►
From YouTube: Kubernetes SIG ContribX 20170628
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).
A
All
right,
we
are
good
to
go
so
this
is
the
28th
control
X,
let's
go
ahead
and
just
start
walking
through
the
agenda,
there's
a
few
other
things
that
I
want
to
do
kind
of
at
the
end,
to
reviewing
1.7
and
talking
a
little
bit
more
about
1.8
and
I
also
want
to
discuss
this
fix-it
net
and
and
I
have
been
talking
about
for
a
little
bit
and
I.
Think
I've
been
to
a
few
of
you
and
start
doing
that,
but
let's
look
go
ahead
and
start
with
George's.
B
B
B
So
let's
say
you're
in
kubernetes
kubernetes
and
you
find
a
bunch
of
issues
and
let's
say
you
look
at
Cube
admin
bugs
want
to
kick
those
out
of
core
and
go
to
file
against
Fornetti
splash
cube
admin.
You
see
in
github,
you
can't
just
say:
oh
okay
select
all
these
and
move
them.
You
have
to
manually
go
file,
a
new
bug
in
the
new
thing
and
then
close.
The
old
one
and
I
found
an
app
that
someone
had
hosted.
B
D
B
D
D
Be
I'm
sure
it'll
be
continuous
problem.
People
don't
know
where
to
file
the
the
issues
and,
if
they're
new
to
the
community,
we
can't
expect
them
to
be.
If
let
me
bug
you
later
on,
if
somebody
can
temporarily
get
me
the
requisite
permissions,
it
really
I
mean
really
talking
about
like
a
page-long,
Python
script.
I
think.
A
We've
got
that
actually
already
Jeff
I
just
put
a
link
in
to
all
the
issues
that
were
filed
like
an
ax
rule
against
testing
for
ax
and
I.
Think
I,
don't
know.
Six
or
seven
months
ago
we
decided
to
move
the
money,
but
from
one
repo
to
another
and
so
all
the
issues
we
want
to
know
the
issues
to
accompany
it
as
well.
So
you
look,
he
actually
just
did
an
all
under
his
username,
so
all
of
them
are
open
by
him,
but
it's
quoting
somebody
else.
A
D
B
C
And
any
subscribers,
but
not
necessarily
people
who
were
on
their
skin,
so
some
states
have
been
more
responses
than
others.
The
other
thing
is
this
has
come
up
for
the
example
site
as
well.
There
are
a
number
of
examples
that
got
kicked
out
into
separate
repo,
but
only
some
of
them
have
owners
or
maintainer.
So
if
like
what
do
we
do.
C
With
examples
that
nobody
has
picked
up,
I
think
just
in
general,
whatever
solution
we
use
as
long
as
there's
a
link
back
to
the
original
issue,
so
you
can
lay
a
trail
that
comes
for
whomever
comes
back
and
wants
to
pick
up
ownership
as
long
as
they
can
script,
it
scrape
it.
Whatever
follow
some
sort
of
audible
chain,
that's
pretty
much.
All
we
can
afford
I
mean
like
tied
back
to
the
original
issue
can
be
extremely
important
for
historical
context.
Sometimes
some
of
these
issues
had
a
lot
of
discussion.
C
A
B
Have
a
cross
referencing,
the
appspot
one
I
used
at
a
cross
reference
link,
and
it
did
all
the
right
you
know
and
ensure
to
you
know
it
pinged
the
original
person
to
know
that
had
been
moved
like
it
did.
A
good
job.
I
was
just
wondering
you
know
we
still
have
4100
issues.
If
that
was
going
to
be
something
that
it
was
going
to
be
an
issue
for
everyone.
Moving
forward.
B
C
Er,
yes,
I'm
sure
it
will
be
an
issue
for
people
moving
forward.
Please
share
your
experience
with
the
community.
Maybe
we
can
connect
to
dishes
here
in
the
meantime,
boring
developing
sort
of
close
of
the
original
issue
copy
pasted
over
to
a
new
issue,
a
political
issue
was
that
mean
by
human
hands
and
strip
that
we
have
that
we
use
for
lunch
data
I.
A
Just
on
the
other
hand,
we're
doing
really
well
Don
getting
sig
labels
onto
issues
now,
whether
it's
the
new
botnet
as
the
comments
asking
for
people
to
add,
like
mention
a
team
or
it's
a
change
in
culture
or
its
Manuel
Manuel
triage.
If
you
take
a
look
at
yes,
you
can
I,
don't
know
if
people
have
Saints
decreasing
rate.
There
are
a
number
of
highly
dedicated
individuals.
I
know
a
few
okay.
We
should
probably
call
them
out
and
say
thank
you.
A
C
Drop
it
by
about
a
hundred
a
day,
if
you
put
in
yeah
exactly
as
we
go
along
so
I
I,
don't
have
something
like
I,
just
sort
of
anecdotally
know
of
some
names
based
on
my
age.
I
think
I
was
trying
to
find
a
way
to
like
scrape
issues
that
had
it
and
then
no
longer
have
it
and
how
that
was
modified.
Some
people
changed
the
labels
directly.
C
A
C
A
So
I
Aaron
correct
me
if
I'm
wrong
here
at
the
Leadership
Summit
as
well,
there's
kind
of
a
push
towards
I
think
letting
that
six
sort
of
govern
themselves
in
a
lot
of
areas
and
it's
I
think
we're
trying
to
not
focus
too
much
on
top-down
leadership
and
specialist
specifying
these
policies.
If
you
have
to
do
this
or
do
it
this
way,
so
I
don't
want
to
tell
sig
how
they
have
to
triage
their
issues,
but
I
think
maybe
we
could
start
to
like
potentially
defined
policies
surrounding
issue
triage
at
the
sig
level.
A
So
something
like
they
have
to
be
looked
at
within
this
period
of
time
and
again,
we'll
need
metrics
and
dashboards
for
this
as
well.
But
does
that
seem
like
a
reasonable
approach
to
define
the
policies
and
what
the
SIG's
figure
out,
how
to
do
the
triage
themselves,
or
should
we
come
up
with
the
suggested
round
robin
and
someone's
a
build
cop
per
seviche
core?
Something
like
that.
C
The
end
of
the
meeting
here,
which
I
think
we
should
key
our
aim
is
like
hey
is,
is
roughly
speaking,
what
we
used
to
triage,
although
I
own
it
I
was
inconsistent
in
assigning
priority
labels,
because
I
sort
of
to
do
needs
sake
to
that
priority.
I
also
separated
a
broad
swath,
yes,
and
some
of
which
slaves
on
what
attracts
adopting
that
recent.
When
I
triage
top
the
issue,
each
state
will
eventually
need
to
make
a
charter
that
defines
in
concrete
terms,
so
it
does
it
does.
C
That
should
be
one
of
the
first
thing,
that's
ratified
by
the
steering
committee
when
that
comes
is
being.
But
again
that
feels
like
that's
going
to
be
something
that
pops
into
existence
sometime
in
August
timeframe
and
people
actually
ratifying
and
defining
getting
stuff
versus
whatever
problem
to
September
timeframe.
So
what
we
do
between
now
and
then
I
think
would
be
those
of
us
who've
done
creating
share
and
creating
our
names
in
terms
of
gaps
or
areas
that
were
unclear.
C
Sort
of
deal
like
the
low-hanging
fruit
is
find
the
things
that
should
have
been
supportive
requests
and
send
them
off
to
the
support,
then
see
if
they've
been
answered,
if
you
can,
but
each
of
this
thing
can
take
an
awfully
long
time
clearly
defined
support
policy
that
specific
versions
of
kubernetes
we
do
in
to
not
support
is
one
thing
we
could
do.
The
food
side
to
that
is
to
actually
be
responsive
to
support
requests
when
they
come
in.
So,
for
example,
there
are
a
number
of
issues
that
are
probably
like
aged
out.
C
If
you
say
that
we
no
longer
support
communities,
one
operation
or
not,
for
maybe
you
once
1.7
comes
ever
I-
was
great.
It's
still
vague
on
that
I
think
that's
a
good
escalate
to
the
community
question
and
just
use
a
blanket
cut,
offer
and
say
what
like,
when
we
support
quests
coming
unity
to
achieve,
sees
now,
issues
that
haven't
had
an
activity
in
a
long
amount
of
time.
C
I
they've
already
been
like
kind
of
things
once
and
slip
in,
and
they
have
an
e
disabled
with
life
and
in
pain
began
by
people
trade
out
and
on,
if
there's
still
no
real
impact
and
need
your
activity
on
that.
Probably
closing
with
the
reason
for
why
it's
because
from
there
and
I
feel
like
that's
kind
of
really
squishy
and
manual,
but
I
think
that
sort
of
takes
us
towards
the
end
of
July,
with
maybe
a
more
reasonable
set
of
issues
that
are
lining
up
with
various
sakes.
Long
events.
A
You
mentioned
is
important:
request,
I
think
is
a
kind
of
a
value
that
we've
had
or
policy
I,
don't
know
if
it
seems
like
the
community's
been
or
the
documentation,
I've
read
so
far.
It
suggested
we
don't
deal
with
support
cases
and
we
want
those
to
be
in
Stack
Overflow,
because
it's
more
discoverable
well,
it's.
C
Slightly
more
than
that,
I
always
tried
to
provide
a
link
to.
Unfortunately,
if
not
the
most
short
URI
I
swear
used
to
be
like
kubernetes
is
smashed.
You
can
really
now
it's
on
your
like
debug
applications
or
something,
but
it
doesn't
say
like
before
you
go
to
stack
overflow
by
the
way
we
also
have
managing
lists
and
we
have
stock
channels
and
if
you're
not
getting
answers
there
then
I
think
that
overflows
the
next
step,
so
they
don't
just
want
to
be
like
a
jerk
and
say
user
support
requests.
C
Give
us
that
overflow
fee,
but
I
do
want
to
like
you
know,
have
you
actually
done
your
due
diligence?
Is
this
obviously
just
a
question,
or
have
you
done
some
amount
of
effort
to
tell
us
how
to
reproduce
the
problem
so
on
and
so
forth?
Generally,
though,
the
link
to
the
document,
as
opposed
to
just
a
one-liner,
that's
a
stack
overflow
done,
seems
to
be
slightly
more
helpful.
Yeah.
B
C
C
Stand
out
to
newcomers
so
I
like
to
just
try
to
make
sure
that
my
interactions
are
explained
and
that
it's
not
just
a
full
stop
end.
The
conversation
make
sure
that
people
feel
like
they
can
continue
in
conversation.
We
open
the
issue
something
to
get
a
little
bit
more
engagement.
If
that's
Queenie,.
C
B
But
this
is
what
I've
been
using
as
part
of
like
going
through
the
cig
labels,
which
is
why
I
don't
know
if
you've
noticed
we've
cut
it
down
almost
half
the
size
it
was
before
once
we
started
to
use
it
actually,
every
day
I
was
like
I.
Don't
need
to
explain
in
half
this
guy
rajendra
morning
we
were
concerned.
It
was
too
long
and
no
one's
going
to
read
it.
So.
C
Held
exclusively
by
Googler
so
has
to
be
like
the
person's
you
have
power
to
shut
down
or
start
back
up
to.
Stick
you,
amongst
other
things,
do
manual
purchases
stuff.
In
parallel,
there
was
like
eight
user
support,
top
sort
of
duration,
who's
in
charge
of
actually
triaging
all
these
users
requests
that
seems
to
the
wayside
con
factor
the
bill
top
that's
sort
of
trying
to
come
back
up
and
so
sick
testing
we're
really
interested
in
making
that
bill
top
goal.
C
Something
that
many
people
just
can
do
so
like
possibly
trusted
individuals
who
are
trusted
not
to
have
like
Google
project
credentials
for
XQ
happens
to
be
running,
gave
it
a
login
to
which
are
your
close
circle
right.
The
user
support
proficient
probably
doesn't
need
nearly
as
much
trusted
credentials,
but
it
does
need
some
amount
of
trust
in
terms
of
having
a
good
base
or
providing
good
interactions
to
users.
C
So
I
think
that,
like
opening
up
look
at
those
irritations,
as
part
of
1/8,
would
be
okay
to
message
to
send
that,
like
we're,
trying
to
actually
document
these
positions,
give
strips
so
that
you
pretty
much
just
follow
the
doc
and
turn
it
or
you
can
write
on
that
to
triage.
So
it's
been
tremendously
helpful
and
so
I
also
see
that
as
a
warning,
both
industry.
B
So
what
I
did,
even
though,
by
the
third
or
fourth
time
you
have
that
memorized
I
kind
of
the
goal
is
to
remove
as
much
information
from
it
as
possible
to
make
it
smooth
and
then
so,
if
anybody
is
still
doing
this
active
triage
and
wants
to
run
through
it
a
few
times.
So
it's
not
just
my
strong
opinion
on
how
to
do
it.
That'd
be
great.
B
A
The
way
say
what
I
was
muted
there
myself,
sorry
about
that,
the
user
on
call
support
document
that
you
linked.
Is
that
going
to
be
in
addition
to
the
bug,
triage
octagon,
or
do
we
want
to
combine
those
through
into
one
thing
so.
B
B
Through
yeah,
like
I,
didn't
know
like
it's,
someone
like
I,
haven't
seen
anyone
mail
a
list
and
be
like
hey.
It's
on
its
user
rotation
support
time,
like
I
figured.
If
someone
was
actually
doing,
I
mostly
just
saw
the
template
and
figured.
It
was
a
good
way
for
the
obvious
ones
to
reuse
the
work
for
closing
issues.
A
B
A
E
This
is
kind
of
the
further
advancements
of
this
conversation
we
were
going
through
and
Signet
work
to
try
to
identify
people
that
are
part
of
different
components
within
cig
network.
So,
under
the
signal
umbrella,
you
give
things
like
DNS
and
the
KU
proxy
and
whatnot,
and
with
inside
of
there
there's
different
people
that
have
different
strengths
and
it's
trying
to
figure
out
how
to
route
both
issues
for
reviews.
We
of
course,
have
the
owners
file
for
other
signing
people
or
at
least
trying
to
entirely
select
people
that
should
review
this,
but
for
issue
tracking.
E
We
don't
utilize
that
so
for
for
that
stuff,
I
was
thinking
of
something
kind
of
like
the
kind,
a
cou
proxy,
for
example,
and
then
trying
to
pick
from
the
owners
file
or
a
list
for
that
type
of
scenario
and
then
kind
of
further
enhances
on
this.
This
is
kind
of
like
the
brute-force
like
the
web
application,
but
ideally
I
think
even
when
we
split
everything
into
different
repos
and
then
you
know,
for
example,
foo
proxies
over
here
DNS
already
a
split
outs.
E
People
are
still
going
to
google
kubernetes
communities
and
land
on
kubernetes
kubernetes
issue
page
and
file,
the
issue
there
anyway
and
nothing's
blocking
them
from
doing
that,
because
that's
just
the
easily
easily
discoverable
link
unless
there's
something
actually
forced
or
put
in
place
to
try
to
have
some
sort
of
direction.
For
example,
if
I
think
that
this
relates
to
cuvee
DM,
then
I
select
the
drop-down
box
of
rubidium
and
the
issue
ought
to
make
is
created
in
coop
ADM
we're
still
going
to
have
to
triage
anything
that
gets
past.
E
E
B
So
I
don't
know
one
of
the
things
I
so
like
I,
don't
know
if
any
of
you
use
Bugzilla
like
one
of
the
things
that
I
don't
like
about
doing.
That
is
that
it
forces
you
to
select
all
these
dropdowns
that
you
might
not
understand
what
each
of
them
mean
before
you
file
an
issue
which
can
be
both
good
and
bad
right.
Maybe
you
don't
want
a
bug
from
a
person
like
me
who
doesn't
understand.
E
Let's
say
the
majority
of
the
time
someone
has
at
least
a
little
bit
of
context.
If
not,
then
we
said
there's
still
going
to
be
a
person
that
goes
through
and
filter
these
things,
but
you
know
that
someone
has
to
go
through
day-to-day
basis.
I
mean,
even
if
all
of
us
on
this
call
take
five
that
probably
one
suit
the
amount
of
issues
that
created
a
day
to
day
and
more
India
the
existing
templates
directly.
E
Yes,
they
should,
but
they
don't
and
time
and
time
again,
we've
seen
the
first
thing
someone
does
when
they
click
the
Create
issue
boxes.
They
delete
everything
that
is
required
in
there,
and
then
someone
comes
back
in
after
it's
been
triage
and
asked
them
to
place
that
back
in
night,
like
what
Googlers
are
using,
what
operating
system
are
you
using.
A
E
Think,
ideally,
once
we
split
a
lot
of
these
things
out
being
able
to
find
the
correct
people
inside
of
a
sig
for
that,
individual
component
will
be
easier.
But
for
now
we
don't
have
anything.
You
just
say
sig
network
and
then
the
issue
triage
er
up
to
the
treasurer
for
the
sig
to
figure
out
who
they
think
is
responsible
for
coup
proxy
or
network
policy,
or
something
like
that.
D
E
A
Problem
like
I,
want
to
push
this
out.
This
does
seem
like
something
that's
pretty
important
and
it's
getting
grown
importance
as
we
move
to
multiple
repos,
because
we
don't
want
people
searching
kubernetes
having
an
issue
with
cuba,
QA
DM
or
with
helm
or
whatever,
and
filing
the
issue
right
into
the
kubernetes,
repo
and
then
sort
of
having
to
recopy
and
then
we'll
we'll
definitely
need
that
bulk
copy
tool
or
something
I
would
just
can
be
annoying
to
run
regularly.
Instead,
we
could
have
people
help.
You
know
issue.
D
I'd
say
first
we're
up
against
the
limitations
of
github
issues,
but
leaving
that
that
aside,
my
suggestion
would
actually
be
a
two-pronged
approach.
One
is
you
know
we
have
the
new
pitbull.
Your
issue
goes
interface
up.
We
get
people
across
the
kubernetes
community,
leadership
to
actually
blog
or
otherwise
put
that
in
social
media
in
order
to
boost
the
little
rank
of
that.
D
The
you
know,
and
just
remind
because
part
of
google
rank
is
how
many
people
have
referred
to
something
so
get
that
URL
all
over
the
place
and
then
put
in
the
templates
for
the
various
projects.
You
know
hey.
Did
you
come
directly
to
this
project
or
you
know?
Did
you
know
about
our
issue
filing
interface?
If
you
don't
use
this
issue
filing
interface,
it
may
slow
down
the
response
to
your
issue.
E
Yes,
I
forgot,
I
thought
I
read
something
that
could
github
recently
added
some
sort
of
support
for
who
it
could
actually
file
issues.
So,
potentially
you
could
close
down
the
issues
say
for
the
bot
that
is
responsible
for
creating
them
through
the
web
interface.
It
can
still
refer
to
them,
but
you
can't
create
them.
C
You
know
I
I
have
a
difficult
time
if
the
kubernetes
projects
dedicated
to
give
up
everything
you
saw,
locator
I
only
understand
them,
you're
open
to
a
significant
amount
of
relationships.
Issue
management-
this
hasn't
been
really
adequate
for
us
for
quite
some
time,
but
I
am
concerns
it
using
something
other
than
this
will
not
fly
the
posterior
Committee,
so
I'm,
trying
to
figure
out
in
this
brave
new
multi
repo
world.
C
We
are
likely
not
the
first
project
that
has
had
to
deal
with
this
to
take
the
other
problem
and
I
would
be
interested
in
a
survey
of
what
other
good
projects
have
done
as
prior
are
I
mean
if
the
survey
is
that's
what
they
all
use
an
external
machine
in
issue
system.
Maybe
that
would
be
better
evidence
or
here
in
committee,
but
they
could
like
just
trying
to
think
of
something
that
stays
purely
from
the
confines
of
github.
You
could
have
the
bot
or
a
human
be
more
aggressive
about
shutting
an
issue
down.
C
It
doesn't
conform
to
the
template,
we're
trying
to
make
a
template
as
easily
and
simply
understood
as
possible.
You
hope
that
have
a
be
machine
possible
or
something
and
half
a
bottle
and
it
as
opposed
to
like
what
happens
now,
where
people
don't
really
file
a
useful
issue,
and
so
it's
just
straight
up
ignored.
It
might
seem
unfriendly,
but
they
would
get
faster
feedback
o'clock
like
no,
not
enough
information.
It's
not
the
information
to
ask
for
clothes
yeah
I'm
going
to,
but
it
would
be
more
immediate
feedback.
It
is
something
else.
C
D
B
Most
of
the
larger
projects
seem
to
have
spun
off
their
tissue
tracking,
but
let's,
let's
not
go
down
that
road,
yet
I
think
Erin.
What
you're
getting
around
to
is,
if
we
just
make
the
bot
be
a
little
bit
more
hardcore
when
it
comes
to
validating
input,
I
guess
right!
If
we
just
match
it,
it
up
a
little
bit
and
then
see
what
happens.
B
C
Me
know
the
horrible
extreme
will
be
like
fill
out.
This
yellow
blob
right
because
you
know
that's
possible
and
it's
kind
of
human
readable
but,
like
you
know,
I
understand
that
they're
kind
of
motivations
that
case
happens
is
all
about
like
how
do
we
do
way
more
than
what
github
offers
right
now,
so
I?
Don't
really
have
a
great
answer:
I
just
I'm,
just
cautioning,
but
I
suspect
the
steering
committee
can
say
like
if
we're
all
going
to
get
cut
or
not.
C
And
so
let
me
constructive
thing
we
can
do
is
attempt
to
document
the
requirements
we
want
like.
Where
is
it
precisely?
Github
is
sufficient
for
us,
and
so
we
can,
you
know,
be
able
to
articulate
what
he's
this
thing
do.
You
might
want
to
bolt
on
or
used
to
circumvent
github
system,
because
I
also
think
it
is
from
the
perspective
of
what,
if
there
was
something
like
get
hat,
get
lab
or
some
other
solution
that
we
could
ultimately
move
to
if
it
met
all
of
our
requirements,
because
this
is
what
one
of
many
areas.
B
C
Okay,
so
it
seems
I'm
not
even
on
the
bottom
part
or
like
I,
said
for
me
again.
It
comes
back
to
like
what's
the
users
first,
interaction
with
the
project
yeah,
so
at
least
at
the
bot
could
close
and
wait
to
an
example
with
something
that's
been
filled
out
well
and
yeah.
That's
why
right
something
at
least
a
little
you
strength
you
right.
B
Or
if
we,
even
if
you
do
close
it,
if
you
explain
it,
look
I
had
to
close
this,
because
it's
missing
these
things
before
human
looks
at
this
we're
just
trying
to
be
efficient.
Please
click
here
and
follow
these
things
out,
Thanks.
You
know
at
least
explain
for
the
person
that
is
nothing
personal
right
and
that,
yes,
the
the
new
the
new
issue.
Template
is
also
land
got
merged
two
days
ago,
I
think
I,
don't
know
if
it's
logged
yet
I.
C
B
Me
see,
I
wrote
it
and
I
forgot
like
it
is:
I
won't
actually
file
a
lot
of
new
issues.
Here
we
go,
I
mean.
C
The
rest
was
having
a
bob
be
super.
Strict
is
there's
only
so
much
engineering
effort.
We
can
have
to
prepping
the
most
friendliest
error
message
that
says
this
is
the
specific
thing
that
you
must
slap
on
and
at
some
point
you
know
you
don't
want
to
writing
a
compiler
or
an
issue:
template
yeah,
but
at
least
being
able
to
link
to
an
example
having
to
bought
be
a
little
bit
more
aggressive
in
closing
issues
and
having
some
kind
of
role
where
humans
do
survey
to
boss.
C
B
D
B
B
E
Wanted
to
remove
the
Bugzilla
I
regretted,
putting
it
in
there,
but
I
wanted
it's
just
use
it
as
an
example.
I
I
know
means
that
I
think
that
we
should
operate
outside
of
the
tools
that
we
have
just
looking
at
ways
to
enhance
so
and
then,
if
we
do
aggressively
your
clothes
at
least,
someone
can
then
just
be
mad
at
a
robot
instead
of
an
individual
person.
That
kind
of
helps
a
bit.
E
C
My
we
have
learners
in
an
ideal
world.
All
those
code
would
be
hierarchically
organized
into
directories
with
the
owners
files
correspond
to
each
Singapore
set
of
people
who
care
in
a
second,
but
that
for
a
favorable
request,
not
so
much
for
issues,
issues
you're,
just
a
lot
of
a
thing,
I,
don't
really
know
which
piece
of
code
specifically
I'm
talking
about
right
now,
I
think
that
great
broader
discussion
to
have
and.
E
I
think
I
imagine
like
this
would
be
more
of
like
the
sig
trios
are
themselves
now
further
triaging
it
to
who
they
believe
is
responsible
for
that
I'm.
Imagine
the
original
author
of
the
issue
would
be
adding
these
unless
they
actually
did
know,
but
how,
aside
from
just
a
sung
in
the
names
and
the
person
triaging
within
the
sig
being
involved
with
the
project
long
enough
that
they
know
the
exact
name,
is
how
can
you
further.
C
B
A
C
At
one
point,
I
think
they're
also
aliases,
you
can
use
an
owner's
Amyas.
Five
I
can
that's
code.
Centric,
that's
not
issue.
Centric
from
the
concern
is,
if
keeps
it
to
create
its
own
set
of
labels
for
the
sub
components
that
it
owns,
but
everything
is
still
living
in
one
big
tree,
we're
going
to
explode
from
one
hundred
and
sixty
something
labels
yeah
now
to
something
even
larger
yeah,
like
the
best
direction
to.
E
A
A
Don't
think
every
single
a
dude
in
that,
so
it's
like
a
partially
filled
out
matrix,
but
that's
what
kind
of
what
you
were
talking
about
with
that
mapping
earlier,
because
I
think
most
of
us
that's
how
we
triage
your
shoes,
we're
like
okay,
yeah
we're
talking
about
it
can
cook
like
scheduler
that
probably
fits
under.
You
know
this
sick
or
something
like
that.
I
see
certain
keywords:
I
yeah.
C
B
C
B
B
D
But
the
thing
is:
we're
going
to
encounter
stuff
more
situations
already
absolutely
trying
to
figure
out
where
an
issue
belongs
right,
because
it's
really
hard
to
look
at
a
set
of
charters
and
figure
out.
What's
not
covered
right,
you
can
figure
out
where
overlap
is,
but
it's
hard
to
figure
out
what's
not
covered.
But
when
you
have
an
issue
when
you
say
hey
this,
an
issue
with
a
vagrant
box
and
no
one
will
take
responsibility
for
it.
Then
you
know
an
area,
that's
not
covered.
A
Let's
move
on
there's
a
few
other
things
that
I
wanted
to
talk
about:
I,
LC
and
I
reviewed
the
one
seven
planning
and
we
had
kind
of
four
major
themes
that
we
wanted
to
address:
consistency,
automation,
Docs
and
metrics
and
feedback.
We
did
a
few
things
pretty
well.
I
was
looking
through
on
a
consistency
side,
there's
a
PR
that
got
open
to
move
a
standard
subset
of
the
labels.
A
I
think
we
there's
a
discussion
about
which
labels
should
be
moved
to
all
sig
to
all
repos
I
think
we
said
we'd
start
small,
but
it
ended
up
getting
stuck
at
some
level
and
I
need
to
follow
up
on
that.
I
know
the
guy
who's
working
on
it,
but
I
think
he
just
had
problems
getting
it
to
run
in
prowl.
As
like
a
like
a
regular
regular
job,
we
did
do
some
work
on
the
bot
consistency.
A
I
know
I've
seen
a
few
things
that
have
changed
the
command,
so
we
don't
have
the
at
Cade
spot
or
that
might
still
work,
but
we
suggest
you
start
using
like
slash
okay
to
test
and
that
way
we're
just
a
little
bit
more
consistent
across
the
board
and
those
aren't
updating
the
docs
as
well.
So
I
think
we
did
well
there
as
far
as
automated
automation,
so
pinging
me
an
active
reviewers
yeah
this.
A
This
had
a
lot
of
discussion
about
whether
we
should
reassign
and
some
people
were
strongly
opposed
to
having
the
bot
reassign
automatically
I
think
it
ended
up
being
that
the
bot
would
find
an
additional
reviewers
for
you
and
reach
out
to
the
author
and
suggest
the
author
ping
that
person
individually
or
assign
them
themselves
just
so
that
we
can
end
people
and
make
it
more
have
more
human
involvement.
But
that
is
done
and
our
templates
got
updated.
I
think
are
the
feature
feature.
A
A
Sorry
I
jumped
a
few
automatically
tree
auditing
issues.
We
need
to
review
this
a
little
bit
better.
I
wanted
to
talk
to
the
tensorflow
team.
We
could
probably
talk
to
docker
just
to
understand
how
like
Gordon
works,
or
you
know
what
what
our
approach
should
be
here
for
for
Auto
triaging,
but
it
looks
like
this
isn't
actually
the
biggest
issue
right
now.
A
A
Yeah
we
still,
we
still
have
some
trouble
with
visibility
into
the
minor
releases.
I
think
that
the
suit
release
team
is
working
on
this
and
I
know
Don
put
some
effort
into
asking
that
all
PRS
and
associated
issues,
whether
or
not
that
was
immediately
helpful-
remains
to
be
seen.
But
there
there's
work,
that's
been
going
on
there
and
hey.
B
I'm
great
can
I
bring
one
thing
about
that,
because
there
was
an
unresolved
issue
that
we
had
about,
that.
That
was
really
vexing.
That's
as
we
split
out
repos
and
issues
start
popping
up
in
other
repositories
seems
like
we're
going
to
have
a
ton
of
duplicate
work
for
contributors
to
enter
in
those
issues
in
say
the
cloud
provider,
repo
and
Azure,
or
in
for
gke
or
not
I,
mean
it's
going
to
be
spread
all
over
the
place.
B
A
I'm
not
sure
that
I
understand
the
issue,
because
I
think
PRS
could
link
to
an
issue
in
any
brief.
Oh
it
wasn't
that
it
needed
to
be
a
kubernetes
issue.
If
you
created
a
PR
code
that
was
going
into
the
1.7
branch,
they
just
wanted
you
to
reference
an
issue,
so
it
could
have
been
in
you
know
in
a
totally
separate
repository.
Is
that
what
you're
asking
about
yeah.
A
A
A
B
A
A
I'm
kind
of
at
a
loss
for
what
we
can
do
there,
I
don't
have
enough
experience
doing
it,
I
think
it
would
have
been
favored.
Something
I
need
to
be
more
involved
with
so
I
understand
the
problem,
a
little
better,
but
my
prospective,
and
we
should
certainly
discuss
a
residence
or
a
tie,
goes
for
share.
Okay,
the
big
thing
from
1:7
that
I
think
we
didn't
we
didn't
get
to
achieve.
Wealth
is
metrics
and
visibility
into
the
projects.
Help
I
still
think.
A
There
are
a
number
of
tools
out
there,
including
smick
you,
including
velodrome,
including
staff.
How
now
including
Pattaya,
there's
so
many
dashboards,
and
they
still
don't
someone
they
don't
answer
all
the
questions
that
we'd,
like
you,
know,
rocky
our
velocity,
and
it
makes
it
hard
for
us
to
understand
the
impact
of
various
policies.
So
our
the
new
bug,
triage
policies,
actually
helping
us
reduce
the
number
of
of
you,
know
open
issues
or
the
the
speed
at
which
issues
are
closed
and
how
fast
PRS
are
reviewed.
A
Things
like
that,
so
41.8
that
was
kind
of
close
one
of
the
thing,
but
I
think
we
could
really
stand
to
to
work
on
it
and
improve
on
and
I
guess.
On
that
note,
maybe
we
can
talk
a
little
bit
more
about
what
we
want
to
do
in
1.8.
If
other
people
have
ideas,
I
think
so
the
bug
triage
is
probably
something
we
want
to.
We
want
to
close
out
issue
triage.
Is
there
metrics
and
project
health?
Is
there
anything
else.
A
Okay,
I've
got
flakes
written
down
here
and
Erin
I
think
we
had
talked
about
like
test
flakes
I
talked
with
aggravation
about
having
dashboards
and
test
grid
so
that
ever
it
was
clear
which
sig
on
every
test.
Every
test
was
own
biasing,
and
this
is
something
that
can
be
done.
Kind
of
an
automated
fashion.
Apparently
Ryan
pitchman
is
just
working
on
this
I
thought.
C
Yeah,
it's
that's
still
a
little
wonky
there's
like
a
test
center
CSV
file
and
then
kind
of
scared
of
section
like
straight
that
down
to
the
set
of
common
prefixes
that
you
can,
then
they
say
getting
ready
task
up
against
with
this
word
and
then
that
could
be
used
to
elsewhere
to
generate
stuff.
We,
in
fact,
at
one
point
about
like
dropping
the
Sigma
in
itself,
into
the
test
text
for
the
test
case
named
as
link
to
avoid
that
is
possible,
really
need
to
work
on.
C
But
all
of
that
tough.
It
is
ultimately
like
the
pre
outing
stuff
that
Ryan
is
working
on
the
to
talk
about
just
written
by
manually.
Human
curated
CSV
file
finding
a
way
to
make
that
better.
We're
involved.
We've
structuring
the
test
code
into
directories
to
clearly
define
the
owner
styles
that
use
SIG's,
not
names
because
tie
back
to
the
book
triage.
Think
for
a
second
like
one
thing,
I
would.
C
Is
trying
to
find
the
relevant
piece
of
code
and
then
we
look
at
the
owners
file
in
the
directory
and
that
would
usually
get
the
user
names
and
then
I
use
those
usernames
and
guess
their
Security's
organization
and
find
out
which
teams
they
belong
to
and
then
those
teams
reverse
engineer,
probably
which
they
actually
came
right.
I'd
love
to
meet
those
worst
I'll
start
to
include
more
direct
mapping
to
cig,
if
possible,.
A
So,
even
beyond
the
donors
file,
he
was
suggesting
that
every
single
test
source
code
file
just
have
a
comment
and
we
can
enforce
this
pretty
easily
right,
like
a
bot
been
Auto
checks
for
this
comment.
That
said,
you
know
owner
equals
this
cig
yeah,
so
I'm
totally
fine
with
that.
The
directory
structure
is
fine
with
me
too,
but
either
way
I'm
just
having
it
super
clear
which
sig
owns
each
time
and
when,
if
a
tent
did
you
know
then
say
like
okay?
Well,
this
tests
not
going
to
pass
we're
going
to
release
without
this
test.
B
C
So
that's
the
super
granular
level
to
take
it
up
a
little
higher
there's,
which
jobs
do
which
take
them
right
now.
Define
signal,
don't
know,
go
signal
for
release
right
people
go
to
look
at
test
great,
and
there
are
a
bunch
of
jobs
that
have
red
and
green
boxes.
For
them.
Tell
me
about
the
the
jobs
themselves,
not.
C
C
The
save
node
team
will
probably
own
all
the
notes
e
to
e
test
jobs.
Things
of
that
nature,
where
it's
going
to
get
a
little
trickier,
are
the
like,
slow,
clicky,
disruptive,
she'll
reboots,
upgrade
jobs
like
finding
six
for
those,
but
that's
another
way
to
we
can
at
the
job
level,
not
necessarily
which
they
did
it
in
test
case
at
and
be
flaking,
but,
like
you
can
expect
a
family.
The
question
is
that
that's
going
to
be
higher
signal
or
like
more
values.
A
A
And
then
the
last
thing
that
I
kind
of
wanted
to
talk
about
was
this
just
fixed
it,
so
it
was
related
to
words
talked
about
for
test
ownership,
but
it
seems
like
a
lot
of
that,
can
be
automated
and
testing
my
skills
with
take
care
of
it
fix
it
work
just
in
case
people
haven't
heard
this
term
before
it's
something
I
think
the
kubernetes
community
tried
last
year.
Maybe
it's
tried
one
one
other
time
since
then.
A
It's
generally
something
that
work
that
can
be
parallelized,
pretty
easily
kind
of
mechanical
work
and
things
that
people
say
it's
not
that
important
will
I'd
rather
be
working
on
a
feature
right
now.
If
given
the
choice,
so
we're
not
going
to
force
anyone
to
participate
into
this,
but
we're
going
to
try
and
make
a
community-wide
effort
and
Google
is
going
to
make
a
like
team-wide
effort,
at
least
to
I,
used
to
focus
on
cleanup.
And
what
do
we?
What's?
The
term
that
samake
me
always
uses?
A
Carry
water
chomp
Wiggin,
carry
water
sort
of
things
like
that.
That
being
said,
some
of
the
mechanical
work
or
the
things
that
I
had
come
up
with,
seems
like
they
can
be
automated
pretty
well.
Do
we
have
ideas
for
things
that
could
strongly
benefit
the
contributor
experience
that
we
could
use
120
or
after
senator,
however,
much
manpower
for
a
week's
worth
of
time
to
get
done?
C
A
A
E
I,
just
following
up
from
the
last
part
for
the
kind,
X
and
or
mapping
teams
to
Auto
responding
to
those
types
of
things.
Would
a
mailing
list
post
be
better
for
discussing
that
Tube
stuff
and,
if
so,
should
I
keep
it
within
Trebek's
or
put
it
directly
to
developers
since
its
kind
of
cross-cutting
developer,
okay
got
it.