►
From YouTube: k8s 1.15 Release Team Meeting 20190603
Description
(Note: Release lead status isn't until about half way through, we went out of order today)
* **Info: [http://bit.ly/k8s115](http://bit.ly/k8s115) **
* **Calendar:** [http://bit.ly/k8s115cal](http://bit.ly/k8s115cal)
* **Team: **[http://bit.ly/115-releaseteam](http://bit.ly/115-releaseteam)
* **Enhancements: **[http://bit.ly/k8s115-enhancements](http://bit.ly/k8s115-enhancements)
* **Retro:** [http://bit.ly/115-retro](http://bit.ly/115-retro)
* **Handbooks: **[https://git.k8s.io/sig-release/release-team/role-handbooks](https://git.k8s.io/sig-release/release-team/role-handbooks)
* **Slack: **[#sig-release on kubernetes.slack.com](https://kubernetes.slack.com/messages/C2C40FMNF)
* **Mailing List: **[kubernetes-release-team@](https://groups.google.com/forum/#!forum/kubernetes-release-team) (the team) [kubernetes-sig-release@](https://groups.google.com/forum/#!forum/kubernetes-sig-release) (the sig)
* **Youtube Playlist: **[https://bit.ly/k8s-sig-release-videos](https://bit.ly/k8s-sig-release-videos)
A
So
the
other,
a
the
other
issue-
that's
been
going
on-
that's
been
going
on
since
last
week
is
the
so
they
bump
in
coordinates
Persian
just
a
bunch
of
CGC
and
one
of
the
kinder
test
to
it
start
failing
and
for
those
we
have
a
couple.
Pierce
open
them
are
possible
solutions
and
at
least
they
listed
them
at
the
end
of
my
my
update,
so
78
302
and
I
was
a
38,
so
the
latter
one
I
think
it
just
needs.
A
Approval
from
a
I
think
needs
in
its
approval
from
Jordan
and
gloomier,
and
they
are
currently
working
on
it
put
the
first
one.
If
anybody
knows
any
of
the
progress
for
cluster
GC,
any
of
the
properties
are
available
and
willing
to
provide
a
cobra
view.
A
please
a
please
Kingdom
I'm,
looking
I've
been
looking
around
the
place
with
in
know
many
responses
a,
but
we
need
somebody
to
do
it
to
look
over
that
PR.
A
So
we
can
either
fix
or
just
get
emergency
if
you
can
actually
clear
stuff
of
the
massive
inform
in
dashboard
and
so
we're
going
back
to
test
grid,
they
there's
a
115
blocking
and
115
blocking.
Looks
exactly
the
same
as
master
informant,
so
they
will
issue
right.
Now
is
a
six
torch
test
that
is
failing
and
other
than
that
is
a
signal
team.
C
A
D
B
E
F
F
B
G
G
C
So
no
major
updates,
the
number
of
PRS
are
so
open,
remains
about
the
same,
so
I'm
going
to
be
putting
together
the
list
of
tickets
for
my
team
to
handle
this
week.
I
think
we're
gonna
see
this
number
precipitously
drop
because
we're
going
to
be
cutting
through
these
milestones.
These
are
hoping
few
hours
that
are
still
in
law.
Some
115
like
a
hot
knife
through
butter,
so
we're
still
at
yellow
but
I
think
that
we'll
be
swiftly
moving
to
green
also,
if
anyone
has
any
PRS
or
issues
that
you
need
attention
on,
please
send
them.
C
B
F
H
I
Destined
for
brow
and
tied
so
I
think
the
post-mortem
is
pasted
and
everyone
can
go
and
look
take
a
look
at
it.
We
had
a
huge,
much
pool,
backed
up
and
we
everything
is
green
right
now,
so
I
think
the
status
should
be
green.
We
slipped
the
code
freeze
by
a
day
and
that's
the
status,
but
we
are
in
code
freeze
now,
yay.
I
H
J
Hey
there
I'm,
currently
working
on
stuff.
Looking
at
the
next
release,
one
of
those
is
I've
opened
an
issue
for
revisions
to
the
shadow
questionnaire.
So
any
of
you
who
are
section
leads
who
feel
that
the
questionnaire
needed
to
be
fixed,
it
was
too
long
was
unnecessary.
It
needed
some
extra
questions.
Whatever
please
comment
on
that
issue,
and
at
this
point
all
the
section
leads
should
have
received
chat
messages
from
me,
asking
them
to
think
about
nominating
a
successor
out
of
their
pool
of
shadows
or
otherwise.
K
K
H
K
H
M
H
Alright
then
Josh
had
mentioned
the
shadow
questionnaire
updates
and
that
he's
paying
lead.
So
I'll
move
on
to
my
update
I'm
gonna
go
with
yellow
in
solidarity
with
CI
signal,
since
we
do
want
to
see
CI
signal
a
little
bit
healthier
and
happier
as
we're
getting
closer
to
the
release
day,
but
it's
good
to
see
that
there's
issues
being
opened
and
it
seems
like
teams-
are
relatively
responsive
for
all
the
the
failures
that
we're
seeing.
So
that's
good
just
to
give
everyone
some
highlights
around
upcoming
milestones.
H
H
June
4th,
as
already
mentioned,
is
the
docks
deadline
and
then
June
5th,
as
mentioned,
is
the
one
15
beta
looking
ahead
to
next
week.
What's
coming
up
is
on
June
10th.
We
have
the
docks
last
call.
June
11th
is
our
planned
date
for
code
thaw
and
June.
11Th
is
also
the
planned
date
for
the
rc1
to
be
released
and
then
June
13th
is
the
plan
cherry-pick
deadline
and
then
on
Friday
we
entered
code
freeze
at
about
5:30
in
the
afternoon
after
prowl
had
stabilized.
H
There
were
some
issues
with
the
github
API,
but
those
issues
appeared
to
be
based
on
looking
through
slack
github
related
and
had
nothing
to
do
with
our
release.
So
we
entered
code
freeze
at
5:30,
again
shout
outs
to
everyone
inside
testing
for
helping
look
through
all
these
different
issues.
Talking
about
the
accept
exception
requests,
so
we
had
three
that
came
in
there
was
I
pv
for
ipv6
after
some
chat
in
slack
they
decided
to
withdraw
their
extension
request
so
that
work
is
going
to
be
deferred
on
for
the
next
release.
H
E
I'm
here
from
from
Nvidia,
so
that
the
feature
itself
we
talked
about
this
a
little
bit
before
you
join,
but
the
feature
itself
was
was
proposed
by
mostly
people
from
Intel
and
I've,
mostly
just
been
doing,
reviews
on
it,
but
I
know
none
of
them
could
make
it
here
today.
So
I
figured
I'd
at
least
come
on
and
speak
to
it
as
best
I
can.
E
The
entire
feature
is
hidden
behind
a
future
gate.
It's
it's
proposed
to
be
an
Alfa
feature,
so
any
of
the
code
that's
already
been
merged
shouldn't
have
any
impact
on
the
build
or
the
current
running,
though
no
of
anything
else
and
the
reason
for
the
extension
was
mostly
because
a
lot
of
the
PRS
depended
on
each
other
and
we
got
stuck.
E
You
know,
merging
the
later
ones
are
even
getting
people
to
test
them
and
see
if
the
functionality
was
doing
what
it
should
expect,
because
we
couldn't
merge
the
other
things
in
and
get
things
re
based
properly.
I
know
this
happens
every
release,
but
that
was
the
main
motivation
behind
the
the
request
for
the
extension
at
least,
and
the
reason
the
request
was
asked
to
be
pushed
out
to
the
end
of
Wednesday
was
because
the
primary
developers
are
based
in
Ireland
and
today
is
a
day
off
for
them.
H
E
There's
there
six
that
are
still
open
from
my
perspective,
the
feature
would
be
good
enough
to
work
in
an
alpha
state
if
out
of
those
six,
four
of
them
made
it
in
so
the
last
two.
If
they
didn't
make
it
in,
it
would
still
be
a
usable
feature,
but
yeah
they're
they're,
both
medium
sized
at
least
one
should
be
large
size.
They
think-
and
at
least
from
my
perspective,
I've
done.
You
know
extensive
reviews
on
it,
but
I'm
also
not
someone
that
has
the
ability
to
add
an
L,
GT
and
label.
H
Okay
and
in
terms
of
just
risk
from
merging
those
like
hypothetically
best
cased
for
PRS,
do
we
kind
of
have
any
sort
of
confidence
level
around
testing
and
how
those
PRS
like
do
they
impact
other
areas
of
the
Kate's
codebase
that
might
have
some
delayed
impact
with
those
not
being
merged
early.
No.
E
E
A
E
E
There
are
some
unit
tests
for
basic
functionality
to
make
sure
that
you
know
each
of
the
independent
core
logic
functions
actually
have
the
inputs
and
outputs
that
they
expect,
but
it's
definitely
not
full
test
coverage
and
there's
not
any
extensive
integration
test.
As
far
as
you
know,
it's
it's
kind
of
hard
to
integration
test
this
feature
because
it
depends
on
having
nodes,
did
have
GPUs
and
other
types
of
devices
on
them
which
isn't
in
our
infrastructure
at
the
moment.
E
N
So
there
are
still
some
open
questions:
they're,
not
they're,
not
that
significant,
but
anything
that
touches
something.
That's
going
to
be
user
visible.
We
have
to
make
sure,
doesn't
get
persisted
or
surfaced
if,
unless
they've
opted
into
the
table
field,
so
yep
I
know
it's
Monday
now
do
we
have
I
know
you're,
not
even
coming
up
you
thud,
so
he
probably
can't.
E
You
know
that
has
any
users
user
facing
impact.
Is
this
one?
Where
that
the
flag
is
gonna,
be
added
to
enable
or
disable
this
feature
and
make
sure
that
you
know
if
they
do
enable
it
and
set
it
to
any
other
valid
settings
that
doesn't
negatively
impact
the
rest
of
the
system
right
so
yeah.
So
if
it
gets
approved,
I
will
take
the
responsibility
on
to
make
sure
that
all
your
comments
get
addressed
and
I'll
pass
that
information
on
for
these
guys.
That
are
it.
N
H
E
E
The
way
that
I
got
involved
from
from
Nvidia
was
we
took
their
work-in-progress
stuff
and
we
have
a
work
in
progress
stack
internally
that
we're
using
to
test
all
of
this
and
make
sure
that
we
can
actually
do
GPU
allocations
in
a
topology
aware
way
using
their
stuff
and
I
gave
a
bunch
of
feedback
to
them.
Saying
hey
this
worked
this
didn't
work.
Can
we
change
this?
Can
we
change
that
and
over
the
last
month,
or
so
they've
incorporated?
E
All
of
that-
and
you
know
through
that
cycle,
it's
now
gotten
to
the
point
where
it
is
and
it's
sort
of
more
minor
things
that
they
haven't
quite
addressed
from
from
what
from
the
feedback
that
I've
given
them,
but
functionally
from
you
know
my
reviewers
perspective
it.
It
does
mostly
what
it
what
it
needs
to
at
this
point,
if.
N
It
if
it's
cleanly
gated
off
and
all
of
the
external
facing
bits,
flags
and
kindig
and
everything
there's
no
impact,
no
surfacing
of
that.
If,
unless
they've
opted
in
then
I'm
less
concerned
about
it,
it
would
be
good
to
be
pretty
confident
that
we
would
actually
get
these
merged
by
Wednesday
too,
because,
like
getting
two
of
these
in
that's
taking
up
time
from
reviewers
and
things
in
a
pretty.
I
H
E
H
I
N
N
E
H
I'm
very
cognizant
that
I'm
sure
there's
a
lot
of
reviewers
who
don't
want
to
waste
energy.
If
we're
going
to
ultimately
defer
this
Kevin,
do
you
think
the
sig
would
prefer
if
we
wait
to
hear
back
from
the
folks
in
Ireland?
Or
would
you
like
me
to
just
give
this
a
response
now,
for
this
exception?
Well,.
H
Right
now,
given
the
amount
of
open
PRS
and
the
fact
that
they
can't
be
merged
simultaneously
I'm
kind
of
leaning
towards
a
no
just,
because
it
seems
that
Wednesday
might
be
a
best-case
scenario,
and
that
doesn't
give
us
a
lot
of
time
ahead
of
code
fo,
which
is
supposed
to
be
the
following
week.
It
just
seems
that
it's
a
little
bit
high
risk
to
merge
that
many
things
this
late.
E
H
H
G
H
G
F
The
problem
we're
having
is
some
of
the
SIG's
scheduler
people
are
out
sick
and
that's
our
so
getting
the
changes
approved.
We've
done
a
pretty
good
amount
of
testing,
so
we
have
a
you
know,
a
fairly
high
confidence
rating
in
the
code
and
that
this
particular
code
path
won't
affect
other.
You
know
other
components,
and
it's
also
a
pretty
important
feature
for
CSI
migration,
and
we
don't
want
that
to
get
pushed
out.
Another
quarter.
F
G
To
this
start,
the
CSF
migration
is
in
progress,
so
there
are
already
work
most
for
it,
but
that's
that's
behind
the
Alpha
feature
gate
to
see
cell
migration
and
we
want
to
move
like.
If
we
don't
merge
this,
then
that
means
that
we
will
not
be
able
to
get
the
CSF
migration
to
beta
by
one-quarter,
basically
because
this
gives
us
time
to
start
deprecating,
features
and
stuff
like
that.
So
like,
for
example,
you
want
to
duplicate
the
volumes
the.
G
If
CSI
driver
is
not
installed
on
a
node,
then
we
don't
want
to
schedule
parts
on
the
node
I
mean
it
is
for
non-migratory
volume.
So
this
gives
us
a
one
deprecation
cycle
to
duplicate
that
behavior.
Similarly,
for
the
other
base
that
and
on
CSN
migration.
So,
but
if
we
cannot
get
this
end,
then
we'll
have
to
push
everything
out
by
1/4.
G
G
Look,
we
have
end-to-end
tests,
for
we
have
some
end-to-end
tests.
We
don't
have
I
think
enter
in
test
that
tests,
the
Alpha
feature
in
the
integration,
but
David
from
Google
is
working
on
getting
that
test
suit
up,
so
we'll
keep
checking
on
that,
and
then
we
have
lots
of
the
unit
tests.
We
have.
We
have
performance
tests
that
that
measure
the
impact
of
this
change,
so
we
have
done
yon
has
done
a
lot
of
performance
numbers
on
how
it
will
affect
the
part
scheduling
with
volumes
with
this
change.
G
The
feature
itself
that
is
changing
its
changing
behavior
of
a
beta
feature,
though
so
the
feature
is
not
be
alpha
its
beta,
but
it's
affecting
our
alpha
feature.
A
beta
feature
is
changing
to
affect
a
feature
that
is
behind
behind
alpha,
so
yeah.
So
this
this
changes
is
beta.
Actually
it's
not
alpha.
So
is.
H
H
F
Think
it's
highly
dependent
on
our
cig,
scheduler
and
review
times.
Worried.
Sick
storage
has
been
active.
You
know
we
have
like
four
or
five
people
actively
working
on
this.
You
know
in
addressing
comments,
I
think
it's
just
you
know
the
sig
scheduler
stuff.
That's
gonna
hold
us
up
so
review
and
sign
off
from
them,
but
we
have.
We
have
been
getting
some
comments
from
them
and
we
have
been
addressing
them
quickly.
It's
it's
just
a
matter
of
if
they
keep
you
know
being
responsible
and
then.
H
G
M
G
Main
concern
was
the
like:
when
we
discussed
the
cap
with
Bobby
and
Ravi.
The
main
concern
was:
it
should
not
affect
the
performance,
don't
like
of
the
Autobot
scheduling
and
we
have
done
the
perforce
measured
measurement.
It's
approximately
like
eight
percent
slower
than
what
was
before
s--
and
so
does.
Those
are
the
numbers
we
need.
The
Bobby's
on
thick
and
PT
or
away
is
set
to
Dexter.
We
need
time
to
review
this
stuff
and
that's
where
we
are.
We
have.
G
F
H
Okay:
okay,
if
we
can
get
a
response
from
SIG's
scheduling
relatively
quickly,
giving
an
idea
of
how
long
it'll
take
them
to
review
this.
That
would
be
helpful
for
me
in
making
a
determination,
I
guess
like
from
my
perspective,
Thursday
feels
risky,
since
that's
less
than
a
week
from
code
saw
and
we'll
have
already
like-
also
cut
our
second
second
beta,
and
it
doesn't
give
ci
signal
a
lot
of
time
to
stabilize
if
we
can
possibly
get
it
earlier
than
Thursday
I'm
a
little
bit
more
open
to
allowing
that.