►
From YouTube: Kubernetes Community Meeting 20170302
Description
We have PUBLIC and RECORDED weekly video meetings every Thursday at 10am US Pacific Time.
https://docs.google.com/document/d/1VQDIAB0OqiSjIHI8AWMvSdceWhnz56jNpZrLs6o7NJY
Demo: Flux; Releases 1.6, 1.5, 1.4; SIG Scalability; SIG API Machinery.
A
A
A
B
C
B
B
Now,
in
this
in
the
program
there's
going
to
be
some
web
user
interface
component
and
those
not
necessary
to
use
flux,
they're,
just
the
bits
of
Club
and
I'm
going
to
use
to
just
shows
up
off
of
it.
Ok,
and
so
before
I
arrived
here,
I
did
a
bit
of
setup
stuff,
and
some
of
our
kids
is
stuff
that
you
would
probably
all
loose
yourself,
and
that
is
mainly
committee
structure
and
I
put
end
on
it
and
the
app
is
so
the
deferred.
B
You
can
see
their
expertise,
Nikki's
crust
for
30,
so
Trey
exciting,
and
it's
in
the
act.
So
there's
just
a
bunch
of
pods.
There
are
some
of
what
you're
talking
to
other
one
and
what
is
actually
doing
is
serving
this
sock
shop,
which
is
a
just
a
demo
dick
and
contain
the
server
cruiser
and
we've
work
go
together,
and
so
it's
actually
fully
functional
in
quite
good,
but
using
your
demos,
because
you
can
do
all
the
sort
of
stuff
edit
to
a
card
when
I
think
our
that
sort
of
stuff.
B
Something
they
did
to
set
up.
First
of
all,
the
all
the
manifest.
C
E
B
B
B
So
can
you
ask
it
to
list
all
the
things
that
can
see,
so
you
can
see
that
kind
of
correspond
to
a
bunch
of
stuff,
that's
running
in
minutes
if
soft
rock-
and
it
also
tells
me
and
rich
of
those
things
and
helps
me
their
containers
and
which
images
those
containers
are
running
am
main
thing
we
want
to
do
with
flux,
usually
is
to
updates
images,
usually
off
the
end
of
a
CI
pipeline.
William
fill
a
new
one,
so
lots
of
this
automation
without
doing
that.
B
B
Ok,
so
that's
telling
us
that
there
are
a
bunch
of
images,
but
it
could
be
running
in
its
ergonomic
test
will
feature
about
the
metadata
of
at
least
from
crab
and
we're
actually
quite
far
behind
on
a
version
that
we're
running
just
amazing
ascending
order
of
the
most
preaching
to
the
least
so
I'm.
Behind
I'm
going
to
release,
I'm
just
going
to
say
for
this
service
truck.
B
B
B
Term
clinton
change
this
file
so
that
it
had
set
aside
own
image.
It's
like
it's
got
anyone
so
that's
cool,
I'm
in
it
there
then
you
go
here
and
probably
will
actually
see
any
differences,
because
I
don't
know
what
change
between
these
images
park
at
the
sands
that
we
discovered
that
actually
the
latest
thing
each
was
terrible
and
we
have
to
go
back
to
a
prior
image.
F
G
B
Wow
and
logos
to
smiles
on
each
other,
and
so
that's
really
the
essence
of
it.
There's
some
kind
of
two
things
a
little
making
their
one
is
making
sure
that
can
you
update
energy
as
it
goes
smoothly,
so
it
changed
it
in
the
right
place
in
the
file
and
figures
out
to
be
there
and
I'm
committed
back
to
get
up
and
the
other
one
is
making
sure
that
that
is
also
sync
with
what's
running
in
your
cluster.
B
B
This
is
really
just
extra
extra,
impressive
stuff
to
make
sure
that
they
moved
from
one
kind
of
known
space
to
another
man's
place
and
the
next
things
we're
working
on
with
clocks
are
instead
of
jumps,
responding
to
genuine
image
versions.
Actually
deploying
new
can
take
when
it
gets
committed
safety.
Major
PR
conti
actually
deploying
that
as
well
automatically.
So
that
so
there's
no
excuse
for
having
things
in
sync
would
get
ya.
D
B
E
Thank
you.
That
was
a
great
demo.
Thank
you
also
to
Jason
for
a
volunteering
to
be
the
note
taker
for
the
work
here
think
we
can
get
into
our
next
topic
here,
which
is
releases
I.
Do
we
have
our
release
manager
Dan
on
the
line
yeah.
F
Everyone
give
you
think
her
good
morning,
depending
on
where
you
are
okay:
let's
just
get
right
into
it.
The
cookie
started
on
Monday,
which
means
that
after
this
place,
the
only
thing
that
she's
being
merged
our
bugs
for
bugs
basically
that
target
home
plate
six,
the
cookies
is
a
hard
deadline,
so
anything
that
was
lgtm
before
the
code
freeze
will
be
merged.
F
We
really
seen
the
FDA
move
back,
because
the
way
that
we
were
doing
it
really
is
dependent
on
features
not
changing
after
a
certain
point,
so
this
I
appreciate
the
large
number
of
exceptions
under
files.
This
is
the
largest
number,
if
exceptions
we
have
gotten
before
sort
of
eventually.
So
we
really
appreciate
people
games
is
dim
and
those
will
be
reviewed
tomorrow.
Sd
milestones
burn
down
meeting
just
for
everyone
who
has
been
such
and
they
should
be
aware
that
they
are
only
rented
historically
am
very,
very,
very
rare
circumstances.
F
I
believe
41.5
only
a
single
deductions
or
scratches
so
just
be
aware
that
probably
your
exception
will
be
denied
and
it
will.
It
will
be
position.
We
bait
on
complexity
and
but
generally
a
proper
will
be
snide,
just
Tessa,
and
so
in
any
issues
with
this
process
should
definitely
be
raised
at
retrospectives.
I
were
constantly
revising
the
way
we
do
things
I'm
just
doing
I'm
just
trying
to
run
code
freeze,
as
it's
been
described
to
me,
but
totally
in
future
releases,
totally
open
to
having
different
processes.
F
Cool
I
think
we
have
a
few
things
chat,
cory.
Gunz.
The
question
was:
what
is
an
lgtm
was
removed
after
every
day,
so,
just
working
off
of
what
other
release
manager
step
says:
curacy
is
it
for
something
like
a
dozen
lgtm
was
removed.
The
mess
totally
buying
orb
is
for
a
very,
very,
very
clear,
specific
feature.
Vectors
are
like
fun
fix
that
scopes
and
its
complexity.
That's
fine!
Is
it
something?
Where
is
unstuck
towards
actually
what's
wrong,
then
the
LG
p.m.
H
So
quick
question:
how
are
we
separating
out
sort
of
features
that
didn't
make
it?
This
is
joe
features
that
didn't
make
it
verses
of
like
because,
right
now
we
have
an
issue
where
the
docker
version
number
changed.
It
was,
you
know
a
little
bit
of
a
surprise
and
the
preflight
checks
for
cube
admin
don't
catch
it.
So
is
that
something
we
should
file
an
exception
for
or
considered
a
bug,
I
mean.
How
are
we
doing
that
I.
F
Think
something
like
that
could
be
considered
above
but
I
think
if
it's,
if
there
is
some
doubt
Charlie
better
to
Solomon,
section
or
bus,
but
we're
nearing
the
end
of
that
period
to
hopefully
get
radiation
as
possible.
I
be
said
that
we
would
not
accept
them
past
eleven
last
night
but
specific
time
but
yeah.
If
there's
anything,
please
try
to
get
something
as
soon
as
possible,
we're
so
sorry
11
tonight
or
service,
ok,
open
middle.
F
F
Raise
them
now
our
detection
process
and
it's
better,
just
a
sim
sorry
to
play
in
today's
lecture
process.
Some
of
these
things
that
are
quick
totally
bugs
and
they're
Buckland's
also
plug
we're
not
trying
to
like
Nick's
here,
but
it's
just
for
trying
to
like
make
sure
sure
that
it
is
actually
pretty
sure
worth
it
and
should
be
a
little
bit
of
it.
F
G
F
Yeah
I
think
it's
a
little
less
clear-cut,
see
if
there's
several
things
that
I
would
consider
features
that
are
in
the
features
here
photo
special
run
changes
I
think
that
we're
going
to
have
to
make
a
case
by
case
to
answer.
Basically,
it's
a
really
good
guy,
but
in
general
this
is
in
the
beach
retreat.
So
that's
definitely
that's
for
sure.
There's
a
few
things
that
like,
for
example,
with
your
adding
test,
is
sort
of
depending
on
how
many
texts
you're
adding
best
to
go
each
way.
So,
yes
and.
F
Yeah
and
it's
not
an
intention
to
take
away
in
lgtm
for
every
base
and
if
we
are
doing
that
we'd,
let
us
know
that
wasn't
intentional
we're
just
trying
to
manage
yesterday
there's
and
if
we
remove
you
from
the
milestone
yesterday
and
you
should
be
in
it-
was
I'm
special
they're,
just
96
open
pr's
in
the
milestone
that
we
had
to
sort
of
get
through.
So
please
reopen
it.
Wasn't
amazing,
first
off
I,
step
cool.
E
I
have
a
quick
question:
I'm
hoping
maybe
Brian
grant
can
eat
you
real
quickly,
although
perhaps
we
should
take
this
off
on
and
discuss
more
in
depth
with
the
next
meeting,
it's
specifically
around
the
use
of
priority
labels
versus
key
labels,
so
at
the
moment
there's
a
document
that
lays
out
all
the
labels
that
should
be
applied
to
issues
and
their
things
like
priority.
Critical,
urgent,
I,
already
important
things
of
that
nature.
E
But
if
you
go
to
the
submit
keys
page
and
read
the
algorithm
that
it
uses
the
short
thing,
it's
the
refers
to
things
like
priority
t0
and
priority
Tibor.
It
seems
like
the
github.
The
lunch
github
program
hasn't
been
updated
to
use
the
priority
labels.
So
when
I
asked
about
that,
brandy
response
was
well
really.
We
should
be
using
a
cue
labels
which
I'm
having
difficulty
finding
a
document
for
shortest
the
TR
that
Brian
putted
me
to
an
Incan
trip,
but
like
I'm,
just
sort
of
unclear
like
what
are
the
labels
were
using
going
forward.
E
I
I
can
speak
to
that.
There
was
a
discussion
back
in
December
about
moving
away
from
priority
labels.
Unfortunately,
to
work
was
rotted
to
not
get
done
by
anyone,
so
the
flake
automatic,
flake
spyler
and
priority
updater
still
uses
the
old
party
labels,
and
so
does
this
make
you
I
still
think
we
should
finish
both
of
those
updates
and
move
to
the
new
model,
in
particular
flakes
I
think
the
party
labels
are
totally
useless.
We
just
need
monitoring
for
flight
frequency
and
that
so
stuff,
our
data
source,
most
of
the
flake
issues,
don't
get
attention
anyway.
I
So
that's
a
whole
separate
discussion,
the
as
far
as
gaming,
the
priority
queue
and
the
reason
we
discussed
moving
to
the
new
labels
is
to
make
it
really
clear
what
the
stairs
were,
where
it
was
appropriate
to
actually
use
the
labels
and
give
us
freedom
in
this.
Make
you
to
decide
what
the
proper
ordering
should
be
I,
don't
believe.
People
should
be,
in
general,
bumping
the
priority
of
their
issues
to
get
them
in
sooner
than
other
people.
I
We
should
be
fixing
the
McCue
stability
and
velocity
problems,
so
I
think
right
now,
it's
you
know:
I
don't
have
anyone
who
has
volunteered
to
go,
implement
the
changes
in
doing
it
in
the
middle
of
the
release,
probably,
as
you
suggested,
probably
not
the
best
idea,
but
417
I
think
we
should
just
cut
over
and
make
a
change
in.
I
didn't
actually
fix
the
root
cause
or
the
problems
and
stop
manually
scheduling
the
queue
and.
E
I
agree
with
all
that
completely
I
think
my
my
question
here
is
just:
can
we
all
agree
to
use
it
the
same
set
of
labels
in
the
same
way,
and
maybe
that's
the
discussion
we
need
to
take
to
the
burndown
meeting
to
decide
what's
most
effective
for
Dan
to
help
triage
the
queue,
but
I
just
wanted
to
raise
it
here,
because
it's
kind
of
it's
been
unclear
to
me.
I'm
not
sure
how
many
other
people
have
been
looking
at
it.
Yeah.
D
E
Cool
can
move
down
to
features
in
a
second
just
one,
last
note
on
the
release
here,
the
subjection
and
on
flakes
we've
been
collecting
necessary.
F
E
I
E
And
educate
is
just
a
special
case
of
the
general
case
of
how
do
you
assign
proper
ownership?
For
you
know,
cloud
provider
like
vendor
mental
level,
test
I
know,
as
we
see
more
as
your
issues
being
filed,
similar
kind
of
problems,
there's
no
necessarily
central
point
of
contacting
their
upsides
to
Brandenburg
who's.
Not
here
today
he.
F
Yeah
just
yeah:
we
moving
to
risk.
Yes,
so
just
identifying
a
person
there.
Okay,
oh
yeah,
moving
into
risks,
biggest
risks.
I
see
right
now,
for
the
reason
is
that
automated
detesting
are
the
best
still
being
set
up.
So
we
don't
have
a
clean
signal
there.
Yet
there
is
frequent
flaking
of
certain
on
upgrades
teach
that
I
also
threaten
the
rose.
E
Cool
yeah
we've
been
trying
so
PR
went
in
for
the
chi.
I
bought
that
labels
way
to
now
start
adding
the
label
for
six
based
on
what's
in
cash
pet
owners
at
csv,
we've
also
gone
back
and
added
the
same
labeling
based
on
that
math
to
the
existing
backlog
of
wait.
So
we
plan
on
publishing
a
week
of
a
week
reports
on
flakes,
so
you
should
ideally
and
we'll
move
this
to
get
heavy
she's,
probably
fairly
seen
here.
E
But
if
you
are
part
of
a
cig,
you
should
ideally
look
at
that
help
out
cup
closed
out
old
flakes
because
it
makes
it
more
difficult
to
try
and
get
a
good
signal
on
the
health
of
the
release
with
all
these
old
site.
Also,
it
does
it
to
be
a
store
general
problem
of
having
far
too
many
issues
to
have
any
one
reason
about
sensible,
epizyme
trying
anymore
with
our
5,000
plus.
E
D
Have
one
quick
point:
we're
not
actually
paying
a
lot
of
flakes
on
a
16
jobs,
yeah
I,
think
the
frequency
of
execution
isn't
high
off
the
trigger
down
because
it's
random.
In
some
cases
we
are
seeing
flakes
on
the
cereal
job
and
in
one
case
we
actually
have
a
hard
failure
and
I
guess
we
have
to
figure
out
a
way
to
fix
this,
but
the
problem
is
it:
some
tests
only
run
on
the
cereal
John
so-called
emerge
shots
cost
the
pr
merges.
D
You
could
actually
broken
something
in
cereal
job
and
we
don't
actually
know
that
until
we
pay
attention
to
that.
So
I,
don't
think,
is
a
good
answer.
I
just
like
people,
maybe
if
you're
watching
the
jobs
for
flakes
the
cereal
jobs
or
maybe
a
little
bit
more
important
that
release,
because
we
could
actually
indicate
progression.
Nansha,
spikes
in
exams,
I'm
also
looking
for
our
help
for
a
couple
of
flakes
that
are
fairly
they're,
not
like
super
consistent.
They
are
flakes,
but
they
exist
in
master
and
16.
D
One
of
them
I
I'd
be
interested
in
hearing
whether
people
think
it'd
just
be
disabled,
actually
failed
all
the
time.
That's
the
restart
API
server
one
it
never
passes
in
the
cereal
job,
so
the
other
one
is
the
reschedule
one
fails
often
enough.
It
probably
deserve
some
attention.
I'm,
assuming
that
the
scheduling
thing,
if
someone
from
chick
scheduling,
maybe
could
look
into
that.
J
F
J
F
E
Yeah
we
do
have
this
general
problem
of
the
occasional
undefined
your
sig
or
yourself,
I.
Think
there's
a
there's
also.
Please
look
at
the
dock
that
link
they're
working
to
develop
metrics
around
flake
we're
running
birds
have
proposed
and
SLA
for
flake,
because
the
voluntary
you
know
triage
has
not
seemed
to
be
effective.
So
the
proposal
links
in
what
are
the
submit
you
will
be
running
given
how
many
flakes
are
currently
detected
by
the
system.
E
G
G
G
We
also
like
to
I'd
like
to
have
a
call
for
their
future
illness
to
start
working
on
the
release
notes.
But
thank
you.
Mother,
a
table
is
not
that
we
will
add
to
the
final
release
and
also
would
like
to
highlight
the
features
that
going
to
be
highlighted,
but
by
a
marketing
activities
that
are
going
to
be
added
to
the
blog
post
and
soul.
G
So
if
your
feature,
oh
man,
if
your
feature
place
prepared
teachers
notes,
there
is
not,
and
also
please
start,
looking
at
the
documentation,
70th
another
item
that
I'd
like
to
highlight
that
we
until
they
we
had
the
DPM
medium
and
one
of
the
gifts
that
we
have
faced
it
they're.
All
the
6pm
representative
is
17
like
absolutely
clearly
defined
for
everyone.
So
we
would
like
to
have
all
this
folder
six
represented
at
the
p.m.
meetings
or,
if
you're
not
able
to
represent
a
secret
DP
ambitions.
Please
reach
out
directly
and
just
being
touch.
H
G
L
Hey
guys
and
we're
having
echo
problems
with
these
are
fixed,
so
I
just
wanted
to
come
and
I've
been
going
up
to
that
meeting
on
bath
scale
at
least
occasionally.
I
think
that
there
is
a
pretty
big
mismatch
between
the
model
that
you're
proposing
and
the
expectations
of
the
techniques,
and
this
is
definitely
a
covenant
issues.
It
requires
more
a
lot
more
discussion.
I
think
my
suggestion
would
be.
Thank
you.
L
L
H
Want
to
back
that
up,
I
think
like
we
should
really
rotate
on
features
and
who's
actually
leading
to
get
a
feature
out,
and
we
should
be
talking
and
presenting
on
that
instead
of
cigs
I.
Think
cigs
are
a
great
place
for
folks
to
meet
together
and
organize
meetings,
but
I
would
love
to
see
us
instead
of
having
cigs
present
at
this
mean
I'd
love
to
rotate.
That
around
features
because,
like
there
are
features
going
in
that,
like
I,
had
no
idea
that
they
were
happening.
They
were
never
mentioned
in
this
meeting.
H
They're
impacting
the
API
I.
Think
it's
a
great
opportunity
to
share
the
good
work
that
people
are
doing
so
that
everybody
knows
the
stuff,
that's
happening
in
the
release
tabs
it
happen
and
I
don't
want
to
follow
anything
specifically
but
like,
like
you
know,
and
a
lot
of
the
figs
actually
never
presented
this
group.
Never
actually
talking
and-
and
it
essentially
work-
you
know,
I
wouldn't
say
hidden,
but
it
operated
from
the
rest,
the
community
so.
A
So
I'm
working
fairly
diligently
to
make
work
our
way
through
all
six
making
sure
they
are
reporting,
Intel
experiences
and
we
have
features
aligned
with
sig
slated
become
in
those
pictures
and
siggy
reporting
makes
sense,
but
that
we
need
to
be
more
diligent
about
me
and,
as
the
features
are
reported
on
on
the
song,
become
models.
If
I
sorry,
the
echo
is
driving
me
nuts
I'm
not
I've
needed
you
again,
it's
out
of
Pepe.
A
Oh
it's
why
we've
been
battling
yes,
so
that
did
we
need
to
be
more
diligent
about
making
sure
features,
get
demo
door
features
get
reported
on
as
opposed
to
just
what
is
this
big
doing
and
the
person
who's
thinking
about
is
I'm
answer
so
I
think
there's
a
balance
there,
because
I
I
don't
want
us
to
pivot
around
futures
in
the
same
way
that
you
just
described
like
moving
cigs
away.
We
actually
have
creatures
designed
to
figs,
so
we
just
need
to
make
sure
that
the
features
are
reporting
up
properly.
I.
M
But
it
sort
of
completes
the
two
things
because
it
sounds
like
the
problem
is,
there
are
actually
features
that
are
happening
in
the
release
that
the
cig
leads
themselves
aren't,
aware
of
which
is
very
surprising
because
generally
the
cigs
is
meeting
and
whatever
features
members
of
that
thing
are
working
on.
My
assumption
is
and
I
think
all
our
assumption
is
that
those
are
getting
discussed
in
those
big
meetings,
so
there
shouldn't
be
anything
totally
new
for
the
big
lead.
L
We
saw
this
at
the
lasting
p.m.
meeting,
where
no
one
needs
a
status.
You
couldn't
actually
go
and
read
the
status
because
people
haven't
been
keeping
up
with
their
documentation.
So
all
of
these
problems
kind
of
tie
together
putting
a
bunch
of
load
on
the
sig
ep
em
on
sorry
I'm.
The
sig
leads
try
to
cover
that
gap.
I
think
it's
really.
I
did.
M
M
Again,
I
don't
think
we're
expecting
the
sig
leads
to
know
every
feature
in
detail
now.
This
is
why
we
have
been
trying
to
have
a
PM
reps
work
with
the
fig
leaves
for
every
for
everything
and
part
of
the
gap
is
that
we
haven't
got
a
PM
rep
for
every
state.
I
think
there's
one
or
two
six,
where
we
need
to
fill
that
and
I
think
we'll
be
able
to
do
that
in
the
next
two
weeks
and
that's
where
we
had
a
gap.
Also
in
the
PM
group
meeting.
M
That's
one
I
agree
with
you
about
prophecies
and
particularly
on
the
template.
The
template
has
been
overly
overly
complicated,
so
we're
trying
to
simplify
that
as
well,
but
on
governance,
I
think
I'm
happy
to
have
a
discussion.
I
think
we
should
have
a
discussion,
though,
are
you
trying
to
say
something.
H
Yeah
we're
super
actually
and
I
apologize
to
everybody
for
that
so
I'm,
going
to
try
and
be
explicit
when
I
talk
here
and
not
not
cause
problems
for
folks
at
this
point
that
I
was
trying
to
make
here.
Is
that
again,
there's
a
ton
of
features
that
are
going
in
that
essentially
get
no
exposure,
even
if
they
get
exposure
in
the
PM
group,
they
aren't
known
by
the
wider
community
and
again
I.
Don't.
H
Specifically,
but
I
would
love
to
see
as
part
of
our
feature
processor
requirement
that,
as
a
feature
is
proposed,
it's
at
least
mentioned
right
as
the
design
is
locked
down.
We
essentially
maybe
have
a
design
meeting
that,
like
we,
we
advertise
and
everybody
can
come
and
actually
see
and
understand.
What's
going
on,
as
the
feature
is
going
to
alpha,
we
actually
talked
about
it
and
we
get
a
demo
in
the
meeting
right
same
thing
as
it
moves
the
data
and
GA.
H
E
E
I
agree
with
that
in
the
abstract,
however,
there's
something
on
the
order
of
30
features
in
this
release
that
we
were
so
if
we
were
to
dedicate
a
10
minute
demo
slot
for
every
feature
for
this
meeting
like
he
wouldn't
get
through
all
the
features
in
time
for
the
release
right
so
sig
p.m.
is
actually
a
fig
itself.
It
seems
like
they
could
probably
be
reporting
out
at
convenient
checkpoints.
So
I
know
that
when
they
went
through
the
there
was
like
a
two
part
series
of
meetings.
E
I
was
the
1.6
feature,
freeze,
meeting
I,
think
where
you
know
we
decided
like
what
was
in
and
out
I
personally
didn't
attendance,
but
I
know
there
were
video
recordings
that
were
posted
to
some
mailing
list.
Someplace
I
think
maybe
the
disconnect
is
that
was
broadcast
to
this
community
seems
like
it
would
have
been
really
awesome
to
have
just
like
a
quick
five
minutes.
Ten
minutes
summary
from
some
representative
from
6pm
to
say:
hey
here's,
what
we
think
1.6
looks
like
all
right,
I,
really.
M
Like
somebody's
comments,
especially
just
the
idea
that
you
just
worked
in
my
mind,
is
that
we
could
use
the
PM
meetings
to
have
demos
for
especially
the
features
that
may
be
falling
through,
and
maybe
only
some
of
them
graduate
to
this
meeting,
because
we
think
that
they're
crossing,
and
so
those
can
graduate
here.
For
the
rest,
we
can
do
a
summary
for
this
community
meeting,
but
that,
but
I
do
want
to
point
out
that
that
puts
a
burden
on
future
owners.
M
N
E
Of
the
different
kind
of
discussions
here
and
would
just
suggest
an
interest
of
time
to
talk
about
these
and
some
of
the
governance
work
that
is
currently
ongoing
or
on
the
contributor
experience
and
sig
p.m.
mailing
list.
If,
if
people
are
okay
with
that,
I
just
get
through
the
rest
of
the
agenda
for
today
well.
C
Know
a
forum
like
this
is
a
lousy
place
to
design
buildings.
You
know
we
really
need
to
get
a
smaller,
more
focus
group.
That's
willing
to
put
a
lot
of
time
and
effort
into
structured,
governor's,
I
think
the
Japan
is
just
one
small
element
of
that.
There's
going
to
be
a
whole
bunch
more.
We
were
going
to
have
this
meeting
Breton
volunteer
to
schedule
it.
He
has
not
volunteer
I've,
not
seen.
E
Remember
taking
here
to
get
the
young
geek
I
have
to
sorry
about
that
brendan
is
in
here.
He
has
been
collecting
results.
He
said,
have
two
groups
of
surveys
and
we
can
hopefully
expect
a
week
and
ask
for
an
update
on
scheduling
from
him
on
the
next
community.
Meeting.
I
just
would
feel
bad
about.
You
know
reassigning
it
mid
miss
like
especially
because
he's
not
here
to
talk
about
how
that's
going
Craig.
A
Has
been
weeks,
but
there
was
an
intentional
discussion
to
wait
until
I
physic
infernity
survey
came
back
in
to
move
forward
with
the
governess
conversation,
so
there
was
an
explicit
plan
to
wait
for
that
to
happen.
That
will
happen
that
closes
on
Friday
and
we
are
going
to
discuss
the
responses
from
that
next
week
on
this
meeting
and
then
I
would
also
in
that
time
expect
that
we
can
pull
together
a
government
schedule,
both
kids
ola
governance,
meeting
yeah.
I
The
other
thing
that
happened
was
there
just
a
few
backpack
conferences
and
other
events
in
a
row
and
federal
holiday
in
the
u.s..
So
I
think
in
fact,
even
the
day
in
time
was
already
sorted
out
by
a
previous
they'll,
build
a
survey
and
we're
just
waiting
for
the
opportune
moment
to
actually
start
meeting.
I
actually
put
together
an
issue
that
collected
a
list
of
the
known
things
have
been
brought
up
that
haven't
been
resolved
by
other
discussions.
C
Vocalist,
I'm
sitting
in
Richards,
gorgeous
I,
stop
thinking
and
discussing
and
I
think.
A
lot
of
this
needs
to
be
done
in
a
formal
little
blonde
yeah.
What
I
think
we're
going
to
need
to
do
is
like
a
day
to
just
lock
in
the
core
stakeholders
and
and
sit
down
to
why
she
have
a
government
structure
in
place,
and
you
know
I
think
the
student
we
start
on
that.
The
benefit
sounds
great
thanks.
I
just
want
to
make
sure
we
making
progress,
I'm
having
to
dissipate
I
think
we
all
care
a
lot
about
this
yeah.
I
And
the
other
thing
I'll
just
call
out
is
we
added
a
number
of
these
finer
grain,
github
teams
and
mirrors
two
groups.
I
just
asked
to
figs.
Try
to
make
use
of
those,
and
the
intent
is
to
create
more
targeted
channels
that
people
can
use
to
watch.
What's
going
on
across
multiple
cigs
with
just
one
fire
hose
/
sig,
it
was
basically
impossible
to
do
that.
So
you
know
with
proposal.
The
hope
is
with
feature
request,
proposals
and
design
documents.
I
G
Sorry
cook
yeah
just
removed
from
here
yet
and
like
full
of
an
hour
current
discussion
again
106
and
it
was
6pm
medium,
so
select
a
second.
This
idea
to
like
to
have
like
they're
super
discussion
regarding
the
governance
and
procedural
justice,
procedural
issues.
The
second
question
did
I'd
like
to
highlight
to
get
in
the
future
same
place.
So,
as
you
know,
we
currently
using
the
simple
that
requires
constantly
updating
the
check
boxes
and
so
on.
So
this
is
the
final
release.
G
Where
I
we
have
decided
to
use
skis
teachers
template
we
like
to
simplify
it
and
I'm,
proud
to
walking
on
the
simple
kind
of
interesting
place.
I
would
go
for
the
new
feature,
simpler
Easter
for
the
future
Oh
magic.
It
was
just
additive
transcendence
edited
like
any
moral
and
to
track
all
the
progress
of
the
feature
in
the
future
station.
G
E
A
A
J
E
Okay,
great
and
I
apologize
to
scalability
and
take
a
PA
and
machinery,
but
let's
see
what
we
can
do
for
an
update
Bob
do
we
have
you,
you
have
made.
L
You
do
that
click
on
my
Christmas
button,
all
right,
still
looking
good
excellent,
all
right,
so
I
just
equip
logistics.
Recap.
I
won't
spend
much
time
on
this.
This
is
really
unchanged,
but
in
case
you
need
it
by
the
way
the
slide
deck
is
in
the
sixth
scale
notes.
So
there's
a
nice
circular
dependency
here.
If
you
care
to
try
to
chase
it,
but
maybe
we
can
I
can
post
this
presentation.
We
need
notes
as
well,
so
just
to
give
you
some
flavor
for
how
things
generally
I've
been
proceeding.
L
We
generally
have
conversations
in
two
or
three
topic
areas.
One
is
the
friend
under
our
noses
than
were
strategic,
longer
running
stuff
on
imprisons
generally
and
then
there's
been
quite
a
lot
of
talking.
I
think
we
all
expect
that
there's
going
to
be
a
continued
conversation
as
we
go
through
both
of
these
types
of
things,
around
benchmarks,
config
and
configure,
sharing
and
results
sharing.
L
So
so
we
have
no
formal
performance
target
upgrade
from
15
to
16.
Part
of
that
is
the
16
ability
targets.
However,
there
is
some
important
stuff
going
on
in
the
16
release.
The
big
one
is
f-ed
tree.
There's
been
a
lot
of
work
actually
over
multiple
releases
to
get
that
in.
If
in
good
shape,
many
clusters-
it
is
the
default
for
16
15
16
upgrades
are
in
good
shape
at
the
sig
meeting
this
morning
we
had
a
pretty
long
conversation
about
the
google
I.
L
Think
Google
is
the
main,
the
main
participants
that
is
operates
with
data
migration
from
16
back
to
15
and
you're,
something
instead,
some
comment
this
morning
that
that
that
thing
was
maybe
caught
up
in
some
of
the
flakes,
but
just
to
point
out
that
that's
probably
a
Google
only
concerned
15
to
16
upgrade
for
most
list.
It's
going
to
involve
an
outage
window,
so
you
could
do
a
snapshot
roll
back
now,
the
so.
Let
me
get
back
to
the
performer
formal
performance
target,
upgrade
ongoing
performances
generally
open
to
the
right.
L
L
Okay,
so
in
terms
of
the
general
consensus
on
our
longer-term
goals-
and
this
relates
back
to
the
5,000
owed
and
beyond
kind
of
weird
maps-
macrina
solos
really
need
to
be
expanded.
We
haven't
aside.
Is
that
there's
so
many
possibilities
here
that
our
path
forward
is
to
pick
one
and
really
concentrate
on
that,
and
also
make
sure
that
we
pick
one
that
useful
in
performance
regressions.
L
L
One
topic
comes
up
over
and
over
again
is
the
need
to
move
away
from
the
orange
shade
move
away
from
to
add,
in
addition
to
the
synthetic
performance
benchmarks,
the
meat
and
f
that
we
used
a
more
comprehensive,
more
real-world
like
sweet.
The
readout
folks
have
done
a
lot
of
work
on
this
area,
so
a
shout
out
to
them
at
some
links
here
in
covel,
there's
a
dock
and
also
a
youtube
video.
This
is
a
area
of
ongoing
work
related
to
this
as
the
depending
on
what
mixed
controller
makes
you
use.
L
Performance
can
be
very,
very
variable.
Some
of
the
controllers
have
been
our
older.
More
mature
have
been
more
tuned,
some
of
the
controllers
on
less,
though
so
that
gets
back
to
being
careful
about
really
large
configurations
and
contributing
all
your
part
of
that
related
to
that
is
I
think
we
recognize
that
there's
a
need
for
some
updated
configuration
performance
guys,
especially
for
a
new
entrance
munity.
L
Also,
the
benchmarking
is
still
needed
for
very
large
clusters,
especially
on
some
of
the
non-google
infrastructure
systems
like
AWS,
and
then
we
do
need
to
step
up
our
game
on
formalizing
our
performance
Davis
for
future
releases
of
equilibrium.
We
will
definitely
take
a
crack
this
for
17.
If
you
go
into
the
future
repos
a
statement.
That's
there
is
not
it's
not
wrong,
but
it's
just
a
bit
out
of
date
and
it
needs
an
update
and
then
on
to
serve
our
third
track
of
conversations
around.
How
do
we
in
all
these
different
environments
with
different
configurations?
L
L
Both
we've
been
working
on
and
we've
got
a
lot
of
support
around
the
community
because
it's
one
of
those
kind
of
cops
across
a
bunch
of
the
projects,
a
long-term
effort
to
try
to
come
up
with
a
better
way
to
automatically
captures
and
space
so
that
you
have
not
just
the
performance
data
from
Iran.
But
you
have
the
configuration
data
from
the
run
and
there's
lots
lots
of
tuning
settings
and
systems
that
we
had
promised
I.
L
Think
before
I
just
make
it,
we
already
had
a
bit
of
a
discussion
about
this,
which
is
the
roles
of
things
and
sig
leadership
is
being
debated,
there's
probably
gonna
be
changed
in
the
works.
Brian
has
been
proposing
some
things,
not
a
covered
in
stock,
so
just
the
just
the
sort
of
pile
on
to
the
previous
conversation,
which
is
definitely
a
big
deal
and
I
think
they're.
One
of
the
special
topics
here
is
there
already
know
releases.
They
kind
of
cross
other
state
boundaries
and
working
out
figure
out
the
role
for
the
growing
power.
L
So
I
think
this
is
a
pretty
complex
topic
on
a
distributed
system.
So
I've
done
my
best
try
to
give
a
consensus
view,
but
I
certainly
suggest
anyone
that
wants
to
jump
in
from
the
6th
galing
meetings
to
jump
in
and
provided
me
further
complexes
additional
computable
or
in
our
I
can
answer
questions.
K
I'm
here
so
I'm,
going
to
give
a
brief
update
of
what
ap
I'm
hearing
has
done
in
16
and
then
plans.
We
have
217
plans
but
not
promises.
So
in
16
we
are
probably
best
known
to
the
great
code
migrating
the
lottery
factoring
a
lot
of
coding
around.
It
ended
up
well.
That
will
not
say
why
we
have
a
generic
API
server
library
that
can
now
be
used
as
stable
base
for
building
keep
stir
and
resilient
service
catalog.
K
Those
are
two
concrete
projects
that
people
care
about
and
we're
trying
to
move
forward,
to
being
able
to
handle
things
like
custom
metrics,
from
keep
stir
for
grabbing
HCA
and
for
service
catalog.
It's
a
missing
feature
and
it's
gonna
be
great
to
have
once
we
have
it.
The
big
thing
we
had
was
around
client
go,
and
so
we
try
to
address
the
day
and
client
go.
We
turned
off
automatic
syncing
until
we
are
able
to
be
sure
that
we
don't
automate
the
client
again.
There
is
a
set
of
thought
that
was
published.
K
K
It's
a
project
that
is
still
in
its
early
stages,
but
it's
needed
to
be
able
to
unify
additional
API
servers
things
like
server
catalog,
so
that
we
don't
have
to
include
it
into
the
core
API
server,
but
that
it
can
be
exposed
as
a
first
class
entity
to
client,
even
if
it
is
not
a
first
class
when
compared
to
the
core
cubed
API
server.
The
implementation
of
16
is
not
ready
for
prime
time,
but
we
will
be
working
on
it
in
17.
K
The
other
thing
that
we
did
in
one
fixed
was:
we
have
changed
the
governor
collection
implementation.
There
has
been
a
long-running
feature
or
lowering
the
issue
for
synchronous,
garbage
collection,
since
there
were
issues
the
original
implementation
around
shortening
we
are,
we
have
merged
it.
Already.
We
haven't
seen
any
additional
flakes
from
the
UGC
and
more
resources
are
now.
Turning
on
the
garbage
collection
links,
which
should
really
help
when
you
direct
citations,
around
deletion,
I
delete
this
and
at
least
things
that
that
it
created
for
me.
So
that's
what
we
did
in
one
sickly
rizky.
K
E
Was
fantastic?
Thank
you
David.
Unfortunately,
we
will
not
have
time
to
get
to
the
questions
will
move
those
to
the
next
meeting.
There
are
a
couple
of
important
notices.
Please
take
a
look
at
those
first
and
foremost.
Stick.
Please
fill
up
the
three
days
developer
happiness
survey
by
tomorrow,
and
there
are
announcements
for
crew
daddy,
stay
at
various
upcoming
conferences
and
some
that
you,
maybe
it
shouldn't
going
to
with
that
I
believe
we
are
done
you
are.
You
can
have
your
one
minute
back.