►
From YouTube: Kubernetes SIG Contributor Experience 20170712
Description
No description was provided for this meeting.
If this is YOUR meeting, an easy way to fix this is to add a description to your video, wherever mtngs.io found it (probably YouTube).
E
A
B
Let's
do
that
in
the
last
half
I
I'll
ping,
Matt
Medan
stays
on
the
line,
but
I'll
ask
him.
He
said
that
hit
some
contributors
for
his
problems
and
and
I
think
it's
something
that
we
should
encourage
is
is
to
get
feedback
like
this.
So
having
a
fun
name
section
might
encourage
that
a
little
bit
more
okay,.
A
Cool
so
we'll
circle
back
around
to
that.
F
And
I
have
is:
what
are
we
actually
trying
to
accomplish
what
the
Help
Wanted
label
is
it
that
we
provide
an
easy
entry
point
for
people
to
do
stuff?
Is
it
areas
where
there's
a
critical
concern
that
needs
to
be
addressed
from
a
project
perspective
or
I?
Guess
I'd
like
to
see
define
the
need
first
and
then
decide
if
the
Help
Wanted
label
is
the
right
vehicle
to
actually
accomplish
what
we
want
or
what
Brian
wants
yeah.
A
D
Yes,
I
do
but
I
think
you
know
with
any
issue.
You
know
which
is
not
already
been
assigned
to
anybody
that
you
know
explicitly
seems
like
can
help
one
day
right.
It's
an
open
issue.
You
know
it's
an
entry
point
for
anybody
in
one's
contribute
right
for
you
know,
with
Help
Wanted
I
mean
one
of
the
things
I
found
useful
for
the
new
contributors
and
from
the
older
project.
Is
you
know
if
we?
D
If
we
also
tag
something
like
glow
and
include
right,
talc
wanted
can
be
a
you
know,
it
can
be
like
a
more
like
an
you
know,
complicated
issue
right.
You
know
where
we
need
help
and
we
may
not
be
right
fit
for
the
new
contributor.
You
know
so,
I
don't
know
if
you
already
have
something
where
we
identified.
D
G
Except
sometimes
done
on
some
things
you
get
to
things
where
there's
been
user
requests
for
a
feature
or
a
change,
but
nobody
who's
currently
contributing
to
the
cig
either.
You
know,
has
an
interest
in
that
or
has
the
ability
to
implement
the
request.
That
is
it
something
with
this.
Yes,
we
actually
want
the
strategically,
but
there's
absolutely
nobody
to
work
on
it,
and
that
might
be
a
substantial
feature
or
complicated.
So
I
mean
there's
another
reason
to
have
the
sort
of
help
on
a
tag.
But
it's
a
very
different
kind
of
Help.
F
Almost
feels
like
the
tag
shouldn't
be
standalone
right.
It
would
need
to
have
like
a
scope
of
difficulty
right
like
there's.
A
difference
between
this
is
a
small
bug
than
a
new
contributor
can
fix
and
yeah.
This
is
a
huge
feature
that
we
want,
but
it
needs
to
be
specked
out.
It
needs
to
be
debated
in
the
community.
You
know
go
through
the
drafting
process.
All
that
kind
of
stuff,
so,
like
I,
feel
like
they're
kind
of
that's
two
separate
goals
or
one's
more
of
a
low-hanging
fruit.
F
You
know
I'm
a
developer
and
I
kind
of
I'm.
Looking
for
small
bite-size
bugs
yeah
versus
you
know,
hey
I
need
to
spec
out
the
next
generation
scheduler
thing
well,
so,
as
you're
saying
this,
it
makes
me
think
that
that
kind
of
help
when
a
discussion
should
be
happening
in
a
sick,
not
in
a
normal
legal.
So
if
I'm
going
to
develop
the
next-gen
storage,
plug-in
and
I
want
to
do
that,
I
want
to
go
to
six
storage
to
get
that
help.
F
Not
just
like
put
a
message
in
the
bottle
out
and
then
creative
community
repos
say:
hey
help,
so
I
think
that
if
we
encourage
the
pattern
that
the
patterns
being,
if
you
have
a,
if
you
have
a
feature
that
you
need
help,
go
find
the
sake
for
it.
And
if
there
isn't
a
cig,
make
one
or
make
working
group
and
if
you're
a
new
contributor-
and
you
want
to
just
start
getting
your
feet.
D
I
hear
another
thought
you
know
to
me
seems
like
every
time
an
issue
is
open,
like
I
was
saying:
that's
Help
Wanted,
so
I
was
wondering
if
even
just
just
a
thought.
You
know
that
every
time
issues
open
its
automatically
label,
Help
Wanted
and
as
long
as
somebody
assign
it
to
it,
you
know,
and
so
also
it
gets
removed.
B
D
B
But
I
think
our
proposal,
where
the
issue
triage
from
a
few
weeks
ago,
is
to
get
it
out
to
the
SIG's
and
defining
a
policy
for
the
SIG's.
As
far
as
issue
triage,
we
don't
necessarily
need
an
individual
to
assign
them
just
because
of
the
rate
at
which
issues
are
coming
in,
but
we
want
to
encourage
people
like
even
new
issue
filers,
to
be
able
to
mention
the
Stig
that
it
should
go
to
yeah.
D
F
On
the
CW
thanks
mum,
what
do
we
want?
This
thing
to
also
suss
out
the
accessibility
of
the
thing
right
I
mean
somebody
with
expert
level
and
put
it
this
way
only
somebody
knows,
but
it's
easy
can
say
it's
easy
right
right,
so
you
would
hopefully
have
whoever
the
triage
person
that
adds
to
help
wanted
label
knows
that
this
is
something
this
is
a
specific
target
for
a
good
user
experience,
so
that.
D
B
H
H
They
are
still
needing
to
be
done,
so
they
haven't
been
like
already
done
by
somebody
else.
You
still
want
to
do
them.
We
actually
have
the
time
to
review
the
changes.
We
actually
have
the
time
to
update
the
docs
and
they
don't
require
multiple
PRS
to
accomplish
or
and
we're
not
no.36
etc.
So,
there's
a
lot
of
me
in
there
and
I
think
we
is
kind
of
the
sake
as
the
Help
Wanted
issue
assigned
to
them.
H
Well,
a
lot
of
this
also
seems
to
be
confused
with
priority,
like
I
think
that
urgency,
the
the
priority
to
the
sig,
assigns
to
an
issues
already
signified
by
the
priority
labels
right
Help
Wanted
is
almost
sometimes
it
seems
like
it's
that
would
be
a
nice
to
have,
but
it's
so
low
down
our
priority.
Keep
nobody
with
the
mistake
is
going
to
work
on
it.
You
need
to
definitely
because
we
also
have
this
label
like
for
newcomers.
H
B
B
B
It's
you
know
some
of
it's
our
own
custom,
automation
that
you
have
to
learn
so
I
think
those
are
the
kinds
of
things
that
we
should
be
doing:
defining
Help,
Wanted,
defining
what
the
label
is
and
making
that
very
easy
to
identify
and
then
making
it
very
clear
to
six
what
we
expect.
What
we
would
like
in
terms
of
having
getting
the
label
applied.
E
F
You
know
I,
think
another
aspect
of
this,
too,
is
that
we
want
to
tear
people
as
possible
or
make
it
someone
available
to
help
Shepherd,
whoever
that
is
the
new
contributor.
Through
that
experience,
it's
one
thing
to
have
documentation,
but
it's
another
thing
to
have
somebody
actually
walk
with
you
and
then
tear
on
it.
E
You
have
clear
instructions
if
it's
once
you
define
this
issue
or
this
tag
or
whatever
something
new
comes
in
and
they
see
it,
they
go.
Okay,
I
want
to
work
on
that.
There
needs
to
be
something
in
that
issue
that
says:
okay
go
talk
to
X,
you
know
whether
it's
a
sigelei
door,
a
cig
whatever
and
and
that
person
needs
to
be
able
to
be
around
at
least
partially
to
say,
okay
I'm
here
you
know,
if
you
have
questions,
let's
walk
through
this
together,
all
through
yeah
I,.
B
Who
is
who
is
working
on
the
new
contributor
onboarding
in
the
past?
Because
it
seems
very
related,
I.
B
B
D
D
H
I
think
we
should.
We
should
make
a
suggestion
to
the
community
one
other
thing
I
wanted
to
put
out
there:
I,
don't
think
you
can
apply
or
remove
the
Help
Wanted
label
via
the
bots,
so
I
am
incapable
of
managing
the
Help
Wanted
label
for
the
two
issues
that
are
assigned
to
my
sig,
one
of
which
is
I'm
looking
at
it
now
is
not
does
not
fit
this
criteria,
but
I
can't
do
anything
about
that.
A
F
F
B
H
B
F
H
H
So
this
is
something
like
I
care
about,
and
I
really
would
like
to
have
the
time
to
work
on
dedicated
blocks
of
time.
However,
I
am
super
in
favor
that
people
want
to
pick
up
his
torture,
carry
it
forward
and
I
recognize
that
a
large
part
of
starting
this
is
to
actually
document
all
the
things
I'm
about
to
ramble
on.
So
it's
just
like
off
the
top
of
my
head,
but
I.
Think
that,
like
we
should,
is
something
about
like
stabilize
the
patient,
so
pick
one
repository
that
suppose
we
had
the
canonical
set
of
labels.
H
I
would
say
it's
good,
your
Nettie's
kubernetes,
then
I
would
make
an
effort
to
document
what
all
of
those
labels
are.
Then
I
would
come
up
with
a
report
that
sort
of
defines
how
many
issues
have
those
labels
applied
figure
out
if
it
makes
sense
for
those
labels
to
be
there
and
try
to
roughly
sort
of
reverse
engineer
like
what's
in
the
labels
mean.
So
we
have
many
many
area
labels,
for
example,
I,
don't
really
know
what
those
are.
You
should
have
a
larger
discussion
about
why
area
labels?
What
do
they
represent?
H
There
are
also
a
variety
of
the
kind
labels,
some
of
which
seem
more
useful
than
others.
For
example,
kind
technical,
debt
versus
kind,
cleanup
versus
kind
enhancement,
there's
also
kind
of
feature
kind,
documentation,
excetera
I'm,
not
sure
where
the
appropriate
place
to
have
the
discussions
on
like
pairing.
These
down
is,
but
we
should
at
least
start
with.
This
is
a
reasonable
set
of
labels
that
are
well
understood
and
you
could
document.
These
are
the
labels
that
seem
to
be
questionable
and
then
I
think
maybe,
as
a
first
pass.
H
Somebody
with
soup
human
labeling
powers
can
like
remove
the
label
from
issues
that
don't
need
it.
So,
for
example,
the
team
cluster
deprecated
do
not
use
label
should
probably
go
ahead
and
die
in
a
fire
unless
there
are
issues
that
don't
have
the
appropriate
thing,
but
I
think
since
in
our
grande,
like
needs
sake,
label
triage,
everything
should
have
a
cig
label
so,
like
don't
think
we
need
those
scheme,
whatever
deprecated
do
not
use
labels
anymore.
H
H
I
have
in
my
head
the
concept
of
there's
like
a
large
github
munzur
right
now
that
can
actually
look
at
labels,
not
my
llamo,
and
make
sure
that
all
of
the
labels
that
are
listed
there
exists
in
the
repository
for
which
it's
running
but
I
find
that
to
be
insufficient
for
needs,
because
you
have
to
run
one
copy
of
Munch
github
for
every
single
repository
that
it's
watching.
I
would
instead
like
to
move
that
to
prowl.
So
there's
one
plugin
turned
on
for
multiple
repos.
That
could
do
the
same
sort
of
thing.
H
Then
the
question
becomes:
do
you
want
to
have
a
label
GMO
file
in
every
single
Brico?
Or
do
you
want
to
have
that
long
pointed
at
one
Thorat
ated
labels
animal
file
in
some
central
veto?
Let's
call
it
repo
intra
repo
created
for
general
repository
reuse
across
the
kubernetes
organization,
and
the
question
comes:
do
you
want
the
bots
to
enforce
that?
Only
labels
in
that
file
should
because,
within
the
repo
so
as
to
discourage
humans
from
creating
brand-new
labels
that
are
undocumented
and
not
part
of
that
file.
H
Or
do
you
want
to
be
the
bot
to
be
permissive
for
certain
repositories
that
are
still
evolving
or
finding
new
uses
that
makes
to
them,
but
not
necessarily
to
other
repos
in
the
project.
So,
for
example,
I
could
see
the
cops,
the
Kaos
Rico
having
labels
or
components
or
whatever
there
are
applicable
just
to
cops,
but
probably
not
for
the
broader
kubernetes
project.
H
I
could
see
the
kubernetes
at
github
that
I/o
repository
having
a
bunch
of
documentation
related
labels
that
are
applicable
there
that
are
clickable
to
the
rest
of
the
project,
so
probably
wanting
to
find
some
way
for
each
repo
or
projects
to
have
their
own
custom
labels.
In
addition
to
the
standard
labels,
and
then
you
know
you
sort
of
steer
people
towards
pull
requests
to
these
file
or
files
as
the
sole
source
of
truth
for
how
you
create
and
remove
labels.
H
You
go
through
the
traditional
pull
request
process
where
people
can
discuss
and
motivation
for,
adding
or
removing
sub
labels
and
I.
Think
that's
enough
rambling.
Does
that
sound
like
sort
of
a
coherent
plan
that
maybe,
if
I
go
back
and
watch
the
recording
of
this
I,
could
just
dough
into
a
proposal
or
issue
or
something?
Yes,.
H
F
I
would
start
with
what
you
set
at
the
end.
First,
a
standardized
set
of
labels
that
consistent
for
everybody
and
then
now,
whatever
is
important
for
that
thing,
they
can
have
their
own
little
custom
set
of
labels.
It
would
be
nice
if
there
was
a
way
that
it's
you
know
we
didn't
let
that
clutter,
I'm,
just
fearing
to
see
like
as
I,
was
getting
rid
of
some
of
these
labels.
I
saw
some
issues.
It
just
had
way
too
many
labels
and
it
was
almost
like
at
that
point.
Are
they
useful.
H
Yeah
I
mean
that
the
key
thing
is
like
I
really
do
think.
There
are
a
lot
of
labels
that
should
go
away
and
I
want
to
have
an
open
discussion
about
that
process,
and
it's
just
unclear
to
me
whether
it's
works
like
taking
a
human
pass.
It
slowly,
cooling
down
the
set
of
labels
or
if
we
start
from
the
standard
set
and
evolve
our
way
out
from
there.
H
B
H
A
D
Thanks
you
I
just
you
know
so
Georgia
and
I.
You
know
we
created
a
PR
for
the
th
guideline.
We've
been
discussing
for
some
time
now
right.
You
know,
based
on
the
last
meetings
in
your
discussion,
so
I
just
want
to
remind
folks.
You
know
it's
an
email.
It's
out
there.
We
had
few
minor
comments.
It's
been
taken
care.
There
are
a
couple
of
any
reminded
comments
that
MIT
needs,
maybe
a
little
more
feedback.
So
you
know
you
guys
can
whenever
you
get
a
chance
and
take
a
look
I
just
wanted
to
mention
here.
F
So
this
just
just
I,
wanted
to
put
this
in
your
in
your
mind,
because
it's
like
a
multiple
complex
thing
as
part
of
SiC
cluster
life
cycle,
one
of
the
items
that
came
out
of
the
meeting
was
hey.
How
come
there
is
no
less
defined
way
to
do.
Local
development
of
kubernetes
and
I
realized.
That's
a
huge
topic
right.
You
have
like
bigger
boxes,
and
things
like
that.
You
know.
Basically,
the
idea
with
what
the
ask
kind
of
came
out
of
the
sig
was
like
it'd
be
really
nice.
K
E
No,
you
know
there
needs
to
be
I,
understand
the
opinion
that
we
eat
we're,
trying
to
have
no
opinion
for
anything
well.
This
is
something
we
need
to
have
an
opinion
on,
and
hatred
I
have
an
opinion
on
needs
to
work,
and
it
needs
to
work
every
time
or
at
least
more
than
50%
of
the
time,
because
I
have
I've
tried.
A
F
I
would
like
to
help
the
things
that
makes
this
one
complicated
is
like
I,
don't
have
any
max
in
the
house
so
like
I,
don't
know
how
to
do
any
of
the
max
stuff.
I
can
do
the
Lenox
stuff,
maybe
test
a
little
bit
of
Windows.
So
in
fact,
one
of
those
things
where
you
need
kind
of
yeah
you
need
rough
life
or
expertise.
E
F
Distro
people,
like
all
sorts
of
like
I,
would
know
how
to
do
any
of
this
on
Fedora,
like
that's
just
not
my
expertise
and
it's
complicated
enough
to
you
know.
F
F
If
it
like,
it
feels
like,
we
need
to
start
having
a
strong
opinion
question
that
this
is
a
equivalent
to
the
la
brea
tar
pit
of
issues
and
that
we
we
need
to
be
extremely
careful
in
this,
but
I
would
say
the
pattern
that
I've
seen
socializing
many
a
development
group,
I've
managed
into
continues
ation
and
all
this
stuff.
Is
it
the
way
you
do?
This
is
to
provide
a
use
case
or
a
user
experience
sort
of
so
compelling
that
people
want
to
use
it
and
then
add,
drives
adoption.
F
F
So
what
I
would
say
is
let's,
let's
think
about
what
the
mo
incredible
user
experience
could
be
to
just
you
know,
grab
a
repo
start
tacking
on
it
and
have
it
run
your
tests
and
get
immediate
CI
feedback
and
all
the
great
things
about
a
solid
user
experience
and
then
see
what
we
can
do
to
engineer
that
or
partner
around
engineering
it,
because
otherwise,
there's
no
hope.
I'm
I
have
scars
yeah,
it's
just
so
such
a
complex
topic
that
I
only
wanted
to
bring
it
up
for
people
to
start
thinking
about
it.
G
It
kind
of
seems
like
we
would
just
need
to
have
a
separate
document
per
platform
and
have
owners
of
each
platform
document,
because,
among
other
things,
not
only
is
there
the
issue
of
a
very
different
experience
on
Mac
or
very
different
rates
and
Windows.
There's
also
the
fact
that
they
on
the
platforms
and
the
availability
of
specific
tools
on
those
platforms
changes
constantly
so
like
my
team
could
commit
to
keeping
a
document
on
how
to
do
this
on
fedora
and
sent
us
updated.
G
F
L
Think
if
we
start
somewhere
and
produce
a
great
developer
experience
on
that,
whatever
that
requirement
is,
if
anyone
wants
to
add
and
pour
to
other
platforms,
let
them
do
that,
but
let's
get
something
great
and
I
think
people
will
use
it
like.
You
know,
I
think,
to
this
day,
doctor
runs
in
a
VM
like
let's
just
stick
with
something
that
works,
which
is
probably
going
to
be
a
VM
and
if
someone
happens
to
run
iOS
on
their
machine
so
much
the
better.
A
Great
I,
don't
think
I
see
max
Elliot
on
the
call
yet
I'm.
J
So
the
basic
form
of
the
story
is
for
this
past
release.
We
wanted
to
write
some
tests
that
verify
that
a
disruption
budget
would
work
in
concert
with
stateful
sets
and
the
node
upgrade
script.
Both
the
one
myth
is
part
of
open
source
for
GCE
clusters
and
then
also
the
one
that
runs
internally
on
GK
and
eventually,
of
course,
any
upgrade
any
node
upgrade
that
is
out
there,
but
those
two
platforms
are
covered.
Now
we
wrote
the
tests,
we
wrote
the
PRS
and
we
started
to
get
input
immediately.
J
I
assumed
the
input
was
from
somebody
who
was
empowered
to
be
an
approver.
I
was
pretty
sure.
I
figured
that
out
from
the
beginning,
but
that's
one
thing.
First
of
all,
it's
not
obvious
to
somebody
who
is
not
a
member
of
the
project
that
person
who
began
well.
First
of
all,
the
notion
of
owners
and
approvers
isn't
obvious
at
all.
If
you've
never
done
this
before.
Second
of
all,
even
if
you
kind
of
get
that
idea,
the
person
who
is
really
vocal
on
your
PR
may
have
nothing
to
do
with
it.
J
So
after
going
back
and
forth
with
him
sort
of
pleasing
and
placating,
this
person,
I
then
discovered
near
the
end
of
the
quarter
that,
in
fact,
this
person
couldn't
be
my
approve.
It
doesn't
really
matter
who
it
is,
because
I
think
this
tail
could
happen
to
anybody.
The
person
who
I
chose
me
my
retriever
was
Daniel
Smith
and
he
didn't
like
what
we
had
in
mind.
Basically,
the
problem
was,
we
wanted
to
add
a
bunch
of
images
like
little
test
binaries
that
would
run
in
kubernetes
and
talk
to
let's
say
Etsy.
J
He
talked
to
Cassandra
talked
to
MySQL
during
the
upgrade
process
and
clarified
that
they
stayed
reached
for
the
whole
time,
and
the
concern
was
this
doesn't
belong
in
the
main
repo.
The
question
why
I
had
was:
where
does
it
belong
and
every
example
we
were
given?
We
would
sort
of
like
go
there
with
our
little
story
in
our
bowl
and
be
like.
Please
can
you
take
our
code,
like
all
I
want
to
do
and
I
work
for
Google
and
I
work
on
this
project.
J
I
want
to
give
you
this
and
it
clearly
as
part
of
kubernetes.
Somehow
no,
it
doesn't
belong
here.
No,
it
doesn't
belong
here.
So
imagine
if
I
didn't
work
on
this
project
as,
like
you
know,
a
big
corporation
that
was
really
involved
with
it.
They
won't
like.
Not
only
will
they
not
take
it,
they
can't
tell
me
where
the
proposals
are
often
like.
Maybe
you
need
to
own
the
repo?
Oh,
you
don't
understand
how
to
like
create
a
new
kubernetes
rapport.
J
You
need
to
learn
that
parts
of
process
too
so
I
think
the
problem
here
is
that
it's
pretty
easy
to
just
and
we're
just
now.
Coming
to
this
point,
where
we're
trying
to
draw
really
firm
lines
around
what
goes
in
each
repository,
particularly
the
main
one,
and
it's
very
easy
to
come
in
and
run
afoul
of
that
and
have
and
basically
had
nobody
there
to
help
you
I.
J
Found
it
to
be
pretty
exhausting
and
difficult,
and
we
still
don't
really
have
a
clear
answer,
as
somebody
is
working
on
the
project,
I
cannot
imagine
what
it
would
be
like
if
you
were
just
trying
to
make
this
one
quick
change,
it's
significant,
but,
like
so
I
guess
that's
the
tale
of
woe
like
I.
Don't
know
if
I
really
convey
to
you
how
much
it
sucked,
but
imagining
me
is
not
me
doing
the
same
thing.
It
would
have
been
like
err,
I
never
would
get
going
and
we
need
to
help
people.
J
The
answer
should
not
be
about
gatekeepers,
saying
yes
or
no
right
it
should
be
about.
There
should
be
some
way.
I
could
have
known
in
advance
where
it
was
all
going
to
go.
I
could
have
identified
correct,
approvers
easily.
It
really
should
have
been
easy
and
nobody
should
have
just
said,
I'm
not
sure,
but
I'm
sure
that
there's
a
different
place.
It's
not
my
place
that
your
code
can
go.
B
That's
that's
all
I
got
questions
so
there's
like
a
few
fundamental
problems.
One
is
like
it's
very
and
I'm
just
repeating
some
of
those
stuff,
you
said,
but
it's
for
clarity.
Yeah
owners
is
really
hard
to
understand
for
a
new
contributor
so
like
once,
we
detect
new
contributors.
This
is
probably
something
we
want
to
explain
since
such
a
big
part
of
the
PR
process
and
who
the
relevant
owners
are
for
the
work
that
you're
doing
make.
That
really
explicit.
J
I
started
to
get
it
wasn't
actually
the
assignee
it
was.
It
was
a
car
Gatos
and
he
just
happened
to
you
know
he
was
neither
assigned
nor
a
potential
approver.
He
just
started
to
comment
on
the
PR,
which
is
cool
like
I,
really
wanted
her
to
say,
but
the
thing
is
it's
so
easy
to
believe
that
you're
now
having
the
Cobra
view
right,
because
why,
wouldn't
you
be
in
and
part,
maybe
I
was
fooled
in
part,
because
I
actually
did
think
he
was
probably
like
I
thought.
I
saw
his
the
name
earlier.
J
I
thought
he
was
in
the
owners
file
in
question.
It
wasn't
I,
don't
know
how
to
signal
that
more
than
we
do,
but
at
some
point
there's
just
too
much
text
like
I
think
that
the
bot
says
a
lot
of
things
and
yet
it's
pretty
easy
to
not
read
them
all.
Ignore
it
yeah
go
ahead,
though
you
were
trying
to
restate
some
stuff.
Okay,.
B
B
The
second
thing
was:
where
should
my
code
lives,
so
you
started
like
most
people
I
imagine
within
the
main
repo,
and
then
we're
told
a
we're
trying
to
like
burn
things
out,
move
it
somewhere
else,
but
no
one
told
you
where
it
should
live
and
that
was
difficult
to
search
as
well.
This
is
a
problem.
I,
don't
think
we
have
discussed
discussed
much
in
the
past
here,
but
it
sounds
like
it's
going
to
become
more
and
more
as
as
we
try
and
shard
the
main
repo.
B
B
B
J
J
J
So
with
on
me,
and
that's
one
thing:
I'm
going
to
try
and
do
this
week
to
contact,
suggest
city
and
say:
look
if
we
create
this,
would
you
guys
be
willing
to
be
responsible
for
it?
We
can
move
these
new
images,
plus
these
existing
test
images
from
the
main
repo
put
them
all
in
one
place.
I
don't
want
to
own
it
right
and
I.
Think
that's
a
burden
we
don't
want
to
put
on
contributors
is
like
well.
J
J
H
L
For
the
issue
of
wear
code
lives,
I
think
we
do
have
an
answer
which
is
still
evolving,
which
is
that
we've
devolved
a
lot
of
power
to
SIG's
SIG's,
have
ownership
of
repos
and
sub
trees
of
the
main
repo,
and
we
need
to
encourage
contributors
to
like
find
a
home
in
a
cig
for
new
work
that
doesn't
obviously
live
somewhere
and
I.
Guess
I.
Think.
J
That
is
a
particular
problem
with
this
project,
because
and
I
ran
into
a
lot
of
problems
to
trying
to
get
people
to
understand
what
it
was.
We
were
testing
the
features
that
we
were
testing.
Our
crosscut
write
the
pause
disruption
budget.
When
we
wrote
it
seemed
like
something
related
to
the
scheduler,
but
now
it's
actually
relevant
mostly
to
workloads,
and
we
were
also
in
a
certain
sense.
J
So
I
I
see
what
you're
saying
but
like
when
you've
got
a
feature
like
this.
That
doesn't
really
seem
to
have
clear
ownership,
it's
hard
to
say
it's
important,
but
like
it's
the
emergent
behavior
of
all
the
things
working
together
that
we're
verifying
was
even
important.
Then
we
test
relapse.
We
wanted
to
say,
look
at
you
if
you
run
Cassandra
like
so,
you
can
expect
it
to
work
through
node
upgrade
so
like
I,
don't
know.
H
All
right,
this
does
sound
familiar
because
this
is
I
couldn't
find
the
specific
PR,
but
I
do
remember
a
PR
that
I
saw
on
a
couple.
Other
folks
saw
and
I
do
remember,
pushing
back
against
it
in
the
context
of
it
does
cross
cut
a
lot.
It
was
testing
of
real-world
applications
on
a
kubernetes
cluster
and
making
sure
that
a
specific
feature
worked.
J
J
There
was
also
this
message
that
this
test,
maybe
doesn't
belong
here,
but
we
are
also
not
sure
where
it
does
belong,
even
though
we
can
all
kind
of
agree
that
it's
important
to
have
a
signal
somewhere
sure,
maybe
it
doesn't
need
to
block
merges,
but
it's
not
as
if
this
is
unimportant
work
so
and
so
I.
Don't
necessarily
need
answers
to
my
questions.
F
A
couple
of
things:
well,
actually,
three
things
come
to
mind.
First
thing
is
it's
good
for
us
to
identify
areas
on
these
user
experiences
where
you're
getting
conflicting
information
like
tab
sources
spaces?
You
know
like
a
where
are
you
running
in
the
buzzsaw
of
opinion
versus
actual
technical
implementation,
details
that
affect
the
way
things
are
done,
because
that's
a
huge
problem?
F
F
And
lastly,
just
in
terms
of
your
issue,
I
think
this
is
something
we're
going
to
see
more
and
more
as
we
close
get
broken
out,
but
there
is
nothing
to
prevent
someone
to
set
up
their
own
CI
pipeline
and
everything
else
against
mastered.
It's
like
their.
If
you,
if
there
are
instances
where
you
want
to
to
run
a
specific
set
of
tests,
we
ought
to
make
it
easy
for
somebody
to
set
up
their
own
test
infrastructure.
So.
J
About
that
last
thing
like
there
is
nothing
to
stop
me
is
in
a
sense
true,
but
then
the
next
thing
you
said:
where
is
we
ought
to
make
it
easy?
That's
the
devil
right
like
it's
strictly
possible,
but
like
among
them
now
I
have
a
million
more
things
to
think
about,
learn,
understand
and
maintain,
and
I
Alan
I
totally
can't
take
everyone's
test
in
the
long
run,
but
yeah,
but
I
felt
like
if
I'd
had
to
cross
that
bridge
as
well.
I
would
have
really
even
crazy
yeah.
F
I'm
not
necessarily
saying
that
it's
easy
or
even
the
right
one,
sir,
but
the
thing
that
the
thing
that
Aaron
brought
up-
and
this
was
something
that
we
we
agonize
over
it.
You
know
we
agonize
over
at
the
sig
leadership
thing
and
we
agonize
over
every
single
day
is
how
do
we
keep
this
project
from
becoming
the
Hydra?
Where
you
you
have
so
many
things
that
are
crossing
so
many
lines,
and
so
many
groups
that
that
essentially,
the
the
velocity
starts
training
toward
zero
because
nobody
can
can
negotiate
it.
F
H
Yes,
yeah,
please,
no
Jenkins
I
need
again
you're
begging.
A
question
at
state
testing
is
desperately
trying
to
work
towards
where
we're
using
we're
trying
to
get
the
entire
stack
of
will
be
used
to
test
kubernetes
and
then
to
be
completely
open
source.
The
two
pieces
that
are
open
source
right
now
are
tested
and
the
code
that
scrapes
results
from
GCSE
and
it's
formatted.
This
is
ruled
by
tester
on
the
test.
Grading
Terminator
are
both
App
Engine
apps,
so
this
could
be
view
written
from
figure
Nettie's
that
would
be
created
for
right.
H
Now
they
like
the
source
for
them,
is
open.
You
just
happen
to
need
in
haven't
to
run
them.
We're
getting
to
the
point
where
you
can
use
prowl
to
you
can
use,
extend
up
route
on
your
own
kubernetes
cluster
and
use
it
to
run
tests.
I
think
that
this
corner
you're
going
to
see
is
CEO
and
open
shift
origin,
be
fantastic
Cruces
concept
of
using
this
to
test
more
than
just
communities,
but
multiple
projects
and
have
multiple
repositories.
H
I'm
furthering
I,
think
we're
opens
the
idea,
people
contributing
to
get
their
own
jobs
through
the
process.
That's
starting
to
get
hunter
documented,
but
could
certainly
use
some
help,
and
the
other
thing
I'll
highlight
I
think
that
that
unfortunately
ran
into
a
case
like
as
a
guinea
pig
for
one
of
these
big
cross-cutting
across
concerns.
Like
youth.
Is
this
you?
H
It
is
correct
that
this
is
something
that
will
come
up
more
often,
and
probably
what
would
be
useful
in
this
scenario
is
to
get
the
appropriate
heads
together
to
figure
this
out
for
the
first
time
so
that
we
have
the
appropriate
template
to
use
going
forward
that
it
sounds
like
what
happened
was
getting
pinballed
around
between
a
bunch
of
people
who
were
stakeholders,
but
you
never
got
the
complete
set
of
stakeholders
together
to
to
craft
like
what
we
would
like
to
do
the
next
time.
This
comes
up.
That
sounds
fair
you're
nude.
J
That's
absolutely
right
and
and
I
would
I'd
be
willing
to
help
sit
down
to
make
sure
that,
like
with
whoever
those
stakeholders,
are
to
kind
of
write,
a
draft
of
what
we
think
we
should
tell
people
the
next
time.
This
comes
up
yeah,
the
other
thing
I'd
be
in
supportive.
If
we
can
stop
it,
I
don't
know
how
much
work
it
would
be.
I
think
the
concierge
idea
is
great
place.
That's
what
who
I
felt
was
missing
so
I
would
I
was
loved
for
there
to
be
such
a
first.
So.
H
On
that
front,
what
I
was
going
to
say
is
something
I
have
been
organically,
doing
I'm,
really
bad
at
marketing
or
documenting.
This
is
I
just
sort
of
troll
through
and
look
for
PRS
that
have
the
needs
of
k2.
Like
replied,
that's
representative
of
somebody
who's,
not
a
member
of
the
kubernetes
organization,
so
they
have
no
right
privileges,
they're
likely
a
first-time
contributor
to
this.
So
right
away.
H
I
can
flash
okay
to
test
to
help
push
things
forward,
to
generate
some
CI
signal
which
might
encourage
the
reviewers
to
look
at
it,
and
also
that
we
can
times
subscribe
to.
What
that
looks
like
moving
forward
and
I
can
help
like
triage
to
the
right
sake
or
code
things,
but
I
agree.
That's
not
necessarily
the
same
thing
as
a
formal
role,
and
also
it
wouldn't
have
helped
Matt,
because
Matt
is
a
member
of
the
kubernetes
organization
and
a
member
of
the
team,
but
right
privileges
on
the
repo.
H
L
I
wanted
to
ask
something:
this
is
very
much
something
I've
hit
a
lot
in
my
recent
contributions,
which
is
we
now
have
the
the
concierge
feels
like
the
role
of
assignee
and
now
what
we
have
is
we
have
multiple,
designees
and
so
funny
doesn't
really
mean
anything
anymore,
like
no
one's
responsible
for
a
PR
anymore.
So
I
guess
I
was
wondering
why
I
we
have
what
I
want
multiple.
L
I
mean
that's,
that's
fine,
but
I
mean
the
problem.
Is
that
then
they
sit
there
for
like
three
months
with
everyone's
assuming
the
other
guys
do.
A
girl
is
doing
it
and
I
think
that
if,
if
we
have
that,
we
can
just
if
there
one
think
it's
over
the
holiday
for
like
an
extended
period,
then
right
and
you
always-
this
has.
H
Been
codified
into
an
issue
that
I
think
carrot
filed
into
the
testing
for
repo
that,
unfortunately,
we
and
testing
don't
have
time
to
work
on,
but
you
would
basically
want
the
bot
to
identify
pull
requests
that
have
had
assignees
given
to
them,
but
the
assignees
have
done
nothing.
So
the
pull
request
itself
is
stale,
not
because
the
author
is
in
commenting,
but
because
the
assignees
are
doing
nothing.
H
What's
the
appropriate
response
here
today
the
response
is
looks
like
the
pr
to
fail,
we're
going
to
start
nagging
and
if
it
doesn't
have
any
activity
with
the
90
days,
we
closed
it.
A
thought
is,
we
could
try
rotating
to
two
other
random
assignees.
The
question
would
be
at
what
velocity
should
we
do
that?
Should
we
do
that
or
owners?
Should
you
do
that
for
approvers
right
now?
H
There's
this
weird
thing:
we're
like
we
don't
actually
notify
approvers,
and
the
hope
is
that
somebody
who
is
less
bandwidth,
packs
who's,
just
a
reviewer,
can
make
sure
that
the
pr
looks
good
and
then
it
gets
bounced.
It
gets
manually
bounced
to
an
approver
by
way
if
somebody's
notifying
a
person
right
so
like
there's,
there's
room
for
improvement
and
there
are
issues
open.
L
If
you
want
to
help
work
on
it,
well,
I
mean
I.
Think
I
would
love
to
work
on
it,
but
right
now,
so
when
I
think
we,
what
I
would
think
we
should
do
is
we
should
keep
track
of
who
isn't
doing
their
reviews
and
remove
them
from
approvers
or
reviewers
or
whatever
it
is,
and
there
are
some
theories
like
if
the
hands
wet
people
who
are
like
notorious
and
I'm,
probably
among
them,
for
like
not
doing
reviews
for
like
months
at
a
time
right
now.
L
H
Was
something
we
had
asks?
What
I
don't
know
if
there's
a
fix-it
going
on
right
now,
but
like
one
suggestion
I
made
in
the
past,
because
I
feel
like
we
collectively
and
I'm
talking
about
the
whole
project,
I
feel
like
we
don't
do
the
greatest
job
of
measuring
the
before
and
then
see
your
izing,
what
the
actor
is
going
to
look
like
and
then
measuring
the
after.
H
We
just
go
ahead
and
make
decisions,
and
so
I
think
this
is
a
great
case
for
if
you
could
actually
help
us
write
the
script
or
the
reporting
tool
that
identifies
the
stale
the
state
look
who
is
sale?
That
would
give
us
some
actionable
data
right
because
I
agree
with
you.
I
have
that
feeling,
based
on
my
experience
as
a
human
but
I
also
don't
want
to
like
do
the
subjective
thing
and
just
call
out
people
right.
I'd
rather
have
a
more
open,
objective
process
that
we
could.
L
Guess
I'm
wondering
like
yeah:
we
could
try
it
yes,
I
guess
there
are
two
things
I
might,
one
of
which
is
like.
We
need
the
accusation
brush
we're
like
the
metrics
brush
and
then
we
need
to
know
who
to
apply
that
brush
to
and
I
guess
you're
saying
we
should
make
the
brush
before
we
have
someone
to
apply
the
to
make.
F
If
you
can
pass
a
link
to
that
tag,
or
whatever
it
was
Aaron
I'm
interested
in
looking
to
seeing
or
I,
would
love
to
socialize
to
maybe
experience
reviewers
be
like
hey.
These
are
people
who
don't
have
write
access
and
should
be.
You
know,
as
we
onboard
people
I'd
like
to
get
them
reviewing
and
onboarding
people
in
seconds
with
that
I
think
that'd
be
a
nice
URL
to
socialize
yeah.
Let
me
just
drop
it
in
the
docs
here,
just
whenever
it
does
have
to
be
right.
Now.