►
From YouTube: Kubernetes SIG Storage - Bi-Weekly Meeting 20200827
Description
Bi-Weekly Meeting of Kubernetes Storage Special-Interest-Group (SIG) - 27 August 2020
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
A
A
So
this
is
the
spreadsheet
copied
from
the
1.19
spreadsheet,
so
we'll
just
go
over
and
decide
what
we
are
planning
to
do
in
1.1.20,
the
first
one
csi
online
offline
resizing
volume,
expansion
fixed
issues.
So
I
think
the
plan
is
to
keep
the
scene
beta
in
1.20.
So
priority
is
one
come
on.
Does
this
look
alright,.
A
Okay
and
the
next
one
is
recovering
from
resize
failures,
so
we
didn't
make
it
in
1.19.
So
we
tried
to
bring
this
one
to
half
about
1.20,
I'm
wondering
about
priority.
Should
this
one
be
increased
because
we
are
saying
we
want
to
hold
our
phone
expansion
for
one
more
release,
I'm
hoping
to
get
this
one
at
the
alpha
feature
in
1.20
and
then
maybe
in
1.21,
both
one
those
two
can
one.
This
one
can
go
beta
and
then
williamson
pension
can
go
ga.
So
how
does
this
priority?
A
B
A
You
next
one
is
new
design
for
volume
file
permission
handling.
C
Yeah,
I
don't
know
if
ben's
going
to
have
time
on
that.
Is
anyone
interested
in
working
on
this?
I
think
the
high
level
item
here
is
the
volume
permissions,
how
they're
handled
and
kubernetes
has
kind
of
been
ad
hoc
and
a
little
bit
of
a
mess,
and
we
wanted
to
redesign
that
anybody
interested
in
picking
up
that
design
feel
free
to
speak
up.
If
not,
we
can
leave
it.
Unassigned.
C
A
All
right,
the
next
one
is
file
permission
handling
for
windows
looks
okay,
so
it
looks
like
deep
is
working
on
this
end
of
when
downloading
status
has
moved
to
1.20
okay,
so
we
will
just
leave
it
as
a
design
and
then
probably
two
is
okay.
C
Is
jing
or
deep
or
kk,
or
anyone
from
that
team
on
online
want
to
get
confirmation.
E
So
this
is
kk.
I've
not
been
attending.
E
C
A
A
So
we
want
to
move
this
to
ga
1020
priorities
too.
Does
this
still
stay
like
this
or.
A
A
Okay,
next,
one
issues
related
to
assuming
volumes
are
mount
points,
so
this
is
a
black
fix.
We
have
nd
here.
So
there
are
looks
like
there
are
pr's,
I'm
going
to
1.20
any
update
on
this.
One
would
be
just
keep
this
as
the
same
priority.
2
bug
fix.
F
Michelle,
oh,
hey,
sorry,
so,
yeah
there
are
a
couple
of
pr's
out
to
sort
of
address
various
different
aspects
of
this
issue.
So.
A
F
Expect
we
can
make
progress
on
this
in
120.,
I'm
not
sure
if
we'll
be
able
to
fix
every
single
issue,
but
but
I
think
we
can
definitely
make
some
progress
on
it,
but
yeah.
So
andy
has
a
big
pr
out
that
we
had
postponed
to
120
and
then
I
think,
there's
someone
else
that
has
a
second
pr
out.
There's
two
pr's
I
know
of
but
there's
probably
more
problems.
F
A
Okay,
next
one
storage
capacity
tracking
do
we
have
patrick
here,
so
that
is
so
that
yeah.
So
that's
a
alpha
feature
in
1.19.
F
E
F
C
Yeah,
let's
leave
it
as
alpha
that
way,
we
don't
unnecessarily
require,
like
all
the
docs
work
and
all
these
other
things
that
need
to
happen.
A
A
Next
one
pvc
inline
fm
volumes,
so
this
one
is
also
a
new
1019
feature
in
alpha
state.
So
should
we
try
to
bring
this
to
beta
in
1.20?
This
is.
F
A
Patrick's
plate:
do
you
know
anything
about
this
michelle
saad?
Should
we
keep
this
like
this
or
are
there
any.
F
I
think
I
don't
recall
there
being
like
substantial
design
concerns
that
we
need
to
figure
out.
There
were
maybe
like
two
issues
that
we
need
to
resolve,
but
I
think
it
could
be
feasible
target
beta
in
in
the
120.
A
Time
frame,
okay,
oh,
should
we
also
look
at
this
one
together
with
this
other
csi
inline
volume
feature
trying
to
see
if
we
can
combine
them.
Is
there
something
like
that?
Let's
remember
seeing
something:
some
kind
of.
F
That
would
be
a
good
thing
yeah.
That
would
be
a
good
thing
to
to
figure
out
the
the
whole
api
story,
and
if
we
can
consolidate
the
two
features
at
all
into
a
single.
F
A
C
Sorry,
real
quick:
can
we
change
eventually
to
now
for.
C
A
A
Let's
see
if
there's
something
else,
that
we
need
to
change:
okay,
just
this
one,
all
right
cool!
Thank
you,
okay,
spreading
over
failure
domain,
so
I
think
this
would
still
be
staying
designed
because
this
right
now
kind
of
coupled
with
this
next
one
warning
group
api,
so
yeah.
So
this
one
is
still
yeah
still
stays
in
the
designs
today.
A
If
we
can
actually
figure
out
like
if
we
can
actually
couple
the
students
who
think
maybe
one
can
move
faster,
otherwise,
I
think
there's
a
lot
of
issues
we
need
to
discuss
first,
so,
let's
find
out,
we
just
keep
it
design
priority.
Two,
I
think
that's
fine,
and
next
one
is
a
csi,
auto
3,
more
iscsi
driver
finish.
What
is
this
status.
A
Michelle,
do
you
know
this
one,
do
we
do?
We
have
a
anyone
walking
on
this
already.
F
Oh
my
scuzzy
yeah
there
were
some
people
that
reached
out
to
me
that
were
interested
in
in
helping
with
the
iscsi
driver.
I
I
need
to
reach
back
out
to
them
to
see
if
they're
still
interested
but
otherwise,
like.
G
Michelle
this
is
kidding
just
trying
to
understand
the
scope
of
work
here.
What
needs
to
be
done.
F
Yeah
so
basically,
we
have
the
like
sort
of
a
common
iscsi
driver
for
csi
and
we're
just
looking
really
looking
for
an
owner
to
help
drive
and
maintain
the
work
that
needs
to
be
done
for
this
driver.
I
think
right
now
it
like
it
builds,
but
I
don't
know
if
anyone's
actually
tested
it.
F
G
Here
I'm
currently
working
on
the
nfs
provisioners
setting
up
the
ci
and
those
things.
Maybe
I
can
take
a
look
at
that
part
of
the
aspects
for
this
csi
driver
for
his
qc
property
choice.
A
F
Yeah,
I
just
need
to
yeah.
I
just
need
to
hit
the
button
on
it
to
archive
this
repo.
Okay,
I
need
to
get
around
to
doing
that.
A
So
I'm
not
sure
how
much
work
is
remaining
for
this
one
in
progress.
This
is,
I
think,.
C
A
And
okay,
so
karen
you're
working
on
this
next
two.
What
is
this
that
is
for
those
two
is
still
some
remaining.
G
Line
number
19:
I
think
it's
a
ci
related
work,
that's
still
pending,
but
I
think
we
have
gotten
into
a
motion
of
raising
prs
and
the
new
contributors
that
are
helping
they're
able
to
review
when
we
are
getting
for
the
next
line
item.
There
was
a
cla
issue.
I
just
before
this
call.
I
raised
a
new
pr
that
fixes
it.
G
A
Great,
thank
you.
Next
one
is
the
volume
snapshot
name.
Space
transfer
is
mike
still
interested
working
on
this.
H
Yeah
hi
I'm
still
alive
and
interested.
Unfortunately,
I
have
no
update.
A
Okay,
so
do
we
still
so
we
keep
this
as
a
design
in
1.20?
Is
that
it
looks
okay,
okay,
thanks.
A
A
Should
we
try
to
like
add
those
one
condition
to
status
instead
of
keeping
in
as
the
events
in
pvc
and
bring
into
beta.
C
Off
the
top
of
my
head,
I'm
not
sure
let's
have
the
discussion.
Okay,
maybe
a
couple
of
design
meetings
to
talk
about
what
we
want
to
do
sure
we
can
preliminarily
target
a
beta
and
then
readjust,
if
needed.
Okay,
sure.
I
I
I
Yeah
I
mean
I
I
yes,
I
want
to
keep
working
on
this
okay,
but
I
don't
have
visibility
into.
Is
my
free
time
for
the
next?
This
guess
is
four
months.
We
have.
A
A
A
Okay,
is
anyone
on
the
call
interested
in
working
with
the
band
on
this
feature?
Basically,
it's
it's
already
our
feature,
but
we
need
to
implement
a
populator
right
now,
it's
in
the
it's
only
in
the
api,
but
we
need
the
implementation
to
bring
this
to
data.
I
What
what
we
need
is
some
changes
to.
I
have
a
design
document
somewhere
about
changes
to
the
external
provisioner
side
card
so
that
we
could
do
correct
error
reporting
when
people
refer
to
a
data
source.
That
is
not
valid,
that
that's
that's
one
of
the
important
missing
pieces.
A
Do
we
need
another
cab
for
this
to
address
those
or
it's
all
right,
indentation?
What.
I
I
believe
we
planned
on
amending
the
existing
cap,
but
a
separate
cap
could
also
work.
I
don't
I
don't.
I
guess.
That's
that's
what
the
community.
A
J
J
A
Oh,
my
this
one
is
crossed
out
I'll
figure
out
later
all
right.
Next
one
is
cozy.
Do
you
have
anyone
from
that
team?
C
I
can
speak
on
their
behalf,
so
they
are
planning
to
move
to
alpha
this
quarter.
They're
wrapping
up
the
design
in
the
next
few
weeks,
hopefully
call
to
action
here,
is
if
anyone's
interested
in
helping
implement
this.
They
need
folks
to
help.
C
So,
if
you're
interested
in
helping
code
this
and
helping
design
it
helping
shape
it
in
any
way,
you're
interested
in
object,
storage,
apis,
you're
interested
in
helping
develop
a
potentially
new
standard.
Here,
please
reach
out
to
jeff
or
sid
or
myself,
and
we
can
worry,
shang
or
any
of
the
other
tech
leads,
and
we
can
help
you
put
you
in
in
touch
with
the
right
folks
and
get
you
going.
A
A
C
Yeah,
I
think
we
keep
it
in
beta
and
then
the
line
item
should
be
aligned
with
the
the
other
feature.
A
A
Okay,
next
one
is
ifs
group
supporting
csi.
We
added
this
one
is
the
alphabet
sharing
1.19,
so
in
1.20
trying
to
bring
it
to
beta
priorities.
Two
does
this
look
right?
Is
christian
still
interested
in
working
on
this.
I
A
Thank
you,
so,
okay,
next
we
have
a
few
items
on
csm
aggression,
okay,
so
the
core
okay
planning
is
to
move
into
ga
from
outside.
Do
we
need
anyone
to
work
on?
Do
we
need
a
w
to
work
on
this
right.
C
Yeah,
so
we
need
an
owner
for
this.
The
core
csi
migration
code
has
been
sitting
in
beta
for
a
while.
Ideally,
we
want
to
get
that
moved
to
ga.
C
F
A
Okay,
so
the
next
one,
okay,
we
sphere
deviant,
is
divianco.
A
Okay,
it
looks
like
it's
not
okay,
so
this
is
the
beta
yeah,
so
we
will
keep
it
to
beta
because
we
do
have
that
requirement
for
these
two
versions
we're
trying
to
resolve
it.
If
that
get
resolved,
then
you
know
we
could
have
a
target
data
for
ga,
but
right
now
we're
still
working.
So
so
maybe
that's
beta
for
now.
A
A
C
Think
it's
still
alpha.
There
was
they're
trying
to
move
it
to
beta
last
quarter,
but
didn't
make
it.
A
B
Does
it
require
that
fact
that
it
uses
fs
group
to
mount
a
file
with
the
permissions
like
an
fs
group
is
not
supplied
via
csi
parameters?.
F
That's
a
good
question.
I
would
follow
up
with
andy
on
that.
A
Okay,
thank
you,
okay,
so
next
one
g
c
e
okay,
so
this
one
I
think
it's
a
beta,
maybe
1.17,
and
then
in
1.20,
we'll
try
to
go
to
ga
okay.
So
we
have
someone
work.
Oh,
this
is
two
okay.
Looking
for
yeah
this
one,
okay.
A
All
right
next
one
is
aws,
it's
a
migration,
so
this
one
also
want
to
go.
Ga
do
we
have
do
we
have
this
other
mat
here?
Do
you
confirm
this.
A
A
So
also
you're
not
sure
so
immediately.
K
No,
in
fact,
as
I
said
last
time
there,
the
openstack
cinder
driver
is
being
used.
I
I
don't
think,
there's
any
further
action.
We
really
need
to
take
on
this
one.
A
A
K
F
F
I
think
we
have
to.
We
have
to
look
up
who
worked
on
the
feature,
moving
it
to
beta
it's
in
the
enhancements
issue.
A
A
Okay,
so
next
few
items
see
seek
scalability.
C
Before
we
move
on
for
csi
migration,
I
created
another
tab
at
the
bottom.
There,
okay.
A
C
And
I
think
one
thing
I
wanted
to
clarify
was
currently:
we
have
two
feature:
flags,
csi
migration
and
csi
migration
complete
so
far,
the
csi
migration
flags
have
been
mostly
moved
to
beta,
except
for
azure
file,
I'm
actually
missing
cinder
and
staff
over
here.
I
need
to
add
those,
but
the
question
is
about
the
csi
migration
complete
flag
and
what
the
plan
is
when
those
are
going
to
be
moved.
C
F
Yeah,
that's
my
question.
So
the
purpose
of
the
csi
migration
complete
flag
is
to
is
to
say
my
kubernetes
nodes
are
at
a
version
where
csi
migration
is
on
and
cannot
be
turned
off.
F
So
I
think,
to
move
the
flag
to
ga
has
to
be
done
releases
after
the
csi
migration
flag
is
ga
whether
or
not
the
flags
move
from
alpha
to
beta.
I
I'm
not
sure
that
has
to
be
so
strict,
like
probably
we
should
have
moved
all
these
flags
to
beta.
F
But
basically
this
this
complete
flag
controls
the
compilation
of
the
entry
plug-in
into
the
actual
kubernetes
code.
So
it
actually
like
we'll
take
it
take
out
the
entry
plug-in.
F
Yes,
okay,
I
think,
and
it
it
also
removes
fallback
logic
right
now.
There's
some
fallback
logic
in
attach
detach
controller
to
see
if
the
nodes
are
on
a
version
that
doesn't
support
migration
or
migration
is
not
enabled
this.
This
feature
flag
controls
that
fallback
logic
and
if
we
turn
this
on,
then
that
fallback
logic
is
removed.
So
it
assumes
that
all
the
nodes
have
migration
enabled
okay.
C
Yeah,
so
we
might
want
to
add
these
to
to
the
spreadsheet
as
well
or.
A
Okay,
okay,
just
add
those
notes
in
for
each
one
of
those,
then
okay,
yeah.
C
A
A
All
right,
so
our
next
one
is
the
immutable
secrets
and
config
maps.
Okay,
so
the
plan
to
go
ga
does
this
remain
the
same.
So
we
have
I'm
not
sure
how
to
pronounce
this.
We
have
this.
D
C
Yeah,
let's
leave
it
assigned
to
him
and
we'll
get
verification
if
he's
still
gonna
work
on
it.
This
quarter.
A
Okay,
so
we'll
try
to
bring
that
to
ga,
then
the
next
one
is
see
apps
just
hpv
security
pacifist
will
not
be
auto
removed.
So
okay,
so
this
one
there
is
a.
I
think
there
is
a
cap
right
that
has
been.
E
Hi
this
is
kkk
here,
so
I've
updated
the
like
the
kept
section
here,
the
cell,
with
the
comment,
because
I
have
lost
edit
permission.
So
if
you
could
update
that
with
the
new
caps,
you
know
like
that.
One!
That's.
E
Cool
I'll
give
it
a
shot
so
latest
update.
I
I
don't
have
a
meaningful
update
since
the
last
time,
but
the
latest
update
that
is
there
is
that
we
have
some
sort
of
consensus
building
in
the
cap
and
you
know
like.
There
are
a
few
comments
which
I
need
to
add
to
the
kept
back
and
we
should
get
going
so
I'm
still
under
active
work.
E
A
All
right,
thank
you!
So,
oh
just
so,
this
is
going
to
you're
trying
to
bring
to
alpha
and
then
priority
three
okay.
Next
one.
E
Yes,
I
think
design
would
be
good,
because
there
is,
there
is
like
dependency
between
this
and
the
resize
feature
itself.
A
E
A
Oh
next
one
is
the
so
execution
hook
yeah.
So
we
we
had
a
meeting
and
then
there
was
some.
We
got
some
review
comments
from
signaled
and
then,
but
I
think
right
now.
A
What
we're
going
to
do
with
this
feature
is
we'll
try
to
have
a
alpha
version
of
the
api
changes,
but
for
the
controller
side
changes
we
will
try
to
implement
that
outside
of
cubelet
to
begin
with,
and
then
eventually
we
want
to
move
that
inside
the
cubely,
but
that's
a
longer
term,
so
sushant
and
I
will
be
working
on
a
cap,
so
I
mean
so
far.
We
do
have
some
something
like
google
doc,
but
we
want
to
have
a
formal
cap
for
this
yeah.
A
J
C
So
one
more
thing
I
wanted
to
call
out,
I
think
a
month
and
I
talked
about
volume
resizing
and
the
remaining
work
there.
I
think
what
we
wanted
to
do
was
call
for
action
and
see
if
we
could
get
more
people
to
help
with
that
is
anybody
on
the
call
interest.
D
In
helping
with
that
functionality,
I
can.
E
Help
sad
because
I
it's
directly
anyway,
in
line
with
the
stateful
set
work
as
well.
Awesome.
B
Awesome,
thank
you
kk.
I
have
yes,
we
have
some
bug
fixes
planned
for
120
like
and
then
we
have
obviously
the
recovery
from
this
failure.
Cap
and
we'll
I'll
try
to
split
the
work
that
could
be
done,
parallelly
and
write
some
some
stories
in
the
project
folder,
and
then
we
can
go
from
there
cool.
We
have.
We
have
a
project
code
for
each
feature
in
github
and
yeah.
You
can
pick
it
up.
We
can
pick
up
from
there.
E
So
I
have
an
update
on
the
file
permission
handling
I
just
ping
jing
and
deep
offline,
so
zing
is
going
to
post
the
issue.
So
basically,
we
are
looking
for
some
input
from
windows
expert
with
more
file
permission,
related
expertise.
E
So
zing
is
going
to
post
on
sig
windows,
the
specifics
of
the
questions
and
I'll
try
I'll
help
follow
up
on
like
getting
getting
a
person
to
get
involved
here.
A
A
A
See
what's
the
chat,
what
is
this
one.
A
Okay,
yeah
so
yeah,
so,
okay,
we
want
to
take
a
look
at
this
yep.