►
From YouTube: Kubernetes SIG Storage - Bi-Weekly Meeting 2023-02-23
Description
Kubernetes Storage Special-Interest-Group (SIG) Bi-Weekly Meeting - 23 February 2023
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
Hello,
everyone
today
is
February
23rd
2023.
This
is
the
kubernetes
36
meeting.
So
today
we're
going
to
go
through
1.27
planning.
We
have
passed
the
enhancement
freeze
deadline.
The
next
deadline
will
be
code.
Freeze,
that's
March
14th.
A
A
Okay,
so
the
so
the
API
changes,
those
are
working,
progress.
A
A
And
the
the
next
one
is
voting
group,
one
group
API,
so
yeah
this
the
cat
was
merged.
Css
back
was
just
merged
as
well.
So
we
now
started
implementation,
I'm,
hoping
to
get
get
the
coding
minted
in
1.27.
C
Yeah
I
I'm
here
yeah
we
should
cut
a
CSI
spec
release.
Question
really
is:
when
do
we
want
to
do
it?
Is
there
any
other
pending
changes
that
we
want
to
wait
for.
C
Is
a
there's,
a
CSI
spec
meeting
next
Wednesday?
Should
we
discuss
it
there
and
see
if
there's
anything
else,
if
not
go
ahead
and
cut
sure.
C
C
Sure
yeah
I
can
send
out
a
message
to
CSI
community
and
see.
A
A
Is
masaki
or
takafumi
here.
A
A
And
the
next
one
is
the
change
block
tracking.
Do
we
have
anyone
Eva
or
anyone
talking
about
this.
A
A
And
the
next
one
is
River
wants
access
mode,
there's
Chris
here.
A
And
the
next
one
is
deeps,
runtime,
assisted
mounting
is
deep
on
a
call.
A
A
A
This
is
the
runtime
classes,
assist
demand,
okay,
I.
D
To
sort
of
revamp
it
into
the
way
that,
like
the
CSI
drivers,
could
be
I,
guess
and
not,
CSI
drivers
will
have
to
have
a
new
capability
to
say
this.
They
support
this
feature,
but
they
wouldn't
have
to
directly
interact
with
the
forgiving.
The
the
new
acronym.
E
D
A
E
D
E
D
D
A
Yeah
I
haven't
not
seen
him
in
this
meeting
for
a
while
either
okay,
maybe
he's
busy
okay
in
the
next
one
see
the
proxy
for
Windows
transition
to
privileged
containers,
so
I
think
Mauricio
has
a
cap.
A
And
the
okay:
next
one
is
a
node
expansion
secret
E3
test,
so
we
did
get
the
cat
merge.
There
is
the
E3
test
out.
There
are
some
comments
there
and
there's
some
failure
on
YouTube,
so
I
pinned
Luis.
He
said
he's
going
to
take
a
look
at
tennis
team
review.
F
Well,
not
much,
but
I
haven't
done
anything.
However,
I
am
untangling
it
from
the
volume
reconstruction
cap.
Let's
track
later.
F
A
And
next
one:
okay:
now
we
have
CSM
migration,
remove
entry,
plugins
1.28
ones.
So
next
one
remove
entry,
AWS
eps,
so.
G
Yeah
hi
folks,
so
there
is
a
PR
open
for
this
now,
which
I
will
link
in
the
chat
so
that
Jan
reviewed
it
and
noted
that
the
code
changes
looked
good.
So
thanks
Jan
for
taking
a
look.
There
is
an
ongoing
discussion
related
to
concerns
with
removing
the
Legacy
cloud
provider
in
that
PR
I've
added
that
to
the
Sig
cloud
provider
agenda
for.
F
I
can't
go
to
that
meeting
next
week.
So
is
there
anybody
who
goes
to
SQL
provider
meetings
or
who
has
some.
H
I
think
from
AWS
one
of
our
team
members
will
be
attending
so
either
Eddie
or
or
me.
One
of
us
will
be
there.
We
can.
We
can
probably
raise
that
if
anybody
else
would
like
to
join
from
six
storage.
That
would
be
great
as
well.
F
Yeah,
like
from
our
perspective,
we
can
remove
the
cloud
provider,
we
can
remove
the
volume
plug-in.
We
have
no
issues
with
that,
so
if
they
can
manage
the
documentation
around
and
whatever
they
need,
because
I
noticed
they
have
their
old
caps
to
remove
the
cloud
providers
and
they
have
their
own
feature
gates
for
that,
and
we
are
kind
of
removing
their
code
providers
now,
which
is
probably
bad.
You
shouldn't
do
that.
F
So
from
our
point
of
view,
we
can,
we
don't
need
them,
we
can
remove
anybody,
can
remove
them,
it's
up
to
them
to
define
the
workflow,
how
to
remove
them
correctly
if
they
want
feature
Gates
if
they
want
beta
or
whatever,
we
don't
basically
care,
but
we
do
care,
but
it's
up
to
them
to
Define
how
to
remove
the
cloud
providers.
That's
the
message
I
want
to
somebody
to
carry
there.
H
Yeah
I
think
Eddie
Eddie
knows
more
details,
but
my
understanding
was
that
the
concerns
are
around
the
fact
that
not
all
the
cloud
providers
are
quite
ready
to
duplicate
the
entry
version.
At
this
point
and
and
so
if
you,
if
we
remove
the
the
entry
cloud
provider
code
in
entry
cloud
provider
code,
then
that
that
might
impact
their
entry
storage,
specific
provider,
storage
provider,
specific
implementation
as
well.
So
it's
so
that's
probably
why
there
is
some
pushback
around
that.
F
The
storage
implementation
is
not
used,
we
still
have
the
volume
plug-ins,
but
they
are
disabled.
Everything
has
migrated
to
CSI,
at
least
in
okay,.
F
F
May
have
other
features
than
storage
in
the
cloud
providers.
I
know
the
label
notes
and
they
have
to
external
code
providers
now
which
do
the
same
job.
Basically,
and
maybe
that's
not
gaas
I
see
I,
don't
know,
okay,
I,
don't
know
so
it's
up
to
them
to
define
the
pace
and
when
when
they
want
to
remove
it,
but
if
you
don't
block
it
like
from
our
perspective,
they
we
don't
need
the
cloud
providers,
at
least
those
that
are
scheduled
to
be
removed
in
this
release,
which
were
GC
and
AWS
I
believe
and
Azure
disk.
H
Yeah,
so
so
yeah
so
I
think
from
our
side.
If
I
understand
it
right,
you're
saying
that
we
should
be
pushing
to
get
that
removed.
Is
that
is
that
accurate,
or
are
you
stating
that
we
can?
We
can
hold
off
on
it.
F
H
Yeah
yeah,
it's
right,
I
think
I.
Think
the
main
concern
is
that
this
already
a
PR
out,
then
the
longer
that
it
stays
in
review.
You
know
somebody
will
probably
have
to
do
that
work
again
within
the
latest
Branch
or
something
so.
F
A
Thank
you,
okay,
so
that's
this
one,
and
and
we
also
have
remove
entry
Azure
disk
I
pinned
Andy.
He
said
he
will
be
working
on
that.
F
We
can
remove
the
volume
plugins
from
like
btg
slash
volume,
slash,
AWS
or
Azure
disk
that
that
we
can
remove
on
our
behalf.
We
don't
need
to
ask
anyone
so
that
that's
something
we
can
do
in
127.
A
A
A
Let's,
just
let
us
know
you
know
to
hear
from
them
next
next
meeting
give
us
a
update
on
this.
A
A
And
next
one's,
so
this
didn't
really
go
in,
so
we
should
probably
do
we
just
cross.
This
out
is
that
the
cap
did
not
make
it
right.
It
didn't.
A
And
next
one
control
volume,
mode
conversion
between
source
and
Target
yeah.
This
one,
the
the
E2
test,
is
already
in
cut
merge.
So
it's
basically
it's
just
documentation
now
and
blog.
A
So
so
next
one
is
this
project
we
called
with
seek
apps
PVC
created
by
stifleset,
will
now
be
Auto
removed.
This
map
here.
A
E
A
B
A
I
think
that's
all
we
have
shot
here,
yeah,
okay,
going
back
here.
Let's
see
what
do
we
have
so
we
already
discussed
this
issue.
A
Oh
this
is
I.
Don't
know
if
Akash
is
here.
Natasha
is
not
here
today,
okay,
so
this
is
the
front
seat
and
Akash
for
cozy
side.
So
basically
asking
if,
if
there's
any
plan
to
write
cozy
drivers
for
Google
cloud
storage
and
AWS
accessory-
and
you
know
people
from
Google
and
AWS.
H
G
A
They're
they're
easy!
That's
why
we're
asking
here
because
we
don't
have
anyone
from
AWS
attendee
meetings
there
right.
So
if
you
could,
maybe
you
could
check
internally
see
if.
H
C
Yeah
same
same
thing
from
the
Google
side,
I
need
to
follow
up
with
GCS
folks
I'll
check
with
Matt
Carey.
If
we
can
find
a
contact
and
get
an
answer.
F
Okay,
so
in
the
past
we
were
meeting
every
Monday
and
every
Wednesday
about
CSI
implementation,
but
we
turned
the
Wednesday's
meeting
to
issue
triage.
So
we
go
through
some
six
storage
issues
reported
on
GitHub
and
we
would
welcome
new
contributors
or
anybody
who
is
willing
to
help
with
issues.
We
have
quite
a
lot
of
them,
so
we
meet
every
Wednesday
at
10
AM,
which
is
one
hour
later
than
this
meeting.
F
We
use
the
same
Zoom
core
as
this
I
link,
the
notes
that
we
use
and
the
project
board
that
we
use
to
trash
issues
and
you
can
notice
there.
There
are
a
lot
of
issues
that
are
marked
as
hell
wanted,
I'm
afraid
that
all
the
easy
issues
were
already
fixed.
So
we
have
pretty
complicated
issues,
some
of
them
waiting
for
a
couple
of
years
for
a
fix
and
simply
like
we
are
out
of
Manpower.
We
need
some
people
who
would
like
to
contribute.
We
can.
F
We
can
explain
those
issues
and
show
you
where
to
look,
how
to
fix
it
where
to
when
to
fix
it,
but
we
don't
have
enough
people
to
like
do
the
real
work
and
fix
these
issues
so
anybody's
welcome.
You
will
need
some
go
experience
and
I
think
we
will.
We
could
teach
you
the
rest
but
yeah
these
issues.
They
are
complicated.
A
A
Okay,
let's
see,
do
you
have
anything
else,
I
think!
That's
all
we
have
here.
H
I
I
had
a
couple
of
I
had
a
couple
of
quick
things
that
I
wanted
to
mention
yeah,
so
we
recently
have
been
asked
about
an
upstream
kubernetes
issue,
I
just
linked
that
in
the
in
the
in
the
meeting
chat.
This
is
something
that
that
has
been
brought
to
our
attention
by
more
than
more
than
one
customer,
and
we
are.
H
We
are
not
sure
at
this
point
like
what
the
best
way
to
fix
it
is.
There
is
also
a
corresponding
AWS
CBS
CSI
driver
issue
that
I
haven't
linked,
but
it
it
essentially
points
to
the
to
the
Upstream
code.
So
I
believe
this
was
something
that
Heyman
had
indicated,
that
there
may
be
a
workaround
for
and
I
just
wanted
to
check.
If
folks
had
had
any
thoughts
regarding
this
in
terms
of
you
know
what
the
right
fix
would
be,
and
if
there
was
any
plan
to
fix
this.
B
A
G
H
Particular
node
and
that
limit
is
not
being
respected
due
to
some
kind
of
a
race
condition
and
it's
it's
seemed
from
the
discussion
that
there
may
be
a
workaround
that
Heyman
suggested,
which
was
to
initialize
that
CSI
node
object,
I
think
with
a
value
of
zero
that
that
might
prevent
the
you
know
more
parts
from
being
scheduled
on
that
node
I
just
wasn't
sure
if
folks
had
evaluated
and
determined
whether
that
was
the
right
approach
and
or
if
further
work
was
being
planned
for
that.
H
I
think
this
has
been
well
I,
don't
know
about
other
drivers,
but
I
know
I
know
for
a
fact
that
there
are
multiple
customers
who
have
who
have
seen
this
issue
so.
F
Yeah
I
think
it's
a
generic
how
to
scalar
issue,
but
I.
Don't
think
anybody
is
actively
working
on
it.
A
Yeah,
so
this
is
something
that
we
probably
need
to.
You
know
go
through
this,
and
you
know
maybe
need
to
find
your
own
owner
to
book,
covers
yeah
I.
E
H
Suggest
the
right
fix,
you
know,
I
think
the
the
challenge
that
we
have
is
that
we
don't
necessarily
have
expertise
in
the
Upstream
code
base
at
this
point
and
we
don't
want
to
make
a
fix
which
only
ends
up
causing
more
confusion
and
problems
later
on.
So
so
you
know
it
would
be
good
if
somebody
could
at
least
do
the
analysis
and-
and
let
us
know
what
the
fix
is
and
then
at
that
point
we
can
think
about
potentially
implementing
something.
So.
A
H
A
H
Don't
know
I,
don't
know
if
you'll
be
able
to
do
it
in
time
for
the
next
one,
but
but
we'll
we'll
try
to
look
into
this
from
our
side.
So.
E
A
H
A
Okay,
so
just
this
note
here.
H
Yeah
and
then
the
other
question
was
actually
for
Matt,
but
I
think
he's
not
here.
We
we
had
some
follow-up
questions
regarding
the
the
provision
IO
cap
proposal
that
that
Matt
is
driving,
okay
and
I.
Think
we
posted
our
questions
on
the
cap.
H
We
got
answers
to
some
of
them,
but
I
think
there
are
still
some
questions
outstanding
and-
and
it
would
be
great
if
we
could
get
some
answers
on
those
as
well,
because
some
of
our
decisions
about
you
know
how
we
would
like
to
go
about
implementing
this
would
be
determined
by
you
know
what
those
what
those
responses
are
so
so
yeah
so
that
that's
that's
basically
what
I
wanted
to
mention.
It
would
be
great
if
you
could
see
some
some
traction
on
that
was
that
was.
A
Yeah,
maybe
just
to
see
if
Matt
wants
you
see
if
you
want
to
organize
a
meeting
to
discuss-
or
you
want
to
just
to
discuss
that
through
the
slack.
H
Yeah
I
think
we
can
probably
try
slack
and
and
reach
out
and
see
if
you
know,
there's
any
update.
Otherwise,
we
I'll
attend
the
one
of
us
will
attend
the
next
meeting
and
and
we'll
we'll
bring
Matt
at
that
point.
If
he's
here.
A
Oh
okay,
so
basically
just
another
dress,
so
maybe
yeah
okay.
So
we
can
just
ping
him
just
to
ask
him
to
address
those
through
the
cap
review
as
well.
Yeah.
H
I
think
I
mean
the
reason
why
I'm
asking
about
that.
That
is
because
I
think
those
questions
are
related
to
some
fundamental
challenges
with
implementing
the
cap
in
its
current
form,
and
you
know,
we
would
like
to
see
if
it's
possible
to
modify
the
cap
to
account
for
those
those
scenarios
essentially
and
that's
why
we
were
asking
about
that.
So.
A
Okay
sure
yeah
so
just
pass.
Let's
note
here,
ping
him
as
well:
let's
see
it's
a
Sony
Sony's
was
not
here
and
the
Sun
is
another
contact
person
for
this
I'll
just
drop
him
a
note,
and
maybe
he
maybe
he's
busy
with
other
things.
Okay,
yep.