►
From YouTube: Kubernetes 1.12 Release Team Meeting 20180907
A
All
right,
the
Google
Doc
is
in
its
normal
place,
I'm
guessing
if
you're
here,
you
know
where
it
is,
but
I
will
paste
into
the
zoom
there
just
for
reference.
As
always,
this
is
a
public
meeting.
We
are
recording.
This
will
post
to
youtube
right
after
the
meeting.
Thank
you
all
for
joining.
We've
got
our
burn
down
to
run
through.
If
you
could
put
your
name
in
the
document
for
attendees,
that
would
be
awesome.
A
First,
just
our
initial
sort
of
level
set
today
is
our
Doc's
PR
deadline.
We
have
our
beta
2
coming
on
Tuesday
a
week
after
that
should
be
our
c1
and
a
week
after
that,
the
released
target
showing
this
yellow
I'm
starting
to
feel
slightly
better.
The
this
week
started
out
relatively
rocky
on
a
CI
front,
but
we've
made
some
good
progress,
we'll
get
into
that
a
little
bit
later
in
more
detail,
since
I
saw
Cole
join
the
actual
first
name
that
went
by
I
wanted
to
make
sure.
B
A
B
A
A
Let
me
have
here
one
two:
three,
four:
five:
six
we've
got
a
list
of
nine
issues
right
now
that
we're
tracking,
almost
everything
was
read,
I
would
say
at
the
beginning
of
the
week
and
a
large
part
of
this
was
insufficient
attention
and
triage
on
our
part.
Honestly,
and
we've
really
stepped
up.
Thank
you
to
ice
and
go
in,
especially
for
for
starting
to
try
some
initial
conversations
there
and
the
folks
at
Google
who
have
been
helping.
A
Across
this
list,
the
main
thing
that's
worrying
me
still
is
the
the
broader
issues
around
gke
and
then
the
tape
by
notes
the
taint
nodes
by
condition
feature
so
on
the
latter.
The
owners
in
China
so
I'm
not
expecting
to
get
an
update
today
versus
on
Monday,
but
there
is
progress
there
and
the
situation
is
responsive,
but
on
the
gke
front,
I
shall
not
sure
who
all
is
in
the
room
with
you
there,
but
could
we
maybe
get
an
update?
Yes,.
D
So
there's
Barney
went
yeah
and
said
so
they've
been
looking
into
the
GK
failures
and
they
they
are
also
the
112
really
Shepherd
cookie,
so
yeah
I.
Let
them
yeah
update
the
status
on
that
front.
E
E
A
E
A
Like
the
change
I've
been
merged,
so
this
is
something
that
was
also
going
on
in
the
very
first
listed
issue
there.
I'm
sorry
ii
listed
issue
there
that
michelle
I
was
looking
at.
We
had
a
discussion
yesterday
in
the
suggesting
slack
channel
and
we
can't
see
why
that
one
wasn't
picked
up
either
now
that
one
because
it
is
a
111
to
112
shift,
it
might
have
been
something
related
to
how
one
11
has
picked
up,
but
it
was
not
clear
I
and
then
for
for
the
for
that.
E
A
D
A
A
Pr
is
six
eight
three
eight
six
and
then
could
you
could
you
drop
a
link
to
which
particular
page
on
test
grid
you're
on
because
it
can
be
confusing
to
see
and
then
that
way
we're
each
looking
at
the
the.
E
D
That
is
why,
because
I
don't
see
this
job.
F
A
E
A
G
A
A
G
A
G
C
F
A
Okay,
so
I
think
we've
hit
on
the
main
things
that
I
wanted
to
discuss
there
on
signal.
We've
got
pretty
much
I
believe
every
single
issue,
we've
gotten
good
responsiveness
in
the
last
36
hours
or
so
so
at
least
there
we're
making
the
connections
and
we're
making
progress,
and
we
have
things
that
yes,.
D
E
So
I
think
both
of
them
may
be
root
cause
to
this.
Then
there
is
another
failure
where
we
could
not.
We
are
not
collecting
master
logs
as
part
of
that
test.
There's
an
improvement
we
need
to
do.
Susie's
were
not
able
to
vote
cause
that
so
essentially
name-space
deletions
were
timing
out
as
part
of
the
test,
and
we
could
not
collect
enough
logs
and
we
need
to
master.
H
C
Yeah
I,
linked
from
I
linked
an
issue
or
I've,
been
trying
to
keep
track
of
every
single
gk
e,
Chavez,
failing
and
now
just
pruning
down
to
updating
every
day
when
looking
at
which
gke
jobs
would
be
blocking
the
release.
So
that's
across
master
blocking
master
upgrading,
112
blocking
and
as
of
Cole's
comment
yesterday
yeah
the
clusters
are
coming
up,
but
it
seems
like
tester
like
there's
a
new
set
of
failures.
Each
and
every
single
run.
There's
no
like
one
test
that
is
consistently
failing
anymore.
C
So
it's
kind
of
unclear
to
me
why
gke
tests
would
be
so
much
flakier
than
GCE
tests
in
this
regard,
but
that
kind
of
seems
to
be
worth
like
I,
don't
know
if
you've
been
looking
at
those
specific
things
and
you're
saying
these
are
the
causes
for
all
of
those
sundry
failures
or
if
we're
gonna
be
left
figuring
out
why
random
tests
are
failing.
After
this.
A
Okay,
thanks
yeah,
so
I
guess
on
Monday,
we'll
have
to
evaluate
that.
If
we've
gotten
a
couple
other
key
things
addressed,
what
are
we
missing
based
on
flakiness?
Are
there
major
issues
that
are
being
obscured
by
that,
and
can
we
get
it
root
causes
or
to
what
extent
do
we
need
to
remove
tests
here
from
blocking.
D
A
A
The
the
other
network
ish
one,
the
load-balancing
one
on
there
is
discussion
in
the
issue
there
around
disabling
some
tests
that
have
been
flaky
and
the
I.
Don't
know,
I
didn't
see
confirmation
yet
and
the
issue
as
to
whether
that
had
happened,
but
the
failures
have
gone
away
so
either
either
we've
got
the
flaky
wobbling
or
they
did
remove
them,
but
I'm
looking
for
confirmation
there,
those.
D
A
The
the
owners
basically
acknowledged
that
in
the
issue
and
said,
let's
disable
these
so
I'm
looking
to
see
for
that,
they
aren't
disabled,
that
we
get
confirmation
on
that
and
then
that
it's
not
just
yet
another
flake
today
that
we're
on
the
the
green
side
of
the
flake.
I
Okay,
so
I'm
actually
finalizing
the
status
on
the
on
the
list.
What's
going
on
today
is
that
every
single
open
issue
is
triaged
as
critical.
Urgent
I
actually
did
one
of
those
myself
mainly
because
it
is
related
to
the
flaky
test
issue
in
scalability.
I
We
actually
connected
on
that
and
we
are
now
back
tracking
it
on
the
CSI
and
the
CACI
issue.
Sorry,
co-op,
CSI
and,
and
so
I
just
decided.
It
was
critical
urgent
because
it
is
being
worked
on
in
the
context
of
release
blocking
tests
and
overall
there
are
two
fewer
critical,
urgent,
open
bug,
pure
bug
issues,
I've
been
doing
some
CI
snooping
and
bothering
people.
I
So
while
they
look
active
I
am
also
not
sure
on
the
timeline
for
a
resolution,
and
we
have
a
new
issue,
but
it
looks
like
it
has
an
open,
PR
and
everyone's
right.
On
top
of
that,
that
is
my
main
report
for
right
now,
I'm
I'm,
going
to
see
what
happens
if
I
shake
more
of
these
loose
I
have
pretty
good
response
when
I
paint
people
directly
on
the
CI
issues.
So
here's
how.
A
That
you
got
the
response
on
the
scaleability
front
is
particularly
good
that
the
time
zone
Delta
between
most
of
those
folks
who
are
in
Poland
and
many
of
the
rest
of
us
is
often
leads
to
to
some
gaps.
But
at
least
today's
status
seems
to
be
that
they're,
aware
of
a
number
of
things
and
actively
working
so
that
time.
I
A
And
into
the
work
week
as
well,
once
we
got
Shyam's
attention
on
the
110
release
and
got
him
feeding
status
updates
like
they
were
working
on
things,
but
we
just
weren't
hearing
about
them.
So
once
we
closed
the
communication
cycle,
it
seems
like
that
that
makes
a
big
difference.
If
there's
any
of
these
issues
that
you
want
others
of
us
to
help
poke
on.
Let
us
know
specifically
issue
number
where
it
stands,
who
needs
ping
next,
yeah.
A
I
A
J
Yeah,
you
guys
can
hear
me
yep,
yeah,
so
I,
so
we
took
a
lot
of
trees
like
you'd,
like
you
said,
Tim
and
I'd
initially
put
features
in
red
after
the
Wednesday
meeting
since
then
I'm
bringing
that
back
to
yellow
the
features
that
are
currently
at
risk
there,
ten
of
them
I'm
leaving
them
at
risk
because
we're
currently
Docs
or
do
so
pending
the
doc
reviews,
I'll
sweep
the
features
again
at
the
end
of
the
day
and
see
if
see
if
that
stuff
has
been
cleaned
up,
but
we
have
gone
from
sick,
see,
total
60,
total
features
down
to
47
the
stuff.
J
J
A
A
Yesterday,
I
sent
out
I
think
seven
warning
notices
on
things
that
basically
had
zero
updates.
Yeah
we've
gotten
one
response
back
overnight
on:
oh
yeah,
that's
done
and
code
and
Doc's
are
all
done.
Just
hadn't
documented
any
of
it.
So
that's
one
off
the
list
have
a
couple
or
have
the
remainders
that
I'm
looking
for
responses
today
and
then
I've
got
a
couple
worries
the
again.
A
L
J
A
J
So
just
so
sorry,
sorry
just
a
note
for
communications.
I
know
the
blog
posts
are
active,
I,
don't
know
if
Caitlin
or
Kristen-
maybe
yes,
okay
around
the
call
so
I'm
gonna
re
review
the
blog
post
at
this
point
because
I'm
not
sure
that
the
features
that
we
highlighted
are
actually
in
the
release
anymore.
D
I
Quinn
go
ahead
and
go
after
you
yeah
as
a
taking
notes
issue.
If
you
look
at
I,
looked
at
it,
it's
it's!
It's
it's!
It's
in
my
bug,
issues!
That's
one
of
the
orange
ones
because
there's
like
there's
like
three
open
pull
requests
that
are
sort
of
meeting
related.
One
of
them
is
experimental.
Maybe
that.
A
And
the
the
owner,
whose
name
I,
don't
know
how
to
say,
because
it's
like
random
numbers
and
letters,
their
death
handle
I'm,
hoping
that
we
can
get
a
solid
update
from
them
on
Monday
and
I'll
ping
them
via
email
directly
as
well.
To
get
a
sense
of
that.
We've
had
some
similar
issues
and
past
releases
and
they
have
been
super
responsive.
So
oh
wait.
A
A
D
A
Some
things
are
actively
being
removed,
but
something
like
this
if
we
have
to
remove
the
taint
node
by
condition
stuff,
because
there's
already
so
many
patches
in
place,
this
is
gonna,
be
really
difficult.
So
when
a
lot
of
things
are
like
that
it,
the
threat
of
removal
is
sort
of
the
the
bat
to
try
to
get
more
attention
on
fixing
things,
because
the
forward
path
is
presumed
to
be
easier
for.
H
A
One
if,
if
next
week
like
if,
if
we're
gonna
update
from
sin,
saying
like
this
is
problematic,
then
that
the
questions
gonna
be
okay,
sin:
how
do
you
roll
this
back
to
a
safe
state
and
that's
I
suspect
going
to
be
complicated
than
this
particular
issue
for
the
others?
I
have
a
whole
lot
less
worry
about
needing
to
actually
do
that.
J
I
J
So,
just
as
a
note,
we
continue
to
have
this
this
fake,
stick
that
we
wave
at
people
about
removing
things
like
and
I,
feel
and
a
lot
of
a
lot
of
instances.
Removal
is
probably
harder
than
just
moving
forward
on
it,
so
it
just
as
a
release
team
and
as
sig
release.
We
need
to
push
forward
on
having
policies
around
removal
and
that
should
be
brought
up
in
the
next
segment.
This.
A
D
C
L
Mean
even
if
we
stop
the
madness
to
merge
code,
ID
right
at
the
tail
end
of
stuff,
that
will
be
like
a
big
deal.
I
think
you
know
right
now.
It's
like.
If
you
see
the
spikes,
you
will
see
spikes
right
before
the
deadline
and
in
if
it's
they're
able
to
somehow
spread
out
even
that
activity,
it
will
be
much
better
than
you
know,
trying
to
come
up
with
artificial
means
to
you
know:
stick
and
carrots
and
stuff
like
that.
Yeah.
K
M
I
try
to
stand
still.
I
am
in
a
very
rural
part
of
Minnesota,
but
hello
Doc's
has
not
really
moved.
We've
removed
features,
so
the
things
we're
tracking
has
gone
down.
So
in
that
respect
it
was
probably
good,
but
otherwise
Doc's
have
not
really
moved,
which
is
concerning,
but
it
actually
I'm
not
too
too
concerned
because,
like
Sprint
and
Verizon
were
in
the
early
days.
This
is
mostly
a
nights
and
weekends
thing.
M
For
me,
too,
the
hashtag
day,
job
stuff
and
so
I'll
be
looking
at
at
yes
at
this
evening
and
and
of
course,
this
weekend,
getting
very,
very
much
more
in
depth
into
these
specific
things
that
we
need
to
review
as
a
Docs
team.
So
expect
those
19
outstanding
features
to
be
getting
pinged
and
I.
Don't
really
have
a
carrot
or
a
stick,
so
there
will
be
nothing
for
me
to
swing
or
offer
except
exceptional
documentation,
but
I
do
know
that
the
faster
we
get
this
stuff
in.
We
now
have
three,
no
wait.
M
One
two
three
three
got
three
languages
that
we
translate
the
release
Docs
into
so
I'd
like
to
be
able
to
get
those
in
as
fast
as
possible.
I
didn't
realize:
I
had
the
opportunity
to
meet
the
our
Korean
translators
this
past
weekend
and
they
work
super
hard,
but
they
don't
work
around
the
clock,
and
so
so
them
having
extra
days
to
be
able
to
translate
the
stuff
before
the
releases
makes.
A
Sense
and
I
suspect,
you're
gonna
find
when
you
you
get
a
little
into
it
this
evening.
This
weekend,
I
think
there
have
been
a
number
of
new
pr's
open,
so
people
are
seeming
to
start
responding
and-
and
this
is
also
one
of
those
things
again
like
lacking
a
formal
stick.
It's
often
around
now
that
we
start
complaining
and
people
are
because
it's
code
freeze
and
they're
freed
up
from
coding
to
some
extent
or
something
they
start
doing,
Docs
a
bit
more,
but
it's
it
has
traditionally
what
laggy.