►
From YouTube: Kubernetes SIG Storage - bi-weekly meeting 20210225
Description
Meeting of Kubernetes Storage Special-Interest-Group (SIG) - 25 February 2021
Meeting Notes/Agenda: -
Find out more about the Storage SIG here: https://github.com/kubernetes/community/tree/master/sig-storage
Moderator: Xing Yang (VMware)
A
Right,
hello,
everyone
today
is
february
25th
2021.
This
is
the
kubernetes
sojig
meeting.
So
today
we
will
go
over
our
planning
spreadsheet
to
check
the
status
before
that.
I
just
want
to
remind
everyone.
Code
freeze
is
coming,
it
is
march
9th,
so
it's
coming
soon
yeah.
So
if
you
are
working
on
feature
make
sure
if
there
is
a
co-developed
entry,
they
must
be
merged
by
march
9th
to
make
it
into
1.25
release,
and
then
there
are
also
two
other
things
that
I
want
to
mention
quickly.
A
The
first
one
is
that
so
there
is
a
we
can
now
do
this
opt-in
feature
block,
so
that
means
every
feature
owner
if
you're
working
on
a
feature.
If
you
want
you
write
a
blog
for
that
feature
in
1.21
release,
you
can
choose
to
do
so.
So
a
please
pin
side
michelle
young
or
myself.
A
So
we
can
add
your
feature
in
this
spreadsheet.
Then
the
deadline,
I
think
it's
we
need
to
add
that
by
march
1st,
and
then
you
need
to
write
up
the
blog
draft,
at
least
by
march
31st
for
review.
So
that's
the
first
one.
A
Yeah
just
they,
they
just
need
to
have
this
one
tracked
there
by
match.
First,
I
think
it's
offline.
Let's
say:
if
you
add
it
there,
then
you
somehow
somehow
it
didn't
make
it
a
match.
Nines.
I
think
that
you
can
remove
that.
I
don't
think
that's
a
problem
yeah.
So
if
you
yeah.
A
Thanks,
okay
and
then
the
next
one
is
that
we
are
so
basically
this.
The
steering
committee
requires
every
seat
to
do
this
annual
report,
so
we're
working
on
that
and
the
pr
will
be
out
soon
for
review,
so
that
will
be
sent
out
in
the
sixth
order,
mainly
list
all
right.
A
D
Yeah,
so
we
are
still,
I
started
the
feature
development
actually,
but
we
had
a
meeting
yesterday
to
discuss
the
fs
group
and
csi.
We
needed
james,
so
be
it
kind
of
blocked
on
the
csi
spec
change
at
the
moment.
D
A
E
B
B
D
A
Right,
the
nine,
so
you
need
to
have
that
chain
before
the
ninth.
D
G
A
Next
one
is
the
csi
online
offline
resizing
fix
issues
command.
Do
you
have
any
update
on
this.
D
Yeah,
so
there's
a
pr
by
song
in
and
for
an
external
resizer
that
I've
been
reviewing
and
there's
a
pr
in
kubernetes
main
entry
as
well
that
changes
the
that
pvc
being
deleted
where
resize
is
pending,
that's
the
one
that
you're
trying
to
fix
for
other
things,
the
allow
volume
expansion
or,
like
the
the
sc,
moving
that
field
from
storage
class
to
pv
or
recovery.
It
requires
cap,
so
I'll
target
next
service.
So
that's
the
status.
A
A
Thanks
so
okay,
so
we're
still
yeah
we're
staying
data,
okay,
so
the
the
those
other
caps
are
also
needed
before
we
move
this
to
ga
right.
D
Yeah,
okay,
at
least
we
had
a
call.
We
discussed
that
this
resizing
cannot
go
ga
without
recovery
from
research.
Oh.
A
D
A
Thanks
next
one
csi
entry
read
only
handling
okay,
I'll,
say
chang
you're
working
on
this
with
humboldt-
or
I
don't
know
if
rambo
is
a
yeah.
C
Yeah
I
haven't
got
it
so
homo
was
able
to
reply
to
me
a
couple
a
couple
of
weeks
back,
but
I
haven't
gotten
a
hold
of
him
ever
since,
but
on
my
side
I
should
be
able
to
have
time
to
work
on
it
within
the
next
week
or
two
if
he
doesn't
he's
not
able
to
get
to
it,
but
yeah.
This
one
is
not
bound
by
code
freeze
because
it's
in
the
external
provision,
I
think
boy.
A
A
Next,
one
is
the
storage
capacity
tracking
for
pot
scheduling.
Do
we
have
patrick
here.
G
G
Yes,
the
the
I
think,
spec
changes
would
potentially
imply
api
changes
in
kubernetes.
A
Okay,
thank
you.
So
we
need
to
get
this
see
this
back
release
resolved
next
week.
I
think,
because
then
yeah
march
9th
is
coming
the
week
after
next
week.
A
Yeah:
okay,
next
one
is
the
pvc
inline
ephemeral
volume,
so
this
is
also
patrick's
misha.
Do
you
know
if
this
one
has
any
pending
things.
G
I
think
john
has
been
reviewing
a
couple
opr's.
I
think
otherwise.
This
seems
in
good
shape.
Right,
I
don't
think
there's
any
major
pending
things
remaining.
Is
that
right,
yawn
who
disappeared.
A
All
right
next,
one,
okay,
so
spreading
over
failure
domains.
So
right
now
no
update.
This
one
depends
on
the
next
one
or
in
group
api,
which
we
did
have
a
review
meeting
just
yesterday.
So
we
got
some
good
feedback.
Our
schedule
follow-up
view
meeting
on
that.
G
F
G
A
Thanks
and
then
next,
the
next
two
nfs
provisioner
and
fs
client
professional-
do
we
have
a
covering
here.
G
I
think
karen
sent
an
email
update
this
morning.
G
I
think
he
said
the
subdir
one
is
done
and
complete.
There's
like
a
new.
G
Version
released
nfs
client,
oh
this
one's
after
okay,
yeah
that
one's
complete
and,
I
think
released,
I
think,
he's
still
working
on
the
nfs
provisioner.
One.
G
Okay,
oh
john's
back
john,
do
you
know
the
status
of
the
ephemeral
volumes.
H
D
Yeah,
I
reviewed
that
actually
and
that's
waiting
for
seconds
like
it
it
it's
it's
it'll
get
merged
in
today
or
tomorrow.
It's
just
very
like
a
small
change.
The
second
instrumentation,
like
we
decided
to
like
not
create
too
many
time
series,
and
it's
and
patrick,
made
the
change,
so
it
should
get
into
there
tomorrow.
F
Yeah,
I
think
that
the
next
step
for
this
is
we
need
to
have
a
meeting.
I've
put
a
link
to
a
doodle
inside
the
main,
stick
storage
planning.
So
if
you're
interested
in
attending
that
meeting
feel
free
to
include
your
email
address
and
select
the
times
that
work
for
you
most
of
these
are
scheduled
in
the
afternoon
next
week,
eastern
time,
but
there
are
a
few
in
the
morning,
so
we
need
to
get
together
and
discuss
how
best
to
move
forward
with
this.
D
Christian,
did
you
want
to
you?
Should.
A
I
did
that
later.
Yes,
this
is
he
just
added
this
one,
while
we
were
pinned
so
yeah.
F
F
A
D
A
All
right
thanks,
open
house
yeah,
so
there
is
actually
a
pr
out,
it's
being
reviewed,
so
we're
getting
some
feedback.
I
think
waiting
for
some
feedback
from.
A
I
don't
know
if
this
person
is
in
signal
or
sick
instrumentation,
so
just
trying
to
get
some
feedback
from
him
and
then
one
decide
how
to
proceed.
There
are
just
some
discussions
on
how
to
implement
this,
so
there
are
several
ways
I
just
want
to
find:
what's
the
big
best
way
and
the
next
one
okay,
so
this
is
the
one
in
populator
then.
B
All
right,
I'm
off
me,
so
the
this
cup
was
approved
provisionally
by
tim,
and
we
had
a
follow-up
meeting
actually
this
week
where
he
he
showed
up,
and
we
talked
about
what
to
do
about
some
of
the
thornier
api
issues.
B
B
So
I
have
to
like
reformat
the
old
cap
and
write
a
new
cap
and
make
an
exception
request
and
they've
said
that
they'll
grant
the
exception
request.
But
then
I
have
to
do
all
that
and
get
the
code
done
and
do
it
by
a
week
and
a
half.
So
it's
a
lot
of
work
to
try
to
you
know:
do
this
the
right
way,
but
I'm
that's
what
I'm
working
on.
G
Thank
you,
oh
ben,
can
you
go
ahead
and
point
me
to
the
code?
That's
problematic.
I
want
to
take
a
look
at
that
because
just
just
to
see
if
we
can
save
you,
some
work.
B
Okay,
yeah:
I
can
dig
up
the
admission
controller
stuff
that
that's
the
source
of
all
these
problems
and
I'll
pay.
Em
you
on
slack.
G
A
A
Yes,
I
think
team
reviewed
it
and
then
this
is
still.
This
is
it's
not
merged
in
the
cabinet,
merch
team
and
not
anyone.
The
team
had
some
comments.
There
team
had
to
join
the
meeting,
so
I
think
there's
just
some
discussions
going
on
for
this
sad.
Do
you
have
any
update
for
this.
E
Cozy
so
cozy
missed
the
kept
deadline.
E
They
were
able
to
get
in
touch
with
tim,
to
get
some
feedback
and
they're
going
to
continue
to
work
on
the
feedback
that's
been
given
and
try
to
come
to
an
agreement
on
how
to
proceed.
There's
going
to
be
a
meeting
immediately
after
this
meeting
to
discuss
some
of
that.
If
you're
interested
feel
free
to
attend.
A
Thanks,
oh,
I
think
it's
a.
I
think
we
asked
her
someone
to
come
talk
about
the
gateway
api.
Remember
that
was
the
last.
E
Yeah,
the
networking
folks
have
a
gateway
api
that
they're
to
present
at
some
point.
A
Next,
one
is
the
change
block
tracking
is
found
here
yeah,
so
I
think
we
reached
out
to.
I
think
it's
I'm
not
sure
how
to
say.
I
bought
a
burkha,
and
I
say
your
name
right.
So
yes,
I
broke
okay,
thanks
yeah,
so
you
said
you
reached
out
to
some
some
engineers
to
take
a
look
to
review
that
stock
right,
but
you're
still
trying
to
get
some
feedback
from
them.
B
A
Worth
yeah,
so
I
let's
actually
talk
about
this
in
our
next
data
protection,
good
meeting.
A
Interest
yeah:
there
are
some
additional
issues.
So
let's
talk
in
that
meeting,
okay,
yeah!
I
was
going
to
talk
about
that
in
this
meeting
yesterday,
but
then
there's
something
else
comes
up,
so
I
think
so
we
need
more
time
yeah.
So
so
the
the
feedback
we
want
from
on
our
broker
is
is
really
regarding.
A
A
G
G
Yeah
he
has
a
provisional
cap
out.
Okay,
I
mean
I,
I
took
a
look
at
it.
It
looks
fine
for
a
provisional
cup,
it
just
sort
of
states
the
problem
and
so
he's
working
on
adding
more
design.
Details
to
that.
I
I
H
I
am
here
I
posted
the
pr
to
translate
from
csi
to
entry.
Is
anything
else
missing?
Oh
yeah,
that's
the
that's!
A
bug
fix.
I
For
that's
par
that's
kind
of
part
of
this,
but
for
this
one
we're
looking
to
update
the
the
the
pv
label,
the
topology
label,
it
used
to
be
affiliate
domain
beta
label,
but
now
we're
updating
that
to
the
ga
label.
So
right,
yeah,
that's
something
we
we
need
to
follow
up
and
also
on
the
azure
other
editor
file.
We
also
need
to
add
another
pr.
I
H
D
All
right
did
we
deprecate
the
azure
files,
behavior
change,
that
we
were
planning
to
sorry
if
this
was
like
the.
D
G
I
don't
think
we
did
andy
did
have
a
pr
out
for
it
and
I
think
it
merged
already,
but
it
doesn't
have.
It
doesn't
say
that
in
the
release
notes,
we
can
go
back
and
ask
andy
to
add
it
to
the
release,
notes
but
yeah
this.
That
item
should
be
tracked
as
part
of
azure
file
migration.
Maybe
not
this
one.
A
I
I
think
leila
you're
gonna
work
in
on
azure
right.
G
A
All
right
next
one
is
the
csm
migration,
csm
migration
core,
so
who's
working
on
this
w
or
mad.
I
Yeah,
so
for
the
core.
Last
two
weeks
we
had
all
the
matrix
pr
merged.
So
now
we
have
the
matrix
for
the
csi
migration
I'll
talk
more
about
the
matrix
in
tomorrow's
csm
migration
meeting.
I
A
Okay
thanks
next
one
is
csma
question
vsphere,
I
don't
know
if
divia
is
here,
so
I
think
we.
A
With
the
sick
architecture,
and
then
I
think
we
we
actually
kind
of
got
an
exception
from
them.
We
can
actually
deprecate.
So
dvr
has
a
pr
out
that
added
some
duplication
notices.
So
I
think
that
one
does
anyone
still
have
any
concerns
on
that
or
can
be?
Is
that
good
to
be
merged.
A
Okay
yeah,
so
I
you
know,
okay,
I
think
andrew
also
approve
it
okay,
so
with
that
case,
then
I'm
going
to
go
ahead
and
merge.
It
then,
because
I
see
I
think,
there's
a
I
forgot.
Who
is
that
there
is
a
look
good
to
me
also
from
some
some
folks
as
well,
so
all
right,
thanks
and
next
one
all
right,
so
this
one
is
crossed
out
and
as
you
file
okay.
So
whichever
you
want
to
talk
about
this
one
here
or.
I
Oh
yeah,
so
for
azure
desk
another
file,
I
asked
andy
and
he
said
the
plan
is
to
on
by
default
in
1.22
in
open
source
and
the
ga
migration
in
1.93,
both
for
either
desk
and
error
file.
G
This
is
called
across
out.
No,
no
sorry,
I
mean
two
columns
in
column,
l.
A
G
I
think
the
other
is
that
we
should
update
the.
We
should
update
the
release,
notes
for
the
pr
that
flipped
the
feature
gate
to
also
deprecate
the
behavior.
That
is
talking
about.
A
Okay,
so
so
java,
you
are
working
on
this
you're
going
to
update
that
or
maybe.
D
D
D
This
is
just
one
that
releases
to
capture
the
behavior
like
both
of
this
stuff.
Okay,.
G
Yeah,
I
guess,
can
you
help
ping
andy
to
to
talk
about
it,
since
I
think
you're
familiar
with
exactly
what
needs
to
be
deprecated.
D
I
I
I
can
update
so
okay,
we're
still
planning
to
on
by
default
in
1.21,
so
currently
we're
working
on
fixing
the
cube
up
script
and
other
than
that
we're
good
you're.
Okay,.
A
One
okay:
do
we
have
an
update
targeting?
Is
this
just.
G
I
think
they're
yeah,
I
think
they
don't
have
any
work
to
target
for
121,
but
they
will
be.
They
are
working
on
being
on
by
default
by
122..
G
A
Strike
through
all
right,
thank
you.
Next
one
is
openstack,
so
this
is,
I
think
young
is
working
already.
H
Yeah,
well,
I'm
mostly
tracking
what
they
are
doing.
They
are
fixing
their
seat,
ci
jobs
and
actually
I
am
testing
how
the
migration
works,
but
I'm
still
not
sure
we
can
catch
the
deadline
in
this
release
for
being
enabled
by
default,
because
there
are
still
still
some
ci
missing
some
testing
missing
and
I
don't
have
the
openstack
that,
for
example,
with
multiple
zone
topology
zones.
A
All
right
thanks
all
right.
The
next
one
is
the
staff,
see
the
migration.
So
this
is
the
humble
sad.
Do
you
get
any
update
on
this
from
humboldt.
E
No
he's
having
some
personal
family
stuff
that
he's
going
through
right
now,
so
I
don't
think
we'll
get
an
update
on
this
for
a
bit.
Okay,.
G
He
seems
to
be
oh,
that's
great.
I
A
Okay,
so
I
put
the
question
up
I
see.
Sometimes
he
is
writing
messages
on
slack
too,
so
I
thought
he
might
be
back.
I
You
let
me
let
me
check
real
quick.
I
A
Okay,
so
next
one
is
angry,
graceful,
no
shout
out
yeah,
so
I'm
so
updated
the
cup
and
then
I'm
actually
doing
some
experiments.
I
just
based
on
you
know
what
the
cap
has
proposed,
made
some
changes
to
the
code
and
see
how
how
that
behave
so
yeah
right
now,
it's
still
in
progress,
so
I
think
you
need
to.
A
I
need
to
pin
jin
on
something
because
there's
a
things
I
think
she's
very
familiar
with.
A
So
when
the
note
comes
back,
we
also
need
to
look
at
how
to
clean
up.
So
I
think,
there's
a
certain
thing
that
looks
like
it's
not
cleaned
up
at
that
time.
So
I
just
want
to
check
with
her
and
and
see
what
she
thinks
yeah.
So
this
is
working
progress
and
next
one
enable
username
space
in
kubelet.
So
your
ids
get
shifted
all
right.
Okay,
do
we
have
any
update?
Okay,
so
herman
is
reviewing
this
come
on.
Do
we
know
anything
about
this?
One.
D
I
haven't
had
chance
recently
to
review
it
actually
from
after
this.
Anybody
else
has
cycles
triggered
they'll
be
much
appreciated.
Actually,
okay,
it's
mostly
sig
node
is
doing
the
work,
but
we
actually
don't
get.
A
All
right
yeah,
so
if
anyone
else
wants
to
help
his
review
yeah,
please
take
a
look
kind
of
the
link
to
the
cap
here.
Okay,
the
next
one
is
immutable
secrets
and
config
maps.
This
one
is
done.
Okay
and
oh
next,
one
is
pvc
created
by
staple
set,
will
not
be
auto
removed,
isn't
that
here.
G
Matt
is
actively
working
on
this.
Okay.
A
Thank
you,
the
next
one,
one
extension
for
safer
said.
I
guess
this
one
is
probably
still
an
update,
like
I
said,
yeah.
Okay,
thank
you,
and
next
one
is
content,
notifier,
okay,
so
yeah.
We
we
had
a
meeting
with
team
and
jordan
to
discuss
their
consensus
on
the
apis
and
sean
chan
is
working
on
updating
the
cup.
A
Next
next
one
is
uto's
mom
split
into
a
new
repo.
I
think
srini
is
working
on
this.
Do
we
have
any
update.
A
A
Okay,
thank
you.
Next
one
is
the
prioritization
on
voting
capacity
yeah.
So
coffee,
probably
not
here,
michelle
yeah,
there's.
G
A
there's
a
pr
out
and
it's
being
reviewed
by
me
and
also
some
scheduling,
scheduling
folks.
So
it's
in
progress.
A
Okay,
you,
I
think,
that's
all
we
have
on
the
spreadsheet
so
going
back
here.
We
have
this
one
item
from
christian.
Okay.
So
christian,
do
you
want
to
talk
about
this?
One
briefly.
F
Yeah,
so
there
has
been
a
kept
submitted
with
the
design.
Ahmad
has
reviewed
it
and
raised
a
couple
of
concerns.
The
main
thing
I
want
to
discuss
is
this:
design
is
based
on
the
design.
First,
transferring
persistent
volume
claims
across
namespaces,
however,
that
design
has
yet
to
be
merged.
There
have
been
four
caps
proposed
over
the
past
two
to
three
years.
F
I
don't
believe
any
of
those
have
been
merged
and
the
author
of
the
most
recent
cat
has
submitted
has
submitted
this
feature
in
cubert,
so
there
is
a
controller
inside
q
vert,
which
uses
object,
transfers
to
request
data
cluster
scopes.
I
believe
data
volumes
and
pvcs
in
there.
However,
we
don't
have
anything
in
kubernetes
right
now,
so
to
transfer
volume
snapshots.
We
would
need
to
implement
the
controller
and
get
this
api
print.
F
The
discussion
I
wanted
to
have
was:
should
we
attempt
to
mirror
a
design
that
is
currently
pending.
Should
we
attempt
to
make
this
a
one-off,
since
it
is
theoretically
doable
inside
of
the
external
snapshotter,
without
worrying
about
pvcs,
or
should
we
attempt
to
implement
the
controller
to
transfer
the
storage
objects
in
a
more
general
capacity
and
then
extend
it
to
volume
snapshots
eventually,
which
was
the
original
goal,
which
would
involve
reviving
those
previously.
A
A
F
A
F
E
B
Yeah,
so
so
this
was
another
thing
that
came
out
of
the
feedback
on
the
the
volume
populators
cap.
When
we
were
looking
at
what
kind
of
metrics
make
sense
for
volume,
populators
tim
said,
you
know
it
would
be
nice
to
know
like
how
long
it's
taking
for
volume
populators
to
do
their
thing,
and
I
thought
about
it,
and
I
I
thought
that
well,
the
same
is
true
for
like
cloning
volumes
for
cloning
snapshots
for
creating
empty
volumes
like
it
would
be
nice
to
have
metrics
on
how
long
volume
creation
takes.
B
You
know
from
an
end,
user's
perspective
and
the
the
thing
that
knows
that
is
going
to
be
the
the
entry
pvc
bind
controller
right.
So
the
the
number
that
you're
really
interested
in
is
how
much
time
elapsed
from
the
time
the
user
created
the
pvc
to
the
time
the
pvc
became
bound
to
apv
and
then
you'd
like
to
be
able
to
break
down
those
elapsed
times
by
data
source
kind.
B
So
you
can
say
like
okay,
empty
volumes
are
taking.
You
know
five
seconds
on
average
to
create
clone
volumes
or
taking
10
seconds
on
average
to
create
or
whatever
and
then
and
if
things
suddenly
start
to
get
slower
like
an
operator
could
say,
hey
that
you
know
all
of
a
sudden.
My
my
you
know,
volume
clones
are
taking
three
times
as
long
or
something
that
might
be
an
interesting
piece
of
metrics
to
to
observe
that
it's
sort
of
a
it
feels
like
it
should
be
a
core
feature
to
to
get.
G
G
We
do
have
that
today
we
do
have
these
metrics,
but
the
problem
is
they're,
currently
not
broken
down
by
data
source.
So
I
think
what
we
need
to
do
is
look
at
the
current
metrics
and
and
see
how
we
can.
You
know,
divide
that
and
add
that
information
about
data
source
into
it,
okay,.
A
B
A
So
that
including
the
basic
there's
a
whole
operation
right,
so
it's
okay,
so
it's
not
okay
to
start
and
until
it
is
created
and
banged
or
I
don't
know
what
is
the
end
end
of
that,
it's
just
checking
up.
G
D
G
G
I
also
the
way
that
we
implemented
the
total
time
metrics
in
for
pv
bind,
isn't
I
think,
the
most
scalable
way
like
it.
It
basically
has
to
you.
You
have
to
like
inject
all
sorts
of
exit
points
all
over
the
code
for
it.
I
recently
learned
of
this
other
project
called
cube
state
metrics.
I
think
I
don't
know
if
that
might
be
worth
investigating
and
see
if
that's
a
better
place,
to
sort
of
put
these
sort
of,
like
tracking,
when
states
of
kubernetes
objects,
change.
A
G
Have
a
link
for
yeah,
let
me
look
it
up,
but
it
might
be,
I
think,
as
part
of
this
investigation,
I
think
we
should
kind
of
consider
all
of
these
different
options.
A
G
We
already
have
it,
we
would.
I
would
go
and
consider
all
of
the
options.
I
think
currently
the
the
total
metrics
that
we
have
right
now.
There's
a
couple
cons
to
it.
So
that's
why
I
potentially
suggest.
Maybe
we
should
look
at
cube
state
metrics
too,
but
but
I
think
we
should
definitely
look
at
all
of
it.
D
A
D
B
Okay,
well
yeah.
I
just
I
wanted
to
sort
of
suggest
this
as
a
possible
core
enhancement,
because
I
I
don't.
I
didn't
plan
on
doing
anything
in
my
work
on
the
metrics
for
why.
But
there
will
be
metrics
in
the
library
that
handles
data
populators
or
volume
populators.
The
populators
library
will
have
its
metrics,
but
those
aren't
the
metrics
people
really
want
based
on
the
feedback
to
the
cab.
I
Okay,
michelle
the
the
total
one
matrix
are
you
are
talking
about
is
that
the
volume
operation
total
seconds.
A
Yeah,
so
this
looks
like
a
it's
actually
under
the
kubernetes
arc,
so
so
I
think
that
consists
of
injuries
in
it.
I
think
it's
kind
of
considered
injury
if
it's
under
kubernetes
og-
okay-
yes,
I
don't
know
if
this
one
actually
can
actually
watch
like
wanting
snapshots.
I
think
that
was
the
question
from
somebody
the
other
day.
B
A
B
No,
I
was
going
to
say
that,
like
I'm
going
to
focus
on
this,
the
the
data
popular
specific
stuff.
B
Didn't
feel
data
popular
specific
to
me,
which
is
why
I
just
wanted
to
suggest
that
someone
looks
at
it
because
I
I
do
not
plan
on
looking
at
it.
Okay,.
A
B
G
You
wanna,
do
you
wanna
open
an
issue,
a
feature
request
issue?
Then
we
can
kind
of
ask
ask
for
help
on
it.
B
Yeah
yeah,
that's
not
a
bad
idea
yeah.
I
mainly
wanted
to
bring
it
up
here
just
to
see
if
how
much
was
already
there,
because
I
didn't
know
that
we
already
had
like
a
metric
that
was
close
to
what
we
wanted.
That
just
doesn't
have
the
extra
dimension
to
it
and
I
don't
know
if
you
can
just
change
an
existing
metric
to
add
a
dimension
or
if
you
have
to
add
a
new
metric
for
backwards
compatibility
reasons.
I
don't
have
a
lot
of
metrics
experience
in
general,
but
I
know
that
people
are
taking
much
harder.
B
Look
at
it
now
with
the
product
or
production
readiness
reviews
metrics,
are
become
a
much
bigger
deal
for
every
feature
so
yeah
I
can.
I
can
I
just
file
a
feature,
request
issue
that
would
go
on
the
kk
repo
yeah.
Okay
sounds.