►
From YouTube: Kubernetes SIG Release - 2019-06-18
Description
A
We
are
recording
all
right,
so
hello,
hello,
everyone.
This
is
the
June
18th
bi-weekly
sig
release
meeting.
This
is
a
meeting
that
is
recorded
and
on
the
internet
for
posterity.
So
please
be
mindful
of
what
you
say.
Please
be
sure
to
adhere
to
the
kubernetes
code
of
conduct,
which
is,
as
Claire
says,
is
tantamount
to
don't
be
a
jerk.
So,
let's
get
started,
we've
got
a
few
things
on
the
agenda,
including
I.
Think
the
first
thing
that
we
should
cover
is
for
release
team
members
that
are
on
the
call.
A
And
as
we're
going
through
the
meeting,
if
anyone
is
interested
in
taking
notes
or
willing
to
take
notes
because
I'm
going
to
gap,
a
lot
feel
free
to
it
would
be
appreciated.
So
the
cig
scalability
we're
supposed
to
that's
a
12-hour
test,
we're
supposed
to
hear
back
from
them
by
tomorrow
morning,
right,
okay,.
A
C
A
Should
push
back
and
and
delay
the
the
release
retrospective,
depending
on
the
results,
I
think
that
we
should
hold
it
in
place
until
we
know
for
sure
what's
going
on
and
I
think
that
will
we'll
find
that
out
tomorrow,
if
we
determined
that
the
tests
are
something
that
are
going
to
block
us
for
longer
or
we
need
to
determine
some
fix
or
scalability
needs
to
determine
some
fix,
then
potentially
we're
looking
at
next
week.
But
right
now,
let's
pretend
like
we're
safe
right.
A
All
right,
so
the
next
one
is
the
removal
of
the
test
info
roll
from
the
release
team,
starting
in
1/16,
so
I
think
we
are.
We've
been
moving
at
good
pace
on
doing
that
stuff.
We
have
most
of
it
documented
I've,
seen
PRS
open
to
update
the
relevant
handbooks
right.
So
for
those
who
are
not
aware,
the
testing
for
role
has
been
mostly
automated
away,
which
is
awesome,
I,
think
that
should
be
the
goal
for
a
lot
of
these
roles
where
possible.
A
So
the
remaining
the
remaining
duties
of
the
test
info
role
are
transferring
to
the
branch
manager
position
and
the
end
bug
triage,
so
I
think
we're
still
working
out
some
of
the
details
of
what
is
being
moved
into
bug
triage,
but
there
are
open
PRS
across
I.
Think
I
reviewed
the
one
for
bug
triage
and
there
is
a
branch
manager
won't
open
coms
had
an
update,
I
think
the
columns
update
was
signed
off
on
and
there
might
be
one
more
that
I'm
missing
feel
free
to
call
it
out
if
I'm,
forgetting
something.
A
B
I'm,
a
worrywart
and
I
worry
were
putting
too
much
responsibility
in
a
small,
smaller
set
of
people.
I
I
I
wish
the
test
of
a
rolled.
Currently
it
was
a
more
proactive
read
out
on
the
health
of
test
infrastructure,
but
we'll
see
if
that
is
a
problem
over
time
and
if
so,
we'll
figure
out
how
to
make
sure
we
we
have
that
in
within
the
team
in
the
process.
D
I
mean
it
really
hasn't
been
for
several
release
cycles
and
instead
the
focus
has
been
on
that
being
responsibility
of
cig
testing.
D
D
D
F
B
Were
talking
about
removal
of
the
test
and
parole
from
the
release
team
and
whether
that
leaves
a
gap
on
proactive
reporting
of
testing
for
status
into
the
release
team,
but
everybody
pretty
is
comfortable
with
that,
not
being
something
that
was
tracked
super
closely
anyway
lately.
So
if
it
becomes
a
an
issue
of
a
gap,
we
will
address
it
in
the
future,
but
in
the
meantime,
I
think
everybody
is
ready
to
to
say
that
primary
tasks
are
automated
yeah
kill
it
with
fire.
A
So
alright
cool
that
sounds.
We
sound
happy.
We
sound
happy
ish.
We
sound
tentatively
happy
alright,
so
the
release
115
retrospective
we've
tapped
Christine,
who
is
on
the
call
it
has
been
showing
up
to
the
sig
p.m.
meetings
and
overall
interested
in
sig
p.m.
we
figure
it's
a
time
to
finally
give
Jace
a
break
on
doing
the
retro
emceeing.
So
welcome
Christine
to
the
call
and
thank
you
for
picking
that
up.
Yeah.
A
Retro
doc
is
linked
in
the
in
the
meeting
notes.
Please,
if
you
were
on
the
release
team,
if
you're
around
the
release
team,
if
you
have
touched
anything
for
1:15
and
you
have
opinions
about
how
we've
been
doing
things,
please
drop
notes
on
the
the
the
retro
doc.
The
retro
is
scheduled
to
happen.
This
Thursday
June
20th
at
the
community
meeting,
so
that's
1
p.m.
Eastern
and
shift
for
your
time
zones
appropriately.
A
B
Alright,
so
we
briefly
earlier
said
like
should
we
delay
and
so
I
I
kind
of
feel
like
because
we
normally
span
or
often
span
multiple
meetings?
If
there's
other
things
that
happen
in
this
coming
week,
then
we
can
follow
up
with
a
focused
analysis
of
those
things
and
I
suspect.
Given
the
state
of
discussion
around
what's
going
on
this
week,
we
were
going
to
do
that
anyway,
even
beyond
the
retrospective.
So
well,
we
need
to
get
a
better
handle
on
how
we
interact
as
a
release.
Sig
with
the
scalability
sync
yeah.
A
So
I
had
I'd
mentioned
that's
Christine
and
Jase
to
that,
like
I,
think
we
should
have
it
anyway,
regardless
of
whether
we
release
yeah
but
but
yeah,
it
is
gonna,
be
continuing
discussion.
I
know
that
Aaron
may
have
some
opinions.
I
know
there
was
a
thread
going
on
regarding
like
revert
approvers
in
sig
scalability
and
that
their
charter
dictate
certain
things,
but
also
that
sig
scalability
is
I
under
resourced.
I
guess
is
the
right
term
for
it
or
they
don't
have
the
right
coverage
at
times
so
Aaron.
If
you
have
opinions
too
wait.
B
Really
just
talked
about
two
that
I
think
are
conflated
one.
We
have
an
issue
this
week
and
we'll
we
can
carry
on
with
the
retrospective
I
think,
because
99.9%
of
this
release
is
done,
we
can
discuss
all
of
the
things
related
to
that
bigger
picture
is
in
addition
to
the
115
release
being
delayed.
There's
this
other
discussion
around
how
and
why
does
scalability
gate
the
release,
and
how
can
we
improve
that
process?
I
think
that,
yes,
that
came
up
in
115,
but
it's
also
come
up
in
multiple
other
releases.
F
Perhaps
they
should
use
a
different
version
of
core
DNS
that
was
like
known
to
work
at
scale,
whereas
the
newer
version
of
core
DNS-
that's
used
by
default,
not
quite
known
to
work
at
scale.
That
to
me
is
like
one
workaround
for
how
we
could
get
there
and
I
mean
I
have
I've
certainly
discussed
how
but
I
feel
like
the
scalability
jobs.
F
Do
not
meet
the
criteria
to
be
like
hard
release,
blocking
they're
in
a
board
called
informing
because
they're
slow
or
they
don't
meet
the
criteria
that
we
as
a
community
have
established
for
blocking
now,
I'm
sure
they'd,
a
different
opinion
for
me
that
informing
and
not
blocking
I
think
that's
really
a
conversation.
We
as
a
community
yeah.
A
F
Generally,
you
don't
have
to
go
too
far
down
this
radical
generally.
When
this
comes
up,
I
always
win,
6k,
Lily's,
slow,
slow,
slow
test
comes
up.
I
always
try
to
figure
out
like
how
could
we
have
caught
this
sooner
with
a
faster
test?
Could
we
have
chained
it
to
be
like
more
aggressive
with
its
thresholds
or
whatever,
and
so
like?
We
do
have
like
I
think
a
hundred
no
test.
That's
in
the
release
blocking
dashboard.
F
D
Wouldn't
one
thing:
I've
asked
sig
scalability
for
for
awhile,
but
I
haven't
been
able
willing
to
participate
in
actually
implementing
it,
so
it
hasn't
happened
is
to
have
accessible
trending
information
because,
like
just
because
we
don't
have
it
I
can't
prove
that
this
is
the
case,
but
it's
quite
possible
that
well,
this
problem
didn't
rip
the
threshold
of
the
hundred
node
test.
We
might
have
seen
a
blip
in
hundred
nodes
test
of
increased
latency
that
didn't
quite
trip
the
threshold
and
right
now
we're
not
getting
that
information.
D
B
Cool
I
still
parently
feel
like
we
should
separate
it
and
have
it
as
a
formal
focus
area,
because
I
think
it's
gonna
take
longer
than
just
the
retro
to
resolve
and
I.
Think
like
for
a
while.
I've
wanted
to
attend
this
big
scale
ability
meeting
but
as
best
I
can
tell
they've
had
one
this
year
so
like
and
I,
don't
I
I
would
guess
they're
not
going
to
attend
the
retro,
so
I
feel
like
we
need
to
make
it
a
focus
where
we
create
a
meeting
to
talk
to
them
and
figure
out.
So.
E
G
G
Project
to
provide
a
lightweight
packaging
mechanism
for
Cooper
to
these
projects,
that's
something
we
felt
for
GE
and
GK
on-prem,
and
it's
something
we're
starting
to
explore.
Using
four
cops
and
a
couple
other
tools,
so
I
don't
know
how
much
interested
there
is.
But
that
is
something
that
we
found
a
lot
of
use
in
at
least
internally
at
Google,
for
releasing
kubernetes
as
like
a
distribution.
So
do.
G
So
the
idea
that
we
came
up
with
we
went
through
a
lot
of
iterations,
so
Swee
excited
to
kind
of
like
customize
actually,
but
with
a
little
bit
different
focus.
We
decided
to
base
everything
on
CRT
is
so
one
of
the
core
of
it.
Is
this
kind
of
very
simple
see
our
CRT
that
wraps
CRTs
so
and
really
what
is
providing?
Is
it's
just
a
versioned
named
bunch
of
objects,
so
these
are
all
unstructured
objects
and
it
comes
with
a
CLI
to
to
manage
these
things
and
kind
of
like
customize.
G
As
a
plugin,
especially,
what's
that
settles
down
kind
of
like
customized,
it
builds
ultimately
a
we'll
call
it
unwind
component
that
we
can
ship
around
combined
together
and
multiple
components
that
make
up
the
distribution,
so
it
I
think
the
most
interesting
thing.
Is
it
strips
with
this
CLI?
So
I'll
do
this
right
now,
so
here
we
are,
let's
see,
do
you
see
your.
G
F
G
G
H
G
It's
been
a
while
okay,
so
look
at
something
else,
so
I
think
one
of
the
one
thing
that's
really
powerful
about
this
idea
of
everything
is
CRD
is.
Is
we
took
a
look
at?
Well,
you
know,
go
template.
You
know,
help
us
Disko
template
idea,
but
we
just
made
like
a
CR
D.
That
was
like
a
Co
templating
CRD,
which
was
like
it's
very
simple.
So
this
actually,
instead
of
just
using
normal,
go
templates,
we
apply
them
like
customize
as
patches.
So
here
we
have.
G
G
F
G
So
you
can
see
that
we
have
now
our
completed
component
and
well.
You
can't
see
it
because
oh
I
think
that
is
it
the
namespace
yeah.
E
G
Template
highest
and
then
it
was
hashed
on
top
of
it,
and
so
that's
how
we're
doing
GK
ng
count
premise.
We
have
these
like
little
patch
templates.
We
can
apply
it
at
one
time,
cool
all
right.
So
then
that's
that's
kind
of
the
whole
idea,
and
you
know
really
the
part
that
I'm,
like
really
bullish
about
promoting,
is
kind
of
this
very
basic
format,
because
that's
like
a
really
nice
there
change,
you
know
once
we
can
like
speak
to
something
that
is
a
bunch
of
versions.
These
objects,
then
it's
much
easier
right
to
land.
G
I
Would
think
this
group
is
probably
most
would
probably
be
most
interested
and
be
the
bundling
part
where
you
take
all
the
version.
Components
produce
a
bundle.
That's
like
okay,
this
is
a
kubernetes
release
from
all
these
things,
like
definitely
a
bunch
of
patching
tools,
and
this
is
a
pretty
neat
one,
but
the
part-
that's
probably
super
interesting
here-
is
that,
like
you
could
imagine
doing
this
upstream
and
saying,
instead
of
all
the
stuff,
that's
in
the
cluster
derp,
we
have
like
all
those
add-ons
and
stuff.
G
Exactly
exactly
and
one
of
the
things
we're
doing
internally,
which
is
like
quite
a
bit
more
complicated
and
I,
don't
have
an
example
this
yet
is
we
actually
have
a
dependency
resolution
system?
That's
built
into
this,
so
we
have
like
basically
requirements,
which
is
again,
you
know
very
simple:
it's
basically
like
a
go
style
of
goemon
style
requirements,
so
we
actually
have
a
bunch
of
these
components
to
go
there
and
we
have
some
requirements
like
okay,
these
114
just
go
to
your
component
repository
grab.
All
these
things
assemble
yeah,.
G
F
G
Yeah
I
mean
it's
just.
This
is
just
unstructured
objects.
It
supports
anything
I'm
right
now
in
G,
K
we're
using
to
both
add-ons
and
core
control,
plane,
stuff
and
we're
also
starting
to
use
it
for
even
other
more
esoteric.
Things
like
images
like
we
pass.
A
CR
D
for
representing
less
images,
see
Rd
for
representing
cube,
config,
environment,
config
or
blending
startup
scripts.
Just
cuz.
It's
like
a
very
nice
generic
way
to
version
configuration.
G
B
A
couple
things
that
resonate
for
me
within
this,
and
especially
looking
at
the
watching
back
the
video
from
cube
Connie
you
so
the
that
you
make
a
point
of
differentiating
build
from
packaging
I
think
is
really
important.
We
we
have
multiple
things
doing
building
and
we
have
a
looser
handle
on
how
we're
doing
packaging
today
so
to
tighten
that
up
and
conceptually
have
those
separate
and
and
follow
them
with
a
qualify
and
a
release.
B
Don't
know
I
like
the
idea
of
being
able
to
like
this
pattern
of
declaring
what
you
expect
and
and
if
it's
there
you
use
it
and
when
the
I
could
see
this
tie
into
all
sorts
of
validator
tools
on
what
we
build
and
are
used
and
those
validators
tying
into
promoters
and
yeah.
This
could
be
the
start
of
pattern
if
nothing
else,
the
having
more
of
a
component
manifesting
agreed-upon
system
there
around.
What
is
a
component.
How
do
you
articulate
his
name
and
version
dependencies
like
in
a
machine?
B
G
B
A
All
right
right
along
we've
got
Jeffy.
Let's
talk
about
the
release,
notes
website,
hello.
I
So,
as
of
1:15,
we
are
split
brain
between
the
changelog
and
the
markdown
file
that
is
released
as
well
as
the
website
1:16.
We
will
be
going
completely
to
the
website
for
just
the
changelog,
the
you
know,
massive
list
of
PRS
that
are
typically
categorized
by
sig.
However,
we
have
all
these
little
dot
releases
in
between
115
and
116.
I
I
F
I
A
Yes,
I
think
excellent.
Yeah
I
think
that
we
would
want
a
delineation
of
it's,
not
it's
nice
to
have
a
rollup
of
what
115
zeros
and-
and
it
would
be
nice
to
as
expected
in
most
anal
changelogs,
to
have
a
delineation
between
the
the
patch
releases
right
so
having
a
separate
changelog
for
that
I
think
is
fine
personally,
but
so.
I
Now
that
we're
following
that
along
who
should
generate
that,
should
it
be
the
release
team
of
115
continuing
to
do
it
for
all
the
dot
releases?
Should
it
be
the
patch
release
manager?
Could
we
potentially
set
it
up
so
instead
of
a
nago
just
generating
the
markdown,
it
also
invokes
the
release,
notes
tool,
generates
the
JSON
and
creates
PR.
A
B
So
right
now
the
patch
release
process
includes
auto
generation
of
release,
notes
of
the
the
Delta
portion
and
I
think
that
ought
to
continue
the
specific
tool
in
in
doing
so.
I'm
I'm.
Okay,
with
that
being
changed,
and
for
a
patch
release,
it's
trivial
for
us
to
to
trial
something
and
show
that
it
presents
equivalent
enough
output,
I.
B
I
All
right
so
just
thinking
of
timing
for
115
one
we
will
probably
still
have
to
generate
those
manual
and
I
have
no
problem
tackling
that,
because
it
is
just
a
one-liner
and
that
buys
a
little
bit
of
time
for
us
to
try
and
actually
do
the
right
thing
and
get
it
into
the
release
tooling.
So
it's
automated
does
that
sound
good
yeah.
B
We
can
we
can
compare
like
we
can
run
the
current
one
and
run
an
alternative
one
iteratively
until
we're
happy
with
it.
It
doesn't
have
to
actually
happen
at
a
release.
Point
is
as
soon
as
one
14-0
ships.
We
will
have
a
PR
right
after
that,
that's
merged
onto
the
branch
and
at
that
point,
you'll
you'll
have
something
that
a
release.
B
A
K
K
So
the
process
for
that
right
now
so
for,
let
me
turn
on
my
video
really
quickly:
mariette
hi
everyone,
so
yeah,
I'm
Amy.
The
situation
for
that
is
image.
Promoter
I,
think
is
being
blocked
largely
by
ete
tests
and
so
currently
in
Linus,
Arbor
I
think
his
last
name
is
and
I
are
going
to
work
together
to
put
together
a
plan
of
what
kind
of
criteria
we
need
in
order
to
call
or
eating
a
test
done.
F
I
think
the
other
blocker
we
had
if
you
want
to
help
Amy
or
align
us
out
on
this,
come
on
down
the
jump
key
to
the
Kate
Center
working
or
the
other
blocker
was
we
can.
We
have
concerns
like
what,
if
something
went
wrong
and
the
promoter
accidentally
deleted
everything
in
the
repos
that
I
was
trying
to
sync
to
like
what
would
our
dr
or
recovery
strategy,
P
and
so
thorough
and
the
end
testing
is
one
prong
of
that
approach.
K
Something
to
add
to
that
that
Justin,
Santa
Barbara
I,
mentioned
in
our
last
infrared
grill
meeting,
is
that,
like
this,
space
is
still
very
like
malleable,
in
the
sense
that
we
can
add
criteria,
for
instance
like
restricting
the
the
abilities
of
the
image
promoter
so
like.
If
we
wanted
to
restrict
it
to
not
being
able
to
delete
things.
That
is
an
option
that
we
are
allowed
to
discuss
as
well
or
not.
A
lots
do
able
to
discuss
so
so.
A
Continuing
the
call
to
action
kind
of
the
image
like
the
the
container
image
promoter
right
now
is
one
of
those
things
that
will
unlock
a
lot
of
work
for
everyone.
So
if
you're
interested
or
if
you're
able
to
help
on
this,
please
attend
the
the
working
group
Cates
infrared
meetings,
a
huge
huge
thanks.
F
To
Amy
for
stepping
up
to
contribute,
it's
very
much
appreciated.
It's
also
kind
of
mutating
into
like
the
artifact
image
promoter.
At
this
point,
because
I
think
Justin
has
a
PR
to
like
add
support
for
promoting
arbitrary
binaries
and
then,
when
we
talk
but
like
we
talked
about
container
promotion
right
now,
I
think
it
can
only
pull
from
G
cRIO.
There
may
be
some
specific
api's
or
something
and
I
would
love
it.
If
somebody
wants
to
figure
out
a
way
to
have
that
talk
to
other
registries,
say
the
github
package
registry
or
seize.
A
And
one
more
time,
this
is
something
that
is
critical
to
the
project
for,
for
everyone,
like
you
know,
for
I
know
for
cluster
API
AWS.
We
want
to
be
able
to
publish
images
in
a
and
a
CNC
F
owned
repository
for
Azure.
We
wanted
to
be
able
to
do
that
to
like
that
is
blocked
on
container
image
promoter
right
now.
So
again,
if
you
are
able
to
help
with
this,
this
is
one
of
the
biggest
things
that
we
can
get
done
in
the
project
in
the
near
term.
So
a.
B
Wonder
if
that's
something
that
we
could
follow
and
in
multiple
tools
or
flows
here
as
opposed
to
deleting
and
but
so
you
have
you'd
have
to
prove
that
you
safely
moved
the
thing
to
a
separate
area
versus
delete
it.
But
if
there's,
if
the
code
definitely
has
no
deletions-
and
you
can
trust
that
the
move
happened
because
move
is
off
to
copy
and
delete
behind
the
scenes,
I
wonder
if
that
gives
us
a
little
safety
margin
or
if
there's
been
discussion
in
that
direction.
So.
I
One
thing
I'll
point
out
is
that,
even
with
like
test
logs,
we,
you
know
they
were
doing
all
the
time
we
detail
the
stuff
out
just
because
you
would
consume
near
infinite
storage
like
if
we
actually
had
very,
very
regular,
build
of
Cabernets
artifacts
like
can
actually
be
quite
a
bit
of
disk.
So
just
something
to
consider.
F
That
there's
been
concrete
or
documented
discussion
in
that
direction.
Tim,
but
yeah
I.
Think
that's
where
we're
all
sort
of
thinking
is
a
good
stopgap
is
to
like,
let's
not
trust,
the
robots,
to
delete
things.
Bad
things
happen
when
that
happens
automatically.
Maybe
let's
find
a
way
to
allow
humans
to
sort
of
periodically
review
what
is
safe
to
delete
and.
B
F
Great,
like
I,
will
say,
like
I'm,
really
appreciative
of
folks,
like
Hawking,
who,
whose
paranoia
meters
are
set
up
here,
whereas
I
would
rather
push
us
through
the
flood
so
that
we
can
like
merge
and
iterate
and
actually
use
this
stuff.
So
I
think
like
making
sure
we
have
that
conversation
and
come
to
an
agreement
is
great.
A
So
we've
yeah
we've
actually
had
questions
about.
Like
you
know,
can
we
can
we
with
the
tooling
that
exists
today
at
least
forgive
her
for
debs
we've
had
questions
about
like
oh
well,
can
I
provide
my
own
down
like
download,
link,
base
rate
or
you
know
or
choose
where
I
push
these
things
to
write.
Part
of
the
like
part
of
the
publishing
processes
is
not
visible
to
us
right
now,
so
I've
been
I've
been
working
on
like
retooling,
some
of
it,
which
it's
kind
of
in
its
infancy.
But
if
you
want
to
see
that
yeah.
F
I've
noticed
you
and
Tim
are
working
on
overlapping
things.
I
will
say
to
that
point
like
I.
Don't
know
that
we
have
an
actuator
discreetly
to
talk
about
repositories
for
apt
and
Debian
packages
right
now
in
the
context
of
the
working
group,
but
like
you
can
need
to
stand,
one
up
like
come
talk
to
us
and
we
can
make
sure
you
have
the
space
to
do
that
and
I
know
you
do
okay,
great,
an
awesome.
I
F
F
The
way
we
can
unblock
ourselves
is
to
make
sure
you
have
the
right
playground
so
that
you
can
demonstrate
hey,
look
ma,
I
can
build
a
package,
all
the
way
from
source
and
I
can
push
it
and
I
can
promote
it,
and
then
you
know,
then
it
all
it
takes.
Is
the
figuring
out
the
signing
bit
and
flipping
some
things
around,
but
so,
let's
make
sure
you
have
the
room
to
develop
everything
you
need
to
I.
Think.
A
So
I
also
I
dropped
the
pole
on
Twitter
just
for
funsies
as
we're
releasing
the
115
RC.
Just
ask
a
question
to
the
group
like
who
is
actually
consuming
alphas
betas,
our
seas
of
kubernetes,
like
as
it
exists
today,
right
and
and
over
50
percent
of
the
poll
said
they
didn't
know.
They
existed
right
so
moving
us
to
a
point
where
we
can
enable
people
to
consume
pre
releases
of
kubernetes.
Like
brings
us
to
a
point
where
we
can.
We
can
test
that
we
can
have
more
people
testing
this
before
it
gets
out
the
door.
F
Would
this
be
in
the
context
of
reworking
the
Debian
and
RPM
packages
to
to
build
packages
of
alpha
and
beta
builds
partly
yeah,
because
I
think
like
what
happens
today,
correct
me:
if
I'm
wrong
is,
we
do
actually
cut
alpha
and
beta
builds,
but
we
don't
actually
create
Debian
packages.
That's
correct,
Rana.
H
F
A
There
was
a
guy
who
who
said
that
on
Twitter,
essentially
I
was
like
that's
cool.
That's
not
the
question
I
asked
so
moving
on
he's
like
why
would
I
ever
test
in
production
and
like
okay,
alright
you're,
not
the
audience,
I
was
looking
for
this
charity
major
says
you
have
to
and
she's
right,
I
literally
like
blink
to
charity
measures
like
article
afterwards.
A
Anyhow,
anyhow,
I
think
I
think
it's
awesome
that
we
have
movement
in
this.
We
are
you
know,
sig
leads
have
to
sit
down.
I
have
a
bunch
of
project
boards
up
that
we've
put
together.
Some
of
the
stuff
is
curated
at
this
point.
Some
of
it
needs
to
be
reworked
for
116
I.
Think
release
engineering
is
the
name
of
the
game
for
116.
Personally,
like
a
personal
focus
of
mine,
I
know
it's
the
same
on
Tim
and
Caleb's
mind.
A
A
Will
be
a
dedicated
meeting
for
release
engineering
I'm
just
waiting
until
we
close
the
books
on
115
I
have
to
spin
up
the
release,
the
licensing
sub
project
meetings,
as
well
as
the
release
engineering
sub
project
meetings,
but
there
will
be
emails
and
all
that
stuff
and
doodles
out
to
make
sure
that
we
find
appropriate
times
for
people
meet.
So
no
but
tuned
are.
B
A
A
D
We
want
to
actually
see
how
we're
doing
particularly
like
I
have
some
questions
where
we
had
a
couple
of
sections
where
none
of
the
shadows
were
ready
to
step
up
and
become
leads
this
time
around
and
I
want
to
find
out
whether
there
was
you
know
some
kind
of
failure
and
process
there
or
whether
it's
just
how
things
worked
out
the
and
so,
as
as
part
of
this
I
thought,
we
should
now
start
actually
making
it
standard
to
ask
both
leads
and
shadows
how
the
release
went
from
a
mentoring
perspective
like
we
get
a
retro
on
how
the
release
went
from
a
release
in
kubernetes
perspective,
but
we
don't
really
get
a
feeling
of
how
how
how
did
mentoring
go,
how
did
training
people
up,
go,
etcetera,
etc?
D
A
Josh
I
know
that
you
are
going
to
be
traveling's,
yeah
I'll,
be
in
Shanghai,
okay,
right
right,
all
right,
so
that
will
be
out
soon,
but
we'll
also
be
doing
something
to
keep
in
mind
if
you
are
a
lead
or
you're
moving
into
lead
position
for
the
next
cycle,
we're
I
think
I.
Think
one
of
focuses
should
be
looking
at
people
who
are
going
to
be
willing
to
lead
in
the
next
cycle.
That's
one
of
the
questions
on
the
survey
yeah.
D
D
We've
supplied
a
lot
of
guidance
there
to
people
who,
who
are
new,
leads
Pinkley
if
they
didn't
come
in,
be
the
shadow
process
the
and
so
so
some
of
the
things
that
I'm
asking
about
in
terms
of
questions
like
did
you
know
if
you
look
at
those
questions
of,
did
you
do
this
or
that
or
the
other
thing?
These
are
things
that
we
can
actually
put
somewhere
in
the
release
team
documentation
to
say
hey
these
are
all
things
shatter
should
be
doing.
One.
B
A
You
know
one
of
the
questions
I
think
around
like
the
docs
handbook
or
something
was
you
know
if
you
like,
you
had
to
be
like
I,
think
it's
specified
that
you
had
to
be
an
org
member
to
be
a
shadow
right
and,
like
I,
opened
an
issue
for
that
and
said
like
you,
do
not
have
to
be
an
organ
member
to
like.
We
hope
that
you
were
moving
towards
that
right,
but
you
know
that's
something
that
someone
said
they
didn't
feel
like.
They
could
fill
out
the
shadow
application
because
they
were
an
ardent
org.
A
A
D
Of
the
hand,
books
do
have
a
requirement
section
now
that
was
one
of
the
things
that
that
I
guided
during
the
last
cycle
is
to
make
sure
that
that
requirement
sections
got
added
to
all
of
them,
but
that
doesn't
mean
that
it
doesn't
mean
that
the
that
the
list
in
that
requirement
section
is
actually
good.
If
you
follow
me,
oh
for
sure,
I've
ever
viewed.
A
D
A
A
F
A
thought
on
a
pattern
of
shadow
engagement
that
I
have
enjoyed
witnessing
would
be
the
shadow
sort
of
looks
at
their
responsibilities,
that
the
lead
is
handling
and
figures
out
which
parts
of
those
problems
could
be
better
solved
by
a
tool
or
automation.
I
think
Kathryn
did
such
a
good
job
with
that
that
we
sort
of
eliminated
a
whole
bunch
of
human,
toil
and
I
feel,
like
Jeff,
has
done
a
phenomenal
job,
sort
of
transforming.
What
release
notes
are
and
I
think
that's
sort
of
a
great
way
of
encouraging
that
pattern.
F
This
is
just
my
account
like
it's
amazing
that
we're
onboarding
all
these
people
and
we're
adding
more
people
to
the
mix,
but
I
also
just
feel
like
this
is.
This
is
a
lot
of
cooks
in
a
kitchen
and
the
more
like
clarity
and
focus
we
can
provide
to
this
process
I
feel
like
the
more
repeatable
it
will
be
and
I
think
like
people
showing
up
willing
to
help
out
with
that
are
super.
A
Appreciated
I
think
I
think
that
part
of
it
is
also
we
want
to
see
some
level
of
tenacity
with
our
shadows
right
there.
You
know
when
I,
not
to
like
tell
my
own
story
but
like
when
I
was
shadowing.
Igor
I
basically
joined
the
release
team
and
said
like
that,
is
the
guy
who's
been
doing
it
consistently
for
a
thousand
cycles,
and
he
has
had
no
help
right.
So
that's
the
role
I'm
going
to
go,
do
and
I
basically
looked
at
like
dude.
What
are
you
doing
like?
A
A
B
I'm
hearing
Aaron
say:
is
that
the
goal
in
that,
though,
should
be
not
that
you
actually
take
it
on
any
longer
than
necessary
for
you
to
understand
whether
you
can
automate
a
portion
of
it
like
it's
not
about
transferring
the
job
from
person
to
person,
but
each
time
we
do
that
spreading
the
knowledge
so
that
the
people
can
build
more
automation
around
it.
Yeah
I
was
making
a
separate
point
entirely.
F
A
F
Like
see
like
I
have
seen,
features
leads
have
like
an
army
of
shadows,
go
out
and
sort
of
in
parallel
engage
with
a
bunch
of
things,
and
so
that
way
the
toil
gets
spread
between
people,
which
is
cool
but
like
to
the
point
of
a
and
his
wonderful
spreadsheet
that
he
started
way
back
when
we're
still
using
a
spreadsheet
right.
How
many
of
us
really
enjoy
the
experience
of
manually,
updating
a
spreadsheet
and
talking
to
humans
and
stuff
yeah.
F
Are
the
sorts
of
things
that
I
feel
that,
if
I
look
at
the
ideal
engagement
with
a
shadow,
it's
the
areas
where
the
lead
has
given
the
shadow,
sufficient
space
and
understanding
if
they
are
so
inclined
health
improve
the
process
right
there?
It's
by
by
taking
some
of
that
tribal
knowledge
or
whatever
and
encoding
it
into
a
tool
that
we
can
use,
which
is
awesome
so
like?
Thank
you
to
the
people
that
I
mentioned
I.
E
Cycle
yeah
thanks
for
doing
that,
Steven,
basically
I
just
put
a
proposed
schedule.
If
there's
anybody
who
has
an
interest
in
looking
at
it
and
signing
off
on
it,
it's
up
there
for
conversation
and
you've.
Also
put
this
the
shadow
survey.
Application
is
going
to
close
out
this
Friday,
so
we're
plugging
it
predominantly
via
Twitter
and
I.
Think
I've
seen
some
things
in
the
community
slack
as
well.
So
if
you
know
anybody,
please
encourage
them
to
sign
up
to
be
a
shadow
and
we
will
commence
selection
next.
A
A
C
Can
introduce
myself
but
since
I'm
talking
tomorrow
anyway,
at
the
at
the
ball
or
not
tomorrow,
on
Thursday
but
yeah
anyway,
my
name
is
Christine
I
work
at
a
cotton
company
called
the
node
and
we're
spinning
up
our
own
Carini's
engine.
So
that's
that's.
Why
I've
been
getting
more
involved
in
in
the
SIG's
in
the
community
and
and
slack
and
all
that
stuff,
so
yeah,
that's.
K
I
guess
I'll
follow
Christine
as
well
as
hey
everyone,
so
I
I
think
I
like
spoke
earlier
in
the
meeting,
so
my
name
is
Amy
Chen
I
work
at
VMware
I'm.
It
system
software
engineer
here
and
currently
internally
I
target
the
upstream
cluster
API
efforts
and
that's
also
partially
the
reason
why
I'm
working
on
helping
unblock
the
image
runner
because
we
would
love
to
have
that
functionality
and
not
have
to
bother
just
in
Santa
Barbara.
Every
time
we
can
want
to
make
a
release
as
well.
So
it's
good
to
me
all
of
you.
J
Yeah,
maybe
I'll
do
the
follow
up
heights.
First,
the
west
hotel
with
no
Wi-Fi.
So
sorry
me
to
save
bandwidth.
I've
been
poking
around
to
cough
space,
but
yeah
mirroring
or
shadowing
Cherith
for
the
last
release
so
very
exciting.
To
also
see
the
rebuild
of
all
the
release
tooling.
At
some
point
in
time,.
A
Okay,
so
stay
tuned
for
updates
on
the
what's
happening
with
115
release.
Hopefully
we'll
have
something
in
the
morning
morning:
Eastern
Time,
maybe
morning,
someone's
time
I,
don't
know
the
release
retrospective
is
going
to
be
on
Thursday
during
the
community.
Reading,
that's
1:00
p.m.
Eastern.
Any
questions
ask
the
mailing
list
cig
release
all
that
good
stuff
feel
free
to
reach
out
to
the
leads.
Personally,
if
you
need
to
have
a
great
day
and
see
you
in
two
weeks,
okay,
happy.