►
From YouTube: Kubernetes SIG Storage 20170525
Description
Kubernetes Storage Special-Interest-Group (SIG) Meeting - 25 May 2017
Meeting Notes/Agenda: https://docs.google.com/document/d/1-8KEG8AjAgKznS9NFm3qWqkGyCHmvU6HVl0sk5hwoAE/edit#heading=h.wb235u6mtrz
Find out more about the Storage SIG here: https://github.com/kubernetes/community/tree/master/sig-storage
Moderator: Saad Ali (Google)
Chat Log:
09:05:15 From SIG Storage : https://docs.google.com/document/d/1-8KEG8AjAgKznS9NFm3qWqkGyCHmvU6HVl0sk5hwoAE/edit#
09:08:04 From SIG Storage : https://github.com/container-storage-interface
09:08:18 From Luis Pabón : quick question, why was it moved?
09:08:23 From Luis Pabón : To move it to the CNCF?
09:11:31 From Luis Pabón : cool!
09:18:59 From Luis Pabón : i can check...
09:22:30 From Luis Pabón : https://github.com/kubernetes/community/pull/524
09:25:09 From Luis Pabón : TPR are v1beta1
09:25:59 From Luis Pabón : https://github.com/kubernetes/client-go/blob/master/pkg/apis/extensions/v1beta1/types.go#L104
09:27:00 From Luis Pabón : They are wondering if they should move it from extensions/v1beta1 to something tentaively called "apiextensions/v1beta1"
09:27:09 From Luis Pabón : in 1.7
09:28:12 From Luis Pabón : ....
09:28:32 From Luis Pabón : My recommendation.. create a TPR as alpha.. would be a great way to start
09:32:10 From Luis Pabón : ....
09:33:24 From Luis Pabón : My opinion, with TPRs there is very little need to add things to core. I use them in Quartermaster, and they are great. TPR does not mean that it cannot be in production, it just means that a container called something like "kube-storage-snap" handles does objects
09:35:46 From Ardalan Kangarlou : Should all vendors get behind the same 3rd party resource for snapshot API objects? Otherwise, each vendor may define a custom API object and custom controller, which results in a lot of complexities for customers using different backends.
09:37:14 From Luis Pabón : @ardalan, that depends. There is nothing today that stops vendors from providing that. It would have to be that the container handling the TPR is provided on the normal kubernetes distribution would be the accepted object.
09:38:16 From Luis Pabón : The recommended model would be the most used one. But there is nothing that would stop a Vendor from creating their own controller and attaching to Kube, then provide the customer with documentation on how to use it.
09:38:30 From hekumar : heads up - we are not going to have time left for discussing other items on agenda
09:39:48 From Jason Singer DuMars : +1 for TPR
09:39:57 From brad childs : +1 to TPR
09:41:04 From Luis Pabón : Look at Quartermaster as an example: TPR plus plugins
A
Right
this
is
the
bye
week.
Bi-Weekly
meeting
of
the
storage
special
interest
group
group
for
kubernetes
today
is
May,
25
2017.
As
a
reminder,
this
meeting
is
public
and
recorded
today
on
the
agenda.
We're
going
to
go
over
static
planning
for
the
cig,
we're
going
to
take
a
look
at
PRS
that
people
think
have
not
been
getting
attention.
A
If
you
have
any
PRS,
please
that
into
the
agenda
I'm
going
to
paste
the
agenda
duck
into
the
chat
window,
then
we're
going
to
go
over
some
design
reviews
and
finally,
we're
going
to
talk
about
onboarding,
Docs
and
process.
If
you
have
anything
else,
you
want
to
add
to
the
agenda.
Please
add
it
to
this
document,
just
paste
it
into
the
chat,
and
let
me
share
my
screen,
we'll
start
going
over
the
planning,
doc.
A
B
B
A
A
A
A
C
D
My
things
I've
been
having
trouble
with
zoom
on
a
new
laptop
Saad,
so
I'm
on
my
old
one
right
now
and
it's
doing
something
strange
anyway.
There's
not
a
big
update
to
report.
John
is
still
on
paternity
leave
and
I
am
working
on
two
issues
that
one
that
Jay
submitted
about
testing
the
storage
plugins
to
make
sure
they
work
with
larger
write
sizes
and
read
sites
and
that
they
create
site
files
of
the
correct
size
similar
to
the
mount
testing.
We
do
in
volumes,
go
where
we
run
through
the
dozen
plugins.
D
It's
going
to
be
similar
to
that.
But
it's
going
to
do
more
stressful
I/o
and
then
michael
has
a
a
similar
PR
about
streaming,
iOS
and
making
sure
pd's
work
correctly.
But
I
was
going
to
do.
I
was
going
to
expand
his
issue
a
little
bit
and
and
cover
all
the
plugins
and
make
sure
they're
working
in
streaming,
io
of
various
sizes,
like
I've
sort
of
architected,
the
way
of
doing
that
test
and
I'm
just
going
through
the
details
of
actually
getting
it
to
work.
Now,
often.
D
E
A
Thank
you
in
the
chat
Louis
asked
why
the
CSI
doc
would
be
moved
into
the
repo
it's
mainly
to
make
it
easier
to
collaborate.
Now
that
we've
kind
of
moved
out
of
the
rapid
iteration
phase,
the
idea
is
that
more
people
can
help
contribute
to
it
and
it's
good
to
have
it
in
a
place
where
you
have
revision,
history
and
everything
and
cool
next
up
lab
provider,
storage,
metrics
that
was
already
done,
snapshots
odd.
Ten
string
joined
us
or
Tomas
on
the
line
I'm
on.
F
A
F
Been
many
discussions
and
we
captured
them
in
my
text,
but
we
haven't
agreed
completely
on
the
API
itself,
turns
out
that
perhaps
we
might
try
to
do
something
slightly
different,
so
now
I'm
trying
to
work
on
something
like
Google
concept.
That's
it
like
I
would
write
something
that
actually
do
the
snapshots
and
we
can
discuss
on
something
like
MVP
and
we
can
discuss
about
it
and
from
from
this
specific
things
try
to
try
to
come
with
a
final
proposal.
So.
F
A
A
In
particular,
they're
thinking
about
things
like
you
have
a
stateful
set
that
gets
upgraded
from
one
version
to
the
next
snapshotting
is
probably
something
that
they
want
to
incorporate
into
that
workflow,
and
they
want
to
make
sure
that
the
design
that
this
group
has
come
up
with
aligns
with
how
they
would
use
it
from
from
their
end.
Basically,
the
end
user
perspective,
so
I
believe
talking
to
Jing
she's,
going
to
set
up
a
presentation
to
the
cig.
A
Apps
group
early
in
June
I
believe
the
first
week
of
June,
because
this
coming
week,
Monday
the
holidays
called
Monday,
is
the
first
next
meeting
of
the
gaps
group
and
so
she'll
present
something
to
them.
That
said,
we're
probably
not
going
to
get
an
alpha
for
1.7
and
that's
okay.
Depending
on
what
we
decide,
whether
this
is
going
to
be
a
third
party
resource
entry
out
of
tree,
it's
possible
that
we
could
work
on
this
complete
got
a
band
from
the
primary
kubernetes
released
so
keep
carrying
on
their.
So.
G
That
we're
necessarily
off
track
for
it
either.
You
know
there
was.
There
was
a
lot
of
discussion
that
the
face
to
face
about
you
know
one
object
versus
two
objects,
and
then
we
had
kind
of
a
little
bit
of
a
derail
on
if
we
should
go
with
that
sub
resource
or
not,
maybe
I
think
we've
all.
You
know
we
kind
of
regrouped
and
we're
back
to
like
the
two
object
design
that
we
discussed
pretty
extensively.
G
Yet
the
face
to
face
I
tomash
as
prototype
or
is
working
on
the
prototypes
for
the
API
objects
and
for
the
controller
nine
element
is
going.
You
know
well
and
by
you
know
all
means
we
should
have
at
least
a
proof-of-concept
here
in
the
next
couple
of
days.
It's
you
know
it.
It's
it's
one
of
those
things
that
every
time
we
have
a
call
like
there's
so
much
feedback
and
excitement
about
it
becomes
difficult
to
determine
where
we
really
are
in
the
development.
G
I
think
that
we
are
still
shooting
to
get
this
code
done
by
next
Wednesday
and
we
have
McCain
tracking
it.
We
have.
You
know
other
people
who
have
mentioned
that
they
would
like
to
help
participate
in
it
and
we
do
have
a
sheet
tracking
it
if
other
people
want
to
go
and
sign
up
for
like
testing
or
documentation
or
other
bits
of
it
I,
but
so.
A
I
think
the
only
problem
is
in
order
for
us
to
get
this
into
the
kubernetes
codebase.
We
need
to
get
sign-off
not
just
from
this
creep.
I.
Think
we've
now
arrived
at
consensus
within
the
big
storage
group
about
what
it
should
look
like,
but
it
turns
out,
you
know,
there's
there's
more
actors
within
our
space
right.
We
have
to
get
sign
off
on
it
where
the
API
should
live
and
whether
this
design
works
for
the
larger
prefer
for
first
two
gaps.
A
Basically,
so
we
basically
need
two
more
sign
offs
to
happen
before
this
code
can
get
in
and
the
first
sign
off
the
API
sign
off.
We
may
be
able
to
get
within
the
next
week,
but
the
second
one
from
say,
gaps
looks
like
we've
missed
the
boat,
because
the
next
meeting
is
going
to
be
after
code.
Freeze.
A
Api
changes
whether
the
API
changes,
whether
snapshots
are
going
to
live
as
part
of
the
core
kubernetes
api
or
whether
it's
going
to
be
third-party
resources.
So
this
is
kind
of
a
tension
within
api
trying
to
figure
out
how
much
is
enough
for
the
kubernetes
core
code
for
for
the
kernel,
basically
for
the
nucleus
and
what
what
we
want.
Every
new
object
that
we
add
has
a
higher
bar
for
why
it
should
be
in
the
core
and
so
I
think
for
for
snapshots.
A
A
E
E
E
A
Okay,
we
can.
We
can
check
on
that.
The
other
part
that
falls
out
of
this
is
like.
Okay,
even
if
you
have
the
API
out
of
trees,
is
the
third-party
resource.
The
second
question
is
well:
where
does
the
logic
that
drives
this
list?
If
it's
going
to
be
part
of
the
core
kubernetes
PV
PVC
controller,
then
it
needs
to
be
checked
in
an
entry.
G
G
C
A
If
we
can
do
that,
I
mean
that
would
be
wonderful.
It
would
give
us
two
things.
One
is
the
ability
to
go
to
alpha
quicker
and
two.
We
don't
have
to
be
tied
to
the
kubernetes
lifecycle
and
then
in
the
future.
It
kind
of
leaves
the
opening
for
us.
If
we
decide
that
we
want
to
bring
it
back
in
tree.
We
want
it
to
be
a
first
party
resource
and
can
do
so
in
future.
Yeah.
G
I
A
I
mean
so
I
think
it'll
be
part
of
the
messaging
about
what
we
want
to
call
it.
If
we
call
it
alpha
or
a
prototype,
we
have
the
technically.
We
have
the
ability
to
deprecate
it
and
move
it
around,
but
in
reality
you're
absolutely
right.
If
we
actually
have
customers
that
start
depending
on
the
behavior,
we
kind
of
get
stuck
with
having
to
maintain
backwards
compatibility.
So
it's
something
we
should
think
they're
a
good
point.
I
think.
G
A
Yeah
we
just
have
Michael
join
in
I'm,
going
to
do
a
quick
recap
and
we're
talking
about
snapshots.
The
discussion
around
snapshots
is
whether
we
make
it
for
1.7
or
not.
The
big,
open
questions
right
now
are
one
around
the
API,
whether
they
API
should
be
part
of
the
core
communities
core
or
it
should
be
a
third
party
resource
and
then
falling
from.
That
is
whether
the
controller
itself
should
be
part
of
the
core
community
code
or
be
outside
of
the
communities
core.
A
And
then
the
third
question
is
around
say:
gaps,
approval
and
so
I
think
folks
would
still
like
to
get
this
in
41.7.
But
we
were
one
week
away
from
countries
the
first
question
around
whether
it
should
be
in
tree
out
of
tree.
If
we
go
out
of
tree
it
kind
of
decouples
us
from
the
kubernetes
release,
you
can
have
the
controller
out
of
tree.
A
You
can
have
the
API
as
a
third
party
resource
and
theoretically
around
the
time
that
1.7
comes
out,
you
could
have
an
out
of
tree,
something
that
you
can
take
the
tires
on
the
something
that
y'all
brought
up
is
well.
If
you
do
that,
you,
you
know,
people
are
going
to
start
using
it
that
way
and
if,
in
the
future,
you
want
to
bring
it
into
the
kernel
into
the
core
kubernetes
codebase
you're,
going
to
have
this
backward
compatibility
issue.
But
Brad
is.
A
E
E
We
talked
about
the
API
being
so
important
here
because
we
have
to
be
able
to
you
know,
recreate
a
snapshot
or
possibly
move
it
to
a
different
namespace,
which
is
something
that
doesn't
happen
today
in
kubernetes
and
so
doing
that
outside,
of
course,
seems
I.
Don't
know
that
just
seems
like
something
we'd
fundamentally
want
to
have
within
coop
there.
Anyone.
G
Have
said
they
can
intubate
it
before
moving
into
something
that
we
actually
promised.
The
support
right
I
mean
like
there's
a
huge
benefits,
in
my
mind,
said
especially
on
something
that's
been
so
talked
about
and
debated
in
somewhat
controversial
to
incubating
it
outside
of
the
core
API
I,
and
that
we
can
see
what
works
and
what
doesn't
and
if
it
is
broken
or
if
there
is
issues
with
it,
then
it's
not
part
of
our
core.
We
can
deal
with
it
and
then
put
in
decor
after
it's
pretty
well
bed
in
the.
A
Other
thing
to
keep
in
mind
is
that
the
component
that
you're
talking
about
Aaron
about
transferring
ownership
between
main
spaces,
that
component
I
don't
believe,
is
planned
to
be
delivered
in
the
alpha
version
and
when
it
is
going
to
be
implemented,
you're
right,
it
should
not
be
a
third
party.
It
should
not
be
an
external
controller
it'll,
be
something
that
has
to
be
something
that
kubernetes
controls
all
right.
A
K
K
For,
if
reasons
one
is
that
the
one
of
the
main
users-
and
you
know,
consumers
of
snapshot
should
probably
be
asking
and
they
have
not
signed
out
on
it
yet
and
to
you
know
it
has
gone
through
a
lot
of
conversation
and
back
and
forth
and
I.
Don't
know
anyone
yet
feel
certain
that
we've
locked
down
the
API
and
I
think
the
only
thing
the
world's
certain
about
is
that
snapchats
are
cool
and
that
we
want
to
make
some
progress.
Is
there
any
other
large
reason?
K
K
The
designation
between
core
and
outside
of
core
is
there
a
strict
definition.
I
just
looked
up
the
definition
of
alpha
and
when
you
know
it's
alpha,
if
you
look
at
it
has
five
points,
the
version
name
is
called
it's
labeled
alpha,
it
could
be
buggy.
The
support
of
the
feature
could
be
dropped
at
any
time
without
notice
and
the
API
can
change
in
very
big
backwards.
Braking
compatible
way
incompatible
ways,
and
it's
only
really
meant
to
be
used
for
short-lived
clusters
and
not
corruption.
K
So
to
me
you
know
if
we
change
that
API
heavily,
you
know
if
anyone's
using
snapshots
can
1.7
and
starts
relying
on
those
snapshots
to
you
know,
or
this
API
to
be
around
that.
It's
kind
of
you
know,
shame
on
them.
That's
really
not
the
state
that
it's
in
that's
what
we
have
at
alpha,
so
I'm
not
worried
about
that.
Much.
K
I
think
also
that
you
know
it's
something
to
think
about
for
later,
but
as
it's
not
an
out,
it's
someone
in
alpha
feature
I
feel
like
that's
also.
You
know
not
something
that
could
be
a
main
focus
right
now.
It
feels
like
right
now
making
sure
that
the
consumers
of
the
apps
team
and
that
the
API
is
in
solid
shape
should
really
be.
K
G
I
think
the
argument
is
just
that.
That's
the
way
we've
done
stuff
in
the
past,
and
so
that's
what
people
are
most
familiar
with,
and
even
though
it's
kind
of
weird
that
this
is
an
experimental
you
know
like
if
the
process
to
experiment
with
stuff
is
somewhat
experimental
is
what's
causing
a
little
bit.
B
B
The
other
thing
is
that,
if
you're
talking
with
like
like
moving
objects
between
namespaces,
then
that
is
something
if
you
are
going
to.
Let
me
we
can
push
it
for
next
delayed
or
something,
but
if
they
are
going
to
do
it
from
get-go,
then
that
has
to
be
subjected
an
encoded
gay
as
well
to
support
that
kind
of
cooperation.
So,
but
we
can
definitely
go
without
that
feature.
Moving
snapshots,
eliminating
pictures
when
there's
not
a
block
of
god,
so.
I
K
If
people
start
depending
on
it,
then
putting
it
in
court
and
we
don't
want
them
to
put
it
in
court.
Maybe
another
signal
to
do
so
and
putting
it
outside
of
court
is
another
way
of
saying
hey.
You
know
we're
making
progress
on
this,
but
please
don't
depend
on
us
yet
I
don't
feel
yet
I
mean.
Does
anyone
feel
really
confident
yet
that
we've
nailed
down
the
API
and
that
it
won't
change?
No.
I
K
I,
don't
know
okay
good
so
then,
and
by
the
way
this
is
I
think
this
is
going
to
be
a
very
pivotal
feature
going
forward.
So
I'm
glad
that
we
are,
you
know,
I'm
very
glad
with
the
approach
that
we've
taken
to
it
and
I
guess
the
more
we
talk
about
it.
The
more
my
vote
is
maybe
to
keep
it
out
of
court.
For
those
two
reasons,
one
I
don't
see
a
reason
to
put
it
in
and
two
it
would
provide
another
signal
to
customers
and
other
people
in
the
world.
A
There's
I
need
to
be
figured
out
right.
One
is
whether
snapshots
meets
the
bar
to
be
something
that
is
part
of
the
core
kubernetes
kernel.
The
bar
is
going
to
be
very
high.
We
need
to
provide
justification
that
this
is
broadly
useful
and
that
you
basically
cannot
run
a
kubernetes
cluster
without
it
I
think
there's
hundreds
of
objects
that
have
been
introduced
since
the
inception
of
kubernetes
and
there's
a
big
push
back
now
to
say
very
careful
about
what
we
adopt
into
the
core
and
anything
that
doesn't
need
to
be
part
of
the
core.
N
Monica
I
think
snapshots
is
critical
for
equivalently
save
option,
especially
in
production
environments,
we're
creating
a
lot
of
requirements
and
customers,
so
I'm
pretty
sure
we
can
work
out
a
justification
there
in
fact,
I
think
having
out
of
core
having
it
available
to
hand
it
to
customers.
So
all
of
the
snap
chat
can
bed
mister
from
kurtal
in
one
interface,
I
think
it's
going
to
go
a
long
way
in
driving
at
the
adoption.
Just
even
go
in
production
definitely
need
this
one.
So.
A
I'm
not
sure
whether
user
facing
it's
going
to
make
that
much
of
a
difference
between
being
a
third-party
resource
and
part
of
the
core.
As
long
as
the
third-party
resources
installed
on
the
cluster,
I
believe
and
I
might
not
be
correct
here
that
you
could
interact
with
it
the
same
way
that
you
interact
with
first
30
objects,
so
you
should
be
able
to
cube
Codel,
create
snapshots,
and
things
like
that
with.
N
M
O
To
bring
up
I've
used
a
third-party
resources,
a
lot
quartermaster
and
we
use
them
a
lot
in
coral
s,
so
we
depend
on
them
for
production,
so
it
is
absolutely
key,
absolutely
capable
of
creating
applications
for
production
and
it.
Actually,
it
makes
things
much
simpler
because
you
can
write
applications
that
are
outside
the
kernel
and
they're
still
being
part
of
our
overall
picture
to
come.
C
O
M
As
well
as
a
third
party
API,
the
objects
themselves
are
stable.
Then
I
do
I
really
have
objections.
You
know
we
are
running
colleges
to
this
platform
outside,
but
in
that
shim
is
not
so
stable
and
then
the
second
question,
because
eventually
160
core
and
analog
have
some
productions
run
in
the
third
party
is
that's.
O
M
O
K
A
A
Plugins
and
just
add
a
snapshotting
API,
our
interface
to
that
the
existing
plugin.
But
if
we
also
decide
to
make
the
controller
out
of
tree,
then
we're
going
to
have
to
go
with
something
like
the
dynamic
provisioner,
at
least
in
the
short
term
project
figure
out
what
that
means
for
CSI
and
how
discovery
needs
to
happen,
and
things
like
that.
It
makes
things
more
complicated,
yeah.
Q
Q
A
A
A
R
K
I
think
these
are
good
conversations
to
have
offline,
but
I
want
to
make
sure
that
we're
driving
towards
this
decision
is
there
any
reason
to
not.
Have
this
be
a
third
party
resource
I
can't
I
feel
like
what
we've
got
now
is.
We
may
have
made
a
decision,
and
now
it's
like
okay,
now
that
we
made
the
decision.
How
do
we
then
start
accelerating
past
that
but
there's
unfortunately,
some
other
1.7
items
for
this
meeting
that
we
know
we
need
to
cover
for
other
items
too.
K
K
A
G
A
That's
great
okay
and
the
only
other
big
open
issue
around
snapshots
is
presenting
to
the
big
ati
group
and
I'm
going
to
run
this
by
you
Michael,
and
would
it
be
okay
for
them
to
start
work
on
this
external
provisioner
and
then
Jing
has
a
meeting
first
week
of
June
to
present
to
say
gaps.
If
there's
any
major
changes,
they
can
incorporate
it
into
their
prototype.
No.
K
A
A
C
C
A
F
F
Testing
of
the
PR
we
found
the
pair
is
a
condition
where
certain
sequence
of
action.
It
may
happen
that
the
volume
that
is
attached-
the
note
it's
not
in
notes
APIs
school,
okay,
H
a
volume-
is
never
detached,
but
it
requires
the
controller
to
crash,
while
attaching
the
volume
and
the
user
a
delete
support
in
the
meantime.
So
it's
quite
rare
condition.
We've
discussed
this
one
with
testing
I,
don't
know
if
any
user
has
ever
encounter
it.
So
it's
not
everything
but.
A
Primary
issue
that
this
fix
was
intended
to
fix
that
has
largely
been
addressed
or
not.
That's
been
fixed
in
much.
That's
right,
okay!
So,
let's
what
will
move
this
item
over
to
1.8
when
you
do
the
1.8
planning-
and
we
can
finish
the
second
half
of
this
in
1.8
I'm?
Okay,
alright,
thanks
a
lot
yeah
mixed
up,
running
stateful
applications
to
a
federation,
plane
I'm
just
going
to
remove
this
item
it!
Well,
it's
not
started
we're
going
to
leave
it
here,
we'll
move
it
over
to
1.8.
A
So
this
is
a
feature
that
says
we
are
already
have
dynamic:
provisioning,
storage
classes,
storage
classes
themselves
are
constructed
by
closer
administrators,
but
potentially
there
are
some
subset
of
options
that
we
would
like
users
to
be
able
to
control
or
across
her
administrator
would
like
to
expose
to
the
user.
Yon
had
a
proposal
around
this
early
in
the
milestone,
I'm,
not
sure
we
made
too
much
progress
on
it.
A
I
I
Month
of
the
propagation
itself
I'm
granted
likely,
they
are
afraid
it
breaks
things,
so
they
won't
take
very
experimental
and
they
wanted
basically
running
only
in
2d
test,
so
we
can
test
it
before
they
receive
customers.
Okay,
I'm
changes,
new
command
line
options
to
be
black
and
I
posted
at
the
art.
Just
before
this
meeting
I
hope
it
gets
into
1.7,
but
I
need
an
API
change,
so
an
API.
A
Alright,
thanks
next
up
is
improving
the
current
chroot
hack,
which
is
like
an
intermediate
solution
between
what
yon
is
proposing
and
what
we
have
today.
I
believe
Jing
had
some
thoughts
around
this
I'm,
not
sure
how
much
progress
she
was
able
to
make
she's
not
on
the
call
right
now.
So
we'll
just
leave
that,
as
is
for
now,
next
up
is
fixing
the
broken
volume.
Reconstruction,
chakri
I
know
you
ping
me
a
few
times
this
week
and
I
haven't
had
a
chance
to
get
back
to
you
I.
Do
you
have
any
update
yeah.
R
R
A
H
Yeah,
it's
a
it's
still
ready
for
review.
Ramen
had
a
first
pass
on
it,
I
just
rebased
it.
So
it's
just
going
to
the
pipeline.
There,
ok
cool
any.
We
are
and
blocking
issues
I,
not
them
aware.
I
mean
it
would
be
great
it.
You
know
if
there's
anything
major
or
anything
like
that,
I'm
due
to
be
away
from
next
Tuesday,
so
it
would
be
great
to
have
a
bit
of
time
to
fix
them.
M
T
So
I've
been
negotiating
the
best
of
forests
of
Microsoft
and
we
have
people
working
on
this
now.
So
it's
actually
moving
forward
and
Brendan
Burns
is
helping.
Me
coordinate
people
and
I'm
meeting
with
the
team
in
China
tonight
to
talk
about
this
specifically
and
work
moving
forward,
so
we're
still
trying
to
shave
for
a17
inclusion.
M
Data
on
this
is
Kwame
again:
I've
got
a
PM
playing
out,
but
I
would
have
some
follows
with
Nokia
and
because
there
is
was
really
created
again
with
the
identical
code.
All
reviews
should
get
lost
right,
so
you
can
have
a
sync
up
meeting
between
you
and
developer.
I
forgot
his
name,
and
so
we
can
talk
face
to
face.
That's
what
we
want
to
make
it
happen.
I
made
some
general
comments
on
PR
and
hopefully
that
could
be
addressed.
M
One
seven
looks
like
a
stretch
because
the
cycle
API
is
way
beyond
is
over
here
for
some
more
features
at
that.
So
we
can
get
this
addressed
in
quick
turnaround.
Maybe
you
can
still
make
one
seven
okay,
how
do
I
am
how
do
I
connect
with
you
receive?
As
you
can
maybe
find
us
comments,
or
you
can
find
me
on
the
github.
We
can
comment
Lee
on
the
PR
Hogan
on
the
PR
and
then
coming
out
at
a
time.
We
can
just
stop
in
there.
Okay,
great!
Thank
you.
K
Okay,
hey
everybody
I
just
found
out
last
night
from
the
release
manager
dawn
Chen
that
there's
this
something
called
the
internal
1.7
crew
learned
Eddie!
Oh
sorry,
oh
wait!
Wait
this.
We
have
everything
in
the
features.
Repo
I,
think
the
question
that
she
wanted
to
know
all
those
items
that
we
just
had
cracked.
So
we.
A
Should
so
we
have
a
column
over
here
for
whether
the
issue
is
being
tracked
in
the
future
repo
on
the
only
one
that
should
be
tracked,
that
isn't
in
there
is
then,
if
that's
wedging
issue
and
looks
like
we're
not
going
to
have
that,
make
one
point
seven.
So
everything
else
looks
like
it
has
a
feature
and
is
everything
also
tied
into
the
docks?
A
H
K
And
if
they've
been
tracked
in
the
features,
repo
I,
don't
I
think
the
deadline
for
the
features
repo
is
passed,
which
is
why,
if
you
don't
have
something
in
the
features
repo,
it
may
require
an
exception
or
something
that
you
really
need
to
bring
to
the
police
team
and
then,
in
addition,
there
is
the
documentation.
I
think
the
documentation
was
the
real
thing
that
I
think
Don
wanted
to
see
it.
A
kick
start
on.
K
S
Because
I
added
this
one
again,
I
was
hoping
to
talk
about
it
later,
but
I'm
I'm
planning
to
add
a
whoo
volume
plug-in
and
this
story
pretty
much
done.
Testing
it
and
is
I
was
hoping
to
make
it
for
one
place
it
because
I
think
it
the
clogging,
is
a
very
same
plugin.
So
all
it
does
is
that
are
all
the
logic
about
provisioning
deleting
attack
and
attach
it's
gonna
be
handled
by
the
where
by
the
group
API
itself.
S
A
Primarily,
what
Michael
was
saying,
which
is
normally
we
do
future
planning
at
the
beginning
of
the
milestone
ruling
seven.
That
was
a
couple
of
months
ago.
The
idea
is
that
we
create
a
features,
repo
issue
about
what
are
the
features
that
are
going
to
go
into
this
release
and
the
release
team
starts
tracking
them
on
what
all
the
requirements
are
for
that
feature
and
make
sure
that
we
have
the
documentation
and
everything
that
needs
to
go
into
the
release.
A
Notes
for
that
feature
and
for
us
to
get
this
feature
in
now,
one
week
before
code
freeze,
we
would
have
to
make
a
pretty
big
justification
to
the
release
team
to
say
why
this
is
something
that
we
want
to
make
an
exception
for
and
I'd
like
us
not
to
go
through
that
process,
unless
we
really
really
have
to
if
you're,
okay
with
holding
off
21.8,
that
would
probably
be
the
best
approach.
Well
actually,
yet
this
is
on.
K
This
is
sort
of
your
choice,
not
ours
at
all.
This
is
just
sort
of
how
the
process
works.
I
mean
the
way
we
just
described
it.
It's
kind
of
like
this
is
what
you
know
where
we
use
sigfig
it's
kind
of
bigger
than
state.
So
it's
something
to
navigate,
and
if
you
need
more
help
about
it,
I
would
spend
mail
to
sig
release
and,
and
they
will
tell
you
the
steps
and
show
you
what
to
do.
K
Okay,
the
big
question
I've
got
is
everything
that
targeted
from
1.7
is
in
the
features,
release
repo
right
and
if
it's,
if
we
don't
think
it's
going
to
make
it
in
1.7,
and
we
should
probably
update
that
too,
because
I
think
this
is
what
Donna
is
looking
at
now
as
she
is
combing
through,
like
the
you
know,
with
a
spotlight
on
what
you
know
needs
to
be
done.
Okay,.
S
U
S
S
L
Okay,
it's
better
now
yeah!
Yes,
sorry!
We
should
definitely
should
we're
going
to
submit
the
PR
an
issue
now.
Well,
let's
take
the
exception
process
offline,
if
needed
at
all,
so
I
think
I'm,
sorry
we're
late.
We
are
kind
of
starting
to
get
used
to
the
process.
This
is
our
first
major
con.
You
know
changing
the
kubernetes
process,
but
well
we'll
see
if
we
can
make
them
exception
are
not
bubble.
That's
do
the
offline
boom.
A
C
A
A
A
A
Alright
design
reviews
I'm
not
sure
we're
going
to
have
time
to
do
any
of
these.
We
have
two
minutes
left
in
this
meeting.
We
can
punt
these
to
the
next
meeting,
we're
going
to
punt
the
onboarding
Docs
to
the
next
meeting,
and
that's
it
is
there
anything
anyone
wants
to
talk
about
in
the
last
minute.