►
From YouTube: Operator Framework SIG 2019 11 19 Full Meeting Recording
Description
Next Meeting Details: https://github.com/operator-framework/community/projects/2
December 17th 2019 will be the next call
B
A
Alrighty,
if
you
haven't
yet,
if
you
could
go
into
hack,
MD
I
put
the
link
in
the
blue
jeans
and
just
add
your
name
to
the
attendee
list.
A
I'll
take
copious
notes,
but
we're
driving
this
from,
as
we
did
the
last
time
from
the
the
OpenShift
community
page
under
projects,
and
we
have
on
the
agenda
so
far.
Today
we
have
myself
talking
about
some
of
the
upcoming
events
that
we
have
and
that
I'm
looking
for
speakers
or
am
I
am
there
on
one
I'm
in
the
open
shift,
one
yep
figures,
greater
framework.
A
B
A
C
C
C
Okay
sounds
like
mostly
security
fixes.
The
SDK
had
its
0:12
release
about
20
days
ago
and
I
know
that
we're
trying
to
kick
off
a
community
meeting
around
the
SDK,
and
they
can
talk
more
about
that
timeline
and
things
in
that
meeting
looks
like
there's
a
pretty
healthy
stream
of
stuff
coming
into
the
community
operators
and
operator
hub
IO
looks
like
some
updates
to
a
number
of
the
different
things,
which
is
really
awesome.
C
If
you
haven't
checked
out
the
new
CSV
editor
on
operator,
bio
that
is,
live
and
it's
a
really
cool
way
to
graphically
see
your
CSV
and
you
can
even
upload
a
bunch
of
files
from
your
old
CSV
and
you
know,
make
changes
and
then
publish
that.
So
it's
pretty
cool
a
little
tool.
If
you
haven't
checked
that
out,
I
would
encourage
you
to
do
so.
C
A
C
A
Though
I
think
I
have
a
couple
of
things
that
other
than
that
there
was
some
folks
that
got
to
level
4
that
were
awarded
the
videos
from
all
of
that
will
be
uploaded,
hopefully
by
end
of
day
tomorrow
to
the
YouTube
channel
and
I'll
tag
them
all
and
put
them
on
the
operator
playlist
as
well,
so
that
you
can
listen
in
to
get
the
feedback
from
them
on
where
they're
going
with
that.
That
was
pretty
good
and
and
nice
to
hear.
A
You
know
great
case
studies
and
stuff
like
that
I'm
looking
for
people
to
to
showcase
what
their
work
is
that
that
way
too.
So
that's
one
thing
and
I'm
in
particular:
I'm
looking
for
people
who
are
using
building
operators
for
enterprise
use.
So
if
you
are
using
an
operator
and
not
intending
to
ever
open
sources,
you're
building
something
specific
to
your
own
internal
use.
We'd
love
to
hear
some
of
those
use
cases
too.
A
So,
if
you're
willing
and
able
to
get
to
any
of
those,
please
reach
out
to
me
definitely
trying
to
curate
and
set
up
those
schedules
and
those
you
can
find
all
of
them,
except
the
Amsterdam
date
on
the
Commons
org
site
swap
over.
They
are
quickly
under
gatherings.
It
has
the
dates
for
them,
so
we're
here
in
San,
Diego
on
the
18th,
but
there's
one
on
the
29th
of
January,
April,
26
and
again
March
30th
for
or
Amsterdam
and
the
next
coop
con
and
EU.
There
are
lots
of
other
opportunities.
A
A
Popping
back
to
the
next
topic,
which
is
the
update
on
the
CN
CF
TOC.
We
have
scheduled
next
presentation
to
the
technical
oversight.
The
o
stands
for
Oversight
Committee
of
TNC
for
December
3rd,
which
also
coincides
with
my
birthday
so
I'm
expecting
a
present
from
all
of
you
that
we
get
this
one
through
and
the
presentation
details
will
be
in
the
Google
Doc.
That's
in
the
agenda
item
there.
A
A
That
would
be
helpful
too.
So,
there's
always
chatter
on
that.
Is
there
anything
else?
We
should
add
to
that
about
the
TOC
presentation,
Rob.
Anyone
think
that's
pretty
much
it.
They
we've
given
this
presentation
a
lot
so
I
think
at
this
point
it's
just
finding
sponsors
and
that
from
the
TOC
panel
to
endorse
it
as
well,
though
well,
alright.
So
the
next
thing
we
have
and
I
looking
to
see
if
Chris
short
has
joined
us.
A
Maybe
we
don't
see
him
on
there
all
right,
so
we
had
asked
Jeff
Garlin
to
give
us
a
presentation
on
some
of
the
work
that
he's
been
doing
around
using
the
ansible
operators
and
McGruder
and
other
things.
So
I
thought
we
would
give
him
the
floor
for
a
little
while
to
do
that
presentation
and
we
could
come
back
and
do
QA
and
get
updates
from
anyone
else
that
had
them
the
Jeff.
If
you
want
to
take
over
the
screen
and
I,
will
make
sure
you're
unmuted,
so
try
talking
Jeff
or
unmute
yourself.
E
Coming
off
a
cold
for
the
past
week,
so
it
is
not
completely
back
yet,
but
very
close
to
it
supposed
to
do
a
podcast
interview
later
today.
So
I,
don't
know
how
that's
gonna
work,
but
let's
see
this
is
the
first
time
I've
tried
sharing
my
screen
on
blue
jeans,
and
everyone
always
says
that
it's
always
always
screws
things
up.
So
let
me
try
it's
only
allowing
me
to
share
blue
jeans
or
my
entire
screen.
A
There
should
be,
but
I
always
go
with
the
default,
my
whole
screen
and
then
everybody
gets
to
see.
You
know
on.
A
A
A
And
I
also
note
that
there
is
I'm
on
Thursday
here
at
coop,
con
I'm
moderating
a
panel
on
operators
which
has
I've
told
400.
People
have
signed
up
for
it,
which
is
kind
of
intimidating
if
they
all
show
up.
A
That's
almost
as
many
as
showed
up
that
the
open
ship,
Commons
gathering
and
I've
asked
Andre
toast
from
IBM
is
going
to
be
on
then
I'm
facing
out
four
gentlemen
from
Google
there's
a
person
from
lifeand
who
had
a
health
issue,
but
they've
subbed
in
someone
from
like
then
and
I've
asked
Jeff
McCormick
to
join
as
well
on
the
panel
to
give
his
perspectives.
So
if
you're
here
and
the
room
isn't
full
I'd
love
it.
If
you
joined
us-
and
that
would
be
great
too.
A
E
All
right,
sorry
about
that.
That's
act
in
the
six
months
I've
been
doing.
This
I
had
never
shared
a
screen
on
blue
jeans,
apparently
so
lucky
the
I've
been
doing
ansible
stuff
for
years
and
then
doing
kubernetes
stuff
for
a
few
years,
and
when
the
operator
SDK
started,
integrating
ansible
I
thought
that
was
brilliant,
because
now
I
didn't
have
to
learn,
go
at
a
deeper
level,
just
to
write
an
operator,
so
I
started
playing
around
with
a
few
operators
like
I.
E
The
first
thing
I
did
was
a
drupal
operator,
and
this
one
we
I
worked
with
a
guy
who
works
in
Australia
and
does
some
drupal
work
and
we
kind
of
collaborated
on
and
I've
worked
as
he
pulled
in
some
changes.
Her
mind
back
in
the
fourth
and
right
now.
His
his
company
would
like
to
start
using
the
operator
for
their
hosting
for
Drupal,
but
that's
more
of
a
long-term
goal
because
they
already
built
I.
Think
in
like
OpenShift,
3
and
they're
still
running
like
kubernetes,
110
or
111.
E
They
built
a
whole
system
for
it
and
they
want
to
migrate
to
a
new
system.
So
that's
their
long-term
goal
and
my
goal
also
was
to
have
an
operator
because
I
was
operating,
I
was
operating
about
five
or
six
hundred
Drupal
sites
in
a
hosting
environment.
That
was
the
largest
one.
I
hadn't
worked
on
so
far
in
eks
on
Amazon,
and
it
was
getting
really
messy,
the
all
the
manual
stuff
that
we're
doing
it
not
manual
but
like
scripts
that
were
kind
of
tied
together
through
Jenkins
and
other
things
to
manage
all
the
sites.
E
E
It's
kind
of
like
throwing
somebody
into
a
large
Java
app,
instead
of
showing
them
like
a
Python,
the
hello
world,
app
if
you're
trying
to
train
them
on
how
to
do
something,
though,
that
there
was
already
an
example
of
a
memcache
operator.
But
the
memcache
operator
basically
starts
some
memcache
instances
and
that's
not
really
showcasing
the.
What
an
operator
can
start
doing
for
you
versus
what
you
can
do.
Just
using
a
home
chart
or
using
you
know,
rock
kubernetes
manifests
and
pushing
them
into
kubernetes.
E
So
the
McCotter
operators,
a
gulf
where
it
was
to
make
to
make
an
ansible
based
operator
that
uses
ansible
to
detect
a
change
and
react
to
it
and
do
things
similar
to
how
any
of
the
other
operator
types
would
do
it,
but
in
a
simple,
straightforward
way-
and
it's
just
complex
enough-
that
you
can
do
things
that
only
an
operator
could
do
without
having
to
integrate
something
external.
But
it's
also
simply
enough
that
it
could
be
presented
in
a
few
minutes
rather
than
a
few
hours
or
a
few
days.
E
E
And
then
the
crowder
distributes
the
request
to
back-end
memcache
servers,
and
it
does
that
inside
of
kubernetes
through
this
operator,
and
one
of
the
things
that
I've
been
playing
around
with
is
how
to
get
it
to
be
easy
to
deploy
and
one
way
that
I
do,
that
is
I,
actually
have
a
little
Hansel
playbook
that
builds
this
thick
router
operator
manifest.
That's
a
compilation
of
the
the
CRD,
the
roles
and
everything
else
that
deploys
the
operator
into
the
cluster.
E
This
is
with
some
defaults,
so
one
thing
that
I'm
looking
at
is
how
to
make
this
so
that
you
can
customize
it
a
little
easier
for
different
types
of
deployments
like
if
it's
a
name
of
cluster
wide
operator
or
just
operating
in
one
namespace.
That
kind
of
thing,
but
the
way
that
I
have
that
set
up
now
is,
if
you
wanted
to
play
this
into
a
new
kubernetes
cluster.
E
You
can
just
apply
that
manifest,
which
drops
the
operator
into
the
default
namespace,
and
then
you
can
put
Emma
Crowder
instance,
which
by
default,
gives
you
three
memcache
servers
behind
the
crowder
and
then
it
just
sets
it
up
for
you
in
any
namespace.
The
example
here
is
just
in
the
default
namespace
too,
but
by
default.
This
is
a
cluster
right
operator
and
most
of
the
ones
I'm
setting
up
I.
E
Do
that
because
in
most
of
the
scenarios
I've
been
in,
we
wanted
to
be
able
to
have
like
a
database
operator
and
a
application
operator
for
something
that
we're
going
to
host
hundreds
or
thousands
of
instances
of
and
each
instance
would
be
in
its
own
namespace.
If
we
had
to
manage
an
operator
in
each
namespace
and
the
application
that
that
operator
manages
in
each
namespace,
it
wouldn't
really
save
us
a
whole
lot
of
overhead,
and
we
usually
we
do
a
cluster
wide
upgrade
of
whatever.
That
thing
was.
E
So
if
we
were
going
to
go
from
postgrads
nine
to
two
nine
three
or
something
that
would
be
a
cluster
write
operation,
not
not
per
application,
that
we
were
running
I
mean
anyway.
So
this
this
operator
was
created
for
that
purpose.
And
it's
it's
fairly
simple
in
that
you
can
play
with
the
memcache
pool,
how
many
servers
it's
running
and
also
it
can
change
whether
the
pool
is
replicated
or
charted.
E
So
I'm
planning
on
bringing
some
of
this,
the
mini
cubes
and
test
scenarios,
which
are
very
similar
to
the
kind
test
scenarios.
There's
only
a
few
differences
I'm
trying
to
merge
that
all
together
put
it
into
the
default
scenario
so
that
anyone
using
operator
SDK.
If
you
build
an
instable
operator,
you
get
tested
and
kind
or
mini
cube
or
like
a
their
openshift
or
kubernetes
cluster.
E
That's
my
long-term
goal
and
because
I
like
for
me,
I
I
can't
work
it's
hard
for
me
to
work
in
like
new
cloud
clusters
all
the
time,
because
a
lot
of
times,
I'm
grabbing
my
laptop
going
somewhere
and
I-
want
to
be
able
to
bring
the
cluster
up
in
a
few
seconds
and
be
able
to
start
doing
some
debugging
locally,
especially
when
I'm
moving
around
at
a
conference
for
Wi-Fi
is
terrible
or
when
I'm
you
know,
traveling
or
a
lot
of
times.
I'm
I
have
some
time
like
I.
Do
something
with
my
kids
and
I.
E
Don't
have
a
very
reliable
internet
connection,
so
it's
easier
for
me
to
work
locally
in
mini
cube
or
in
in
a
kind
environment
and
I
plan
on
doing
a
couple
blog
posts
about
this
operator
at
some
point.
Talking
about
at
the
molecule
and
mini
cube,
integration
for
local
development
and
I'll
also
be
probably
talking
a
little
bit
about
it,
and
also
working
with
the
tower
team
I'm
trying
to
figure
out
how
we
can
get
this
to
be
an
official
operator
for
ansible
tower.
E
We
already
have
an
official
installer
for
OpenShift
and
it's
like
yeah,
but
you
can
get
on
operator
hub
and
somebody
goes
here
and
they
just
click
on
like
oh
I
want
something
from
Red
Hat
or
is
it
you
know
you
go
here
and
you
click
Red
Hat
and
right
now
you
don't
see
ansible
tower
like
it
would
be
cool
if
all
the
applications
that
we
have
had
an
operator
on
operator
hub
and
you
just
click
on
it,
get
the
information
install
it
in
your
cluster
and
you're
done.
That's
something
that
that
I
think!
That's
that's.
E
My
long-term
goal
for
this
is
this
is
where
kubernetes
starts
being
more
awesome
and
more
useful
for
people
as
let
me
have
these
operators
when
they
have
them
get
supported
and
then,
when
they
get
to
you
know,
number
2
and
number
3
are
the
two
that
I
think
make
the
biggest
impact
number
four
and
five
are
cool
and
awesome
can
get
us
further
in
everything.
But
right
now
most
things
with
kubernetes
barely
do
basic
install
and
when
they
do
they
they
do
it.
E
You
know
sometimes
very
poorly
and
make
it
harder
to
do
the
rest
of
it.
So
I
think
getting
these
kind
of
operators
up
and
running,
seamless
upgrades
and
full
lifecycle
is
going
to
get
it's
going
to
accelerate
adoption
of
not
only
the
kubernetes
and
gulp
and
shift,
but
the
operators
that
we're
building
on
them
and
the
tools
that
were
integrating
with
the
operators
anyway.
I'm
gonna
stop
talking
because
my
throats
about
to
rip
open
but
I
would
love
to
answer
any
questions
or
discussion.
Anything
like
that.
A
Camilla
had
a
an
idea.
A
D
A
F
E
D
A
Anyone
else
have
any
comments
or
questions
for
Jeff
or
is
there
my
my
one
question
for
you,
Jeff
and
I
know
I've
asked
you
this
before
the
the
work
you're
doing
on
the
with
the
for
Drupal
is,
and
that's
looks
like
right
now,
it's
in
your
personal
repo.
Is
there
any
conversation
with
the
drupal.org
wonderful
community
that
we
all
know
and
love
do
move
that
possibly
into
their
repo
and
have
that
be
something
that
is
a
Drupal
wide
there.
E
There
has
been
because
right
now
about
six
months
ago,
we
started
a
kubernetes
CMS
cig
in
Drupal,
which
is
it's
not
like
an
official
kubernetes
sig.
It's
not
an
official
Drupal
sig,
but
it's
basically
all
the
people
that
use
kubernetes
and
Drupal.
We
have
a
slack
channel
on
the
Drupal
slack.
We
have
meetings
that
pretty
much
every
Drupal
camp
and
Drupal
con.
That
happens
and
we've
had
online
meetings
every
other
months
for
the
past
months
and
we're
getting
slightly
closer
to
a
point
where
that
might
happen.
E
But
it's
the
hard
thing
is
like
mainline,
Drupal,
there's
not
there's
not
as
much
P
there's
not
as
many
people
using
kubernetes,
yet
there's
a
couple
hosting
providers
that
are
starting
to
move
to
it,
but
a
lot
of
them
are
still
just
you
know:
lots
of
VMs
on
Amazon
or
wherever,
and
they
don't
really
have
a
plan
yet
to
get
to
kubernetes
I
think
once
we
get
a
couple
more
of
the
bigger
hosting
providers
on
board,
that
could
be
a
serious
conversation
but
I.
You
know
the.
E
Is
an
open
source?
It
could
be.
You
know
three
months,
it
could
be
three
years
who
knows
the
Drupal
community
is
a
lot
less
a
lot
less
structured
than
a
lot
of
the
other
open
source
communities
like
there's,
not
there's
not
as
much
pressure
from
the
top
unless
it's
like
one
of
the
two
or
three
major
initiatives
and
kubernetes
has
not
been
right
now
that
the
major
things
in
Drupal
are
the
ease-of-use
and
updating.
You
know
design
themes
and
things
like
that
to
make
it
more
competitive
with
some
of
the
major
Drupal
CMS
competitors.
A
Well,
I
will
I
know
a
few
of
them,
though.
I'll
see
ya,
I'll
reach
out
to
you
separately
and
see.
If
we
can
do
that,
because
it
would
be
even
if
you're
still
the
ones
supporting
and
everything
if
it
lived
and
breathed
under
the
drupal.org
repo,
it
would
probably
get
more
more
stars
and
probably
more
feedback,
and
so
maybe,
if
there's
the
next,
that
unofficial
cig
meeting,
maybe
we
could
bring
it
up
there
and
I'll
join
you
for
that
call.
A
E
A
Doing
that
one
yet
I
think
we're
just
in
Java
at
the
moment,
I'm
meet
Python,
so
so
yeah
so
I'd
like
to
see
that
happen
and
and
then
get
it
into
operator.
A
How
do,
though-
and
that
would
be
great
and
anyone
else-
has
any
questions
or
feedback
and
around
the
table
here,
because
I
can
do
that,
because
there's
people
in
the
room
with
me
for
a
changed,
we're
doing
a
stew
face
here
at
the
Marriott
there's
a
note
from
Mandana
Vaziri
also
that
there's
another
operator
from
Ida
composable
which
I
hadn't
seen
and
it's
in.
Oh
it's
an
operator
hope
I
should
have
seen
that
by
now
and
I'm.
A
C
A
A
Pausing
he's
gonna
restart,
so
we'll
we'll
pause
for
a
minute
there
and
let
her
restart
and
maybe
give
us
a
few
words
about.
What's
going
on
with
that
one
does
anyone
know
if
that's
helm
based
or
what
is
Jake?
Is
that
a
go
one
and
then
I'll
ask
again
if
you
haven't
yet
added
yourself
into
that
bag?
Md,
please,
please
so
put
it
in
the
chat.
A
Give
everybody
a
minute:
let's
go,
that's
cool.
These
things
are
coming
in
to
operator
hub
faster
than
I
can
notice
them
in
the
beginning,
I
was
tweeting
out
every
time
a
new
one
came
in
like
hey.
Thank
you.
You
know
crunchy
data,
oh
thank
you
Missouri!
Thank
you!
Thank
you,
and
now
it's
like
okay,
wait
a
minute,
so
we'll
have
to
figure
that
out
I'm,
not
even
sure
what
the
password
is
for
the
Twitter
handle
anymore
sure
it's
baked
in
somewhere
by
now.
After
looking
start
doing
that
again,
let's
see
we
have
Madonna
back.
C
A
So
if
she
can't
get
back
in
and
make
sound
that
bigger
and
then
again,
I'm
gonna
put
up
the
I,
don't
see
Chris
short
online,
the
anyone
I
have
any
other
updates
or
anything.
While
we
wait
for
her
to
be
as
she's
got
access.
Otherwise
I'm
going
to
go
back
to
the
calendar
again.
F
Heard
yes,
so
we're.
A
A
Nope
Mandana
I
think
what
we're
going
to
do
is
we're
going
to
let
you
present
at
the
next
meeting,
which
brings
up
our
always
final
fun
topic
of.
When
do
we
meet
next
I,
always
looking
at
the
calendar?
Now,
though,
if
we
did
it
again
in
one
month
that
would
bring
us
to
the
8th,
the
9th
of
December.
Is
that
about
right,
cadence
or
should
we
do
it
on
the
16th?
C
A
A
D
G
G
I
shared
in
my
screen,
okay,
let
me
see
if
I
can
manage
that
so
I
mean
just
to
let
you
know
what
this
is.
This
is
an
operator
from
IBM
research.
It's
it's
in
preview
at
this
point,
so
it
was
published
on
operator,
I
love
yesterday
and
what
it
is
is
sort
of
like
it's
not
actually
managing
an
application
or
anything
like
that
is
kind
of
a
basic
operator.
G
So
what
you
can
do
is
take
an
existing
resource
and
wrap
it
inside
a
composable
objects,
and
then
any
field
can
have
can
be
annotated,
have
a
value
that
says
get
value
from
which
is
sort
of
you
know
indicating
that
you
have
a
you,
have
a
reference
there
and
at
that
point
where
the
controller
operator
does
is
that
it
takes
these
annotations
and
then
it
resolves
them,
and
then
it
deploys
the
underlying
object.
So
so
this
is
useful.
G
For
example,
if
you
have
you
know,
resources
that
need
some
piece
of
data
that
has
to
be
produced
by
the
deployment
of
another
resource
like,
for
example,
imagine
you
have
a
Kafka.
You
have
a
Kafka
source.
Your
Casca
sourcing
is
a
bootstrap,
you
know
URL,
and
so
what
you
have
to
do
is
to
first
deploy
the
Kafka
in
order
to
obtain
this
URL.
And
then
you
have
to
take
this
piece
of
information
and
put
it
in
your
Kafka
source
yamo
and
deploy
reattach
the
source
llamo.
G
So
this
is
sort
of
inconvenient,
because
then
you
have
to
stage
the
development
of
various
resources.
But
if
you
have
composable,
then
you
can
dynamically
configure
them.
So
then
you
can
take
all
of
your
llamó,
including
whatever
operator,
was
deploying
the
the
Kafka
in
the
first
place,
and
your
Kafka
scores
deploy
all
of
that
Yamma
at
the
same
time
wrap
the
appropriate
things
in
composable
and
there's
just
let
kubernetes
do
its
job.
If
that
does
that
make
sense,.
G
G
All
right,
very
good,
so
I
have
an
example
here:
the
calculus
source
example.
That
was
mentioning
so
as
you
can
see,
for
example,
this
particular
resource
has
a
bootstrap
server
field
and
you
have
to
hardwire
those
values.
But
in
order
to
obtain
in
values
you
have
to
first
deploy
a
kafka
and
then
see
you
know,
wait
for
it
to
successfully
deploy,
and
then
you
can
actually
obtain
this
information,
put
it
in
this
yamo
and
deploy
the
llamó.
G
So
what
people
do
when
they
have
solutions
that
are
made
of
many
different
resources
is
that
they
either
have
you
know
some
kind
of
you
know
documentation
that
says.
First,
you
have
to
deploy
this
and
you
have
to
do
for
that
or
they
have
scripts
that
can
be
brittle,
and
you
know
error-prone
and
in
any
case,
once
you
set
up
with
one
of
these
solutions,
it's
very
difficult
to
then
go
through
that
workflow
again,
when
you
have
different
environments-
and
you
have
you
know,
production
and
testing
staging
etc.
G
So
what
we
would
like
to
do
is
to
enable
people
to
just
take
all
of
their
yamo
and
be
able
to
just
deploy
it
at
once,
and
so
what
we
would
do
in
this
case
is
that
we
would
have
this
composable
objects
and
then
inside
it
we
have
a
template
and
you
put
the
calculus
or
other
template
inside
of
it
and
for
the
bootstrap
server.
We
have
this
reference
that
we
call
a
guess
value
from
and
there
you
can
actually
point
to
any
other
kubernetes
object.
G
So
in
this
case
it's
a
secret
that
happens
to
contain
point
as
the
binding
for
that
cactus
service.
That
has
this
particular
data
with
with
the
broker
URL,
and
the
other
thing
to
note
is
that
in
this
case,
sometimes
its
data
is
sort
of
formatting
mismatch,
so
we
also
composable
also
provides
some
data
transformers
that
you
can
type
together.
So
in
this
case
you
can
we
take
a
base64
encoded
data
and
get
drink
out
of
it,
and
that
happens
to
be
a
juicer
and
transform
it
to
an
object.
G
G
The
controller
for
the
composable
is
going
to
take
the
template
and
at
once,
it's
going
to
try
to
get
all
of
the
values
for
the
objects
that
are
being
referenced,
and
this
is
important
because
we
want
to
get
a
consistent
view
of
the
objects
and
not
to
read
them
in
piecemeal,
so
that
we
don't
get
values
that
never
actually
happened
in
a
TCD.
And
so
once
we
were
able
to
resolve
these
all
of
the
all
of
the
missing
holes
in
the
underlying
object,
then
composable.
G
What
it
does
is
that
it
goes
ahead
and
deploys
the
underlying
object
and
it's
sort
of
basically
it's
responsible
for
the
lifecycle
of
the
underlying
object.
So
if
you
actually
delete
manually
the
underlying
object,
because
the
composable
operator
would
bring
it
back
etc
and
you
feel
diseased
composable,
it
would
delete
also
the
underlying
object.
So
that's
that's
pretty
much.
It.
G
A
C
C
All
right
so
I've
posted
a
link
to
this
service,
binding
operator
that
some
of
the
Red
Hat
folks
have
been
working
on
and
I'm
curious
I
think
they
sound
like
they
play
in
the
same
space,
but
have
different
models,
which
is
interesting.
The
surface
binding
is
a
more
of
a
like
a
late,
binding
kind
of
thing,
where
it's
like
a
more
of
a
after
you
have
both
operators
running
or
both
applications
running
that
then
they're
connected
versus
this
is
kind
of
doing
that
all
in
the
beginning.
C
One
of
the
questions
in
the
room
here
was
about
security
of
this,
because
I
think
one
of
the
the
principles
of
the
service
binding
operator
is
that
you
declare
which
fields
people
can
bind
to
versus
this.
It
seems
like.
Is
it
taking
more
of
it?
You
can
find
everything
approach
like
how
do
you
protect
other
applications
data
at
that
point,.
G
So
it
could
be
that
we
could
use
composable
in
order
to
actually
what
service
binding
is
trying
to
do
in
terms
of
security.
That's
a
great
question
right
now:
it's
not
great
as
made
that,
because
we're
using
an
unstructured,
you
know,
objects
to
represent
anything
really,
and
so
in
terms
of
our
back
we're
giving
the
comfortable
operator
all
sorts
of
permissions
that
we
shouldn't
be,
it
can't
do
anything
basically
so
in
the
future.
G
What
you
could
do
is
to
say
that
composable,
you
know
here's
the
list
of
CR
DS
or
resources
that
we're
allowing
composable
to
wrap,
and
then
at
that
point
we
can
give
very
specific
permissions
to
composable
so
that
you
know
it's
not
doing
you
know
it.
Does
it's
not
too
permissive,
so
that
that
would
be
one
way
of
tackling
that
problem,
but
it
is
true
today
the
way
written
it's
completely
generic.
It
uses
the
unstructured
type.
G
H
Yes,
so
you
just
add
something
to
do.
This
is
that
Madonna
and
I
are
also
involved
in
the
discussion
for
the
service
binary
operator.
We've
been
actually
in
the
first
call
that
was
led
by
Kyle.
She
lost
her
from
IBM
and
we're
actually
on
the
agenda
for
the
next
call,
where
we
can
actually
talk
about
an
other
operator
that
we
had
in
operator
up.
H
That
is
the
IBM
cloud
operator
and
the
idea
there
is
also
to
show
some
integration
between
them
and
the
service
binding
operator,
because
then
we
can
potentially
use
the
service
binding
operator
to
create
some
other
binding
service,
buying
the
request
and
basically
inject
that
wait
credentials
into
the
pod,
specs
and
potential.
We
can
also
show
maybe
some
ideas.
H
You
know
how
we
can
actually
something
similar
using
composable
I
think
you
know
the
sig
for
service
binding
is
still
relatively
in
early
stages,
so
I
guess
the
idea
is
to
try
also
to
get
ideas
from
you
know.
Other
work
in
the
community
see
if
we
can
somehow
come
up
with,
are
the
most
stronger
proposal.
H
Overall,
I
think
is
still
something
I
guess
going
on,
but
of
course,
I
think
that
the
main
difference
that
the
service
binding
was
very
focus
on
a
very
specific
scenario,
and
this
one
we're
talking
about
here
is
really
trying
to
solve,
maybe
a
marginal,
a
broader
kind
of
problem,
basically
for
every
kind
of
resource
you
can
have
in
kubernetes.
You
have
a
way
to
now
provide
this
kind
of
dynamic
binding
to
other
things,
and
that
can
be
pretty
much
anything.
So
it's
not
really
a
constraint.
Of
course.
H
There
is
also
this
idea
that
in
this
case,
we
are
kind
of
wrapping
something
versus
not
wrapping
and
injecting
after
the
resource
is
created.
So
those
are
very
different
approaches
in
work
is
may
some
of
me
like
this
or
may
not,
so
we
are
still
basically
trying
to
taste
the
world
that
I
see,
who
basically
have
no
concern
with
this
approach
or
not
right
is
still
relatively
early
for
us.
B
H
A
A
D
A
If
you
have
something
you
want
to
talk
about,
that's
operator
related,
please
reach
out
to
me
and
I
will
try
and
get
you
a
podium
at
any
of
the
operator
at
the
operator
at
the
OpenShift
Commons
gatherings
that
are
coming
or
allow
give
you
an
hour
to
talk
about
it
as
a
novitiate
Commons
briefings
or
help
you
find
a
meet-up
or
some
place
to
put
significant
about
your
wonderful
work.
There's
a
lot
of
them
out
there
now.
So
we
really
want
to
get
that
feedback
to
all
of
the
folks
working
on
the
SDKs
and
I'm.
A
Looking
forward
to
seeing
that
Java
SDK
emerge
soon
as
a
wholly
usable
thing,
so
we'll
keep
you
posted,
maybe
even
put
that
on
the
agenda
for
the
next
time
and
see
if
we
can
move
that
forward
a
little
bit
there
anything
else
from
anyone
in
the
room.
Not
oh,
that's
right,
one
more
thing
all
on
you
and
you
go
yeah
just.