►
From YouTube: Kubernetes 1.12 Release Burndown Meeting 20180912
A
A
B
A
You
all
right,
I'm
gonna,
get
things
rolling
here,
dropping
a
link
to
today's
agenda
in
the
zoom
there
welcome
to
today's
burn
down
September
12th
2018
for
the
112
kubernetes
release.
I
am
Tim
pepper.
Your
release
lead.
This
is
a
public
meeting.
It's
recorded
we'll
be
posting
it
to
YouTube
and
just
a
little
bit
after
the
meeting,
so
as
always
be
mindful
of
that
responsible
and
appropriate
for
things
that
will
end
up
on
the
internet
forever.
A
I
guess
I
could
actually
start
to
video
to,
since
it's
being
recorded,
so
there's
not
just
a
voice
in
a
black
box
on
the
screen.
So
a
first
couple
of
points
on
timeline.
We
did
go
ahead
and
cut
the
beta
to
release
yesterday
as
the
day
started.
My
inclination
was
to
not
cut
it
because,
basically,
we
hadn't
had
any
thing
changed
for
the
positive
in
our
test.
A
Results
and
we'd
had
some
regressions
as
well,
but
as
the
day
progressed,
we
had
a
better
understanding,
at
least
of
the
regressions
and
then
a
on
a
cluster
lifecycle
issue
and
I.
Think
actually,
yesterday,
on
slack
I
had
said
cube
ATM.
But
it's
a
cube
anywhere
issue
that
did
get
resolved
and
a
test
went
green.
So
that's
at
least
a
was
a
positive
and
we
went
ahead
and
built
the
beta.
A
Our
next
release
will
be
the
RC
on
Tuesday,
September
18th
or
that's
the
target
anyway
and
then
a
week
following
that
is
our
release
target
overall
status,
I'm
continuing
to
call
yellow,
because
we
have
a
number
of
concerning
issues,
but
all
I'll
get
into
that
in
a
little
more
detail
in
a
moment.
But
as
of
this
morning,
like
in
the
last
10
minutes,
I'm
almost
willing
to
say
like
well,
it's
still
yellow.
A
It's
been
like
a
yellow,
barely
on
the
red
side
of
the
red,
yellow,
green
speck,
drum
and
I'm
feeling
like
we're
more
squarely
in
the
yellow,
with
a
trajectory,
maybe
even
towards
green
here.
If
we
get
a
few
more
fixes
in
in
the
coming
days,
we
might
be
starting
to
look
a
lot
better,
so
I'm,
actually
I'm
I'm
starting
my
day
relatively
positive
and
and
that's
a
nice
change.
A
Yes,
I've
been
really
worried
by
some
of
the
things
we've
been
seeing
the
last
couple
of
weeks
honestly,
but
that
the
that's
been
so
slow
the
last
week
or
so
is
really
starting
to
make
me
worried
so
I'm
gonna
carry
on
through
the
agenda
feel
free
to
to
interrupt
me
or
add
things
as
you
see
or
have
comments.
I
want
to
mention
on
test
infrastructure,
I'm,
not
sure.
If
we
have
Cole
know
I,
don't
see
him
I'm
not
expecting
any
changes.
Here
we
made
the
transition
on
Monday
to
two
tide.
A
4Kk
and
things
seem
to
be
working
fine
right
now
there
had
been
some
confusion
around
github
error
issues
even
going
into
last
week
ahead
of
tide,
and
then
yesterday
we
had
some
github
issues,
but
I
I.
Don't
think
these
are
related.
Like
part
of
my
brain
worries
and
was
like,
did
we
just
break
github
I?
Don't
I,
don't
think
that's
the
case
and
so
I
think
we
should
expect
some
stability
on
this
front
and
then
the
next
milestone
on
test
infer
that
we'll
want
to
watch
for
one
of
the
the
cool
aspects
of
this
tide.
A
Change
should
be
that
when
we
have
our
reopening
of
the
master
branch
for
113
the
patches
that
have
been
sitting
there
waiting
flow
in
much
more
efficiently
in
the
past,
so
I
know
the
testam
for
folks
are
really
keen
to
watch
that
and
we'll
see
and
we
may
have
hiccups,
but
we'll
we'll
deal
with
them,
as
that
happens,
so
that'll
be
sometime
in
the
next
two
weeks.
Just
glancing
at
my
big
master
calendar
the
specific
day.
I.
A
Think
at
this
point
is
really
kind
of
going
to
be
TBD,
though,
because
I'm
I'm
not
gonna,
vote
YES
on
reopening
master.
If
we
have
the
sort
of
issues
we
have
right
now
we
need
these
things
fixed
before
we
have
hundreds
of
patches
merge.
All
of
a
sudden
and
I
know
that
there
are
some
large
things
that
are
pending
to
is
1:13
starts
and,
and
that's
good
and
great.
But
we
don't
want
that.
A
Complicating
our
lives
just
right
now
and
especially
because
patches
come
in
to
master,
and
then
we
cherry
pick
them
or
branch
fast
forward
to
the
112
release.
Getting
the
CI
signal
on
the
master
branch
is
really
important.
It
gives
us
a
heads
up
if
we
think
we
made
a
change
deliberately
and
it
improved
the
master
blocking
tests.
Then,
when
we
hold
changes
over
to
the
112
release,
we
expect
to
see
the
same
thing
and
we
can,
amidst
all
of
the
flaky
tests
we
have.
A
We
can
get
a
little
more
confidence
that,
yes,
we
deterministically
made
a
positive
change
here
and
we
reproduced
it
there
and
that
that's
really
good
information
to
be
able
to
take
away.
So
that's
just
in
front
of
thoughts
for
the
day,
and
then
that
gets
us
on
to
CI
signal.
Mohammed
said
he
wouldn't
be
able
to
attend
today.
A
I
think
I'm
actually
expecting
a
somewhat
sparse
attendance
and
the
main
thing
really
to
talk
about
a
CI
signal,
so
I'm
gonna
scroll
through
the
list
of
things
I've
only
just
started
triaging
the
day's
failures,
but
we
do
have
a
new
regression
that
I've
noted
there.
So
I'll
look
at
getting
an
issue
open
for
it.
The
jke
costs
112
default
had
a
DNS
issue
and
sig
networking.
That's
new.
C
A
C
A
A
C
A
A
Okay,
so
I'm
gonna
color
that
yellow
then
instead
of
red,
and
we
can-
we
can
follow
up
on
that.
So
that
thing
carries
on
into
the
the
next
topic.
Then
I
do
I
want
to
finish
scrubbing
through
the
list
and
see
if
there
are
any
other
unexpected
things.
But
the
main
thing
in
flight
then
I
think
right.
Now
is
the
the
taint
nodes,
my
condition
work.
It
continues
to
be
ongoing
and
Klaus.
A
His
got
two
things
merged
yesterday
and
is
realized
that
those
things
need
back
ported
to
111
and
it's
unfortunately,
now
seeing
that
we
have
the
the
time
lag
versus
China
again.
My
question
there
is
so
if
we
need
those
cherry-picked
into
one
11
for
n,
it
sounds
like
we
need
a
one,
not
11.4
actual
release.
I
don't
like
the
feeling
that
we're
experimenting
in
the
prior
stable
release,
so
I
I'm,
gonna
drop
Klaus
a
note
and
and
get
a
sense
of
his
confidence
level
and
what
options
we
have
to
do.
A
Some
experimentation
not
in
the
fryer
release
branch,
because
if
we
try
to
pick
these
and
they
don't
make
things
better,
are
we
gonna
have
a
series
of
iterative
fixes
and
backports
to
111
and
turn
our
user
base
through
something
visible
like
that
111
for
111,
5,
111
6
over
the
next
week
or
two
trying
to
resolve
this?
That's
not
that's
not
appropriate,
so
we
need
a
better
plan.
There.
C
Considering
our
our
secret
is
on
Tuesday
I
mean
still
and
and
eat,
and
we
at
least
need
a
couple
of
days
of
runs
and
generally
his
fix
kind
of
or
all
the
stained
stuff
is
touching.
The
upgrade
tests
extensively,
which
take
at
least
13
12
to
13
hours
to
run
in
the
best-case
scenario
I'm
thinking
we
might
have
to.
If
we
are
back
making
a
decision
to
back
out,
it
has
to
be
like
today
or
tomorrow.
Let
me
do
you
like
for
things
to
be
packed
out
and
then
for
things
to
stabilize.
B
For
what
it's
like,
the
upgrade
jobs
they
don't
as
far
as
I
know,
please
somebody
correct
me
if
I'm
wrong,
but
I've
actually
like
chased
through
the
commit
Shaw's,
they
don't
test
against
a
tag
they
test
against
head
of
the
release
branch.
So
in
terms
of
turning
our
users,
our
users
until
we
actually
cut
a
release.
So
if
you
really.
B
B
In
chat
because
I
think
I
don't
know
how
much
okay
so
like
just
just
cherry-pick
and
revert,
and
that's
fine
like
if
we
ever
have
any
concerns
in
the
build
logs,
Kubb
API
server
will
say
what
version
it
is.
It
says
you
can
grep
for
a
cube,
API
server
version,
and
then
you
can
trace
that
shop
back
to
github.
If
you
really
want
to,
but
again,
I
can
double
check
if
it
would
make
people
feel
better,
but
I
feel
pretty
confident
that
the
CI
/
Kate's
stable
one
means
it.
B
Is
the
latest
build
off
of
the
N
minus
1
branch,
so
the
same
thing
will
take
place
when
we
start
doing
when
we
come
out
of
code
freeze
and
we
start
running
against
the
112
branch.
Ci
/
Kate's
beta
beta
stands
for
the
currently.
The
Rachel's
gonna
go
out
the
door,
so
none
of
those
should
be
tied
to
actual
tags.
B
A
I
and
I
guess
I
did
so.
The
thing
that
I
was
immediately
worried
about
was
classes
comment
on
slack,
so
it
tests
against
one
11.3
so
lucky
that
he
explicitly
said
that
made
me
feel
tagged
wise.
But
if
and
then
I'll
go
through
and
grep
the
logs,
also
for
the
version
and
see
and
and
go
step
back
through
to
get
hub
on
the
commit
as
well
and
see
and
try
to
convince
myself
as
well
independently,
because
if
that's
the
case
I'm
a
whole
lot
more
comfortable,
not
so
wanna.
B
See
if
I
can
help
you
put
together
or
either
standing
back
real
soon,
because
I'm
supposed
to
drop
off
the
face
of
the
earth
later
today
after
the
steering
committee
meeting
at
1
p.m.
Pacific,
but
like
I
am
very
very
this.
This
question
comes
up
every
time,
I'm
getting
sickness
constantly
having
this
question,
I
would
like
us
to
definitively
answer.
Where
does
it
upgrade
from?
Where
does
it
upgrade
to
end
of
story?
Hallelujah.
A
Alright,
then,
so
we're
making
some
progress
there,
hopefully
with
Klaus.
We
have
some
roadmap.
There,
then
the
next
one
68
509
ice
and
and
I'm
not
sure
the
name
of
the
person
on
your
left.
B
C
A
C
A
The
onion
okay,
the
the
next
one,
the
horizontal
pot,
autoscaler
instability.
We
have
Solly
actively
debugging
on
that,
so
I'm
I'm
not
super
worried.
That's
the
right
owner.
We
have
their
attention,
so
it's
that
it's
been
around
for
a
while
worries
me
it's
red
but
I'd,
be
surprised
if,
in
the
next
day
to
three
we
don't
get
some
some
resolution
there.
So
the
next
one,
then
six,
eight
three,
five,
nine
all
the
before
sweet
failures
in
the
the
GCE
scale
testing.
C
A
C
A
C
C
A
And
I
do
see
just
glancing
over
those
issues.
I
do
see
some
updates
and
something
that's
been
merged
as
well
and
test
em
fora,
but
only
a
day
ago.
So
since
these
are
longer
running,
it
may
be
a
bit
yet
but
I
guess
again
for
for
the
current
release,
we're
kind
of
in
the
normal
place
that
we
are
around
scalability
we've
got
the
right
people.
Looking
at
these
issues,
we
have
a
little
bit
of
runway.
I
I'm,
worried,
but
I
I
feel
like
this
is
the
I'm,
the
dog
in
the
room
with
the
flames.
C
A
This
is
it's
awkward
with
the
slowness
on
the
the
feedback
loop,
but
we've
got
the
right
people
watching
and
aggressively
working
on
this.
It
seems
so
we'll
continue
to
watch
this
space.
Only
the
there
isn't
one
where
Sean
mentioned.
So
this
was
17
hours
ago
and
not
sure
which
so
he
says
we
should
have
a
run
once
every
weekday
starting
tomorrow.
I
don't
know
if
tomorrow
is
today
now
which
day,
but
today
or
tomorrow
we
might
start
getting
more
feedback.
Then
so
that's
that'll
be
helpful
in
the
coming
days,
then.
E
A
A
So
that
that
might
have
been
part
of
what
happened
and
I'll
reassess
this
the
expectations
there
for
today
and
we
can
see
the
next
one.
Six,
seven,
six
hundred
the
DNS
config
tab
out
DNS
a
little
bit
earlier,
probably
not
related
to
this
one.
But
this
one
the
PR
has
I,
don't
say,
has
merged,
but
it
is
merging
it
is.
It
should
be
merged
pretty
quickly
here
and
if
so,
then,
probably
later
today,
we'll
we'll
see
some
we'll
see
some
progress
on
that,
and
that
would
be
cool
and
it'll.
C
A
A
The
I'm
still
worried
about
the
the
taint
notes
by
condition,
but
if
that
cherry
pick
goes
through
today
still-
and
we
happen
to
see
even
if
we
don't
confirm
I
guess
if
the
Tesco
agree
and
that's
the
the
evidence
without
following
the
audit
trail
of
commits
that
that
would
that
would
be
a
positive
thing,
potentially
that
we
might
get
lucky
on
today
as
well
for
those.
So
we
will
see
on
issues
Quinn
had
mentioned.
A
She
might
not
be
able
to
make
it
today,
not
sure
if
anybody
else
is
triage,
I
don't
see
anybody
on
the
call
who
would
have
I'll
give
a
look
at
those
today
as
well,
but
not
super
worried
just
because
the
the
velocity
there
has
been
relatively
quiet
and
compared
to
our
test
failures.
Those
not
super
worried
there
at
the
moment,
nothing's
feeling
on
fire
next
on
the
agenda,
features
Stephen
couldn't
make
it
today,
but
did
give
an
update,
not
sure
Kenny.
A
If
you
have
anything
else,
you'd
want
to
add,
but
I
feel
like
in
the
status
is
updated
to
Green
there
I'm
actually
pretty
comfortable
here
the
the
press
over
the
last
week,
we've
basically
gotten
updates
on
everything
now,
so
we
know
which
things
have
officially
dropped,
and
basically
everything
else
is
on
track.
So
we've
we've
done
that
final
splitting.
A
We
know
what
we
have
so
from
our
original
66
features,
we're
down
to
forty
half
of
the
things
that
were
alpha
fell
out
and
then
a
little
bit
of
the
things
that
were
built
beta
and
stable
targets
fell
out,
but
it
was
mostly
early
alpha,
stuff
and
I
feel,
like
that's
kind
of
normal,
that
it's
an
aspiration,
get
new
alpha
features
in
so
probably
kind
of
done
on
the
future
tracking
front.
Aside
from
documentation,
tracking
down
and
I,
see
anonymous.
Coyote
is
typing
on
documentation.
Right
now.
Are
you
there
Zach.
F
F
In
scenic,
Sonoma
County,
so
the
dark
status
I'm,
so
gonna
leave
it
at
chartreuse,
one
because
I
think
it's
a
lovely
color
and
two,
because
there
is
still
one
missing
dog
account.
It's
a
feature
for
vertical
pod,
auto
scaling
and
I
was
reading
through
the
comments,
and
it
looks
like
it's
probably
actually
gonna
go
independent
of
1/12,
so
it
is
a
feature
but
not
in
the
sense
of
like
we
are
going
to
be
releasing
it
as
part
of
the
1/12
dock
cycle.
Like
I
said
it
will
have
special
attention.
So
I
made
a
note.
F
If
the
owner
of
this
features
name
is
I'm
not
going
to
try
as
I
was
gonna
have
to
scroll,
but
if
that
particular
owner
says
hey,
yeah,
we're
gonna
go
independent
and
I
will
make
a
pull
request
and
document
this
independent
of
the
1/12.
Then
that
would
make
the
missing
count:
zero
and
our
status
green,
because
we
are
more
than
halfway
complete
with
documentation,
meaning
it
is
merged
into
the
1/12
release.
Branch,
which
means
I'll,
need
to
be
scaffolding,
the
pr
into
master.
Here
soon
there
are
some.
A
Done
one
thing:
I
want
to
note
there
just
kind
of
as
we're
thinking
about
the
transition
forward
as
well
on
that
scaling,
one
so
ended
up
being
implemented
as
a
CR
D,
it's
outside
of
KK.
So
it's
in
that
space
of
the
way
the
system
is
now.
We
don't
really
track
it,
but
we
gotta
watch
that
and
and
figure
out
how
we
handle
this
better
as
as
time
goes
by,
because
this
isn't
the
only
one
we're
also
like.
A
If
first
it
didn't
look
like
there
is
any
code
that
had
changed,
but
people
are
saying
it's
done.
Where
is
it
and
things
are
his
things
split
out?
It
becomes
very
ambiguous
and
then
some
of
those
do
end
up
being
things
that
should
have
user
facing
documentation
and
if
that
user
facing
documentation
is
just
on
some
kind
of
semi
random
github
repo
is
a
piece
of
markdown.
That's
that's
worrying.
So
something
to
think
about
is
an
improvement
and
I
guess.
I'll
put
a
note
to
this
regard
and
the
retro
doc
to
be
think.
F
A
F
Yes
seems
like
we're
following
the
same
cycle:
there's
this
huge
53
million
line
monolith
that
is
now
being
spread
split
into
the
individual
micro
services
of
kubernetes.
Now
it's
going
to
be
impossible
for
people
to
install
this
unless
they
have
gk3
chaos
or
aks,
so,
but
but
all
of
these
independent
things
I
do
think
because
we're
having
the
same
issue
of
like
we're,
not
really
sure
we
fought
like
we'll
we'll
find
out
so
I'm
gonna
put
an
issue
up
on
the
kubernetes
website
to
get
joven.
Nobody
hey.
This
features
completely
undocumented.
F
It
would
go
like
we're
like
we
didn't
know,
it
was
a
feature
or
that
people
even
wanted
to
know
about
it.
So
so
people
are
even
looking
at
the
general
mane
kubernetes
documentation
website
for
things
that
are
these
plugins,
like
external
DNS,
which
is
an
incubator
project,
I
think
it
still
is,
and
it
was
for
a
while,
but
the
metric
server
I
guess
it's
now
part
entry
and
various
other
things
so
I
totally
agree
with
you
Tim,
that's
an
improvement
that
I
could
definitely
made.
A
Alright
then
release
notes
today,
hi
puppy
today.
E
Yeah,
so
we
are,
we
are
green
and
we
are
actively
copy
editing.
The
documents
we
have
now
moved
to
a
daily
schedule
of
adding
in
new
PRS
and
Dave
is
contacting
the
sig
leads
to
to
get
the
major
themes
so
we're
trying
to
get
the
drafts
of
the
major
things
by
the
end
of
this
week,
which
means
we
probably
won't
have
them
till
the
middle
of
next
week,
which
puts
us
pretty
much
exactly
on
time.
So
we're
good.
A
D
Yep
Steven
I
believe
asked
for
an
additional
day
to
review,
so
we're
just
waiting
on
him
for
that
on
the
release
blog,
we
also
haven't
received
any
input
on
the
media
FAQ
document,
so
I'm
gonna,
resend
that
out
today
that
just
needs
to
be.
You
know
we
can
finesse
it
on
our
side
for
what
that
content
looks
like.
So,
if
Tim
and
Steven
could
just
drop
in
some
bullets,
that
would
be.
That
would
be
fine,
maybe.
D
Just
a
note,
because
we're
planning
to
start
pitching
next
week,
which
involves
sending
out
the
blog
post
so
we'll
need
to
try
and
lock
down
kind
of
what
our
best
guess
is
for
the
release.
If
we
think
it's
going
to
move
at
all
so
that
we
can
use
that
as
the
embargo
date
for
the
blog
post,
okay,
so
I
have
it
tentatively
scheduled
for
I,
believe
5:00
p.m.
on
the
25th,
so
I
think
just
if
we
can.
A
A
A
All
right
call
it
a
meeting,
continue
to
be
a
bit
worried
about
test
grid
and
anything
anybody
can
do
to
give
it
a
glance
a
few
times
a
day
and
push
on
issues
and
PRS.
I.
Definitely
appreciate
that
and
that's
my
sole
focus
in
coming
days.
So
thank
you
all
for
all
of
your
contributions
and
pulling
all
this
together.
It's
a
little
stressful
right
now
for
everybody
I
think
watching
it
all
go,
but
it
is
starting
to
come
together
where
we're
doing
it
so
have
a
great
day.
Y'all.