►
From YouTube: Kubernetes SIG Service Catalog 20170227
Description
The SIG meets to view demos of:
- Add broker, provision, bind, unbind, deprovision, delete broker
- Sending parameters on provision and bind
We also discuss remaining gaps to close before Kubecon (new MVP-2 milestone) and some tactical stuff for the next week.
A
So
today
we
have
a
pretty
full
agenda:
I'm
going
to
start
off
by
just
given
the
status
of
MVP
one,
so
we're
worth
fairly
close
to
meeting
MVP
one.
The
gaps
are
that
Michael
Cuba
is
run
into
some
problems
with
getting
his
Jenkins
instance
to
play
nice
with
github,
and
we
lack
some
of
the
tests
that
we
wanted.
A
So
jeff
has
had
a
full
request:
merge
that
adds
the
unit
tests
for
the
informer
based
controller
and
that's
good
as
a
starch,
but
we
need
to
complete
the
complete
the
unit
tests
and
get
the
the
integration
tests
for
the
controller
and
you
use
going
Morgan's
on
an
issue
with
the
e
de
framework
from
upstream.
A
That
makes
it
so
that
we
will
not
be
able
to
to
pull
that
code
in
until
after
we
update
Cooper
Nettie's,
which
we
can
talk
about
later
since
I
see,
there's
a
general
item
for
this,
but
the
functionally
we
are
at
you
can
add
a
broker.
It
gets
the
catalog.
The
controller
pics
that
event
saw
gets
the
catalog
from
the
broker
but
service
glasses
into
the
API
server.
You
can
make
an
instance
and
control
a
pic
set
up
and
does
provision
you
can
make
a
binding
controller
pick
set
up.
A
Does
it
find
Kent
has
a
a
pull
request
in
progress
that
adds
d,
provisioned
and
unbind
to
that,
so
we're
very
close
functionally
and
then
aaron
has
done
a
large
chunk
of
the
work
to
make
the
API
server
work
with
third-party
resource
as
a
back
end,
which
is
pretty
awesome
and
also
a
thing
no
one
to
my
knowledge
has
done
or
ever
so.
That
is
also
pretty
cool.
So
any
questions
on
that
before
I
do
a
demo.
A
A
So
I
am
wanting
a
local
crew,
benetti's
cluster
and
I.
Have
the
excuse
me:
I,
have
our
API
server
and
controller
manager
deployed
as
pods
in
that
cluster,
the
API
server
pod
has
the
API
server
and
STD
in
two
different
containers
and
I've
also
got
this
UPS
broker.
Ups
is
user,
provided
service
and
I
made
a
couple
little
changes
to
that
to
get
around
the
fact
that
when
I
was
getting
this
stuff
to
work,
we
didn't
have
parameters
integrated.
A
So
the
change
that
I
make
to
this
is
that
UPS
worker
used
to
expect
you
to
give
it
a
parameter
on
provision
for
the
credentials
you
wanted
to
get
back
on,
bind
and
now,
if
you
don't
post
any
it
just
makes
some.
So
what?
Let's
take
a
look?
You
can
see
here
that
I
don't
have
any
brokers,
nothing
up
my
sleeve,
by
the
way.
A
What
I'm
doing
for
this
now
is
that
in
my
service,
catalog
directory
I
have
a
I,
have
a
cute
config
that
points
to
our
API
so
running
in
my
local
cluster
and
I'm,
switching
between
Cooper
Nettie's
in
this
pain
here
and
service
catalog
in
this
pain
here.
A
So
when
I
use
running
cube
CTL
with
my
k,
alias
it's
against
crew
benetti's
and
when
I
run
it
here,
it's
against
a
Service
Catalog
API
server,
so
I
am
going
to
create
a
broker
and
if
we
take
a
look
at
this
you'll
feed
at
it
points
to
the
cluster
local
dns
name
of
the
service
set
front.
This
UPS
broker
pod.
A
So
if
I
go
and
I
get
that
broker,
I
should
be
able
to
see
all
right,
perfect,
so
check
this
out.
It's
got
a
ready
condition
on
it.
That
says
true:
it's
ready!
It's
exceptionally
successfully
fetched
the
catalog
from
that
broker.
So
if
I
get
the
service
classes,
we
should
be
able
to
see
bingo
there.
We
go
user
provided
services
in
there.
So
let's
take
a
look
at
that
thing
and
we
can
see
it's
got.
A
So,
let's
take
a
look
at
that
real
quick,
so
this
instance
is
created
in
namespace
test.
Ms,
its
name
is
UPS
instance
and
it
points
to
service
class
user
provided
service,
but
plan
name
default
and
if
I
get
this
instance,
the
controller's
already
picked
up
this
event
for
the
instance
having
been
added
and
talk
to
the
rope
to
the
broker
and
said
go
ahead
and
provision
a
new
copy
of
this
nuisance.
So
if
I
get
oh,
never
mind
so
you
can
see,
we've
got
a
ready
condition
on
here
with
the
true
status.
That's
good!
A
A
Bam
we
can
see,
we've
got
a
secret
called
my
secret
in
namespace
test
enough,
and
this
just
adds
like
the
default
credentials.
That
UPS
berger
will
give
you
back.
If
you
don't
give
it
anything
which
are
just
like
special
key
one
in
specialty
too.
So,
let's
see
alright,
so
we
can
see
special
key
one
special
key
to
and
the
keys
are
just
special
value,
one
special
value
too.
C
A
The
controller
right
now
expects
to
use
an
in
cluster
can
take
it,
which
means
that
it's
going
to
look
for
the
service
count.
So
can
I
you
could
you
could
make
it
run
outside
a
cluster
by
putting
something
in
there?
But
it's
really.
The
controller
expects
to
run
in
a
cluster
inside
okay.
A
So
hopefully,
next
week
we
can
do
a
demo
and
showed
p
provision
or
unbind
and
d
provision
working
I.
Think
Kent
is
very
close
to
that
and
some
C's
at
a
an
appointment
I
now,
if
you
can't
attest
for
that,
but
I
think
he's
on
the
cusp
of
having
that
prb
murderball
I
think
it
just
needs
to
mute
so
I,
that's
what
I
have
got
for
that
I'll
just
take
a
chance
to
say
now
like
what
we
actually
now,
I'm
going
to
I'll
leave
this
for
later,
because
I've
got
another
agenda
item
on
it.
C
So
this
is
very
quick.
Doesn't
let
you
guys
know
that
we
did
talk
about
this
briefly
through
some
other
PRS
issues
and
stuff,
but
IBM
does
have
a
requirement
to
build
communities
on
power
system
and
NZ
systems.
We
actually
do
build
credit
today,
obviously
sure
was
kind
of
law
going
to
be
a
key
part
of
that
as
well.
So
nothing
needs
and,
as
I
chained
other
right
now,
I
think
we're
okay,
but
just
keep
in
mind
that
you
may
see
few
hours
in
the
future.
C
B
Ball
but
I'm
gonna
go
and
share
my
screen
here
already.
Can
everybody
see
this?
Can
anybody
not
see
this?
You
can
make
a
little
bigger
for
a
full
player.
Absolutely
all
right!
So
I've
got
two
terminal
windows
here
similar
to
what
Paul
did
this
journal
window?
Is
the
regular
old
cute
CTL
configured
to
talk
to
the
core,
and
this
one
I'll
make
bigger?
This
is
Q,
can
say
configured
to
talk
to
our
API
server.
B
So
if
we
first
go
and
look
at
the
pod
I'm
going
to
show
that
this
pot
is
not
using
a
CD
just
running
the
API
server
only
and
I
have
started
the
this
pot.
Obviously,
and
it
created
a
broker's
different
plenty
resource
now
generally,
when
this
potty
runs
it'll
create
all
four
of
our
resources,
but
to
make
things
simpler,
if
I'm
just
showing
broker,
I
have
deleted
all
the
other
third-party
resources
to
avoid
a
little
bit
of
confusion
here,
so
I'm
going
to
switch
over
to
the
other
API
window.
B
Here
and
again,
this
is
the
one
configure
to
talk
to
our
service.
Catalog
API
I'll
show
you
what
I'm
about
to
create
pretty
simple.
This
is
the
broker
that
should
take
it
from
our
contribute,
older
and
now
I
have
a
cube
CTL
in
this
directory,
which
is
the
160
alpha.
This
is
necessary
right
now
to
negotiate
version
api
version,
so
after
1.6
release
we're
not
going
to
have
to
use
the
Alpha,
it
will
just
be
the
up-to-date
cube
CTL.
B
Dude
a
chef
so
we've
created
the
broker
now
now
I'm
going
to
get
the
broker
here
and
I'll
print
it
out
as
yellow
boy
talking
and
typing,
not
the
best
for
me.
So
we've
created
a
broker
on
our
API
server,
we're
getting
back
the
ammo
here.
It
just
adds
all
the
regular
metadata
that
we
expect
any
API
server
to
create
now
to
prove
that
it's
a
third-party
resource,
I'll
go
back
over
to
the
core.
Api
servers
cute
Cpl
I'm,
going
to
get
the
broker
here.
B
Printed
out
as
yeah
mo
this
is,
we
ought
to
do
and
now
you
can
see
that
the
core
API
server
has
a
third-party
resource.
That
looks
exactly
like
what
our
API
server
returns.
So
that's
the
sea
in
cry.
We've
created
one
now
and
as
I
showed
already,
we
can
get
the
list,
and
now
we
can
get
the
broker
itself
and
printed
out
as
yeah
mo
is
already
showed.
B
B
So
we
deleted
our
test
broker
here,
just
to
prove
that
it's
been
deleted,
we
can
get
the
broker
again.
This
is
the
operation
on
our
API
server
and
then
to
prove
that
it's
been
bleated
I'll
go
back
over
to
the
core
API
server
here,
I
will
try
to
get
it
and
no
broker
exist
anymore.
So
this
is
as
far
as
the
third
party
resource
support
is
gotten.
B
There
is
a
pull
request.
Number,
let's
see,
number
4
10,
which
is
going
to
implement
the
last
piece
of
tpr,
and
that
is
for
watch
and
watch
list.
The
two
functionalities
that
are
the
final
two
functionalities
that
are
required
to
get
the
controller
to
work
with
this.
So
currently,
if
you
boot
up
an
API
server
in
tpr
mode,
it
will
panic
on
an
ill
pointer
exception,
because
this
keep
this
pull
request
for
TP
are
beige
based,
launch
and
watch
list
is
not
in
yet
so
once
this
in
progress,
PR
is
finished
and
merged.
B
C
A
A
Ok,
kinfolks
odd
see
my
screen
yep,
ok,
so
the
status
of
resources,
for
instance
in
binding,
are
done.
Integration
tests
for
the
API
server
and
some
resources
are
done.
Placeholder
gasps
for
CLI
experience,
I,
think
I
think
we
can
count.
This
one
is
done
because
eat
your
fries
or
16
is
today
and
that's
all
right,
because
we
still
have
an
okay
CLI
experience,
even
if
we
don't
have
described
and
lift
showing
certain
fields.
I
think
it's
fine
parameter
support
in
the
API
VA
has
this
in
progress.
A
I've
transplanted
much
of
the
old
behavior
or
the
the
old
behavior
for
AD,
a
broker
provision
and
mine.
I
Kent
has
unbind
and
deeper
vision
and
I
have
a
cannoneer
and
next
to
update
brokerage
elite
broker,
actually
can't
already
got
delete
broker
in
process.
So
I
cant
appears
to
have
all
of
this
stuff.
A
Integration
test
for
the
controller
I
had
Morgan
Morgan's
name
next
to
this,
and
Jeff
and
Jeff
has
also
been
working
on
a
unit
test.
So
we
can
I
will
go
back
through
and
split
out
unit
test
into
their
own
bullet
later
on,
but
those
appear
to
have
coverage
so
II
de
pattern
and
infrastructure
is
kibbles
in
myself,
and
kibbles
has
run
into
some
trouble,
getting
his
Jenkins
instance
to
work
correctly
with
github,
so
that
might
not
make
it.
A
A
It
looks
like
that
bed
risk
and
then
placeholder
for
API
aggregation
is
still
TBD,
I
hope
to
get
to
that
this
week,
I
don't
think
I'm
going
to
get
that
and
the
neck
a
couple
of
days,
so
that
one
is
at
risk,
but
here's
the
thing
is
that
we've
come
very
close
and
I
think
that
will
probably
be
at
a
place
where
we'll
have
unit
coverage,
perhaps
not
integration,
test
coverage
of
the
controller
behaviour,
but
a
very
close.
What
I
would
like
to
do
and
discuss?
A
Probably
not
now,
since
I
would,
rather,
if
we're
ready
to
wrap
this
up
before
the
end
of
the
hour,
I'd
rather
use
the
time
to
continue
working
on
MVP
one.
What
I
would
like
to
do
is
have
for
perhaps
two
weeks
from
now
a
MBP
to
milestone,
that
is
about
cleanup
and
improving
testability
and
getting
cin
CD
in
there
so
I.
A
I
would
like
to
round
off
any
rough
edges
in
that
MVP
milestone
the
next
MVP
milestone
and
make
it
so
that
when
we
show
this
at
cube
con
in
the
end
of
March,
that
we
have
something
that
you
will
feel
good
to
show
people
from
a
code
and
testing
standpoint.
How
does
that
sound?
A
book
I.
A
D
Yeah
I
want
to
bring
this
up
just
to
see
what
everybody
has
to
say:
rebasing
our
service
catalog.
What's
the
newest
coupe,
and
the
reason
of
that
is
because
right
now
we're
able
to
generate
open
API
expansion
through
our
definition,
but
we're
not
serving
it
and
the
reason
we're
not
certainly
because
I
became
akeli
in
the
open
API
coming
convict
the
Neil,
a
TI
machinery
is
passed,
a
map
that
we
write
to
a
shrug.
D
So
in
order
to
unblock
that,
we're
going
to
have
to
read
page
and
seems
like
we
definitely
sure
if
I'll-
and
I
don't
know
call
you
volunteer
to
do
that,
but
I'm
just
wondering
you
I'm
Brock
right
now
and
if
I
want
to
get
into
this,
is
it
a
crazy
idea?
I
know,
there's
going
to
be
like
a
good
amount
of
work
involving
this.
A
I
have
some
prior
experience,
rebasing
code
bases
on
to
Coober
Nettie's,
and
it's
not
for
the
faint
of
heart.
So
what
I
would
what
I
would
say?
Simon
is
I
appreciate
your
enthusiasm.
I
would
rather
look
for
things
that
you
can
do
that
are
smaller
chunks
to
bite
off
and
that
are
going
to
be
more
easily
actionable.
B
D
A
Okay,
so,
like
I,
said
I
really
like
to
get
to
that
this
week,
and
I
would
say
that
a
rebase
onto
onto
KU
benetti's,
like
16
candidate,
would
be
something
that
I
wanted
to
do
as
definitely
as
a
part
of
MVP
2,
so
know
that
it's
on
my
list
and
I
don't
want
to
make
anybody
else.
Do
it
because
it
will
probably
be
painful
and
I
will
I've
also
spent
some
time.
A
C
A
A
So
if,
if
you're
working,
if
you're
one
of
those
folks
whose
name
was
mentioned,
going
down
to
burn
down
list
and
you
to
switch
off
to
another
project
or
be
out
for
several
days,
I
would
sincerely
appreciate
it.
If
you
could,
let
me
know
or
send
an
email
to
the
mailing
list
or
put
a
note
in
the
flak
channel,
and
if
you
know
you're
going
to
be
out,
we
just
don't
pick
up
new
work
on.
E
A
A
A
A
Sorry,
my
bad
yeah
so
face-to-face
at
cube
cod
I'm,
going
to
be
there
like
I'll,
be
getting
in
it
like
seven
in
the
morning
on
monday,
I'd
love
to
do
a
face
to
face,
especially
for
folks
that
might
have
been
following
the
work
in
the
community
that
haven't
had
a
chance
to
to
go
to
the
last
cute
con
I
I'm
speaking
three
days
out
of
that
week.
So
there's
just
about
no
time
that
will
be
excellent
for
me,
so
whatever
works
best
from
other
for
other
people,
I
can
make
work.