►
From YouTube: Kubernetes SIG Service Catalog 20170213
Description
The SIG meets to discuss informer-based controllers and review the remaining work needed to complete MVP 1.
A
A
A
Right
elbow
all
right.
First
up
Aaron,
you
have
table
wait
on
us,
quick
announcement.
I
mentioned
a
few
times
recruiting.
You
know,
last
week's
meeting
that
working
on
a
third
party
resource-based
API
server,
that
pull
request
is
currently
complete.
It
is
a
pull
request.
Number
3,
38
and
link
for
it
is
coming
into
the
chat
right
now.
A
It
is
big.
It
is
all
necessary
to
go
in
at
once.
So
I'd
encourage
you
to
go
in
and
start
reviewing
it.
A
common
cetera
I
am
currently
testing
it.
Along
with
some
others,
I
believe
testing
it
in
a
real
cluster.
Seeing
what
breaks
what
works-
and
I
would
like
to
get
this
in
as
soon
as
possible
after
its
verified
working
in
a
real
cluster.
So
that's
it
for
me.
This.
B
D
B
D
B
A
A
Server
can
start
up,
I
can
create
all
the
third
party
resource
types
and
a
gat
will
work
given
a
third
party
resource,
that's
already
in
the
cluster,
but
it
kind
of
gotten
to
create
or
any
of
the
other
crowd
action.
So
you
can
get
a
tpr,
that's
already
placed
in
the
cluster
as
our
version
API
and
works.
I,
don't
know
what
you
mean
by
as
our
version
API
boy
I'll
tell
you.
A
A
A
Yes,
but
I'll
I'll
copy
out
that
by
saying,
if
watch
doesn't
work
in
this
PR,
that's
okay,
everything
that
this
PR
will
merge
is
optional
functionality
and
everything
will
still
default
to
SCD.
So
what
I
would
like
to
do
is
get
in
the
majority
of
operations
and
this
PR
and
things
that
don't
work
still
will
be
issues
in
github
and
then
six
to
follow
up.
I,
think
that
is
sensible.
A
B
Yeah,
okay,
Morgan
well,
I
was
using
spatially
I
just
wanted
to
encase.
Anybody
was
worried
about
that
368
PR
ago.
Yeah,
no
I
understand
that
we
missed
out
some
on
some
arguments
and
you
know
we
need
that
to
merge
so
that
we
can
get
the
controller
to
do
its
thing
and
bought
into
is
saying
with
controller
and
the
informers.
And
you
know
those
directories
look
good.
B
They
point
at
the
other
generated
clients
and
the
only
thing
I
can
think
of
is
maybe
we
need
to
make
sure
that
the
editing
gets
generated
in
order
and
we
go
ahead
and
do
that,
but
otherwise
look
good.
We
need
it
so
emerges
as
far
as
the
I've
got
the
basic
client
thing
working
today
for
the
beginning
of
a
test
and
I'm
trying
to
understand
how
to
pass
arguments
and
stuff
to
it
to
make
it
an
actual
like
go
test
or
something
rather
than
don't
want
your
script,
but
getting
there
I've
got
something
that
things.
B
B
A
It's
a
fundamental
building
block
of
controllers,
and
it's
also
a
thing
that
does
not
add
bug
surface
to
our
project
if
we
reuse
it
from
upstream
so
I
have
had
some
like
Walker
bugs
for
one
of
my
other
work
lied
and
some
just
like
soft
things
that
have
kept
me
from
verifying
that
it
worked
yet
I
got
a
building
over
the
weekend.
I.
A
So
we
can
start
supporting
in
the
business
logic
from
the
existing
CPR
based
controller
and
that's
pretty
much
what
I've
got
I'm
happy
to
answer
questions,
possibly
you
know
like
we
can
do
one
of
these
one-off
meetings
and
folks
want
to
go
over
this
thing,
I'm
happy
to
do
that.
Let
me
know,
and
next
up
just
as
piss.
C
A
D
A
Like
api
review
to
take
a
pass
at
it,
but
I
mean
I
think
that
the
concepts
are
pretty
well
baked
there
in
line
with
like
if
I
were
writing
a
proposal.
The
bar
equality
I'd
want
it
to
be
so.
I
would
like
someone
from
api
review
to
take
a
look
at
it,
but
I
think
it's,
I
think
it's
ready
to
get
implemented
since
it's
going
to
be,
especially
since
it's
gonna
be
out.
First,
where.
A
B
A
A
A
Well,
good,
ok,
so
I
believe
it
was
wednesday
at
last
week
we
spent
a
little
time
I
getting
the
gaps
that
need
to
be
close
to
be
at
MVP
one
on
to
this
google
document,
and
so
after
that
Doug
and
this
aaron-
and
I
spent
a
little
time
talking
about-
maybe
folks
that
you
think
would
be
good
to
work
on
these
things,
and
so
I
would
like
to
just
go
through
these
now.
A
A
We
need
integration
tests
for
the
API
server
and
sub
resources.
So
what
is
meant
by
this
is
that
we
need
like
a
goal
eng
test
which
morgan
is
working
on
right
now,
that
will
use
the
clients
to
create
resources,
the
leak
to
update
them,
to
exercise
them
and
then,
due
to
the
sub
resources,
to
update
the
status
field.
For
the
resources
that
have
them,
we
have
this
kind
of
test
in
communities
as
well
and
kind
of
like
a
bread-and-butter
type
a
test.
A
So
that's
what
that
is
place
will
placeholder
for
gas
for
the
CLI
experience.
There
is
so
the
gas
that
we're
talking
about
our
that
the
currently
like
cube
CPL,
the
q,
Pernetti
CLI
tool
have
to
know
via
code
that
has
compiled
into
it
about
resources
and
long
term,
especially
since
the
extension
Patterson
crew.
Benetti's
is
one
where
you
build
a
new
API
server
outside
the
core
I
one
term.
A
We
need
the
API
server
to
be
able
to
originate
information
about
how
to
do
like,
like
which
feels
to
show
when
you
do
a
get
which
feel
too
sure
when
you
do
a
described
I,
don't
think
that
we're
going
to
get
that
for
MVP,
one
I
think
that's
also.
Ok,
because
it's
this
is
really
more
like
a
polish
type
thing
than
a
you
know,
functional
gap
that
we
need
to
close.
So
one
until
I
know
you
guys
have
been
talking
about
this
I
know.
A
There's
a
proposal:
that's
open
that
I,
don't
think
we're
going
to
get
in
16,
but
maybe
we
could
get
in
17
and
Simon.
Your
name
was
put
out
that
someone
that
might
be
interested
in
working
on
that.
Well.
B
E
A
A
Parameter
supported
in
the
API
is
further
than
unaware.
Are
the
open,
serviceworker
API
lets
you
specify
like
a
raw
blob
of
JSON
or
yamel,
that
is
in
order
to
communicate
parameters.
Belay
has
taken
taking
this
one
Simon
you're
certainly
welcome
to
pitch
in
on
it.
I
think
PLA
actually
has
a
full
request
for
this.
That
needs
some
tests,
but-
and
certainly
we
need
test
coverage
from
an
end
perspective
on
this,
but
I
think
Eli's
got
this
one
under
way.
A
We
can
begin
courting
in
the
business
logic,
and
this
we
had
myself
and
Brendan
and
delay
on
I'm
thinking,
based
on
a
conversation
that
I
had
today
that
it
might
be
more
delay
and
we
might
need
to
reassess
who's
going
to
be
on
this
one.
So
I'll
I'll
ask
you
all
to
think
to
yourself
if
you're
familiar
with
the
business
logic,
this
might
be
something
that
you're
interested
in
working
on.
A
Then
there's
implementation
of
the
remaining
behaviors,
which
are
key
provision,
update
and
delete,
broker.
Aaron
I
know
that
you
have
started
working
on
some
of
this
stuff
in
the
context
of
the
CPR
prototype.
We
thought
Kenton
Aaron
would
be
good
resources
to
put
on
this
one,
not
that
you
guys
aren't
human
beings,
but
also
for
you
know
the
purposes
of
getting
stuff
done.
A
Resources
of
is
a
thing
that
is
used.
I
do
not
mean
to
any
in
any
way
to
impugn
your
humanity,
so
I
do
have
my
desk
about
10
tons,
he's
supposed
to
be
able
to
like
chop
a
tree
down
with
his
raw
or
his
bare
hand.
Something
like
that.
So
we
might
need
to
just
peek
and
see:
what's
going
on
inside,
can't
verify
his
humanity,
maybe
get
a
small
one
leader
sample
of
blood,
something
like
that.
I,
don't
know
what
you
mean
about
supposed
to
be
able
to
and
what
sense.
Oh
I.
A
Actually,
we
don't
mock
out
the
the
API
server
in
the
cabin
Nettie's
control
integration
tests.
We
run
an
embedded,
API
server
and
basically
will
pump
resources
into
the
API
server
and
make
sure
that
the
interactions
with
brokers,
open
service,
Rooker
API
go
according
to
plan
and
that
our
resources
wind
up
in
the
right
in
the
right
state.
A
So
we
had
Morgan
and
Jeff
on
this
I
think
that
it's
the
kind
of
work
that
is
easy
to
parallel
eyes,
and
so,
if
we
have
anybody
on
the
call
that's
interested
in
pitching
in
you
know,
don't
hesitate
to
say
something
in
flac.
So
next
one
eet,
EEZ,
there's
biggity
pattern
in
infrastructure
and
what
is
meant
by
that
is
like
one.
Do
we
have
an
eee,
a
system
that
integrates
with
github
and
two
does
the
pattern
of
how
you
write
those
bees
work
in
the
sense
of
like
can
I
get
a?
A
Can
I
get
a
cooper
Nettie's
cluster?
Can
I
get
a
service
worker
api
server
and
controller
integrated
or
running
in
that
cluster?
Can
I
do
the
things
at
a
high
level
like
add
a
broker
and
then
ensure
that
I
can
that
I
can
get
the
services
service
classes
back
from
it
provision
one
of
those
bind
to
it
things
manifest
correctly
in
turbonetics
core
all
that
kind
of
stuff.
So
that's
the
pattern
piece
and
then
the
next
one
is
that
we
need
integrator
I'm.
A
Sorry,
you
need
ete
tests
that
exercise
all
those
different
use,
cases
that
we
go
and
kibbles
and
myself
we're
on
number
nine
Morgan,
Jeff
and
myself
are
on
number
10
we
might
need
to.
A
We
might
need
to
parallel
eyes
or
otherwise
add
people
to
that
since
network
is,
is
fairly
easy
to
paralyze
and
then
number
11
is
the
PPR
storage
back
end
for
API
server.
Erin
is
turning
the
crank
on
that
as
we
speak,
and
the
lovely
tpr
sausage
is
coming
out
of
the
machine
with
a
crank.
So
that's
in
progress
right
now
and
then
see
I
and
CD
integrated
with
github
for
quest
process
with
publicly
available
test
results.
The
test
results
are
publicly
available,
kills
has
a
PR
to
pick
some
things.
Kibbles
are
you
on
come
on?
B
A
So
cables
attached
that
one
and
then
there's
another
place
holder
for
API
aggregation,
which
I
think
is
just
getting
to
the
point,
maybe
where
it's
going
to
be
ready
for
people
to
start
playing
with
it.
So
we
don't
have
anybody
on
that
one
yet,
but
honestly
it
would
be.
It
would
be
really
fantastic
to
be
at
a
point
where
we
felt
like
we
both
had
the
need
to
get
on
it
immediately
and
then
bandwidth
for
someone
to
start
looking
at.
So
that's,
basically
what
we
have
at
this
point.
Any
questions
about
any
of
this.
A
B
A
All
right,
well
so
I
think
that
exhausts
the
agenda
for
today.
If
there's
anything
folks
want
to
raise,
we
have
we
have
a
little
time.
Otherwise
we
can
all
get
back
to
turning
that
crank.