►
Description
Kubernetes Storage Special-Interest-Group (SIG) Object Bucket API Standup Meeting - 01 February 2021
Meeting Notes/Agenda: -
Find out more about the Storage SIG here: https://github.com/kubernetes/community/tree/master/sig-storage
B
B
So
good
morning,
everyone,
this
is
the
first
day
of
february.
B
The
world
is
in
a
really
weird
spot
right
now,
with
with
reddit
taking
over
hedge
funds
and
yeah.
I
just
wanted
to
say
this
on
recording.
So
when
we
look
back
at
it,
we'll
remember
these
times,
so
I
want
to
so
generally
on
mondays,
we
have
technical
meetings.
We
go
over
technical
issues
that
are
facing
at
that
point,
get
updates
on
where
we
are
at
in
terms
of
development.
B
So
I
would
actually
like
to
get
get
an
update
from
chris
who's
here
and
also
maybe
understand
more
about
what
blaine
is
working
on
and
how
how
we
can
be
of
help
to
him.
B
Since
again,
we
have
very
few
people
today
and
other
than
that
I'll
quickly
go
over
where
we
are
in
terms
of
our
overall
plan.
B
So
our
current
priorities
right
now
are
getting
the
demo
out,
and
I've
mentioned
this
many
times
before.
The
demo
is
not
just
a
smoke
and
mirrors
kind
of
demo,
not
just
something
that
works,
but
we're
using
it
as
a
milestone
for
developing
end-to-end
functionality.
B
We
have
this
use
case
working
as
a
right
now,
however,
there
is
one
pull
request,
that's
pending
and
that
pull
request
needs
to
be
merged
in
one
of
the
api
reports
in
the
spec
repo
and
once
that
is
merged,
we'll
have
to
make
some
changes
to
our
existing
code
to
reflect
that
new
api.
B
So
as
of
right
now,
that's
the
one
part
left
in
the
development
before
we
can
say
that
the
demo
is
ready.
Another
constraint
that
we
added
on
the
demo
was
to
have
have
it
be
as
production
ready
as
possible.
What
we
meant
by
that
was
we'll
have
docs.
We
would
have
tested
it
in
our
environments
as
much
as
we
can
and
not
just
manually,
but
also
write
test
cases.
B
So
so
that's
where
that's
that's
the
that's
the
scope
of
the
demo
and
in
terms
of
testing
and
documentation.
We've
made
a
lot
of
progress.
We
have
a
good
amount
of
documentation
present
and
we
have
more
people
working
on
it
and-
and
we
also
encourage
more
people
to
help
us
out
with
any
of
these
efforts,
development,
documentation,
testing.
B
So,
for
the
demo,
like
I
said,
everything
green
here
is,
are
things
that
are
done?
Everything
is
in
yellow
are
things
that
are
being
done,
however,
based
on
how
we
defined
the
requirement
for
the
demo
to
be,
I
think
I
think,
we're
in
a
good
shape.
B
The
next
step
is
api
review,
so
api
review
is
very
important
for
us
before
we
become,
we
can
go
alpha
in
the
kubernetes
ecosystem
and
and
that
will
make
our
project
much
more
visible
to
others,
and
also
it
will
help
us
attract
more
contributors
and
move
faster,
so
for
the
api
review,
the
the
same
pull
request
that
that
we're
waiting
on
for
the
development
is
also
you
know
relevant
here
once
this
is
merged.
B
B
B
I
have
sent
a
link
to
this
on
the
chat
and
this
pull
request
basically
follows
up
on
what
we
had
decided
in
in
in
the
meeting
in
one
of
these
meetings
a
few
few
weeks
ago,
probably
two
or
three
weeks
ago.
If
sad
is
here
so
you've
addressed
your
comments.
Are
you
here?
B
No
you're,
not
okay
and
yeah,
so
shin
I'll
send
out
another
email
to
saad,
hoping
that
he
would.
He
would
take
a
look
at
it
again.
It'll
be
good.
If
you
can
also
take
a
look,
so
it
might
be
easier
for
saad
to
approve
and
follow
what's
going
on.
If
you
also
take
a
look,
I
think.
A
B
B
Okay,
okay,
so
that's
like
next
friday
right,
it's
probably
wednesday
next
wednesday
next
week,
I
think
yeah
next
wednesday,
so
yeah
anyway,
so
yeah!
That's!
Why
that's
why?
This
is
a
high
priority
issue
for
us
and
you
know
we
want
to
get
it
through
as
soon
as
possible.
Any
questions
so.
B
Far
now
I
would
like
to
get
an
update
from
the
people
who
were
working
on
things
already
chris.
What
were
you
working
on
and
where
are
we?
I
think,
you're
working
on
the
csi
adapter
right.
A
Yeah,
that's
right.
Can
you
hear
me?
Okay,
yeah,
okay,
I've
just
been
working
on
doing
manual,
testing
and
validating
the
behavior
of
the
csi
adapter.
A
So
last
week
I
was
having
some
trouble
getting
the
deployment
working,
but
it
seems
like
that's
mostly
resolved
now.
I
did
some
work
over
the
weekend.
B
A
A
Yeah,
it
wasn't
calling
node
capabilities.
A
Yeah,
and
so
I
took
a
look
at
some
other
manifests
that
are
out
there
and
I
made
some
changes
and
it's
it's
working
now
so.
B
Okay,
great,
okay
and
okay,
once
that
is
done
so
do
you
have
have
you,
have
you
tried
having
multiple
buckets
in
the
same
pod
and
did
you
see
if
it
mounts
correctly
and
everything.
A
So
I
managed
to
get
it
calling
the
all
the
appropriate
endpoints
now.
So
it's
calling
like
note
get
capabilities
and
then
node
stage
and
on
stage,
which
was
what
we
needed
to
get
capabilities
call.
So
now
it's
just
a
matter
of
making
sure
that
it's
mounting
the
secret
correctly.
B
Got
it
okay,
yeah?
We
we
should
go
over
the
secrets,
one
more
time
like
the
like
the
file,
permissions
and
yeah,
just
some
properties
of
how
it's
retrieved
and
how
it's
saved
yeah
we'll
go
over
it
once
again,.
A
B
Yeah
jeff,
I
I
noticed
you
were
here.
You
were
working
on
the
cap
right
right,
yeah,
I
think.
Last
we
spoke
yeah
you,
you
left
it
at
a
good
point
and
someone
I
think
I
was
gonna
take
over.
Is
that
right,
just
yeah.
C
Yeah
I
I
got
the
kept
to
reflect
current
api
in
rpc
spec,
because
that
had
drifted
some
from
the
original
cap
that
had
been
merged
and
so
that
pr
was
ready
to
go
but
sid.
Then
you,
you
were
saying
for
api
review.
C
We
need
to
make
additional
changes
to
the
cap
right,
make
it
implementable
to
move
status
from
unimplementable
to
you
know
ready
for
alpha,
and
you
wanted
to
do
that
from
what
I
read
in
this
same
pr,
rather
than
loot
pr
because
of
overhead
of
getting
people
to
review
the
pr
so
that
additional
work
of
some
new
sections
in
the
cap,
some
formalities
around
making
it
implementable,
where
it's
going
to
take
extra
time
for
me-
and
I
think
since
there's
some
time
urgency
here
and
it's
not
a
high
priority
in
my
work
right
now
that
use
to
get
it
done.
C
B
Yeah,
no,
that's
that's
that's
best,
I
think
yeah
and
if
I
need
any
help,
I'll
I'll
reach
out
absolutely
yeah.
So
so
so
you've
done
the
majority
of
the
work.
I
just
have
to
add
sections
for
production,
checklist,
production,
readiness,
checklist
and
whatever
else
is
required
for
making
it
implementable.
I
believe
there
are
a
few
sections
for
it
like
upgrade
path
and.
B
Yeah
so
yeah,
okay,
so
so
this
is
good.
I
was
hoping
srini
and
rob
would
be
here
too,
because
they're
working
on
the
controller
and
provisional
sidecar,
I
was
hoping
also
janus
would
be
here.
He
is
working
on
the
documentation
updates,
but
I
think
today
has
been
a
slow
monday
for
a
lot
of
us.
C
B
D
I
I
work
with
I
work
for
redhead,
I
work
with
the
openshift
container
storage
product
and
I
practically
speaking.
That
means
I
work
on
the
rook
upstream.
D
Project
and
we
currently
use
the
sort
of-
I
guess
what
I
have
sometimes
called
the
like
prototype
version
or
the
like
initial
kind
of
like
concept
for
for
cozy.
D
In
the
the
lid
bucket
provisioner
library
and
have
I
have
worked
recently
with
looking
at
that
library
and
how
to
how
it.
A
D
Cozy
yeah
with
I
guess,
like
mostly
I'm,
just
interested
in
tracking
what
cozy
is
doing
as
far
as
what
you
know
how
the
project
is
going
at
some
point,
I
would
like
to.
D
Yeah
at
some
point
I
would
like
to
integrate
cozy
into
rook
so
that
we
could
have
that
as
an
option.
B
D
Yeah,
I
mean
I
I
guess
like
I
want
to
know
like
what
is
what
is
the
best
way
forward,
like
with
you
know,
with
rook
and
with
cozy.
B
D
And
like
how?
How
do
we
migrate
users,
who
are
using
the
the
object
bucket
claims
to
using
object.
B
C
Ahead
or
just
a
comment,
a
couple
rook
orchestrates,
storage
in
general
and-
and
so
you
know,
it
also
has
crds
for
the
whole
object
store
itself.
C
If
oz
assumes
exists,
you
know
we
don't
we
don't
do
that
in
cozy
and-
and
so
you
know,
it's
sort
of
a
single,
a
single
point-
a
single
control
plane
to
let
administrator
that's
using
rook
to
control
all
their
storage,
including
objects.
So
I
think
I
think.
B
Yeah,
so
you
know
every
time
you
know
that
we've
tried
to
so
I
mean
the
beauty
of
kubernetes.
The
beauty
of
linux
also
is
that
you
have
tools
that
do
one
thing
well
and
they
do
it
really
really
well
ten
times
better
than
everything
else.
So,
if
you
look
at
grep,
it
just
does
filtering
of
strings,
and
it
does
really
really
well.
It's
highly
scalable,
it's
highly
flexible
same
thing
with
kubernetes
like
like,
take
take
the
api
server,
for
instance,
or
take
the
q
proxy
or
cube
scheduler.
B
They
all
do
one
thing
really
really
well
and
that's
what
allows
us
to
scale,
because
you
are
now
integrating
compact
but
simple
pieces
and
it's
easier
to
build
on
top
of
simple
pieces
when
you
want
to
go
to
high
scale.
B
Cosy
is
purely
object,
storage
now,
rook
also
orchestrates.
For
for
for
a
given
crd
that
requests
object,
storage,
rook,
I
don't
know
if
you
can
choose
to
use
either
cassandra
or
mineo
or
some
other
object-
storage
provider,
but
but
we'll
go
provision.
The
storage
for
you,
based
on
based
on
that
crd,
that
you
create
cozy,
does
the
same
thing,
but
in
a
much
more
you
know
we're
going
to
be
more
cube
native
and
I
love
any
of
these
providers.
Object.
B
Storage
providers
like
like
sandro
or
mini
or
others
to
integrate
with
cozy,
so
so
they
they
kind
of
overlap
and
and
rook
does
more
than
just
object.
Storage
because
he's
going
to
be
focused
and
going
to
be
going
to
do
just
object,
storage,
and
do
it
really
well.
C
It's
just
a
common
pattern.
You
have
kubernetes
as
a
highly
scalable
performant
framework
and
people
build
frameworks
on
top
of
frameworks
and
saying
it
is.
B
B
What's
the
common
pattern
at
a
high
level?
Yes,
it's
it's
a
common
pattern
like
writing,
another
controller
or
writing
another
framework
on
top
of
a
framework,
but
then,
but
then
you
know
you
could
say
like
python
and
go
are
the
same
in
that
sense,
because
they're
both
frameworks
for
writing,
programming
or
programming
on
top
of
the
kernel,
so
you
know
I,
wouldn't
I
wouldn't
put
it
the
same
way.
Cozy
and
rook
are
not
the
same
thing.
Well,
I'm
not
saying
they're
the
same.
C
I'm
saying
I'm
I'm
just
trying
to
explain
the
the
the
use
case
or
value
of
something
like
rook
and
then
there's
even
higher
levels
of
of
abstraction
that
can
that
can
overlap.
Look
like
cross
plane,
I'm
just
saying
it's.
It's
it's
good
to
have
rook
on
board,
with
getting
rid
of
deprecated
code
related
to
object,
provisioning
and
to
plug
into
the
native
kubernetes
way
of
doing
it.
B
Yeah
yeah,
the
the
right
solution
would
be
for
rook
to
migrate
to
start
leveraging
cozy
right
once
it
can
be
leveraged
right
once
you
know
no,
we
should
start
integrating
now
because
we
already
have
a
working
product
because
it's
going
to
be
a
continuous
thing,
it's
going
to
be
a
living,
breathing
integration
effort.
It's
not
going
to
be
a
one-shot
thing.
B
Work
and
we
can
get
real
feedback
on
absolutely
so
so
blaine
so
going
back
to
how
we
started
the
discussion,
were
you
able
to
follow
everything
you
were
saying.
D
I
I
think
so
I
I
have
a
a
good
understanding
of
the
lit
bucket
provisioner,
I
think
sort
of
which
is
kind
of
the
idea
from
which
cozy
was
born,
and
I
know
cozy
like
has
expanded
on
those
ideas
quite
a
bit,
and
I
yeah
I'm
I'm
here
to
try
to
figure
out
when
is
the
right
time
and
like
how?
How
best
can
I,
with
the
rook
project,
start
integrating
cozy
into
the
things
that
rick
is
doing?
B
Yeah
yeah,
I
think
I
think
that's
going
to
help
the
project
a
lot
having
having
others
integrated
with
it.
B
My
my
recommendation
would
be
that
you
start
playing
with
cosy
and
and
once
once
you
get
a
hang
of
it
you,
you
can
start
looking
at
different
integration
points,
so
yeah
I've
also
been
involved
with
the
rook
project
in
cross
plain.
I
don't
know
if
you
know
someone
named
bassam
from
yeah,
so
when
he
was
just
starting
up
on,
we
had
a
long
chat.
He
was
asking
me
to
go
work
at
upbound.
B
This
was
like
three
years
ago
when
he
just
had
the
proposal
for
cross
plane
and-
and
I
ended
up
going
to
minaio.
But
you
know
it's
a
small
community.
We
keep
coming
back,
so
I'm
happy
to
see
that
crossplane
has
taken
off
yeah,
I'm
excited
for
it.
So
so
I'm
looking
to
learn
more
about
that.
However,
about
rook
I
have.
I
have
a
decent
understanding
of
it
and
again
through
this
integration
effort,
we'll
we'll
you
know
I'll,
get
a
better
understanding
of
it
as
well.
B
As
you
know,
we'll
be
able
to
iron
out
the
cozy
api
better
as
well.
C
Yeah
and
just
to
plug
blaine
he's
being
modest,
you
know
he's
done
several
prs
in
the
with
the
old
object
library,
bucket
provisioning
library
and
understands
some
of
the
intricacies
of
writing
a
controller.
That's
that's!.
D
D
I,
like
my
my
next
goals-
I
I
guess,
are
sort
of
to
see
in
in
our
downstream
development
like
where,
where
does
this
integration
fit
in
our
priorities?
And
what
sort
of
time
can
I
can
I
have
to
to
to
dedicate
to
it
based
on
those
priorities
which,
which
also
then
would
sort
of
you
know
what?
All
of
that
I
would
include
development
that
might
go
toward
cozy
itself
and
then
also
toward
the
integration
of
of
cozy
and
wreck.
B
Yeah,
I
think
that's
the.
I
think
that
would
be
the
best
way
for
you
to
get
involved.
Then,
if
you
as
a
part
of
integration,
you
might
have
to
make
changes
here
that
way
you
would
get.
You
would
get
a
good
exposure
to
cozy,
but
you're
always
welcome
to
start
sooner.
B
We
have
do
have
some
issues
that
need
to
be
looked
at,
but
you
know,
I
believe,
naturally
integrating
with
the
project
naturally
getting
involved
is
the
best
way
to
do
it
as
you
as
you
integrate
with
rook.
C
Yeah,
I
think,
blaine.
If
you
have
an
opportunity
to
review
some
prs
in
the
in,
we
have
a
couple
of
repo
a
couple
of
repos.
You
know
you
can
get
some
exposure
to
the
code
and
I
think
it
will
look
familiar
to
you.
B
Yeah,
so
a
point
of
where
you
can
get
started
is
I'll.
Send
out
this
link
on
the
slack
channel.
B
B
B
Yeah
yeah
so
like
jeff
was
suggesting
we
have
we,
we
have
things
that
we're
working
on,
but
if
you,
if
you
do
have
experience
with
writing
controllers,
there
is
something
that
is
slightly
lower
level.
The
implementation
of
the
controller
itself,
the
the
logic
behind
how
it
synchronizes
and
gets
new
events.
B
I
would
like
someone
to
take
a
look
at
it
and
and
go
over
it,
because
this
is
an
implementation
that
we
came
up
with
to
speed
up
development
and
just
keep
things
simple
and-
and
I
would
like
more
eyes
on
this
and
and
to
make
sure
that
this
this
works
exactly
as
intended
yeah.
This
is
this
is
one
of
the
things
that
that
I'll
need
someone
who
has
a
good
amount
of
experience,
developing
controllers
to
to
take
a
look
at.
D
Yeah,
I
I
think
I
can
definitely
take
a
look
at
that
and
okay.
That
yeah,
I
also
agree
like
I
think,
trying
to
integrate.
Rook
and
cozy
will
also
do
a
lot
to
to
like
kind
of
see
where
there
might
be
some
some
sticking
points
or
some
pain
points.
B
Reach
out
to
me
on
slack
anytime
and
when
you
start
integrating
or
if
you
want
to
start
trying
it
out
or
you
know,
if
you
want
to
look
into
I'll
I'll,
send
out
this
link
also
to
this
controller
and
also
send
out
what
needs
to
be
done.
B
Someone
needs
to
test
it
out
really
because
this
is
a
this
is
something
we
wrote
on
our
own.
Someone
needs
to
make
sure
that
that
there
are
no
deadlocks,
there
are
no
race
conditions
and
also
that
it
works
as
intended
so
yeah.
So
so
I'll
put
it
out
in
the
in
this
channel
as
you
as
you
take
start,
taking
a
look
at
it,
you
might
want
to
know
how
to
how
to
run
it
or
how
to
test
it.
B
In
those
cases
just
reach
out
to
me
directly
and
and
yeah
I'll
work
with
you
and
and
get
you
up
to
speed.
Okay,
that
sounds
great
all
right.
All
right.
D
Oh
you're,
on
on
the
slack
you
are
w.
B
B
Yeah,
the
first
wireless
interface
now
yeah
with
that
we're
actually
out
of
time.