►
From YouTube: Kubernetes SIG Storage - Bi-Weekly Meeting 2022-02-10
Description
Kubernetes Storage Special-Interest-Group (SIG) Bi-Weekly Meeting - 10 February 2022
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
So
if
you
are
working
on
any
feature,
you
want
to
write
a
blog,
make
sure
that
you
ping
one
of
the
seek
lead.
So
we
can
add
that
in
this
spreadsheet,
so
we'll
start
adding
the
items
in
the
spreadsheet
and
then
and
the
deadline
the
next
deadline
march
23rd.
This
is
the
feature
block
opt-in
phrase.
Basically
by
that
data
we
just
need
to
add
the
entry
in
this
in
this
spreadsheet.
A
If
you
want
to
write
a
blog
and
then
march
30,
it's
cold
freeze,
I
think
the
and
then
there's
this
deadline
april
jobs
that's
later
for
the
dogs
and
blog
must
be
completely
then
reviewed.
B
Hello,
hi,
yes
yeah,
so
the
enhancements
are
on
track.
I
to
move
csr
online
offline
and
like
expansion
in
general,
to
ga.
I
have
one
bug
to
fix,
basically
like
at
least
as
far
as
general
expansion
is
concerned,
which
is
like,
like
the
the
read,
read
many
and
I'm
working
on
it.
I
have
not
actually
started
quite
yet.
To
be
honest,
I
had
it.
I
had
a
pr
that
didn't
get
merged
like
a
couple
of
days
back.
B
A
Okay,
thank
you
yeah.
So
that's
that
one
is
on
track.
Cap
is
merged
and
then
the
recovery
size
failure.
Do
you
have
any
update
for
this?
One.
B
Yeah,
the
cap
is
merged
and
it's
on
track
and
yeah.
So
well.
We
are
working
on
moving
the
feature
get
to
beta,
with
potentially
like
fixing
the
case
that
michelle
brought,
which
is
like
allowing
users
to
go
back
all
the
way
to
the
original
size.
A
A
A
A
A
And
yeah,
I
still
I
don't
have
update.
I
need
to
get
back
to
this
one.
A
I
thought
there
was
a
yeah
and
there's
a
release
out,
so
I'm
not
sure
what
we
okay
so
from
releases
out.
We
need
to
duplicate
ice
cuzzy.
We
break
up
okay,
so
I'm
not
sure
if
impumble
is
doing
this
or
is
it
humble
on
the
call?
A
E
I
think
last
time
I
I
followed
up,
we
were
trying
to
fix
the
build
so
that
it
would.
The
image
would
be
published
to
our
repo.
I
have
not
checked
to
see
if
that
was
fixed,
yet.
F
Yeah,
I
think,
we'll
need
to
make
andy
the
point
here.
Julie
is
taking
a
leave,
so
she
is
out
I'm
a
con
contact.
A
Thank
you,
and
next
one
is
a
pvc
volume
snapshot
name.
Space
transfer
is
a
mustafa
or
masaki
here,
so
I
I
asked
masakis.
He
said
he
is
interested
in
working
with
mustafandas
and
they
at
least
they
they
chatted
on
this.
So
but
it's
just
not
immediately
so
so
it's
still
talking
design
for
this
release.
A
Next
one,
since
I
want
to
house
additional
metrics,
there
is
a-
and
there
is
a
pr
I
think,
still
okay,
so
I
need
to
stop
ping
people
to
get
that
pr
merged.
A
And
the
next
one
is
the
volume
house,
programmatic
response,
so
I
need
to
check
with
nick
and
see
if
he's
back.
B
D
B
A
D
B
Actually,
xfs
singing
people
injured,
kernel
engineers
who
reviewed
some
of
that
kubernetes
code.
They
said
it's
not
a
good
idea
to
always
fsk
and
I.
G
B
I
think
we
can
take
this
that
offline,
but
there
was
some
concern
about
that.
It
could
be,
it
could
result
in
problems.
The
mounting
itself
replaced
the
journal
and
tries
to
correct
it.
So
it's
better
to
mount
an
unmount
rather
than
run
fsck
at
every
at
every
so
yeah
like,
but.
D
B
B
That
might
be
a
different
case,
but
the
feedback
we
got
from
kernel
engineers
on
the
issue
where
people
were
trying
to
add
for
xfs
was
like
running
trying
to
like
fsck
and
corresponding
xfs
repair.
Whatever
is
I
don't
remember
what
it's
called?
It
was
not
meant
to
be
invoked
automatically,
and
it's
it's
if,
like
programmatically,
it
will
make
things
worse.
So
we
had
to
revert.
D
A
Okay,
thanks
for
that
update,
so
the
next
one.
This
is
a
volume
populator
pen.
D
Yeah
yeah,
so
with
the
populators
we
are,
I
updated
the
cup
for
the
enhancement
freeze.
So
that's
done
we're
on
the
we're
on
the
list
of
things
for
124
we're
in
the
process
of
reviewing
the
metrics
change.
Seeing
I
don't
know,
I
didn't
see
a
response
from
you
yeah,
oh.
D
It's
fine
yeah
yeah
and
I'm
I'm
also
in
the
process
of
doing
the
the
other
metrics
update
to
the
library.
D
Will
be
testing
and
releasing
okay?
We
have
until
I
guess
the
end
of
march,
but
oh,
but
there
needs
to
be
a
an
entry
change
to
switch
the
feature
gate
and
there
was
a
big
discussion
yesterday
about
beta
features
still
defaulting
to
off.
But
then
we
concluded
that
that
only
applies
to
new
ap
beta
apis
and
the
beta
feature
gates
do
defaults
on
so.
A
Oh
yeah
actually
pull
up
on
that
one.
I'm
not
sure
if
michelle
have
you,
but
do
you
know
anything
about
this?
It's
a
beta
feature
known
as
a
cap
saying
on
by
default.
Jordan
submitted
a
cap,
for
that
is
that
it
was
for.
A
We
can
yeah,
I
don't
know
if
michelle
knows
anything,
but
any
otherwise
we
can
follow
up.
Is
that
offline.
A
Next,
cozy,
okay,
so
I
actually
did
not.
I
didn't
check
the
cap
status
of
michelle.
Are
you
still
reviewing
that
right?
I
have
not.
E
Yeah,
I
think
there's
sid
made
some
updates,
and
so
it's
on
my
plate
to
take
a
look
again.
Okay,.
H
Deep
hey,
so
I
had
a
review
with
signo
this
week
and
basically
the
the
feedback
was
they
want
to
see
the
apis
that
we
discussed
earlier
in
csi,
plum
through
cri
now
so
I'll,
be
just
updating
the
cap,
with
kind
of
like
the
new
suggestions
for
the
latest
suggestions
from
cignon
and
reach
back
to
them.
H
H
Exactly
there
will
be
some
changes
to
csi
which
are
not
as
controversial
yes
and
then
the
runtime
side
changes
signal
wanted
to
see
them
kind
of
propose
through
cri
at
least
and
then
they'll
do
further
reviews
and
iterate
on
that.
So
that
was
the
feedback.
A
H
A
And
send
out
application
notice
for
flux
volume,
so
we
do.
We
still
need
this
is
done,
so
we
don't
need
to
okay
leave
it
here.
Just
even
though
there's
any
follow-up
work,
okay
and
next
one
official
deprecated
cloud
provider.
This
one
also
says
done
so
we
just
leave
it
here
for
now
cs
migration.
Call
boxing
issues:
do
we
have
any
update
on
this
michelle
matt.
E
So
I
think
the
there's
one
bug
fixed
out
that
jaway
has
related
to
volume
limits.
I
think
that's
the
only
outstanding
issue
that
we're
aware
of.
A
And
csm
aggression
this
year,
it's
deviant
here
any
chance
so
right.
So
I
think
we
had
some
discussions
on
slack
on
this
one
right
so
right
now
the
main
issue
is
the
windows
support.
A
So
can
we
start
testing
so
running
problems
with
the
performance
and
there
are
some
concurrency
issues
when
it
runs
concurrent
tests.
So
there
are,
I
think
he
opened
three
issues.
One
is
the
performance
issue
with
the
csi
proxy.
A
There
is
a
temporary
fix,
but
didn't
seem
to
improve
it
in
our
own
testing
and
then
there's
one
issue
with
the
with
the
resize.
A
There
is
a
bug
fix
that
the
deviant
submitted,
that's
still
being
reviewed,
and
then
I
think
the
most
I
think
the
the
the
biggest
issue
is
with
the
read
only
one,
because
we
still
have
not
got
to
the
bottom
of
that.
That's
very
strange
issue.
A
When
running,
multiple
volume,
resize
operations
and
some
of
those
just
become
read
only
for
some
reason,
so
that's
still
so
that's
why
there
are
so
many
issues.
That's
why
we
were
thinking
that
just
for
the
windows
support.
We
just
keep
it
as
alpha.
A
A
So
that's
why
we
are
running
basically
we're
running
mixed
operations.
D4Ga
now,
then
we
range,
but
the
entry
drive,
of
course
never
comes
through,
was
testing
at
all
right.
So
we,
as
a
matter
of
fact,
we
never
got
any
questions
for
any
customers
regarding
windows,
support
with
the
entry
drive
also
without
it
anyways
that
you're
using
that.
So
because
of
that
right
now
we
are,
you
know,
thinking,
let's
just
keep
it
as
a
alpha,
so
we
can.
A
You
know
we
can
continue
discussion
on
slack
just
to
determine
like
you
know
whether
this
is
a
blocking
ga
or
so
what
about
the
so?
Basically
for
in?
Why
not
24,
because
right
now
we're
not
really
going
to
hear
yet,
but
we
in
we're
not
24.
We
target
turn
this
on
by
default,
but
still
stay
in
beta.
A
So
if
we
keep
the
the
windows
support
offer
without
your
problem,
I
guess
that's
more
immediate
question.
I
guess
for
1.24:
do
you
guys
have
any
thought
come
on
michelle.
E
E
So
I
guess
my
I
guess
the
question
is
we
don't
really
know
the
answer,
but
is
windows
for
entry?
Also,
some
have
similar
issues.
A
A
Re
so,
okay,
so
the
thing
is
that
okay,
the
csi
okay
first
of
all.
Well,
maybe
we
don't
have
to
talk
about
csr
proxy
issue,
because
I
think
the
seems
to
me
the
biggest
blocker.
Is
this
read
only
thing
which
we're
not
going
to
try
that
on,
because
we
don't
support
voting
expansion
for
entry
volumes
right.
So
we
won't
even.
B
E
Guess
it
would
be
good
to
get
a
summary
of
all
of
the
issues
and
we
can
like
go
through
it
one
by
one
to
evaluate,
like,
I
think,
the
resize
issue.
I
think
we
can
you
know
say
since
it's
not
it's
not
supported
entry,
then
it's
not
part
of
the
csi
migration
criteria,
but
I
want
to
understand
the
other
non-resize
related
issues.
You
were
seeing.
A
Yeah,
so
actually,
I
think
actually
there
are
three
issues.
Two
issues
are
related
to
the
size.
Only
one
issue
is
a
performance
issue,
but
that's
the
css
proxy
issue
that
it's
not
even
it's
not
a
vsphere
plug-in
specific
issue.
That's
because
the
powershare,
you
know
it
uses
a
lot
of
powershare
that
yeah.
E
J
A
A
A
E
Know,
but
I
like
that
yeah,
but
basically
our
I
think
our
rule
of
thumb
is:
we
only
need
to
match
entry
behavior,
okay.
So
if
entry
behavior
also
has
the
same
performance
issue,
then
we
can
say
that's
fine
for
the
purposes
of
the
csi
migration
feature
it
just
needs
to
match.
It
doesn't
need
to
be
better
yeah,
yeah.
A
H
So
just
a
one,
quick
data
point
there
shing,
so
it
sounded
like
someone
was
doing
a
very
specific
test
right,
like
bringing
up
pod
with
volumes
on
windows
every
second
within.
A
A
node
something
like
that
they
are
like
yeah,
we
have
like
performance
testing,
doing
something,
and
then
we
have
system
tests
and
doing
some
a
lot
of
concurrent
stuff,
a
lot
of
things
coming
together
and
then
and
then
they
are
trying
to
match
that
with
the
linux.
Of
course,
that's
just
impossible.
It
looks
like
windows
is
just
so
much
slower.
H
A
I
A
No
I'm
just
saying
because
kind
of
windows
new
right
so
like
I'm
just
saying
like,
for
example,
the
people
who
are
doing
perf
performance
testing
they're
like
oh,
how
come
it's
so
slow,
just
like
can't
compare
with
other
but
yeah,
but
you
can't
really
compel
them
they're,
just
different.
That's
what
we
learned.
E
A
A
B
A
B
A
E
Yeah,
I
think
this
one
andy
has
a
pr
out
to
bump
this
to
ga.
E
I
think
it's
on
my
plate
to
look
at.
I
think
I
just
wanted
to
double
sync
up
with
andy
and
double
check
that
he's
run
through
all
of
the
testing.
A
Okay,
thank
you.
Oh
one
thing
I
want
to
ask
is
that
maybe
we
should
okay.
Maybe
we
should
also
see
if
people
working
on
csm
migration
if
they
want
to
write
blog
so
humble
said
he
wanted
to
write
a
blog
for
sav,
cesar
migration.
A
A
A
Okay
yeah,
I
remember
there
was
something
that
got
put
up,
or
maybe
this
is
it's
the
next
two.
Okay
just
remember
something.
Maybe.
F
Yeah,
so
so
here
we're
actually,
the
gce
is
still
beta
in
124.
F
So
it's
beta,
but
it's
it.
It
just
went
on
in
gke
in
122,
which
is,
you
know,
still
hasn't
seen
huge
amounts
of
usage,
so
we'd
like
to
see
it
baked
there
a
bit
and
just
given
the
timeline
of
stuff.
That
means
the
125
is,
I
think,
when
we're
gonna
put
it
to
ga.
A
B
Okay,
yeah
by
the
way
for
azure
file,
it's
azure
furniture
disk.
It's
a
little
bit
curious.
As
in
like
the
cloud
provider
entry
cloud
provider
has
to
be
still
enabled
for
azure
file,
even
though
your
disk
is
being
moved
out
of
tree,
so
it
I
don't
know
if
anybody
will
run
into
any
trouble
because
of
that.
But
I.
F
B
Found
it
a
little
bit
a
little
bit
odd.
I
guess
that
azure
disk
will
be
ga
so
cloud
so.
A
Oh
you're,
saying
like
there
are
two
different.
You
think
they
should
be
going
same
way.
What
is
your
I'm
not
getting.
A
B
So
that
the
people
running
cube
control
measure
could
disable
the
entry
cloud
right
altogether.
But.
A
E
A
I
I
B
B
They
could
be
tagging.
The
notes
differently,
like
we
have
seen
like,
maybe
thus
possibly
like,
if,
like
it
depends
on
cloud
provider
like
in
vsphere
case.
We
have
seen
that
if
there
is
some
specific
zone
information,
then
the
engine
out
of
three
cloud
provider
could
tag
no
differently
and
cause
scheduling,
scheduling
problems
for
parts.
I
don't
know
if
it
applies
to
azure,
but
generally
we
don't
want
people
to
run
to
collaborators.
At
the
same
time,.
I
F
A
Thanks
and
next
one
is,
this
is
aws.
I
believe
this
one.
I
think
it's
going
to
be
delayed
right
michelle.
I
remember
there
was
some.
A
And
last
one
is
open
stack.
I
think
this
okay,
so
the
cap
merged
pi
is
also
merged.
It's
just
dogs,
so
probably
probably
going
to
pin
or
pin
dims
and
see
if
it's
a
updating
dog.
C
A
Next
one
is
always
owner.
Reclaim
policy,
d-pack.
A
A
So
the
next
two
secret
protection
and
eu's
protection-
those
are
so
both
are
masaki,
looks
like
usaki,
I
think
he's
not
in
meeting
today,
so
I
think
he
said
he's
still
interested
in
working
on
them,
but
I
think
it's
still.
This
will
be
designed.
B
Command,
no,
I
don't
think
there
has
been
any
movement
on
this.
Somebody
maybe
needs
to.
If
somebody
wants
to
own
this
and
kick
start
the
work
that'll
be
nice
because
that
owner
has,
I
don't
know
yeah
if
you
had
chance
to
sync
with
abhishek,
but
we
haven't
heard
anything
from
him
right.
B
A
Okay,
so
just
ask
the
questions:
if
anyone
want
to
hold
this
chat,
talk
to
your
vermont
about
this
cuz,
I
I
don't
know
what
this
one
is
about.
B
There
was
a
cap,
but
it
was
not
a
proof.
It
was
yeah,
it
was
kind
of
provisional
and
it
needed
a
lot
of
lot
more
work.
A
A
So
so
she
I
changed
with
her
last
week,
she's
still
trying
to
update
the
the
cap
addressing
comments,
but
I
have
not
does
not
look
like
she
submitted
the
updated
cap.
Yet.
B
A
So
you'll
be
saying
that
the
the
cap
does
not
have
that
that
much
detail
you're
saying
that's
the
but
the
the
other
cap
that
you
reference,
you're
saying
that
one
is
more
right.
A
A
Yeah
so
she's
she's
reading
that
the
the
other
cap
that
you
referenced.
So
I
think
she's
trying
to
add
some
of
that
from
this
other
cup
to
this
new
cab.
Just
I
don't
have
not
seen
her
final
version
so
but
I'll
pin
her
again
see
where
she
is.
A
Next
one
is
a
execution
hook
hook
until
notifier,
so
this
one
still
design,
I'm
not
gonna
update
on
this
one.
A
Thanks,
I
probably
should
move
this
one
up
yeah.
I
think
this
is.
This
is
a
six
story
on
me:
yeah,
okay,
all
right!
So
that's
everything
in
the
spreadsheet
going
back
here!
I
okay!
There
are.
We
don't
have
any
other
issues
added
here.
Does
anyone
have
anything
else
you
want
to
discuss
in
today's
meeting.