►
From YouTube: CF-for-K8s [November 2019]
Description
No description was provided for this meeting.
If this is YOUR meeting, an easy way to fix this is to add a description to your video, wherever mtngs.io found it (probably YouTube).
A
It's
already
on
okay,
so
hi
everybody
welcome
to
the
first
cloud
foundry
on
kubernetes
call.
After
all,
continents
have
agreed
on
being
on
winter
time,
I
think
the
last
one.
Unfortunately,
he
got
a
little
bit
railed
by
the
fact
that
the
Europeans
wanted
to
join
an
hour
later
than
the
folks
in
in
the
US.
A
So
I
think
now
is
actually
the
time
where
what
we
discussed
in
in
the
initial
call
where
we
wanted
to
see
what
the
common
interest
of
group
is
in
terms
of
topics
and
as
preparation
for
for
this
call
I
put
in
the
topic,
suggestions
that
we
had
so
far
into
a
slack
poll
and
I
think
cake
also
added
another
another
topic
in
there.
So
I
think
going
by
the
by
the
woods
it
looks
like.
A
Meanwhile,
we
have
a
mature,
but
he
wrote
for
relevant
for
cloud
foundry
on
kubernetes,
with
eight
people
voting
for
that
and
then
I
think
we
even
have
a
tie
with
like
the
dual
approach
of
discussing
both
face
cloud,
foundry
and
container
based
cloud
foundry,
and
if
and
how
old.
This
should
go
in
parallel.
So
I
guess,
as
we
have
a
tie-in
in
those
any
any
preference
which
topic
to
to
start
with
I'm
kind
of
to
suggest
the
relevant
topic.
But
if
there's
other.
A
A
E
Mention
that
I
paint
side,
this
I
think
it
was
last
week
but
I
think
he
might
be
on
vacation
or
out
of
office
so
waiting
for
him
to
get
back
to
start
a
conversation
around
the
pipeline's
that
we
currently
have
for
kids.
Yes,
we
have
built
quite
a
lot
around
that
and
yeah
I
think
there's
just
a
few
things
missing:
to
have
an
entire
flow
that
reacts
on
changes
for
stem
cells
releases
and
gets
you
a
result.
Saying
hey!
This
cube
CF
passes
cats.
E
A
F
B
E
G
B
That's
good,
that's
good!
That's
that
that's
I'm,
really
good
and
I
mean
it
want.
One
thing
that
I
think
is
very
clear
is
that
between
what
we
have
now
and
the
CF
operator
I
will
take
the
safe
operator
very,
very
keenly
so
I'm
I'm
keen
for
that
to
happen
whether
or
not
there's
some
simpler
thing
as
well.
That's
another
conversation
by
I'd
love
to
get
there
I
mean
you
see
what
we.
B
Now
so
so
yeah
this
is,
this
is
one
thing,
I
want
to
say:
I,
don't
know,
I
mean
this
is
kind
of
a
not
for
nothing
but
kind
of
underlines.
The
importance
I
think
I
mean
to
me
anyway.
I
sent
a
1.0.
It
really
is
one
point
I
we
met
out
today
in
the
a
and
that
that's
really
good,
but
you
know
there's
there's
a
line
in
there.
B
We
have
problems
and,
as
we
try
to
roll
that
out
so
I
mean
it,
isn't
the
arena
team's
first
priority,
because
it
isn't
something
we
should
be
solving.
But
if
I'm
honest
I
think
it
is
the
biggest
priority
for
arena
success,
so
I'm
I,
guess
I
I
mean
there's,
there's
no
action
that
other
than
I
really
think
that
getting
to
their
side
the
ideal
that
we
want
to
get
to
I
think
we
all
know
what
that
looks
like,
but
I
think
right
now,
it'd
be
really
good.
B
C
So
I
know:
we've
got
representatives
from
the
networking
program
here.
That's
one
of
the
other
core
component
teams,
that's
contributing,
and
that
is
I
think
maybe
the
farthest
along
in
terms
of
deviating
from
Bosh
package
artifacts,
so
might
be
interesting
to
hear
from
either
you
Gabe
or
a
taco
or
Shannon
about
your
perspective
as
another
contributing
team
that
religious
integration
would
be
helping
to
integrate.
I
H
B
That
kind
of
reminds
me
I
mean
we
have
some
of
my
other
hat,
obviously
is
with
the
garden
team
in
garden.
Is
gonna,
try
and
start
taking
on
the
work
of
I,
really
neither
pod
security
policies
and
that
kind
of
stuff
from
Irene
II,
and
we
have
the
exact
same
problem
even
though
I'm
the
real
EPM
I've,
no
idea
how
to
tell
the
garden
team
how
to
integrate
that
stuff
with
it
really
right
now
and
yeah.
B
B
E
I
G
D
We
all
want
to
release
integration,
but
we
don't
have
anybody
who
is
currently
responsible
for
or
probably
willing
to
adopt
the
responsibility
of
release
integration
for
CF
on
communities,
but
it
has
been
discussed
when
we
had
a
with
him
abital.
We
had
a
big
meeting
for
the
folks
who
contribute
to
the
application
service
and.
A
E
Way,
if
I
could
answer
that,
so
first
I
wanted
to
say
that
what
we're
currently
doing
for
kube
CF
is
exactly
that.
We
have
two
sets
of
pipelines,
one
we
call
pre
publish
that
one
is
used
to
gate
PRS.
It
runs
one
set
of
cats,
sorry,
not
cats,
yet
just
smoke
tests
right
now
on
all
the
PR
that
you're
that
you're
building
it
you
know
to
see
if
it's
green
or
not,
and
then
we
have
a
post
published
pipeline,
that
one
can
actually
run
a
matrix
of
tests.
E
That
involves
multiple
cube
versions
and
multiple
feature
sets
so
say
you
have
cube
113,
114
and
115,
and
you
want
to
test
with
Diego
and
with
Irene
e.
It
should
run
that
entire
matrix
of
of
tests.
So
that's
what
we're
currently
working
on
for
pipelines
and
to
answer
the
question
about
CF
deployment
cube
CF
is
now
tied
to
a
particular
CF
deployment.
Right
now
it's
8.0,
but
we
are
just
bumping
in
this
print
to
12,
12.1
I.
Think,
okay,
thank
you.
F
I
think
what
is
probably
the
the
least
clear
is
like
how
will
release
integration?
Look
like
for
the
non
boshy
will
like
it's
like.
You
can
now
cut
out
of
the
legs
if
operators,
starting
with
with
ireenie
and
but
like
how
are
we
going
to
manage
like
a
handful
of
thumb,
charts
or
whatever
we
are
going
to
use
I?
Think
that's
the
thing
that
we
should
yeah.
E
E
Hoping
that
we
can
use
these
feature
flags,
what
we
call
feature
flags
in
the
helm
chart
you
have
the
main
home
chart,
that's
the
cube
CF
and
then
our
spiked
with
Nats
was
using
nads
as
a
sub
chart
of
that.
And
then
you
have
a
flag
in
your
values,
llamo
saying:
okay,
which
type
of
net.
Do
you
want?
You
want
the
boss
based
one
or
do
you
want
the
native
one?
So
once
you
have
the
feature
flag,
you
can
also
put
that
in
the
test
matrix.
E
B
Yeah
so
I
think
I
think
the
missing
pieces
in
my
head
anyway.
Two
things
one
is
to
have
have
a
repo
I
can
point
people
at
so
so
in
it
really
release
documentation
instead
of
us
telling
people
how
to
deploy
a
CF.
What
I
really
want
one
to
be
able
to
do
is
point
them
to
a
place
with
instructions
for
deploying
CF.
If
you
see
what
I
mean
assessing
that
thing,
good
automatically
have
the
latest
stuff
I
published,
and
one
thing
I
think
we're
slightly
missing
is
in
their
kind
of
Bosch
world.
B
It's
like
Bosch
IO.
We
like
publish
the
latest
release
and
Relling
can
just
pull
that
down.
If
you
see
what
I
mean
and
we
need
some
sort
of
similar
kind
of
control
point
or
whatever,
like
some
way
of
coordinating
so
God
and
can
ship
the
latest
security
there
yeah
moles
and
they
get
consumed
and
networking
can
do
the
labels.
Networking
on
my
leg,
pointing
as
that's
one
thing
I
think
we're
missing.
The
other
thing
is
I
totally
see
where
you're
going
and
I
think
it's
a
good
directions.
B
Like
start
with
its
overall
thing
and
the
flavors
to
configure
different
things
out
and
stuff,
I
would
also
say:
I
think
we
do
need
so
that
there's
like
a
conceptual
overhead,
so
this
whole
Cloud
Foundry
beast.
If
you
see
what
I
mean
and
that's
fine,
if
you
want
the
whole
consentual
overhead,
but
I
think
we
do
need
a
minimal
CF
which
might
well
use
the
same
stuff
like
maybe
it
just
uses
the
operator
and
a
more
minimal
operator
but
I
think
having
a
minimal
CF.
Let's
just
five
things.
B
Early
is
kind
of
important,
like
I.
Think
you
can
you're
saying
to
someone
hey
just
pick
up
this.
This
Cloud
Foundry
thing
I
mean
I've.
Had
this
conversation,
you
say
pick
up
this
crab
sound,
you
think,
I
know
it's
just
this.
It's
like
30
containers
and
I.
Don't
know
how
to
operate
it
and
then
you
say
have
you
seen
miss
do
at
all,
but
it
doesn't
matter.
Thank
you
it's
and
by
anyway.
E
H
Curious
I
know
that
there
are
people
from
the
release
integration
team
here
to
really
listen,
but
I
am
curious
to
like
maybe
surface
some
requirements
from
the
folks
who
are
here
about
what
we
would
want.
One
question
I
have
for
people
is
Bosh
IO.
Will
we
ship,
in
terms
of
box
releases
a
source
code
effectively
to
the
hood,
but
the
sort
of
standard
way
of
deploying
or
shipping
artifacts
in
the
good
news
world
is
container
images.
H
B
This
this
brings
us
onto
another
thing,
actually
I'd
like
to
create
some
controversy,
please.
So
there
there's
this
obvious
problem
that
we
have
that
we
don't
have
enough
requirements
from
real
developers,
telling
us
how
things
should
work
right
and
that
often
causes
problems
like
that.
The
thing
with
cs4
honest
is
it:
it
has
focused
on
CTO
requirements,
not
developer
requirements
and
that's
the
special
problem
right.
B
It
should
worry
the
crap
out
of
us
that
the
first
answer,
when
you
ask
how
most
cloud
foundry
teams
should
ship
things
isn't
cf
the
vast
majority
comments
rights.
All
of
them
take
away.
My
sequel
on
twelve
facts
are
apps
and
if
they're
not
that
should
be
the
fact
that
they
don't
want
to
use
CF
and
we're
not
like
we
should.
We
should
be
actually.
Why
isn't
the
answer
CFL
already
after
that
add
to
CF?
A
B
B
We
probably
have
to
either
each
or
two
I
maybe
have
to
have
a
feature
that
if
you
bind
service,
cube,
config
or
cube
or
service
account
or
whatever
you
get
the
ability
to
actually
hit
the
kubernetes
api
right
now
we
can
just
give
you
the
coop
config
as
an
environment
variable
if
you
bind
service
cuba,
nettings
right
and
because
a
few
of
our
things
gonna
need
to
keep
an
exit
like
networking,
but
other
than
that,
I
don't
I
mean
they're
all
twelve
factor
apps.
Why?
Wouldn't
they
all
go
through
CF.
F
I
think
it's
a
bootstrapping
problem
clearly
but
like
I
agree
with
Jules
that
the
majority
of
stuff
in
CF
should
be
actually
like
possible
as
a
12
factor
app,
so
it
might
be
worth
looking
at
how
we
could
like
do
a
minimal
bootstrap
and
from
that
point
forward
nobody
should
care
about.
Writing
you
moles
about
kubernetes
deployments
right,
but.
B
F
A
C
E
B
I
mean
right
now:
Rini
is
just
the
cool
thing
right
and
I
would
think,
for
example,
like
networking
the
stuff
that
never
seemed
were
producing,
isn't
in
a
really
right
now
another
thing
they're
going
to
remove
everything
from
irini,
so
I
don't
see
why
they
couldn't
see
it
push
what
they're
producing
true
sample.
Similarly,
I
mean
you
AAA
and
I
would
think
they
could
I
don't
mean.
C
B
Exactly
that
exactly
that,
it's
like
maybe
a
core
certain
features
that
we
can
use
and
a
lot
of
them
will
break,
and
maybe
we
do
need
to
add
some
stuff
to
make
this
possible.
It
just
seems
it
just
seems
like
super
and
healthy
like
given.
We
have
a
lack
of
good
feedback
from
developers
and
we
have
a
load
of
developers
like
what
better
way
to
get
empathy
for
people's
problems
using
CS
and
what
they
would
want
to
change
in
human.
I
How
I
wanna
rainy,
but
I
kind
of
not
agree
is
that
on
called
fangy
components,
we
can
push
a
bit
more
on
developers
and
have
more
limitations
for
developers.
Then
we
should
ask
from
the
community
from
the
users
and
users.
Yes
like
we
can
ask
people,
please
have
different
I,
say
health
greatness,
endpoint
and
lightness
endpoint
right
since
we
can
say
I
things
like
this,
then
I
think
from
the
cloud
for
the
community
from
co-founder
component.
Also,
we
don't
need
to
make
people
push.
I
The
source
code
will
push
docker
images,
so
we
can
require
selves
and
docker
images
level
and
if
we
just
get
some
tiny
part
from
Marines
its
get
scheduling
that
might
be
helpful
and
I
sure.
How
is
it
how
important
it
is
from
the
area
nearby
clock,
but
yeah
I
I
am
busy
on
this
concern
said
for
me,
it
feels
that
we
need
something
like
a
father
plus
plus
to
deploy
call
for
the
components
and
then
we'll
have
to
migrate
and
usual.
B
I
Like
from
this
irony
and
current
way
to
a
stateful
sets
and
some
by
the
way
why
we
deploy
applications
because
I
don't
think
they
will
be
one-to-one
mapping
from
the
application,
so
we
are
love
to
see
a
push
right
now
and
how
components
for
the
applications
at
Utica
badges
should
be
deployed.
You
need
more
things
for
composite
of
local
artists
at
least
right
now.
I
You
probably
need
to
specify
that
we
can
policy
already
and
we
shouldn't
wait
for
irony
or
to
all
call
for
the
components
to
be
worried
enough
to
ready
enough
before
we
same
it's
good
enough
for
developers,
we,
you
might
go
with
some
other
way
and
then
at
some
point
your
dream
might
come
true.
That
we'll
do
see
a
full
component
at
fill
navel
throughout
in
ever
since,
but
we
do
far
away
from
Sur
and
maybe.
B
Maybe
it's
the
way
Cephas
first
bootstraps
me,
but
like
this
minute,
like
c99,
compiled
like
a
minimal,
see
that
actually
compiles
with
right.
Maybe
there's
just
like
a
CF
CLI,
it's
just
like
a
bash
script.
You
say:
CF
way
only
has
two
CF
push,
but
it
takes
the
same
input
so
usually
would
do
and
you
use
it
the
same
way,
but
it
does
it
directly
against
arena
or
whatever
I,
don't
know
what,
like
you
know
whatever
again.
B
B
Right
I
mean
maybe
single
user
mode,
like
maybe
it's
the
interface
to
arre
me,
which
I
think
the
problem
should
be
in
this
another
topic,
I've
been
at
the
interface
between
CC
and
arena,
was
just
a
CID.
Cc
was
creating
then
and
CC,
just
added,
like
you
AAA
and
all
the
nice
stuff
and
all
the
other
things
around
it.
Then
then
the
minimal
CC,
if
you
see
what
I
mean,
would
just
create
a
CID.
C
B
C
Know
Zack
has
been
interested
in
promoting
this
line
of
dogfooding
and
reasoning
on
the
component
vendors
as
well,
so
the
I
think
I
might
have
mentioned
last
time
that
the
campi
team
has
been
doing
a
couple
different
tracks
of
work
around
both
container
izing,
Cloud,
Control
or
more
directly
and
starting
to
produce
a
help
chart
for
that,
at
least
in
API
mode.
So
one
they're
also
interested
in
integrating
that
into
a
larger
CF
distribution
for
their
own
CI
integration.
But
then
there
they've
been
interested
in
building
that
with
cloud
native
build
packs.
C
The
sticking
point
for
them
is
that
there's
no
I
don't
think
there's
an
officially
supported
or
integrated
Ruby
cloud
native
built
back,
yet
that's
somewhere
on
the
CNB
roadmap
and
I
think
there's
like
some
early
version
of
it,
but
it
like.
If
you
look
at
the
builders
that
come
with
kpac,
it's
not
listed
in
that
set
yeah
some
think
it's
completely
ready.
Yet.
C
I
mean
along
those
lines,
type
controllers
also
they
have
done
some
exploratory
work,
see
you
integrate
with
KPAC
itself
to
run
staging
tasks,
not
that
that
would
be
required
at
this
level
for
the
kind
of
dogfooding
that
we're
talking
about
for
component
development,
but
it's
at
least
on
their
radar
and
they're
kind
of
figuring
out
where
those
seams
are-
and
you
know
that
could
start
dragging
in
the
k-pax
resources,
as
well
as
the
core
of
that
staging
tasks.
Infrastructure.
A
H
Don't
have
any
pre-existing
strong
opinions
about
that
I
really
like
this
idea
of
dogfooding
I.
Think
if
we
ship
source
code
it
helps
us
all
with
a
number
of
needs,
and
so
perhaps
there's
this
is
design
pressure
on
whatever
we're
building
here
to
be
able
to
consume
and
deploy
from
source
third-party
dependencies
like
SDO,
for
example,
it's
an
interesting
thought.
J
Which
could
run
on
kubernetes?
You
don't
want
to
make
it
dependent
on
CF
right,
because
what
we
really
want
is
more
communities
and
we
want
to
keep
the
user
experience
from
Cloud
Foundry.
So
I
understand
the
dogfooding
part,
because
it
puts
a
lot
of
pressure,
but
they
think
the
priority
should
be
to
run
on
one
kubernetes
and
not
to
run
on
top
foundry
I.
I
H
Like
we're
not
saying
that
it's
gonna
require
the
whole
Cloud
Foundry,
there
I
think
we're
talking
about
using
some
of
the
core
components
that
Cloud
Foundry
uses
as
a
way
of
like
deploying
some
of
these
essential
system,
components
right
so
like.
If
you
wanted
to
build
your
container
image
up
some
other
way,
then
you
wouldn't
need
kpac
right.
B
And
and
if
it's
the
case
that
adding
Cloud
Foundry
to
occlude
minetti's
is
too
heavy
way
away
to
people
to
install
all
the
other
stuff,
that
tells
us
something
about
CF
being
too
heavy
weight
for
people
right.
If
we
wouldn't
use
it
to
deploy
our
stuff.
Why
would
anyone
else
not
have
that
problem
right?
If
I
mean
everyone
else,
has
the
same
problem
of
one
thing
in
three
key
vanities
makes
it,
but
not
wanting
to
build
a
thing.
There's
too
tightly
integrated
to
Cloud
Foundry
right.
B
We
have
to
make
Cloud
Foundry
in
the
same
way
that
helmet
or
anything
like
that,
something
that
people
are
willing
to
install
an
sto
and
everything
like
that.
Something
that's
lightweight
enough
that
people
and
installed
it
on
their
clusters
and
that
they're
happy
to
use
as
an
option
to
the
post,
though
I,
think
and
and
and
to
the
extent
that
it
isn't
I
mean
it
means
to
be
well.
J
B
E
Think
do
we
have
some
I
know
that
I've
heard
anecdotes
of
Cloud
Foundry
being
difficult
to
install
because
it
requires
Bosch
knowledge
and
you
need
a
really
really
large
amount
of
vm's.
But
if
you
have
a
couple
of
helm
installs
even
now,
if
it
says
CF,
how
do
you
know
that
that's
too
heavy
weight
or
that
it's.
B
E
That
consuming
it
from
a
developer
standpoint
is,
can
be
quite
tough,
especially
if
you
want
to
switch
stuff
out
and
especially
for
SCF,
because
it's
going
through
the
whole
bar
thing,
but
as
a
day,
zero
experience
for
a
customer.
It's
really
easy,
really
easy.
It's
like
to
Helmand
stalls
and
you
got
a
CF
15
minutes
later.
E
G
B
It
doesn't
have
to
be
perfect,
but
I'd
say
it
needs
to
pass
and
the
dog
free
test.
That's
all
right,
I'm,
not
saying
these
bees
using
the
world,
but
I'm
saying
if
we
are
willing
to
use
it,
then
it's
too
heavy
right.
If
it's,
if
it's,
if
it's,
if
it's
you
know
heavy,
but
we're
willing
to
use
it
cool.
J
B
So,
given
that
we
use
in
kpac,
we
should
make
sure
that
it's
possible
for
people
to
just
directly
deploy
their
things
simply
without
goes
to
the
CF
stuff,
because
that
what
you're
saying
is
a
requirement
and
a
lot
of
our
teams,
a
lot
of
our
developers
will,
like
the
networking
team
I
think,
should
be
producing
something
and
garden.
Cerny
is
going
to
try
and
produce
something
that
can
be
usefully
used
outside
CF
right
people
should
be
asked
just
adopts
our
pods
security
policies
if
they
want
to.
B
F
B
E
B
Mean
probably
the
case,
in
fact
I
mean
no,
it's
not
only
probably
the
case
right.
It
is
the
case
that
we
will
also
need
a
unified
CF
and
a
way
to
deploy
old
I
would
hope
that
days
today,
developers
just
use
CF
to
deploy
new
versions
and
have
that
developed
productivity
and
again,
in
the
same
way
that
we
will
have
a
requirement
that,
as
well
as
doing
that,
we're
gonna
wanna
package
up
the
whole
thing
and
be
able
to
install
in
places
as
a
hound
chart.
B
Every
CF
uses
no
intimacy
app
users
will
have
the
same
requirement
and
we're
gonna
need
to
solve
that
requirement
from
the
Maya
I.
Think
one
of
the
problems
CF
has
is
that
it
doesn't
have
a
good
answer
to
that
story
of
how
do
I
package
up
all
my
CF
Micra
service?
I
got
this.
It
was
really
fast
to
develop
them
and
it
was
low
barrier.
My
time
to
first
win
was
really
low,
but
how
do
I
package
the
it
all
up
in
a
way
that
makes
sense
and
put
it
in
the
Hound
chart?
D
D
B
That
seems
like
a
problem.
I
I
mean
I.
Think
it's
a
common
problem,
I
think
one
of
the
reasons
people
really
like
docker
and
helm,
and
all
these
things
is
because,
as
well
as
it
being
consumable
by
them,
they
can
give
it
someone
else
they
can
listen
on
their
class
when
it'll
work.
The
same
way.
I
know.
B
E
It's
important
for
each
project
to
be
able
to
stand
on
its
own,
not
because
there
might
be
third
parties
that
are
interested
in
consuming
them,
but
because
I
think
should
have
clearer
API
contract
and
yeah
loosely
coupled
architecture.
I
do
remember
the
latest
in
food
I
think
because
I
remember
thinking
that
having
a
contract
between
the
car,
controller
and
yeah
good
idea,
but
it
didn't
take
off
and
and
I
think
it
also
required
a
lot
of
resources
to
you
know
to
maintain
the
project
so
yeah
I.
E
Think
if
you
know,
if
you
have
all
the
pieces
that
are
good
and
you
also
integrate
them
very
well,
no
one
will
care
about
the
individuals
anymore
there.
They
will
always
want
the
entire
thing.
The
whole
cake
like
if
I,
if
we
have
a
really
good
logging
solution-
and
we
have
a
really
good
scheduling
system-
and
we
have
a
good
thing-
it
will
always
take
the
whole
pie.
B
B
Say
the
finger
so
I
think
about
lattice
lattice,
didn't
work,
I!
Think
we
took
the
wrong
message
from
that
I'm,
honest
I,
think
lattice
was
multi-touch
phones
before
the
iPhone
and
and
we
had
a
multi-touch
phone.
They
didn't
work
and
we
took
the
message
that
there's
no
value
in
multi-touch
funds
and
the
actual
answer
was
a
lot
such
phonesat
and
we
hadn't
done
it
right.
E
B
B
E
B
E
B
D
B
So
they
should
prioritize,
it
I
agree
as
far
as
that.
They
should
prioritize
it.
One
strengths
and
weakness
of
the
way
we
build
software
is,
if
we
don't
intentionally
do
something
we
are
very
good
at
not
doing
it.
You
see
what
I
mean
and
the
whole
the
whole
process
is
based
around
not
doing
anything.
B
You
didn't
intensity
so
that
the
only
way
we'll
ever
spend
any
efforts
on
even
being
possible
that
any
of
these
things
will
be
independent,
consumable
is
intentionally,
which
is
how
it
should
think
right,
but
it
also
means
that
if
we
don't
do
that,
we
won't
do
that.
What
I
would
say
is
it
is
it's
not
the
main
goal
of
any
project
but
I?
Think
every
project
should
be
spending
some
effort,
not
all
laughter.
That's
a
matter
on.
B
There
being
the
possibility
of
an
ecosystem
around
I
think
because
otherwise
they
all
get
replaced
and
then
we
play
catch-up
and
that
seemed
it's
just
strategically
sits
just
with
10%
more
efforts,
there's
a
very
good
chance
of
a
very
good
return.
I
think
it's
worth
more
in
those
tights
and
a
few
projects,
I
I.
H
See
it
like
from
the
networking
standpoint,
what
we're
doing
is
barely
developing
and
anything
new
we're
just
integrating
existing
stuff
like
we're
gonna
integrate
this
existing
open
sources.
The
other
thing
into
this
existing
ireenie
thing
and
just
try
to
a
stirring
it
together.
So
it
works
and
expose
as
many
servers
much
features
as
we
can
in
a
reasonable
way,
and
so
I
don't
see
that
integration
itself
as
being
a
whole
lot
of
value
on
its
own
I,
see
us
working
in
service
of
this
larger
goal
of
the
whole
thing
being
useful
as
a
nexus.
B
Right
you,
a
a
it's
a
big
example.
I
think
kpac
are
doing
the
exact
right
thing
right.
The
fact
that
bill
packs
were
only
consumable
inside
before
you
know
it
was
the
right
thing
for
CF,
but
had
bill
packets
earlier
spend
more
time
making
that
something
that
could
have
had
its
own
thing
around
them.
I
think
that
would
have
been
valuable,
never
King,
III,
don't
know
yeah
I,
don't
have
enough
context,
really
comment.
B
B
Money
right
all
right,
then
now
an
infection
about
it.
I
think
that
we
could
underestimate
the
value
of
a
high
level
API
to
some
of
these
things
that
simplify
stuff
but
I,
don't
know
that
much
about
the
networking
stuff.
So
I
don't
really
know,
but
it
feels
to
me,
like
those
plausible
value
in
something
that
makes
it
easy
to
configure
the
networking
stuff,
because
the
networking
is
that
it's
real
hard,
if
you
haven't
been
close
to
it
for
a
while.
B
Well
I
mean
I
would
say
if
we
split
that
up
right,
like
I
would
like
to
be
able
to
consume
a
public
API.
That
lets
me
associate
routes
with
apps
and
services
without
necessarily
needing
the
whole
rest
of
CF
and
I.
Think
there
would
be
quite
a
good
use,
case,
least,
to
me
right,
I
think,
there's
lots
of
people
with
their
own
stuff
deployed
the
stuff
that
they're
happy
with
that.
Would
like
this
nice,
easy
API
that
does
the
stuff
they're
networking
that
so
that
makes
it
easy
to
do.
Contain
a
networking
configures.
D
D
We
started
out
on
release
integration.
We
all
want
ICF,
we
don't
want
to
know
how
to
deploy
CF
on
kubernetes,
okay-
and
my
takeaways
here
are,
it
sounds
like
Vlad
is
working
on
something
called
CF
q,
where
q
CF
and
they
have
pipelines
and
I,
would
expect
that
anybody
who,
from
pivotal
or
otherwise
wants
to
contribute
to
our
dream
of
having
some
one
place
and
one
artifact
of
way
of
easy
way
of
deploying
CFM
kubernetes
might
contribute
to
collaborate
with
or
learn
from
what
glad
is
doing.
C
Those
are
all
designed
to
integrate
with
I
mean
those
feet
into
CF
operator
right
and
so
I
mean
as
we're
talking
about
like
the
heavy
witness
of
the
system.
If
that's
potentially
part
of
the
concern
is
like
well
how
much
additional
machinery
is
there
that
is
required
to
run
what
we
think
of
as
CF?
Is
this
package
solution?
So
I
guess
one
question
that
I
that
was
coming
to
mind
to
me
as
we're
talking
about
like
okay?
How
are
we
going
to
interleave
lower-level
kubernetes
resources?
C
J
Think
you
still
need
an
operator
for
lifecycle
management.
So
when
some
component
changes
you
want
to
restart
other
components
and
well
I
think
also
the
installation.
The
initial
deployment
is
complicated
enough
so
that
you
don't
want
to
do
it
with
the
with
the
hand
shots
because
of
the
dependencies
and
my
figuration
and
all
of
it.
G
F
Yes,
I
think,
like
many
of
the
things
sorry
I
think
like
many
of
the
things
that
we
currently
see
in
the
bottle
asus
in
CF
deployment
out
the
air
because
of
historic
reasons
right,
our
boss
gave
you
an
interface
to
do
certain
things
in
a
certain
way.
It
has
like
certain
lifecycle.
Hooks
people
tend
to
put
stuff
in
there.
You
have
a
way
to
like
block
until
something
happens,
so
you
do
it.
You
have
a
way
to
have
an
enforced
ordering
in
the
update
like
during
an
update
in
your
manifest.
So
you
do
it.
F
H
Exactly
like
suppose,
we
took
the
mindset
of
kubernetes
native
developer,
who
says
we're
not
going
to
put
an
explicit
dependency
ordering
instead
we're
going
to
expose
a
status
on
our
resource.
That
indicates
whether
we're
healthy
or
not,
and
if
not,
the
reason
why
we're
not
healthy,
and
then
we
can
just
deploy
everything
at
once
and
wait
for
it
to
all,
become
healthy
as
sort
of
components
come
up
and
resolve
themselves
like
if
we
threw
away
all
that
legacy
garbage.
Wouldn't
we
end
up
with
a
simpler
and
easier
to
deploy
system,
yeah.
F
H
F
F
Wasn't
advocating
for
hey,
we
need
to
now
find
a
way
to
build
lots
of
custom
code
to
replicate
this
exact
functionality.
My
opinion
is
exactly
be
the
opposite.
It's
it's
really
that
I
think
we
can't
throw
away
like
most
of
this
stuff
to
some
extent,
I.
Think,
like
all
of
those
scripts,
are
there
like
for
a
reason
right.
People
figured
out
a
way
how
to
operate
their
component
in
like
one
situation
or
another.
Okay,
something
needs
to
be
done
before
you
shutdown,
stuff.
Okay,
there
is
a
drain
script.
Let
me
put
stuff
in
there.
F
Obviously
you
would
do
things
differently
when
you're
creating
a
kubernetes
native
application,
and
that
makes
perfect
sense
right.
I
don't
want
to
like
advocate
for
hey
and
it's
all
build
now.
Custom
go
code
to
replicate
whatever
the
boss
directory
used
to
do
during
an
update.
That's
like
the
exact
exact
opposite
of
what
I
would
like
to
achieve.
I
think
like
one.
E
Thing
that
I
would
add
to
that,
like
all
these
runtimes
features
of
Bosch
are
very
useful
and
they
have
dictated
the
design
of
components
but
I
think
one
main
aspect
of
Bosch
that
we
don't
have
a
counterpart
even
now
with
the
operator
is
the
release
integration
piece
exactly
what
we're
talking
about,
which
is
CF
deployment.
It
gives
you
this
way
to
bring
out
all
the
software
and
make
sure
that
it
works
every
time
and,
as
I
said,
I
haven't
seen
yet.
E
So,
if
anyone
knows
please,
let
me
know
of
a
system
on
kubernetes,
that's
as
complex
as
Cloud
Foundry
being
deployed
with
any
mechanism,
whether
it's
helm,
charts
or
whatever
I
haven't
seen
one
project
containing
so
much
software
being
deployed
with
one
helm,
install
and
having
a
day
zero
and
the
day
one
and
at
the
end
and
secret
rotation-
and
you
know,
configuration
and
cascading
updates
to
get
the
system
up
to
date
and
so
on.
I,
don't
think,
there's
anything
out
there
that
does
that
I.
A
A
There
was
this
like
different
views
on
the
topic
where
I
believe,
like
we
don't
have
kind
of
enough
common
context
to
basically
say
how
to
take
that
part
forward,
so
how
to
take,
like
those
individual
hand,
charts
and
it's
and
what's
needed
to
kind
of
get
them
together
to
something
that
looks
like
see
if
deployment
today
and
if,
if
and
how
that
that
is
required
to
kind
of
integrate
offices.
So
maybe
that's
then
kind
of
the
starting
point
for
for
the
next
discussion
is
as
I
believe.
A
A
Having
said
that,
thank
you
very
much
also
to
an
anonymous
author,
I'm
kind
of
assuming
that
his
sworn
are
taking
notes
there,
but
I
was
like
glancing
at
at
the
Google
Doc
and
it
actually
accrued
accurately
reflects
the
discussion
here.
It's
kind
of
hard
to
follow
the
discussion
and
take
notes
at
the
same
time,
I
think.