►
From YouTube: Kubernetes SIG Storage Meeting 2021-08-12
Description
Kubernetes Storage Special-Interest-Group (SIG) Meeting - 12 August 2021
Meeting Notes/Agenda: https://docs.google.com/document/d/1-8KEG8AjAgKznS9NFm3qWqkGyCHmvU6HVl0sk5hwoAE/edit#heading=h.osrvllkiel1e
Find out more about the Storage SIG here: https://github.com/kubernetes/community/tree/master/sig-storage
Moderator: Saad Ali (Google)
A
Okay,
today
is
august
12
2021.
This
is
the
meeting
of
the
kubernetes
storage
special
interest
group.
As
a
reminder,
this
meeting
is
public
recorded
and
posted
on
youtube.
The
agenda
is
in
front
of
you
here.
The
link
you
can
find
in
the
calendar
invite
feel
free
to
add
to
it
as
we
go
today.
A
We're
gonna
go
over
the
sorry,
we're
gonna
start
the
123
planning
the
122
release
happened
last
week
on
august
4
and
there
were
features,
or
there
were
blogs-
that
sig
storage
helped
write
that
were
published,
including
on
the
windows,
csi
support
and
those
are
starting
to
be
released
as
of
this
week.
A
A
Okay,
so
I'm
just
going
to
go
over
these
and
we
can
drop
the
items
that
don't
have
any
follow-up.
So
first
up
is
delegate
fs
group
to
csi
driver
instead
of
cubelet,
so
this
was
completed
for
the
previous
cycle
as
alpha
hamad.
Do
we
want
to
move
this
to
beta
or
what's
the
next
steps
here.
B
Yeah,
I
think
we
want
to
move
it
to
beta.
There
are
some
some
things
that
we
want
to
discuss,
or
at
least
addresses
the
the
what
happens
if
people
are
using
conflicting
fs
group
like
we
need
to
sort
it
out
somehow,
but
I
think
that's
a
different
design
item
we'll
have
a.
A
All
right
next
item
is
csi
online
offline,
resizing
volume,
expansion,
update
cap
fix
issues
last
cycle.
This
was
marked
for
beta
and
based
on
the
last
status.
This
was
blocked
and
did
not
make
it
do
we
just
move
this
over
to
the
cycle
a
month.
B
Yeah
and
this
and
the
next
so
we
have,
we
had
some
identified
some
risk
condition
with
recovering
I'm
recovering
from
recess.
Failure
is
the
only
thing
that
is
kind
of
blocking,
so
we
have
to
resize
and
go
in
ga
so
well,
I
have
updated
the
cap,
I
open
the
enhancement
update.
I
have
to
update
it
some
more
actually.
A
A
A
I
don't
think
so.
Yeah,
that's
a
good
question
now
so
for
folks
that
are
listening
in
and
are
wondering
how
you
can
participate
or
how
you
can
join
in.
This
is
a
good
meeting
to
attend,
and
you
know
if
there's
a
feature
that
you've
been
hoping
to
be
able
to
help
out
on.
This
is
an
opportunity
folks,
like
yan,
can
help
guide
you
if
you're
not
sure
exactly
what
you
need
to
do.
This
is
a
good
opportunity
to
jump
in
and
work
on
core
kubernetes
code.
A
So
if
you're
interested
reach
out
to
yon
or
myself,
michelle
shane.
C
B
E
A
Okay,
so
if
anyone
is
interested
feel
free
to
reach
out
offline-
and
we
can
definitely
get
this
back
on
the
123
schedule-
next
item
is
csi
entry,
read-only
handling,
so
last
status
here
was
pr,
was
out
for
review.
Anyone
have
the
latest
on
on
this
one.
G
I
A
A
All
right
next
up
is
storage
capacity
tracking
for
pod
scheduling.
This
was
supposed
to
go
to
beta
and
last
status.
Update
was
bug,
fixing
any
updates
on
on
this
one.
G
A
Got
it
so
it
sounds
like
this
cycle
is
going
to
remain
in
beta,
probably
yeah,
so
that
is
fine.
Next
one
is
pvc
inline
ephemeral
volume,
so
this
is
the
older
way
of
doing
ephemeral
volumes
that
works
with
csi
drivers.
E
A
So
if
anyone
else
on
the
call
is
looking
for
context
on
this,
there
are
basically
two
ways
that
we
do
ephemeral
volumes
kind
of
the
the
old
way,
which
was
we
kind
of
created
our
own
version
of
the
csi,
like
interfacing
kind
of
violated
the
spec
and
said
you
know
you
can
instead
of
doing
a
provision
first,
you
can
just
go
ahead
and
do
a
node
publish
volume
and
if
you
do
it
in
a
specific
way
and
the
csi
driver
understands
you
can
create
these
special
csi
ephemeral
volumes
and
you
can
imitate
effectively
kind
of
the
life
cycle
of
the
ephemeral
volume,
just
like
a
secret
volume
or
other
projected
volumes,
and
this
has
worked
fairly
well.
A
It
is
currently
in
alpha
and
we
need
to
move
it
to
beta.
In
addition
to
this
patrick
started,
another
effort
where
he's
trying
to
make
it
so
that
any
pvc
can
be
used
as
an
inline,
ephemeral
volume.
This,
of
course
requires
the
creation
of
the
pvc,
but
the
code
will
automatically
create
the
pvc
on
the
user's
behalf
and
then
automatically
delete
it
when
the
pod
is
deleted.
A
So
we're
trying
to
support
both
modes
number,
seven
actually
introduced
a
new
kind
of
api
field
for
ephemeral
volumes,
and
so
the
required
work
for
number
eight
would
be
to
align
with
that
and
move
it
to
beta
as
well.
So.
A
Oh
sorry,
so
it's
the!
I
think
the
work
is
just
to
align
with
the
existing
api
then
or
the
the
new
api
right.
So,
if
anyone's
interested
in
helping
with
that
effort,
if
you
use
csi,
ephemeral
volumes
and
and
want
them
to
kind
of
continue
to
move
along,
this
would
be
a
good
place
to
jump
in.
A
And
then
for
pvc
inline
alpha
the
beta
storage
capacity
tracking
alpha
two
beta,
okay,
all
caught
up
on
that,
so
I'm
gonna
go
ahead
and
leave
this
item
here.
Csi
ephemeral
volumes-
if
anybody
wants
to
pick
it
up,
they're
welcome
to
next
item
is
spreading
over
failure
domains.
This
was
a
design
for
the
last
quarter.
G
A
A
A
Next
item
csi
out
of
tree
moving
the
iscsi
driver,
fit
and
finish
image,
building
testing,
ci
cd
documentation.
Humble
was
helping
with
this.
I
don't
think
we've
completed
this
work,
so
I'm
gonna
go
ahead
and
move
it
to
123.
A
A
A
Next
up
is
sending
out
deprecation
notices
for
flex
volumes.
Michelle
was
helping
review
the
message
here.
Any
updates.
G
I
think
yeah
michelle
reviewed
it.
She
has
some
comments.
I
think
yeah
also
have
some
comments.
I
think
we
still
have
not
decided
the
issues,
but
we
want
to
say
we
want
to
stop
supporting
some
features.
A
Out
exactly
what
recording
should
be
moving
on
to
the
next
item:
pvpc
volume
snapshot,
namespace
transfer,
so
mustafa
is
helping
drive
this
one.
They
had
a
design
meeting,
trying
to
sort
out
kind
of
the
next
steps
here.
Mustafa
sounds
like
we're.
Just
gonna
move
this
over
to
123.
N
Yeah
I
already
actually
created
a
cap.
In
the
last
design
session,
we
decided
to
focus
on
bbc,
basically
in
volume
snapshot
without
secret.
Yesterday,
in
the
standup
we
discussed
that
we
need
even
from
the
for
the
volume
snapshot,
we
need
the
secret
and
then
we
ended
up
having
saying
the
default
will
be
without
secret,
but
we
can
offer
options
for
beta
yeah
and
I
created
a
cap
already.
So
anyone
who
has
can
review
and
give
me
comments
will
be
appreciated.
N
A
Okay,
cool
next
item
is
csi
volume:
health
for
csi
volume,
health,
the
last
status
was
we
had
some
design
meetings
here,
kind
of
kick
starting
what
we
should
do
next
shang.
Where
do
we
want
to
go
with
this
one
for
123.,
yeah.
G
So
I
was
wondering
if
we
can
split
this
into
two
parts.
I
wonder
if
that
makes
sense
so
because
we
talked
about
we
want
to
have
matrix,
we
want
to
make
the
move
the
volume
house
into
matrix.
So
it
looks
like
that.
Is
I
don't
see
any
objections
to
that.
So
should
we
move
forward
with
bursts
and
make
this
better
and
then
for
the
for
the
reaction?
I
think
that
needs
more
work.
G
I'm
still
waiting
for
nick
to
add
some
use
cases
there,
so
maybe
that
one
can
be
an
you
know
can
be
a
new
feature
gate
once
we
figure
out
what
to
do
like
it's
a
maybe
one-half
reaction
or
something.
A
A
G
That's
yeah,
okay,
so
so
that
well
it's
yeah.
It
is
basically
it's
just
turning
the
current.
Currently
we
have.
We
have
events,
I
don't
know
if
it
makes
sense
to
still
keep
the
events
and
then,
in
addition,
add
the
metrics
or
do
we
just
want
to
copy
an
event
and
not
have
matches
or
not
events,
but
that
maybe
we
can
discuss
later.
A
And
we'll
delete
the
history
on
that
and
say:
split
from
above
target
off
of
4
123
split
into
two
items:
target
beta
for
core
functionality.
A
D
Yeah,
I'm
here
so
the
as
you
know,
the
the
features
in
the
docs
are
in
the
blog
is
scheduled
to
be
posted
in
a
couple
of
weeks,
but
it's
mostly
reviewed
right
now.
What
we're
waiting
on
is
the
out
of
tree
stuff
to
tag
and
release,
and
that's
that's
what
I'm
working
on
so
yeah.
The
other
tree
needs
to
have
an
alpha
release.
A
So
question
is:
where
do
we
want
to
take
it
for
123?
Do
you
want
to
keep
it
in
alpha
or
try
and
target
a
beta.
D
A
Are
we
waiting
on
any
implementations
to
kind
of
validate
this
before
we
move
to
beta,
or
is
that
more
of
a
ga
blocker.
D
That's
a
good
point.
We
would
want
to
see
actual
implement
actual
alpha
implementations
that
do
useful
things
that
are
released
and
supported
by
somebody.
D
D
A
Let's
maybe
keep
it
as
beta
here
and
then,
as
we
get
closer
to
the
kept
deadline,
we
can
make
a
call
of
whether
we
have
enough
data
or
not
to
proceed.
How
does
that
say
absolutely.
A
All
right,
thank
you.
Ben
next
up
is
the
cozy
effort,
and
so
cozy
effort,
as
far
as
I
understand,
is
targeting
alpha
this
cycle
last
status.
Update
was
cap
was
ready
to
review
any
anything
new
from
the
cozy
team.
D
I
I
saw
that
that
the
tim
was
back
in
the
thread.
Talking
about
the
review,
I
don't,
I
think.
G
A
G
Yeah
so
so
phong
and
now
noah
are
going
to
start
a
cap
in
1.23.
A
G
A
Next
up
is
new,
read,
write
access
mode,
chris
status
of
this
and
next
steps.
I
Yeah,
so
it's
it's
in
alpha
now
we
don't
have
plans
to
target
beta
currently
for
123..
Okay,
just
to
have
a
bandwidth
right
now
make.
A
A
Okay,
and
is
there
any
bug,
fixes
or
anything
that
needs
to
be
done
or
any
other
additional
work
that
we
need
to
track
here,
or
can
we
just
remove
it
from
123
and
reintroduce
it
for
124.
A
A
O
Yeah,
so
I
think
the
main
set
of
apis
and
the
proxy
that's
all
done,
as
we
just
discussed
the
the
main
remaining
bit
that
we
would
want
to
look
at
is
like
if,
if,
if
the
csr
plugin
is
using
the
proxy
today
and
later
down,
the
line
privileged
containers
and
windows
do
get
supported.
Can
the
api
provide
a
nice
kind
of
like
a
transition
back
into
that,
so
that
transition
path
is
going
to
be
the
main
thing
we
will
look
at
in
in
the
next
range.
A
Is
that
something
we
want
to
start
this
cycle
or
something
we
want
to
do
in
the
next
one.
D
O
D
C
D
G
O
Right
and
that
that's
going
to
be
the
goal
of
this
like
how
do
we
minimize
the
set
of
changes?
The
the
goal
is
like
you
know,
let's
make
sure
that
if
they
wrote
against
the
proxy
api
that
api
can
also
work
in
the
privileged
model,
but
at
a
minimum
I
think
a
rebuild
would
be
necessary.
A
Awesome
so
mauricio,
are
you
still
working
on
this
or
someone
else.
L
A
On
this
okay,
yep
yeah,
and
then
I'm
gonna
mark
it
as
design
for
this
quarter.
Not
alpha
is
that
reasonable.
A
A
G
Sure
I
do
have
a
question,
though,
so
I
think
the
css
windows
support
is
the
blocker
for
csm
migration
to
be
turned
by
default.
So
now,
if
we
have
this
one
ongoing,
should
we
wait
for
this,
or
should
we
still
go
ahead
and
try
to
add
that
support
for
that's.
G
I'll
go
ahead.
Go
ahead!
Sorry,
obviously,
because
we're
trying
to
add
that
support
at
the
end
of
the
year,
but
I'm
not
sure
what's
the
timeline
for
this
moving
to
privileged
container
support.
So
so
I
don't
know
if
we
should
go
ahead
and
continue
with
the
cs
epoxy,
or
should
we
wait
for
this
printed
container?
I
guess
that's
the
question.
O
So
I
think
in
ga
now
is
a
good
time
like
since
that
that
is
a
stable
release.
You
can
base
your
changes
on
that
and
then
that
and
then
down
the
line
as
and
when
privileged
containers
or
windows
does
get
too
stable.
O
At
that
point,
you
will
get
that
transition
on
ramp
to
privileged
container.
So
so,
I
think,
caesar
proxy
being
stable
is
a
good
point
to
kind
of
start
designing
and
then
making
the
changes
in
the
in
the
plug-in
side,
as
gcpd
has
done,
and
and
then
consider
the
transition
when
that
becomes
ready
through
privilege.
J
And
I
think
deep,
one
of
the
goals
I
think
of
this
transition
to
privileged
containers
is
to
try
to
retain
api
compatibility
with
cst
proxy
as
much
as
possible
to
minimize
the
like
amount
of
work
needed
to
transition
between
the
models.
O
A
All
right,
speaking
of
csi
migration,
let's
start
talking
about
that
draw
dryway!
I'm
glad
to
see
that
you're
here
so
csi
core
bugs
and
issues
is
this
gonna
be
carrying
over
or.
A
So
go
ahead
and
get
that
moved
over
next
item
is
deprecation
of
entry
drivers
that
will
not
migrate,
remove
plugins
that
were
previously
deprecated.
A
P
Yeah,
I
guess
another
thing
is
in
the
next
coming
1.23
release:
do
we
want
to
like
officially
deprecate
all
the
cloud
provider
entry
plugins
or
like,
like
I
I
I
I
know
in
1.27.
P
P
Meaning
here
like
we
send
out
emails.
E
A
Okay
seems
reasonable,
so
officially
deprecate
cloud
provider,
plug-ins,
send
out
emails,
add
release,
notes,
etc.
So
that
seems.
P
P
Change
so
like
when
the
plugin
is
used
there,
there
will
be
a
warrant,
a
warning
lock.
P
A
Okay,
send
out
emails,
add
release
note
code
change
for
warning.
This
would
be
effectively
bug
fixed
level.
P
Yeah,
but
not
all,
not
all
the
like
the
cloud
provider
plugins
like
gc
and
aws
and
azure
right
now,
there's
only
a
few.
I
think
it's
non
club
provider
plugins
that
got
deprecated
okay,.
A
Okay,
I'm
gonna
put
your
name
on
here:
jaway,
yeah,
sure
and
matt.
Are
you
interested
in
helping
review
this.
A
P
Yeah,
the
biggest
is
risk
right
now
is
windows
support
right
as
long
as
we
have
that
it'll
be
all
good
and
there's
another
thing
about
nfs
waveform.
I
think
I
sent
out
email
already
asking
for
help.
A
Okay:
next
item
is
azure
disk,
an
azure
file.
Where
are
we
with
those.
P
P
Think
I
think
either
file
is
already
beta.
A
Azure
file
is
already
beta.
Okay,.
E
P
I
matthew
want
to
give
an
update.
E
Yup
yeah
sorry
just
trying
to
find
the
unused
button.
Yes,
so
where
this
is
target
for
123,
the
complication
we
ran
into
is
that
there
are
we.
There
are
currently
sort
of
load-bearing
pd
tests
in
the
gce
or
independent
suite.
That's
run
on
gce,
for
example.
Stateful
sets
assume
anything
that
assumes
a
storage
class
existed
because
of
the
whole
cloud
provider
extraction
thing
though,
people
didn't
want
to
turn
on
the
cpsi
driver.
E
On
by
amma
default
in
the
generic
end-to-end
test,
so
instead,
after
talking
to
folks,
we
have
a
strategy
that
we're
going
to
basically
remove
all
of
these
load-bearing
tests
from
the
generic
upstream
test
and
instead
the
cloud
provider
gcp,
proud
tests
are
going
to
get
upgraded
to
like
as
release
blocking
as
cloud
provider.
E
Gcp
could
do
so
with
this
plan
that
on
unblocks
turning
it
on
by
default
in
123
and
if
anyone's
interested
in
the
details,
there
is
a
cloud
provider
gcp
meeting
this
afternoon,
pacific
time,
which
you
are
welcome
to
join.
You
can
ping
me
for
the
details
on
that
and
we'll
get
into
a
discussion
of
this
in
detail.
There.
A
What
are
they
going
to
do
for
the
stateful
set
tests
or
the
other
tests
like
that?
Just
use
like
a
generic
csi
driver
or
something.
E
No,
I
mean
the
the
plan
of
record
is
to
just
remove
them
from
the
upstream
release.
Blocking
things
like
we'll,
probably
have
to
add
a
new
ginkgo
feature
tag,
or
you
know,
however,
and
instead
they'll
be
run
from
the
cloud
provider.
Gcp.
E
Yeah-
and
it
actually
may
be
possible
that
a
few
of
these
tests
don't
actually
need
a
full
thought.
What
mike
could
use
a
csi
host
flight
or
could
could
use
yeah
a
s
host
path
thing
like
because
oftentimes
the
tests
don't
actually
require,
for
example,
just
to
be
moved
across
nodes.
However,
that
is
probably
like
a
thornier
thing,
so,
like
we've
explicitly
kind
of
deferred
that
as
something
that
isn't
going
to
block
turning
on
migration,
yeah.
A
Okay,
thank
you
for
the
update,
matt
and
just
to
confirm
the
alpha
beta
designation
is
gonna
remain
in
beta
this
cycle
or.
E
Yes,
it's
going
to
stay
in
beta,
but
be
on
by
yeah,
on
by
default
and
and
just
just
confirm.
The
reasoning
here
is
that,
if
you
ga,
you
can't
switch
it
off
if
you
have
it
on
by
default,
it
still
gives
people
the
option
to
turn
it
off,
and
it
seems
wise
for
this
feature
to
have
that
be
the
case
for
at
least
one
release.
You
know
just
in
case
there
are
unforeseen
problems.
A
Got
it
all
right,
cool
sounds
like
a
plan.
Thank
you.
Matt
next
item
is
aws
windows
support
on
by
default.
If
ready
was
the
last
cycle
and
still
working
on
windows
support,
so
it
seems
like
the
same
story
here.
P
Yeah,
I
think
last
time,
matt
mentioned
windows.
Work
is
all
done
right
now.
It's
all
testing
I'll
confirm
with
him.
Our
next
meeting.
Okay.
L
A
Okay,
moving
on
from
migration
next
item
is
update
topology
labels
from
beta
to
ga
for
vsphere.
This,
I
believe,
is
no
longer
relevant.
It's
already
done,
yeah,
okay,
I'm
going
to
go
ahead
and
remove
that
then
always
honor,
reclaim
policy.
G
So
we'll
try
to
do
ava
in
window
23.
F
A
And
then
secret
protection
finalizers
to
prevent
deletion
of
secrets.
Asaki
was
working
on
this.
Q
Yes,
I
opened
another
cape
for
generic
in-use
protection
in
april
machinery
and
as
a
result
of
discussion
there,
the
conclusion
is
using
validation
hook
to
block
the
rate
request
itself.
A
A
Okay
sounds
good,
let's
target
alpha
and
then
we
can
reassess
as
we
get
closer
to
kept
free
state
sounds
good.
Thank
you.
Masaki.
The
next
set
of
items
are
co-owned
between
sig
storage
and
other
sigs.
First
up
is
sig
auth
user
id
ownership
and
config
maps
and
secrets
preserve
default
file
mode
bit
in
atomic
writer
and.
A
P
Yes,
we're
we're
actually
gonna
have
a
meeting
today
to
discuss
it.
Okay,
yeah
so
we'll
have
a
design
for
one
point
on
history.
A
Sounds
good
next
item
is
enable
username
space
and
cubelet
so
that
uids
get
shifted,
rootless
mode,
monty
we're
helping
review.
This
looks
like
you
got
merged,
won't
work
with
csi
drivers
by
design
when
it
goes
beta.
We
need
to
figure
out
how
to
how
it
works
and
to
end
should
keep
an
eye
on
it.
P
A
O
A
E
Yes,
we
should
be
well
yeah.
I
haven't
had
a
chance
to
get
back
to
pushing
through
the
implementation.
The
api
changes
are
all
approved,
and
so
that
should
be
easy.
A
Awesome
and
we're
gonna
call
it
alpha
this
quarter,
yep
cool,
all
right!
Thank
you!
Matt!
Next
up
we
have
volume.
Expansion
for
stateful
sets.
A
Okay,
fingers
crossed
on
that
one
next
up
is
the
kubernetes
utils
mount
that
needed
to
be
moved.
We
got
no
updates
on
this
for
a
while.
Let's
go
ahead
and
move
it
to
123.
A
So
it
looks
like
a
scheduler.
Configurable
change
is
merged,
it's
part
of
122..
Do
you
know
what
the
status
is
for
123
or
the
plan.
J
I
don't
think,
there's
a
plan
to
move
at
the
beta.
I
can
double
check.
A
Okay,
I'm
gonna
go
ahead
and
mark
it
as
beta
and
put
a
comment
here
and
double
check
if
this
is
moving
to
data
or
not,
if
not
drop
it.
Okay
and
finally,
we
have
csi
service
account
token.
This
was
g8
last
quarter,
so
I
think
we
can
go
ahead
and
remove
it
from
here.
O
Yeah
I
have
one
this
is
I
I
just
opened
this
issue
2857.
O
A
And
deep
I'll
put
you
down,
as
the
owner
here
sure
sounds
good.
M
A
Yeah
no
worries.
Let
me
finish
this
one
real,
quick
and
then
we
can
go
back
to
that.
M
I'm
also
interested
so
I'm
not
sure
what
but
I'll
certainly
keep
an
eye
on.
It
sounds
good.
A
Patrick
we'll
put
you
down
as
api
reviewer.
A
All
right,
thank
you
all
now,
let's
jump
back
to
doo
doo
doo,
ephemeral,
aph,
okay,
so
we
got
number
seven
and
eight.
M
So
for
ephemeral,
inline
volumes
there's
not
much
that
actually
needs
to
be
done
anymore.
There
was
one
thing
brought
up
by
someone
else
using
this
feature
about
the
controller
that
didn't
have
a
configurable
number
of
threads,
but
that
has
been
addressed
or
is
going
to
be
in
122..
M
So
I'm
wondering
actually
whether
this
should
be
whether
that
is
ready
to
move
to
ga.
So
we're
talking
about
line
number
seven.
A
M
Yes,
I
see,
and
I
think
the
only
thing
missing
is
collecting
feedback,
I'm
not,
and
I'm
not
sure
what
we
have
an
official
process
for
that,
whether
I
need
to
put
up
some
questionnaire
or
what
what
we
do,
whether
it's
just
based
on
our
experience,
what
we've
heard
from
users
it
it
certainly
seems
to
be
used.
Otherwise,
people
wouldn't
wouldn't
ask
for
this
configurable
configurability,
for
example,.
H
J
I
would
go
back
to
the
cap
and
the
graduation
criteria
and
see
if
we
meet
all
the
check
marks.
Okay,
I
think
you
know
there's
the
usual
stuff
of
metrics
scalability
and
all
of
that
yeah.
M
So
yeah,
let's
try
make
sure
that
we
have
a
look
at
it
and
see
whether
there
is
actually
anything
missing
for
storage
capacity
tracking.
I
think
we
do
still
have
outstanding
questions
around
integration
with
cluster
autoscaler.
I'm
going
to
do
some
work
with
that.
A
Got
it
so
this
is
mostly
bug
level
work
this
this
quarter.
Yes,
okay
and
then
csi
ephemeral
volumes,
the
existing
api.
We
put
a
call
out
for
help
here
if
anybody's
interested,
they
can
help
jump
in
and
help
update
the
api.
A
I
think
we
need
the
api
to
be
updated
first,
since
it
needs
to
be
aligned
with
the
new,
the
new
way
of
doing
things
and
then
once
we're
there.
I
think
at
least
one
quarter,
maybe
two.
A
Yeah,
it's
interesting
because
there's
a
lot
of
folks,
I
think
that
have
written
these
csi
ephemeral
volumes,
but
I
think
most
of
them
are
outside
of
this
sig.
So.
M
J
J
A
J
I
mean,
I
think
yeah
matt
has
also
volunteered,
but
probably
won't
be
able
to
get
to
it
until
later.
So.
A
Okay
sounds
good
to
me,
so
I
think
with
that.
That
concludes
the
123
planning
session.
Oh
sorry,
go
ahead.
B
A
B
Okay,
one
was
it's,
I
I
I
think
they
got
lost
in
the
non-recursive
volume
ownership
in
120.
We
moved
it
to
beta
item
number.
Five
in
120
spreadsheet
item
number
five.
B
B
The
other
one
was
christian,
was
owning
it.
I
don't
know
if
christian
is
still
available.
I
I
don't
think
so,
though,.
D
B
Matt
do
will
you
have
cycle
to
do
the
the
flipping
the
feature
kit
you're,
the
owner?
It
shouldn't
be
beyond
the
flipping
the
future
gate.
I
think.
E
Yeah
I'll
give
you
a
ping
on
slack
and
figure
it
out.
Thank
you.
A
All
right
and
let's
go
ahead
and
switch
over
to
our
agenda
doc,
design,
reviews,
vermont
added
csi
drivers
that
directly
want
to
talk
to
kubernetes
api
and
has
dependency
beyond
csi
driver
haman.
Do
you
want
to
save
this
for
the
next
cycle
since
we're
over
time?
A
A
A
But
the
current
situation
is,
we
don't
have
anything
planned
right,
not
as
far
as
I
know
shin.
I
don't
think
we
have
anything
planned
right.
D
G
So
there
is
this,
so
there
is
a
like
the
contributor
summit.
They
do
have
like
this
meet
and
greet
even
at
least
we
could
right
there.
But
if
we
want
to
do
this
face
to
face,
then
we
need
to
actually
we
need
to
plan
early
because
I
think
they
have
some.
They
say
they
have
like
two
rooms.
But
if
you
want
the
room,
then
we
can
talk
to
them
early
problem.
Is
we
don't
really
know
who
are
going
yet
right?
So
I
think
we
will
be
going
this
cycle.
G
Oh
you're,
not
going
okay,
so
I'm
planning
to
go,
but
I
also
still
need
to
get
approval
of
my
company.
I
got
it
since
you're
waiting
for
that
yeah.
So
maybe
we
can
just
start
this
agenda
sheet.
If
anyone
plan
to
go
put
your
name
there,
then
we
can
see
how
many
people
are
attending.
A
Yeah,
I
think
it
will
be
difficult
for
the
europeans,
since
I
think
there's
still
restrictions
on
travel
into
this
us
so
yeah.
I
think
this
will
be
pretty
lightly
attended
but
yeah.
Let's
revisit
this
at
the
next
meeting
in
the
meantime,
folks
who
are
planning
to
attend,
please
throw
your
name
in
here
and
if
we
get
enough
of
a
quorum,
sinching
will
be
there.
Maybe
we
can
organize
something.