►
From YouTube: Kubernetes SIG Service Catalog 2018-10-08
Description
Agenda: https://docs.google.com/document/d/17xlpkoEbPR5M6P5VDzNx17q6-IPFxKyebEekCGYiIKM/edit
* Check with the group on current plan for custom plans and reuse of external id
* Chat about implications of AWS moving away from brokers to custom operators
A
Hello
and
welcome
to
cigs
service
catalog
October,
8th
edition,
I'm
Carolyn
vans,
like
one
of
the
service,
catalog
chairs
and
I'll,
be
facilitating.
The
meeting
today
feel
free
to
add
your
name
to
the
attendee
list
so
that
we
all
know
you
were
there
or
lurk
whatever
it's
fine,
I'm
gonna
run
through
just
top
to
bottom.
A
Catalog
of
this
service
can
only
do
the
service
instance
that
I,
you
don't
need
to
know
all
the
right
parameters
and
things
like
that.
But
what
the
way
it
works
is
we
copy
it
and
we
change
only
the
the
kubernetes
name
and
the
external
name,
and
we
leave
the
external
ID
alone,
and
you
know
Jay
and
I
were
talking
about
this
last
week.
That
I
think
that's,
ok,
cuz
I
haven't
seen
anything
blow
up
yet,
but
that's
not
really
like
the
best.
B
A
A
So
what
I
mean
what
I
will
end
up
looking
like
is,
if
you
queried
at
CD
use,
you
know
whatever
ignore
SP
cat
or
anything
like
that
for
a
minute.
You
queried
it
and
said,
give
me
plans
that
have
this
external
ID,
for
example-
and
usually
it's
some
really
long
doing
that
you
know
abc123
something
but
much
longer
you
may
get
duplicates
and
what
will
be
different
about
them
is
operator
defaults.
A
One
will
have
been
provided
by
the
broker
and
it's
flagged
as
being
owned
by
a
Service,
Catalog
and
then
another
one
won't
have
that
owner
reference
and
it'll
have
extra
parameters
set,
and
maybe
a
friendlier
name
and
things
like
that.
It
says
it's
like
hey
dad's
use
this,
or
this
is
our
custom
database.
If
one
useful
corruption
or
whatever
things
like
that,
but
it
will
have
a
duplicate
external
ID,
I,
don't
think
service
kinda
like
gets
mad
about
that,
but
I
don't
think
the
broker
cares
because
essentially
it's
the
same
service
under
the
hood.
C
I
actually
I
posted
in
the
catalog
service,
slack
channel,
like
I,
was
done
in
biscayne.
What
does
it
take
to
creator
and
I
was
right
and
so
I
came
across
previous
templates,
and
somebody
had
actually
done
one
in
Ruby
they're
connected
at
the
github,
and
it
actually
wasn't
too
bad
like
to
really
understand
how
to
create
an
OS
B,
because
I
think
when
you
look
at
this
face
value,
a
lot
of
people
interpret
it
well.
I
have
to
use
the
azure
one
or
the
AWS
one
or
it's
like.
C
C
I
can
very
see
that
there's
sort
of
like
a
marketplace
right,
a
broker
marketplace
and
you
could
just
throw
knees
as
brokering
drop
it
in
a
dozen
and
whether
it's
just
a
generic
one
or
whether
it's
something
to
civic
to
like
I,
don't
know
I've
broken
the
Salesforce,
for
example,
use
as
an
example
right,
yeah
I
definitely
can
see
that
and
I
think
that's
pretty
powerful
about
this
about
this
product.
A
lot
of
people
today
when
they
look
at
it
and
I've,
talked
to
people
about
it.
In
my
organization
they
go
what's
that.
A
A
D
She
has
a
marketplace
command,
but
it's
not
really
showing
me
all
the
brokers
and
the
services
provide
it's
more
like
show
me
the
things
I
can
buy
on
the
system.
It's
not
really
tied
to
the
abstractions
of
brokers.
Class
I
mean
it
sort
of
is
plans,
but
it's
not
it's.
It's
less
of
a
concern
of
like
show
me
which
servers
are
connected
to
the
platform
and
more
like
show
me
the
things
I
can
attach
to
my
app
okay.
A
Yeah
so
I
guess
we're
using
the
marketplace
in
two
different
ways.
One
is
what
services
are
just
available
on
my
cluster
right
now,
that's
all
kind
of
already
been
configured
by
the
operator
and
then,
if
you're,
an
operator
and
you're
shopping
for
what
are
fun
things
I
could
expose
from
herb.
You
know
provide
my
users
if
you're
not
familiar
with
those
would
be
kicked.
This
is
kind
of
a
neat
thing.
Oh
he's
in
the
agenda
just
yeah.
A
Has
a
list
of
some
brokers
and
if
you
know
of
other
brokers
that
we'd
like
to
add
here,
please
please
put
them
on
as
well,
and
then
one
that's
not
on
the
list
but
hopefully
could
be,
is
a
mini
broker
my
feet,
my
favorite
broker,
cuz
I
wrote
it
it's
essentially
a
a
broker
that
makes
things
using
home
charts
on
your
local
cluster.
It
just
lets
you
work
with
service
catalog
without
actually
charging
money,
so
one
of
the
cloud
providers
and
and
paying
them
stuff.
It's
not
production.
A
C
E
Actually,
so
it's
a
discussion
happens
to
be
right
up
my
alley,
so
the
CF
marketplace,
what
that
is,
so
it
is
a
command
and
what
that
does
is
it
lists
the
services
that
are
available
to
instantiate
I,
believe
in
the
service
catalog
world
to
call
those
service
classes
right
in
the
plans,
yeah
and
then
there's
plans
that
are
the
plans
that
are
associated
with
those
services.
So
it
does
show
you
it
doesn't
list
the
service
brokers
per
se,
but
it
lists
the
services
and
plans
for
those
services
that
are
available.
A
All
right,
so
the
next
thing
up
on
our
agenda
was
how
many
of
you
saw
that
AWS
Amazon
has
decided
to
not
these
brokers
anymore
and
they're
actually
made
a
custom
operator
for
s3
and
everything
else,
and
it's
basically
an
AWS
operator
and
it's
really
cool
from
a
UX
perspective.
You
can
use
the
Sierra
DS
and
you
can
have
a
Sierra
tea,
that's
specific
to
a
an
s3,
oh
gosh,
Cena
and
I.
Don't
use
Amazon,
so
I
can
like
basically
like
buckets
whatever
containers
or
I.
A
Think
RDS
is
something
that
they
want
to
be
able
to
do
and
instead
of
having
like
what
we
do,
which
is.
We
have
like
a
service
instance
and
it's
all
very
kind
of
generic
at
the
moment
in
Service
Catalog
they
actually
use
C
or
D.
So
it's
a
specific
kind
for
it.
And
it's
is
it's
like
a
little
richer,
UX
experience
and
there's
a
lot
of
people
talking
about
what
that.
What
that
really
means
for
OSB
and
Service
Catalog,
and
is
that
where
people
should
be
moving
to
in
general
and
I
was
curious.
A
A
A
So
what
I
was
thinking
about
was
like
what,
if
we
moved
to
CRTs
I,
know
Scott
prototype
this
a
bit
and
I.
Think
at
this
point
nearly
everything
we
need
is
out
of
the
way
and
it's
been
implemented
by
API
machinery
that
we
could
really
try
it
again
and
move
to
CRTs,
and
then
that
got
me
thinking,
though,
that's
just
one
thing
right:
that's
a
back-end
implementation
change.
A
A
This
is
like
ivory
v2
gamal
that
I'm
making
up
right
now
in
front
of
you,
but
you'll
have
like
a
cluster
service
class
name,
and
then
you
have
to
come
up
with
whatever
it
is
you're
like
my
sequel,
foo
and
you
have
to
put
in
the
class
the
plan
too.
So
it's
the
same
thing
plan
name
three,
something
like
that
right
and
then
you
have
primers,
and
this
is
actually
where
all
the
interesting
flags
go
for
making
that
service.
A
A
And
those
likes
put
this
out
really
weird
high
port:
the
pine
doesn't
even
exist.
You
can't
change
all
these
various
things
and
the
same
thing
with
service
bindings
as
well.
It's
super
generic
and
you
have
to
kind
of
deal
with
some
internals
about
how
Service
Catalog
works,
and
you
don't
spend
a
lot
of
time.
Thinking
about
the
real
domain
of
what
you're
working
with,
which
is
one
of
the
cloud
providers,
databases
or
you
know,
objects,
store
solutions.
A
Sorry
mega,
my
sequel
are
us
I'm
sure
as
a
plan
for
matter
you
can
set
parameters,
but
if
we
take
what
we've
been
working
on
with
the
plan
default
proposal,
where
an
operator
can
set
up
default
values
for
custom
plans,
you
can
actually
take
it
one
step
further
and
just
do
something
like
this,
where
you
can
say
one
of
my
sequel
server
instance.
This
is
what
I
want
to
call
it
and
here's
what
here's
a
parameter
that
I
want.
You
don't
say
any
of
the
underlying
details
of
the
broker.
You
didn't
have
to
say.
A
As
our
my
sequel,
you
don't
have
to
say
what
the
plan
was,
and
maybe
you
didn't
have
to
say
back
up
true
because
the
operator
already
set
that
for
you
and
you
get
some.
That
looks
a
lot
closer
to
the
CR
DS
that
you're
seeing
with
AWS
is
operator,
and
you
can
do
something
even
super
simplified
like
let's
say,
for
example,
I'm
an
operator
and
I
have
a
mega.
My
sequel,
server
instance,
somewhere
and
dad
was
just
want
to
able
to
randomly
create
databases,
as
an
operator
in
a
namespace
I
could
have
made
a
broker.
A
That
says
when
you
ask
for
these
I'm
gonna
give
you
a
random
database
attached
to
this
server
instead
of
everyone.
Getting
a
new
server
and
all
the
defaults
are
set.
All
they
have
to
do
is
pick
a
name
and
they
can
start
using
it
like.
You
could
literally
get
it
down
to
this
size,
and
it
looks
a
heck
of
a
lot
less
complex
and
generic
and
intimidating
than
what
we
have
today.
A
It
would
require
a
ton
of
work
on
our
part,
like
one
moving
the
CRTs
to
there's
one
more
thing:
we
need
from
API
machinery,
which
is
dynamic
and
former's,
which
would
allow
us
to
pull
and
watch
for
any
kind.
So
the
kind
instead
of
just
looking
for
service
instance
and
watching
for
changes.
You
have
to
be
able
to
watch
for
my
sequel,
database
or
blob
container,
or
you
know
whatever
it
is,
that
they
could
eventually
be
registered,
but
they've
got
all
the
plumbing
for
it
and
they
have
an
open
issue.
A
It
would
like
to
implement
it,
so
this
is
something
a
path
we
could
run
out
go
down
when
we
take
CRTs,
we
implement
it.
We
maybe
put
something
in
front
of
our
controller
that
watches
for
special
CRTs
and
then
turns
it
into
service
instances
or
things
just
go
directly
to
the
broker
from
these
Sierra
T's,
like
I,
think
we
can
implement
this
and
it
like
a
stepwise
fashion,
without
rewriting
a
ton
of
service
kind
of
long
and
try
this
out.
A
C
A
All
right,
so
what
I'm
gonna
do
is
maybe
make
a
couple.
I'm
gonna
turn
this
into
a
real
proposal,
but
it's
gonna
be
a
couple
different
things:
one
is
the
move
to
see
are
to
use
figuring
that
out.
Let's
look
at
our
milestones,
real
quick,
so
we
look
at
our
b1
milestone.
I,
don't
remember
if
we
had
Sierra
T's
in
it
or
not,
it
was
kind
of
a
contentious
thing.
That's
something
one
remember.
A
A
A
B
A
I
I
don't
know,
I,
don't
think
that
we
were
too
keen
on
trying
to
push
off
one
Oh
even
further,
for
to
be
honest,
whatever
goes
into
one
Oh
is
what
we
work
on
right
now.
If
you
look
at
what's
in
one,
oh,
it's
whatever
people
are
excited
about
and
if
nothing
has,
if
some
things
haven't
worked
on
the
while
it's
cuz
no
one's
keen
on
doing
that
work
mean
it's
not
terribly
necessary
for
one.
Oh,
maybe
that'd
be
a
good
thing
to
talk
about
next
time,
but
here
we
go.
B
A
A
B
A
Yeah
and
then
we
have
an
umbrella
issue,
as
people
have
been
working
on
getting
all
this
stuff
implemented
in
SVG.
Yet
I
have
been
maybe
a
little
too
aggressive
and
breaking
down
these
issues
into
really
small
pieces,
so
lots
of
people
could
could
work
on
it.
We've
had
like
three
different
people
working
on
things.
We've
been
keeping
track
of
even
little
gaps,
sort
of
cropped
up
in
between
everything,
so
I'll
be
working
with
those
contributors
to
kind
of
hammer
out
the
little
last
details
and
as
we
can't
but
I
think
we're
super
close
yeah.
A
Cool
so
yeah,
oh,
oh,
you
make
another
proposal
that
just
kind
of
talks
about
this
and
links
off
to
what
we
need
to
do
if
someone's
interested
in
running
with
a
CRD
prototype.
Let
me
know
otherwise
I'll
probably
jump
on
that
next
and
just
try
to
get
it
to
the
point
where
we
see
how
awful
is
this
gonna
be
I.
A
Did
and
then
work
is
kind
of
reprioritize
things
about.
You
asked
me
to
carry
that
torch
a
little
bit.
Okay,
but
I'm.
Just
super
keen
on
getting
this
in
cuz
I
hate,
I
hate,
responding
to
things
like
the
AWS
operator
with
we
have
plans
to
be
this
cool,
but
we
just
haven't
gotten
there
yet
so
I'd
like
to
just
get
us
to
that
state
as
quickly
as
possible,
but
yeah.
Maybe
next
week
we
can
talk
about
some
0.2
and
100,
and
things
like
that,
like
we
really
don't
want
CRTs
in
there.
A
So
that's
a
really
rambling
thing
thanks
everyone
for
bearing
with
me,
if
you
have
any
PRS
that
you
want
review
just
from
week
to
week
and
you're
not
getting
the
attention
it
needs
and
you're
pinging
me
I'm,
not
reachable
or
whatever
feel
free
to
just
put
them
in
here.
I
put
in
a
couple
that
are
pretty
much
ready
to
roll
what
I
care
about
most
is
fixing
our
home
chart.
Someone
reported
a
bug
where
we
messed
up
the
indication
for
EML
and
it
broke
our
counter.
A
B
A
A
Just
best
efforts
is
always
helpful
and
that's
what
the
approvers
are
there
for
is
to
be
that
final
catch-all
of
a
smart
person
who's
familiar
with
the
project
to
find
any
gotchas.
So
all
right
Nikki,
would
you
like
to?
Is
it
Mickey
or
Mike?
Here,
oh
my
god,
Mikey
okay,
can
you
introduce
yourself
yeah.
E
I'm
Mikey
bolt,
like
I,
said
I
work
at
pivotal.
We've
been
a
a
you
know
in
in
the
past,
we've
been
solely
focused
on
cloud
foundry,
so
we
have
an
on-premises
found
cloud
foundry
solution
that
we
sell
as
part
of
that
I've
been
working
with,
like
I
said,
is
V
partners
or
other
software
vendors
to
get
things
like
service
brokers
and
their
services
available
to
cloud
foundry,
workloads
relatively
I.
Guess
it's
less
recent
than
it
might
be
about
about
a
year
ago
we
added
a
kubernetes
offering,
and
so
you
know,
I
think.
E
Having
the
background
with
you
know,
a
lot
of
service
broker
work
with
different
companies
and
having
our
new
focus
on
kubernetes
I
think
could
hopefully
make
some
contributions
here
and
and
pivotal
is
definitely
interested
in
seeing
what
we
can
do
with
making
services
available
on
kubernetes.
In
the
same
way,
you
know
develop
your
driven
way
that
they
are
available
on
Cloud
Foundry
right
now,
so
I
actually
was
hoping
to.
E
A
Yeah,
if
you're
looking
for
things
I,
do
try
to
keep
things
with
Help
Wanted
and
good
first
issue
you.
So
if
you
want
to
look
at
some
of
them
or
actually
there's
a
ton
that
are
just
ready
to
grab
and
are
pretty
reasonable,
so
go
through
there,
and
if
you
can't
find
something,
let
me
know
and
I
can
turn
out
more
open
on
my
project
management
hat
and
make
more
issues
there's
a
ton
of
work
if
you're
in
if
you're
looking
for
stuff.