
sig.
You
can
ask
the
sig
chair,
sig
TL,
to
find
someone
mm-hmm
to
actually
review,
because
sometimes
seemingly
innocuous
changes
can
totally
change
the
meaning
of
the
test.
C
So
please
do
you
call
in
a
domain
expert
and
it's
not
just
sufficient
to
determine
necessarily
what
is
being
covered
by
the
test
and
whether
that
should
be
covered
by
conformance,
like
you
would
say.
Oh
yeah,
definitely
whether
this
pod
is
working
correctly,
should
be
covered
by
conformance,
but
you
actually
also
have
to
look
at.
How
is
it
testing
that,
because
there
are
certain
behaviors
that
are
not
guaranteed
to
be
portable
or
stable,
so
you
actually
need
to
look
at
what
mechanism
is
easy
to
to
evaluate
that
behavior?
C
C
Sometimes
that
can
surface
differences
between
different
providers
as
well.
If
they
run
for
a
while
or
different
providers,
but
at
least
it
will
determine
whether
they're,
stable
or
flaky,
but
we've
had
some
cases
where
some
tests
had
tests
have
looked
at,
say,
event,
contents
or
other
other
aspects
that
are
not
guaranteed
to
be
portable
or
have
used
specific
container
images
that
are
not
not
really
portable,
so
so
that
stuff
can
get
pretty
hairy.
C
The
general
advice
there
would
be,
if
you're
not
sure-
and
it's
not
really
covered
by
the
officially
documented
conformance
to
criteria
in
the
community
repo
just
raise
the
issue
and
ask
for
guidance
for
other
people.
And
if
we
come
up
with
some
new
general
principle
to
apply
to
the
conformance
test,
we
should
add
it
to
the
documentation
just
to
make
it
obvious
to
other
people
who
are
reviewing
the
test
in
the
future.
A
So
we
have
a
very
small
number
of
reviewers
in
this
list
and
and
I
haven't
even
seen
Annabelle
in
the
community
for
a
very
long
time
should
we
and
he's
enlisted
in
his
approver.
Should
we
revise
this
list
currently
with
people
who
are
on
the
call
who
are
well
established.
People
who
want
to
you
know
be
part
of
this
process.
We.
C
Should
yeah
there
were
some
people
involved
in
early
on
as
we
were,
implementing
the
or
updating
how
the
conformance
suite
automation
worked.
So
that's
why
I
FML
was
in
that
list.
You're
right
as
far
as
I
know
he's
not
working
on
this.
Currently
we
did
have
a
number
of
volunteers
to
help
with
the
review
process.
I,
don't
I
can't
see
everybody
who's
on
the
call,
but
there's.
A
A
lot
of
people
on
the
panel
they
should
have
access
to
the
document,
so
I
think
the
approvers
has
to
earn
their
stripes,
but
I.
Think
reviewers
I
think
it's
good
to
have
multiple
reviewers
on
this.
Yes,
so
if
folks
want
to
volunteer
for
reviewers,
please
add
the
details.
The
docs
and
I
would
happily
do
a
PR.
After
this,
we
need
to
just
update
that
list
of
reviewers
and
modify
it
you're
not
even
listed
here
as
approvers
Brian
yeah.
B
Listed
listed
as
an
approver
just
for
the
conformance
test,
theta
right
like
but
listing
yourself
as
a
reviewer
and
a
directory
called
conformance,
doesn't
necessarily
mean
you're
going
to
be
automatically
applied
as
a
reviewer
on
any
PR.
That
has
the
area
of
conformance
labels,
because
again,
that
label
can't
really
be
automatically
applied
to
any
PR,
except
those
that
touch
that
conformance
test
list,
which
that
is
where
Brian
is
an
approver
yeah.
C
And
we
could
add
some
special
automation
for
that
like
for
a
while,
we
had
automation
about.
If
you
type
start
go
files,
then
it
gets
labeled
as
an
API
change.
If
a
file
that
contains
some
conformance
at
code
were
touched
by
someone,
we
could
automatically
label
that.
But
then
we
don't
have
a
way
to
connect
labels
to
automatic
review
lists.
Reviewer
assignment
right
now
far
as
I
know,
correct.
A
B
Okay,
I
see
what
you're
talking
about.
No,
you
don't
have
support
for
that.
So
I
think
like
people
who
are
interested
in
updating
our
tooling
to
support
that
kind
of
functionality.
Those
are
the
people
I
that
should
be
in
the
owners
file
in
the
conformance
directory.
There's
like
some
bit
of
code
there
that
walks
the
goat
all
of
the
files
and
looks
at
what
they're
doing
syntactically
and
I
think
we've
talked
about.
B
Could
we
maybe
update
that
to
link
some
of
our
conformance
tests
to
make
sure
that
they
don't
do
some
of
the
things
that
we
know
they?
We
don't
want
them
to
do
to
ease
some
reviewer
burden,
so
I
think
there's
lots
of
stuff.
We
could
do
to
improve
our
tooling
and
that
that's
a
great
place
for
that.
But
I
did
feel
like
review
bandwidth
for
all
of
the
conformance
tests
separately,
like
putting
those
names
in
that
orders.
File
will
help
with
that
yeah.
A
A
F
Yeah
I'm
just
saying
this
particular
meeting
I
wanna.
We
should
probably
try
and
keep
focused
on
I'm
I'm,
making
sure
educating
people,
including
me
on
how
to
go
throughout
and
do
this
go
about
and
do
this
as
opposed
to
how
do
we
update
the
tooling
to
sign
things
automatically
change
the
way
we
track
things
so.
A
A
If
we
want
to
increase
coverage
and
get
better
coverage
for
existing
behavior,
that
we
currently
have
there's
probably
a
test
for
most
of
the
core
things
that
matter
but,
like
Brian,
said
understanding
the
implications
of
what
it
needs
and
making
sure
that
you're
applying
all
the
semantics
so
before
I
display
they
promoted
this
individual
test.
I
asked
a
question
right,
so
this
is
a
common
case
of
most
of
the
peers
that
we're
seeing
most
of
the
peers
that
we're
seeing
inbound
are
basically
we're
going
to
take
this
test.
A
What
we've
had
for
a
long
time,
which
basically
tests
that
you
know
a
shared
volume
between
containers
and
we're
going
to
promote
it
to
be
a
full
conformance
test,
and
the
question
I
had
was
that
this
leverages
empty
Durer
and
the
question
I
had
was
what
does
53
mean
of
Windows
and
I?
Didn't
have
more
questions
of
how
to
share
volumes
between
containers,
work
on
Windows
because
I
don't
actually
know
the
history
of
those
questions
without
me,
actually
going
through
the
details
of
the
dock.
So
I
just
wanted
to
ask
some
questions
here.
A
First,
and
there
was
some
response
back
and
then
I
said:
if
it's
ignored,
why
is
it
not
Linux
Emily,
so
I
think
the
one
thing
that
Brian
tried
to
stress
earlier
on
I,
don't
know
if
he's
still
in
the
Co
is
that
yeah
make
sure
that
we
go
through
the
due
diligence
of
promotion
and
ask
all
the
questions
up
front.
Yeah.
C
It
could
be
like
in
this
case.
You
know
you
got
a
response
from
someone
from
sick
windows,
which
is
great,
there's
also
a
cap
that
goes
into
gory
detail
on
what
Windows
does
and
doesn't
support,
and
why
and
that
can
be
used.
You
know
if
you're
interested
in
doing
more
reviews
of
Windows
related
things
in
the
future.
It
would
be
good
some
more
domain
experts
read
up
on
that
and
build
up
their
knowledge.
C
Yeah
I,
just
posted
a
link
to
the
conformance
tests,
documentation,
which
includes
some
of
the
criteria
and
whatnot
you
know
empty
jurors
is
a
good
example
where
that's
a
feature
that
should
be
portable,
or
you
know,
at
least
in
all
cases
was
obviously
portable
until
until
Windows
came
along
except
there
are
certain
properties
that
might
not
be
like.
You
can
request
a
memory,
medium
integer,
so
effectively
Ram
a
ram
disk,
and
you
know
that's
an
example
of
one
thing:
that's
not
supported
on
Windows,
so
through.
F
All
the
tests
Brian
if
I
yep,
sorry
in
this
particular
case
so
was
it's
that
I
think
that
the
comment
from
the
windows
think
was
it
an
empty.
Tourner
works
fine,
but
this
meat
medium
memory
doesn't
do.
We
is
this
test
setting
medium
memory
and
if
it
is,
would
we
split
that
then
into
two
two
as
one
that
just
the
basic
emptied
or
the
basic
shared
volume
functionality?
And
then,
if
we
want
to
test
that
testimony,
but
was
just
almost
more
of
a
unit
test
for
the
yeah.
C
So
that's
a
good
question.
We
already
went
through
a
ton
of
these
and
for
every
case
where
it
exercised
some
functionality
that
wasn't
available
in
Windows
like
medium
equals
memory.
We
documented
the
reason
why
the
test
didn't
work
on
Windows
and
we
tagged
it
Linux
only.
We
did
not
do
the
next
step
of
well.
Maybe
we
should
have
a
Windows
test
and
part
of
the
reason
for
that
was
Windows
is
not
covered
by
conformance,
but
we
would
like
Windows
to
be
tested
right.
So
many
of
these
tests
don't
how
to
quit.
C
We
describe
what
they're
trying
to
test,
so
some
reverse
engineering
needs
to
happen,
or
we
need
to
go
back
to
the
author
of
the
test
to
figure
out
what
we're
really
trying
to
test.
Some
of
them
are
somewhat
non-obvious.
Some
of
them
also
reuse
code,
to
make
it
easier
to
stand
up
various
test,
fixtures
and
they're
using
exercising
functionality
that
they
don't
actually
test
so
that
some
of
those
things
we
may
need
to
clean
up.
We
haven't
really
done
that
level
of
detail
just
because
we
have
so
much
uncovered
stuff.
C
A
Question
I
had
with
regards
to
like
the
windows
work,
because
this
overlaps
with
the
conversation
is
a
lot
of
the
existing
tests,
use
Linux,
specific
containers
and
some
of
the
windows
tests
use
Windows,
Server,
2000,
19
ish
based
style
containers.
So
the
question
I
have
is,
if
we're
going
to
promote
things
that
would
apply
to
this
military
OS
deployment,
doesn't
matter
for
to
make
sure
that
you're
running
Windows
Server
containers
or
is
there
a
generic
base
that
we
should
be
basing
the
images
on.
E
Yeah,
so
I
think
that
the
right
solution
is
that
for
ones
that
are
for
containers
are
used
in
conformance
tests.
We
should
make
sure
that
we
get
a
multi
arc
image
published
for
them,
and
so
we've
we've
done
that
for
many
of
them,
busybox
is
one
example
where
you
know:
we've
got
a
Windows
equivalent
for
that
that
runs
busybox
and
we're
using
I
can't
remember
the
layer.
E
E
C
So,
even
before
we
get
the
profiles
officially
defined
and
have
windows
has
officially
part
of
conformance.
We
do
want
as
many
of
the
tests
as
possible
to
be
able
to
run
on
Windows,
and
we
want
to
prevent
accidental
accumulation
of
more
tests
that
can't
run
on
Windows.
So
I
view
that
as
sort
of
a
specific
sub
effort
that
will
need
to
be
undertaken
by
a
set
of
people
to
try
to
figure
out
mechanically.
What
do
we
need
to
do
and
have
in
place
to
ensure
that
we
don't
get
part
even
farther
behind
there?
C
C
Os
is
this
really
well
speaking
of
speaking
of
multi
arch
in
theory,
I
think
most
of
the
tests
can
run
on
multiple
architectures,
but
in
practice
I.
Don't
I'm
not
aware
of
any
providers
that
have
certified
that
are
not
x86,
although
I
could
easily
be
wrong,
I,
don't
know
of
k3s
certified,
x86
or
arm
I.
A
C
To
run
but
I
think
something
to
be
great
to
do
is
to
try
to
figure
out
the
most
effective
way
of
onboarding.
The
new
reviewers
may
be
picking
some
less
tricky
test
and
and
going
over
that
since
Aaron
looked
at
a
bunch
of
the
ones
in
the
in
review
column,
maybe
pick
one
of
those
and
walk
through
how
you
did
the
review
and
what
questions
got
asked
and
see
if
it
can
be
moved
to
the
knees,
approval,
column,
okay,.
A
C
A
C
I,
don't
monitor
slack
unless
people
ping
me
explicitly,
but
sometimes
it's
good
they're,
gonna
you're
gonna
be
asking
a
domain
expert.
So
that's
gonna
depend
on
the
area
of
the
test
and
sometimes
you're
gonna
be
asking
about
general
conformance
criteria,
and
that
would
be
cigar.
Ch
mailing
list
would
be
my
preference,
but.
A
C
C
A
So
we
have
a
bunch
of
action
items
I
think
for
folks
were
on
the
calls
for
being
new
to
this
effort.
Taking
a
look
at
the
requirements
we'll
try
to
update
the
owners
file
with
people
who
are
experienced
and
updating
and
we'll
walk
through
an
example.
Does
that
seem
like
a
reasonable
next
steps
right
now,.
B
That
sounds
good
to
me.
One
other
thing
I
just
wanted
to
point
out
for
who
do
we
contact
win
or
whatever
when
it
comes
time
for
a
PR
to
you?
It
like
looks
good
to
me.
I
want
somebody
like
Brian
or
Clayton.
The
only
two
people
right
now.
If
you
can
approve
the
promotion
of
a
test
to
conformance
I,
try
to
use
the
conformance
mailing
list
and
I
point
them
at
the
needs.
B
E
A
Adding
it
there
makes
a
ton
of
sense,
because
if
we
just
have
one
stop
shopping
for
the
details
of
what
the
conformance
tests
are
and
just
modify
that
doc,
that
would
probably
be
helpful.
We
could
just
use
that
at
the
top.
We
need
even
have
it
to
the
top
of
this
document
as
like
here,
the
main
documents
that
we
you
use
as
part
of
our
evaluation.
E
And
so
who's
gonna
take
the
first
action
on
adding
that
section,
because
I
want
to
make
sure
I
get
the
you
know
list
of
you
know,
here's
things
that
you
should
watch
out
for.
If
you
have
questions
around
whether
or
not
it's
going
to
work
on
Windows,
here's
the
right
way
to
contact
somebody
to
get
a
second
review
on
it
to
see.
A
E
A
Think
the
the
checklist
for
Windows
is
the
one
that
I
asked
the
question.
I
think
there
there
is
enough
details
from
the
last
time.
I
looked
at
this
that
outline
what
were
the
criteria
so
and
we
did
a
review
last
time
and
we
wanted
to
refine
it
even
more
okay,
so
I
think
I
think
we
can
audit
that
and
update
as
necessary
and
talk
about
it
in
this
conversation.
Okay,
as
an
actual
I
know
for
this
group,
I
highly
recommend
reading
this
list.
A
B
E
H
E
Are
here,
okay,
did
you
want
to
start
the
first
PR
on
that
or
should
I
take
a
first
first
try
it.
Oh.
H
E
A
A
B
So
and
the
reason
I
say
not
that
one
just
to
try
and
maybe
skip
this
meeting
I
feel
like
this
meeting
isn't
the
place.
We
should
be
trying
to
scope
out
what
validation
Suites
are
and
how
they
differ
from
overall
conformance
tests
and
whatnot
I
feel
like
this
is
about
how
can
we
most
effectively
Shepherd
forward
reviews
of
conformance
tests
and
promoting
conformance
tests?
I
think
there's
other
discussions
best
take
place
either
at
stake
architecture
or
the
conformance
working
group.
The
scenes
you
have
to
conform.
It's
working
good
meeting
tomorrow.
A
All
right
so
promote
continued
to
run
time,
each
be
verifying
termination
message,
I.
Think
I
look!
So,
let's
take
a
look
at
it
here.
Real
quick
first
part
is
the
that
test
conformance
data,
which
is
I'm,
hoping
it's
all
automated
for
what
it
rips
through
the
conformance
it.
That
was
the
changes
that
MML
made
a
long
time
ago.
So
it
takes
a
look
at
what's
been
changed
and
conformance
it,
and
then
it
basically
plops
that
data
into
this
so
it'll
raise
the
bar
signal.
Every
time
that
we
see
an
update
here.
A
A
A
A
I,
don't
know
if
anyone
else
noticed
that
I
saw
that
everyone
trying
to
push
things
in
to
114
at
the
end,
I
think
pushing
performance
should
happen
at
the
beginning
of
the
cycle
where
possible,
and
during
that
last
2
weeks
it
should
be
totally
Crescent
like
we
should
not
try
to
promote
anything
like
even
2
weeks,
or
maybe
it
up
to
come
freeze
if
it's
really
important,
but
it's
better
to
err
on
the
side
of
being
conservative
and
promotion.
In
my
opinion,
Tim.
G
I
Group
timber:
this
is
before
code
freeze
that
I,
don't
think
the
last
two
weeks
of
a
release
or
the
time
to
change
the
conformance
definition
of
a
release.
It
creates
a
bunch
of
other
challenges
for
people
qualify
and
it's
it's
the
worst
possible
time
to
change
the
definition
of
what
kubernetes
is
and
I
also
like
features
going
into
a
115
release.
I
G
F
B
As
a
former
release,
lead,
I
don't
want
to
see
the
addition
of
a
bunch
of
new
tests
right
at
the
end,
I've
gotten
used
to
watching
all
of
the
CI
signal
generated
by
the
existing
tests.
So
if
you
want
to
place
an
additional,
if
you
want
to
like
slightly
change
the
name
of
a
test
or
add
a
tag
that
says
conformance
I
think
I'm
fine
with
that.
But
if
you
start
writing
a
whole
bunch
of
brand
new
tests
like
that,
absolutely
doesn't
fly
during
code
freeze
in
the
run
up
to
code.
B
I
That's
like,
what's
the
rush
to
promote
something
to
conform,
it's
like
I
think
that
I
would
agree
with
the
general
goals
we
want
tests
for
the
conformance
we
want
stable
tests
for
the
conformance
I
I
feel
like
we
should
not
create
a
sense
of
urgency
about
promoting
conformance
tests,
because
it's
part
of
the
release
process
I
feel
like
it's.
The
good
tests
are
part
of
the
release
process
and
having
good
signal
from
them
and
conformance
is
an
additional
step
where
we
are
growing
the
definition
of
kubernetes.
B
A
I
I
think
that's
a
great
idea.
I
mean
like
the
other
reason
that
I
was
so
worried,
but
this
is
like
there's
just
a
lot
of
stuff
going
on
at
the
end
of
the
release,
that's
more
important
is
getting
out.
A
good
kubernetes
release
is
the
most
important
thing.
I
want
to
not
add
pressure
to
the
review
and
end
of
release
process
to
change
the
definition
of
kubernetes.
It
feels
like
we
can
do
that
as
a
two-phase
commit,
but
I
would
totally
agree
with
you.
Tim.
I
B
D
B
A
B
I
D
there
are
two
groups
of
people
were
talking
to
here.
There
are
a
group
of
people
here
who
are
reviewing
conformance
tests
and
approves
and
stuff
and
so
like.
If
you
don't
review
things,
then
they're
not
going
to
land
in
the
release
and
that's
great,
but
from
the
perspective
of
where
do
you,
people
invest
their
time
in
energy?
We
don't
want
people
investing
their
time
and
energy
in
these
tests.
So
how
do
you
communicate
that
prior
to
countries
well.
A
G
A
A
Good,
so
let's
take
a
look
at
this
test
and
exactly
what
it
does
so
create
a
pod
with
LA
with
ink
with
n
container
containers
output
is
recorded
in
the
log
and
container
exits
with
an
error
when
the
container
is
terminated,
termini
termination
message
must
match
the
expected
output
recorded
from
the
containers
log.
The
the
grammar
of
the
sentence
drives
me
a
little
crazy,
but
that's
ok,
the
does
anyone
think
from
reading
that
description.
A
G
A
G
H
I
Simple
test,
so
real
quick
do
we
have
the
right-
and
this
is
more
a
meta
question.
I
just
want
to
be
100
percent
certain.
We
have
the
right
process
for
ensuring
that
things
that
are
marked
Linux
only
in
conformance
have
a
have
a
prioritization
and
an
urgency
to
make
sure
that
they're
on
Windows,
like
what's
the
what's
the
meta
process
around
that
are
we
comfortable
with
it
cuz
like
just
like
when
you
brought
this
up
Tim?
This
should
absolutely
be
required.
I
I
Yeah
Kate
and
I
didn't
cats,
that's
obviously
a
bit
buggy,
so
I.
Look
at
this
and
I
think
the
fact
that
this
only
works
on
Linux
is
breaking
a
fairly
fundamental
part
of
a
kubernetes
system,
and
so
it
just
means
that
this
is
a
bug
that
windows
does
not
support
this
and
so
I'm
just
trying
to
I'm
trying
to
make
sure
them.
We
have
a
what
uses
as
a
checkpoint.
We
have
a
good
process
that
ensures
that
sig
windows
adequately
prioritized
when
they're
not
actually
fulfilling
the
goal
of
queue.
For
instance,
all.
G
A
We
did
it
did,
but
the
question
that
is
here
is
whether
or
not
the
test
itself
the
behavior
and
how
the
tests
execute.
Could
that
be
modified
so
that
to
ensure
that
this
statement
still
is
true,
so
I
think
the
problem
here.
Is
that
the
statement
of
what
is
saying
and
it's
the
implementation
details
of
how
it's
actually
done.
It
might
be
two
different
things:
yeah.
F
E
It's
something
that
we're
working
on,
but
I
can't
give
a
specific
date
on
it.
Yet
we
need
to
migrate
over
to
container
D
and
then,
depending
on
the
situation.
There
may
also
need
to
be
a
future
windows
release
to
fix
that,
and
so
it's
something
that's
that's
being
tracked,
but
I
can't
tag
it
with
a
particular
milestone.
Yet
and.
I
Is
it
unreasonable
for
us
to
put
that
supporting
tracking
info
in
this
to
some
degree
like
should
we
require
bread,
crumbs
from
the
conformance
committing
performance
description
back
to
an
overarching
issue
that
identified
the
implementation
gaps
when
it
is
something
that
is
should
be
implemented,
but
it
isn't
vs.
can
never
be
implemented.
F
E
Mean
so
I
guess
the
could
be
versus
versus.
Never
is
a
very
difficult
distinction
to
make,
because
if
it's
something
that
requires
a
change
in
Windows
and
you're,
looking
at
a
ship
cycle
that
might
be
six
months
to
two
years
away,
I
mean
it's.
Do
we
want
to
just
have
an
issue?
That's
open
for
the
next
two
years.
Well,.
I
I
F
F
Ones
I
know
about
are
right
if,
if
there's
some
things
around
partially
qualified
domain
name
look
upstairs
if
host
host
network
is
not
supported,
and
that
is
required
for
some
things
like
pod
to
node
communication
I,
believe
it
was
so
there's
a
few
like
underlying
things.
So
we
can't
not
a
file
can't
do
host
networking.
E
E
A
So
one
thing
that
I
could
take
as
an
extra
in
amount
to
this
to
is
to
understand
why
the
behavior
of
this
requires
a
single
file
mounted
into
a
container.
I
I
don't
see
why
it
should
it.
It's
a
ship
report
termination
message
from
log
output
if
termination
res
apology
fall
back
to
logs
I'm
here
is
sets.
B
So,
if
I
can
go
back
to
the
meta
point,
I
feel
like
just
the
sheer
fact
that
we
put
Linux
only
in
the
tag,
and
it's
also
a
conformance
test-
should
be
priority.
Enough
and
I
feel
like
the
discussion
of
looking
at
the
list
of
conformance
tests
and
how
many
of
them
are
marked.
The
Linux
only
is
a
Sig
Windows
problem
if
they
want
to
get
to
a
world
where
a
cluster
with
Windows
nodes
is
certified
as
conformant.
We
need
to
cycle
through
that
list
and
figure
out.
A
I
think
I
think
for
the
first
time
we
should
not
bike
on
this
repeatedly,
I
think
for
the
first
time
getting
people
on
the
same
page
seems
reasonable,
but
I
do
think
this
needs
to
be
documented
in
gory
detail.
So
that
way,
every
time
you
actually
list
out
the
Linux
only
portion,
you
can
go
to
a
link
that
lands
on
a
page
that
so
it
actually
has
a
description
of
why?
Because
right
now
it
is
totally
unclear
to
me
without
without
the
knowledge
from
Patrick
and
I.
A
B
A
I
think
we're
violent
agreement
I
think
we're
just
going
through
a
gory
detail,
one
thing
at
a
time
and
I
think
so
long
as
they
have
a
documented.
We
have
a
link
that
goes
back.
That
gives
a
breadcrumb
to
saying
like
this
is
why
I
think
that
should
be
sufficient.
This.
This
comment
here,
in
my
opinion,
is
not
something.
G
H
A
A
B
A
I
want
to
I
want
to
keep
us
on
track,
because
we
got
t-minus
three
minutes
and
want
to
make
sure
that
we
at
least
get
through
one.
We
can,
we
can
say
what
things
we
saw,
what
things
we
didn't
see.
So
the
first
thing
we
definitely
saw
is
that
the
version
you
see
the
updated
we
talked
about
potentially
adding
some
more
documentation
about
when
we
should
promote
things
when
we
should
not
promote
things
we
also
talked
about.
It
definitely
needs
the
description
of
why
oolitic.
Suddenly
we
said
that
this
conduct
is
not
sufficient.
A
We'd
like
to
ideally
point
to
a
location
which
is
back
in
this
conformance
document,
which
actually
outlines
the
reasons
why
these
things
fail.
So
that
way
that
has
a
good
breadcrumb
of
information
history.
We
agreed
that
this
test
that
should
be
promoted
that
makes
a
ton
of
sense
I
wanted
to
go
on
to
the
next
test.
It's
got
the
same
comments
that
I
had
before
to
get
clarification
to
make
sure
that
arise
on
the
same
page,
create
a
pod
within
a
container
containers.
Output
is
recording
the
log
and
the
container
exists
sucessfully
without
an
error.
A
E
G
A
Okay,
we
got
t-minus
like
30
seconds
there.
Any
last
comments
folks
would
like
to
make
with
regard
to
this
review.
I
think
we
have
plenty
of
action
items
to
follow
up
in
this
document.
So
if
you
haven't
already
sign
yourself
up,
I
know,
Patrick
is
on
the
on
the
list.
I'm
on
the
list
and
I
think
we
need
to
follow
up
with
Bryan
and
Clayton
about
what
interval
makes
sense
for
burning
through
that
backlog
of
visa
scribble
all
right.
Any
last
questions.
Complaints
concerns.