►
From YouTube: Cloud Foundry for Kubernetes Monthly SIG [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).
B
B
B
All
right
cool
thanks
should
we
so
basically
we
have
two
options.
Now
we
can
either
if
Shannon
knows.
If
and
what
happened
at
inception
for
the
railing
team,
we
could
talk
about
this
or
switch
to
like
the
next
topic
within
the
next
most
votes,
which
was
CRD
all
the
things
proposed
by
Jules,
who
also
isn't
here,
but
a
few
of
the
Iranian
people
are
here.
So
potentially
they
want
to
talk
about
it.
B
B
D
B
E
Hi
so
I
think
the
question
here
is
like:
should
we
do
everything
CRD
based
or
not
like?
Should
the
clock,
controller
or
copy
create
CR,
DS
and
rainy
is
acting
as
a
controller
based
on
the
CR
DS
to
deploy
applications,
and
if
we
should
go
that
way
or
not
like
pros
cons,
I
think
that
should
be
the
discussion.
That
area.
B
E
A
F
A
G
E
G
D
Expectation
is
that
if
a
statement
was
made
no
database,
it
it's
referring
to
the
the
platform
system,
components
not
needing
their
own
database,
but
leveraging
the
kate's
API
for
state
and
whatever
data
store
is
backing.
The
kate's
API
represents
a
database
in
a
way,
but
the
statement
is
that
cloud
controller
doesn't
have
its
own
database
as
distinct
from
the
kate's
API
datastore.
Exactly.
E
D
G
F
F
A
Wasn't
thinking
about
it
from
from
like,
let's
replace
a
big
database,
I
was
thinking
higher
level
like
you,
don't
want
to
have
two
sources
of
truth
right
when
you
create
a
route
in
using
the
controller,
you
store
that
information
somewhere
in
the
database.
If
you
then
use
this
tea
or
something
else
to
create
a
route
in
the
kubernetes
world,
you've
duplicated
that
information,
and
you
always
need
to
keep
them
in
sync.
A
D
B
E
Like
so
all
we
did
and
we
were
able
to
finish
it
in
one
day,
but
what
we
did
is
we
just
kept
the
cloud
controller
code
and
it
just
created
CR
DS.
We
just
took
the
Jason
and
just
created
us
here,
D
with
with
a
application
Jason,
and
we
created
this
Rini
controller.
There
was
just
like
listening
to
it
and
like
yeah
watching
them
and
whenever
there
was
popping
up
something
we
just
read
it
and
created
the
app
on
kubernetes.
Alright,.
D
E
Main
work
was
basically
on
the
arena
side.
We
didn't
have
to
do
much
on
the
cloud
controller
side.
It
was
just
like
a
post
to
create
a
C
or
D
on
kubernetes
we
didn't
like
I,
went
down
the
way
to
replace
databases
or
anything.
We
still
got
information
from
the
clock,
control
the
database
from
the
clock,
controller
and
just
create
see
our
DS.
A
E
F
Whatever
to
say
is
what
I
see
with
CI
Jesus
that
it's
very
hard
to
create
them
and
it's
slow
to
carry
them
everything's
a
synchronous,
you
don't
have
any
locks
unless
you
implement
them
yourselves.
So
I
think
this
is
quite
a
challenge
to
make
something
which
is
stable
and
robust.
I
know:
I
I
see
how
how
some
objects
belong
to
the
resource
right.
Some
objects
belonging
to
the
in
the
cluster
with
their
counterparts,
but
I
think
there
is
value
in
even
in
duplicating
data
and
syncing,
when
you're
ready,
I
have
I
have.
H
You
know
the
Empress
got
no
clothes.
Think
about
see
all
these
so
they're.
Definitely
terrible
and
no
one's
allowed
to
point
out,
but
I
mean
we.
The
ship
has
sailed.
If
you
see
what
I
mean
like
the
whole
architecture
is
based
on
cles
anyway,
right
like
under
the
covers
it's
all
that
stuff
anyway,
and
there
are
advantages
to
them
as
well
as
disadvantages.
H
You
know
there
is
completely
true
right,
like
they're,
SED,
isn't
very
scalable,
yet
there's
some
slowdowns
as
soon
as
you
do
anything
async,
which
you
eventually
do.
You
have
to
worry
about
the
entirety
of
distributed
systems
to
make
it
work,
but
pretty
much
all
the
services
are
going
to
be
using
operators,
we're
probably
going
to
start
using
k-
at
some
point
to
do
stuff.
H
H
H
H
H
So
and
this
deadline
is
it's
a
YAG
meeting?
If
you
see
what
I
mean
right
like
it's
like
right
now,
the
reality
is
berries
of
connect.
Right,
like
that's,
that's
the
word
weigh-in
and
we
could
try
and
abstract
that
away
and
the
initial
kind
of
thought.
Okay
I
was
too
abstract
that
way,
but
the
more
I
think
about
it.
The
more
after
my
do.
We
want
to
abstract
away
the
kubernetes
or
do
we
want
to
abstract
away
being
sad,
kubernetes
resources
that
get
created?
If
you
see
what
I
mean
I
guess
it
can
ativ?
H
A
Not
thinking
about
it
necessarily
from
a
technical
perspective
or
I
think
the
technical
perspective
is
very
important
too,
but
from
a
product
or
project
standpoint.
If,
if
someone
has
like,
maybe
some
data
point
on
whether
we
expect
the
entire
planet
to
be
kubernetes,
because
you
would
talk
about
hundreds
of
communities,
100%
right.
A
H
So
so
you
would
tie
it
to
communities
only
in
the
sense
that
it
was
previously
tied
to
Diego.
If
you
see
what
I
mean
it
would
always
be
possible
just
like
when
it
was
previously
tied
to
the
a
go
for
us
to
introduce
any
schedule,
including
X's,
because
code
is
code
right,
we
can
always
change
it,
and
the
question
is
whether,
with
whether
we
need
to
do
that
effort
now,
I
maintain
that
abstraction.
H
Now,
if
you
see
what
I
mean
or
if
we
just
need
to
not
block
ourselves,
introducing
abstraction
the
future
and
and
I
mean
like
I
said,
I
mean
it's
the
if
you
ain't
gonna,
neither
think
in
my
head
right
like
it's.
We
don't
need
to
architects
for
that
right
now.
We
just
need
to
keep
the
code
non
sucky
so
say.
If,
if
and
when
we
all
wake
up
from
our
dream
in
realize
that
kubernetes
is
awful
and
something
else
is
very
cool
thing.
H
D
Jose
I'm
interpreting
that
the
the
question
not
as
which
scheduler
at
Cloud
Foundry
is
using,
but
the
the
the
safe
fear
that
by
refactoring
all
the
system,
components
to
interact
with
a
weather
data
store
by
a
particular
API,
the
kubernetes
api
that
we're
Cloud
Foundry
to
a
particular
infrastructure,
and
it
means
that
Cloud
Foundry
couldn't
be
as
easily
operated
on
VMs
or
bare
metal
or
some
other
no
destination
infrastructure
is
did
I
interpret
that
correctly.
But.
A
A
H
Well
I
mean,
as
far
as
I
know,
there's
no
member
of
the
foundation
that
doesn't
have
a
keeping
at
these
offerings
right
and
as
far
as
I
know,
there's
no
member
the
foundation
that
doesn't
have
a
kubernetes
offering
that
runs
everywhere.
They
run
their
cloud.
Foundry
offering
I
could
be
wrong
about
that.
But
I
think
that's
depends.
D
Or
what
it's
worth
as
a
product
person
from
pivotal
pivotal,
is
investing
heavily
in
in
participating
with
the
CF
f,
to
move
cloud
foundry
to
run
on
kubernetes
in
response
to
feedback
from
our
commercial
customers
who
have
given
us
feedback
over
the
last
year
or
two
that
they
are
concerned
that
cloud
foundry,
if
it
doesn't
run
on
kubernetes
I,
may
be
falling
behind
in
its
relevance
right
now.
This
feedback
may
be
based
on
fear,
uncertainty
and
doubt
that
our
customers
themselves
have
that
they
are
attracted
to
a
new
shiny
thing
that
is
kubernetes
dr.
dr.
D
It's
real
and
it
seems
that
kubernetes
is
becoming
ubiquitous
infrastructure
yeah.
As
much
as
it
concerns
me,
it
feels
like
we're
refactoring,
a
very
functional
yapper
platform
that
delivers
real
business
outcomes
and
rewriting
it
for
an
immature
scheduler,
and
it
comes
with
a
great
deal
of
risk
yeah.
It
feels
like
we're
we're
it's
inevitable
that
these
are
all
right.
We
have
to
solve,
because
this
is
where
the
market
is
going.
H
Yes,
I
for
what
it's
worth
you
you
could
be
speaking
straight
out
of
my
brain,
like
I,
yeah
I,
said
the
exact
same
concerns
and
many
of
the
same
instincts
yeah.
A
lot
of
kubernetes
strikes
me
as
horribly
overcomplicated
and
pretty
pretty
crazily
heavy
and
very
hard
to
abstract
your
way,
out
of
which
is
the
biggest
concern
like
it's.
Then
it's
idiomatically
necessary
to
couple
yourself
to
it.
If
you
see
what
I
mean,
unlike
unlike
most
things,
kubernetes
seems
to
value
a
lack
of
abstraction,
it's
like
Ruby
anti-patterns
as
a
service.
H
If
anything
you
shouldn't
do
with
already
on
Rails,
is
what
you
should
do
in
keeping
Etsy
like
exposure,
a
data
model
as
your
API
that
scale
and
but
it's
exactly
what
you
said
Shannon
right
like
did.
This
is
where
people
are
and
I
think
I
think
there's
huge
value.
We
can
often
simplifying
it,
but
I
think
we
have
to.
You
know,
meet
people
where
they
are
and
if
you
see
what
I
mean
yeah.
C
I
think
different
conversation,
though,
is
like.
Would
we
at
a
certain
point
in
time
kind
of
let
the
kubernetes
api
kind
of
shine
through
the
Cloud
Foundry
user
experience
and
I
think
I've
written
that
in
one
of
the
documents
initially
I
thought
this
would
be
a
good
idea.
I
guess
I'm,
not
so
sure
anymore
about
that
I'm.
H
Excellent
I
mean
I
mean
that
what
it's
worth
like
so
I
when
I
go
back
and
forth
the
thing
that
keeps
bringing
me
back
to
you.
It's
a
good
idea
to
hide.
It
is
what
I'm
hacking
on
my
own
could
deal
if
that
most
bad
time,
I
use
stuff
like
plate
right,
like
I
use,
stuff
like
that
and
glitches,
is
not
better.
If
you,
let
me
start
messing
with
you,
vignetting
right
like
it,
you
know,
I
do
think
that
there
is
a
high
level
and
low
key.
H
Luckily,
actually,
like
I
was
using
rocky
the
other
day,
it's
so
much
easier
to
stand
up
a
quick
graphic.
You
labr
and
Haraki,
then
spinning
up
the
kubernetes
cluster
right.
You
couldn't
leave
a
space,
it's
like
one
kick
deploy
to
Heroku
and
it's
deployed
with
the
database
attached,
and
that
is
a
lot
easier
than
applying
a
ham
chart
and
notes
and
stuff.
H
D
Also
I've
been
developing
an
opinion
based
on
repeated
signal
that
it
may
be
valuable
to
have
an
expert
mode
if
you
will,
for
initially
for
a
privileged
user
like
a
form,
engineer
or
security
stakeholder
who
can
interface
with
lower-level
api's,
but
over
time.
If
we
could
solve
access
control
for
resources,
kubernetes
native
resources,
then
we
could
expose
this
to
more
users.
You
know,
for
the
time
being,
our
abstraction
solves
access
controlled
for
us,
because
hootenannies
itself
hasn't
solved
this
problem.
D
I
D
I
H
H
Why
I'm
kind
of
motivated
towards
the
CID
thing
they're,
like
I,
think
we
still
want
Cappy
on
top
because
of
because
it
gives
us
Roberts
accent
role,
because
it
gives
us
the
thoughts
or
API
because
it
gives
us
the
you
know:
I
don't
want
to
give
people
expert
mode
right
now
and
maybe
not
ever
but
I
think
starting
to
put
in
those
CR
DS
one.
It's
the
expert
mode.
It
lets
people
start
experimenting
right,
notes,
less
people
start
building
higher
level
things
on
top
of
stuff,
so
I
think
that's
tough
right!
Now.
H
H
F
The
I
remember
there
was
like
manual
for
posh
how
you
could
get
into
the
PostgreSQL
database
to
delete
locks
and
I
would
just
wanted
to
say
what
would
I
see
with
the
operator
is
that
you
wouldn't
need
that
manual.
All
our
CID
suggests
there
and
we
actually
think
we
documented,
which
one
is
safe
to
edit,
because
the
operators
on
the
kubernetes
operators
they
have
this
the
humans
right.
F
H
Yeah
I'll,
say
out
loud
I.
Think
operators
are
operators
scare,
the
crap
on
me.
You
know
it
was
just
like
it's
the
opposite
of
declarative.
It's
like
imperative
logic,
completely
custom
to
reach
thing,
but
when
it
blows
up
halfway,
I,
no
idea
why
anyone
supposed
to
do
I
get
into
a
bad
state
in
your
operator.
I
have
no
idea
what's
supposed
to
happen,
and
it's
it.
F
Helps
me
to
think
of
the
whole
thing
as
an
event-driven
boss:
it's
not
so
different
than
been
using
RabbitMQ
right.
You
sent
events
out
there
and
you
get
stuff
back
and
you
these
operators
sitting
on
that
bus
and
they
do
something
and
often
you
it's
really
hard
to
see
what
the
state
machine
is
going
to
do.
Yeah
exactly.
F
H
B
So
this
a
real
thing
that
happens:
I
mean
our
people
like,
even
if
you're,
not
using
CF,
if
you're,
not
using
K
native
right
you're,
just
building
your
stuff
with
plain
Kade's.
Is
this
a
thing
that
people
use
cubes
ETL
to
modify
their
parts
which
actually
maintained
by
I
a
deployment
or
a
stateful
set
and
then
are
like
surprised
at
what
happens?
Is
this
a
thing
because
my
feeling
is
that
it's
not
right,
I
think
at
some
point
people
know
their
API.
B
They
know
their
interface
and,
of
course,
like
there
is
stuff
happening
in
between
and
like
you
can
look
at
it
if
you're
interested
in
it,
you
can.
You
can
understand
it
if
you
want
to
but
like
mod
making
modifications
to
it
and
then
saying
like
whoops
I,
didn't
know
it's
like
modifying
it's
in
a
bit
service
and
then
saying,
like
oh
I,
didn't
know,
I
I
wasn't
allowed
to
change
it.
I'm.
Sorry.
A
H
Yeah
yeah
just
yeah
anyway.
This
is
just
like
the
populace
in
the
back
of
my
head
like
and
like
that
feels
all
very
well
in
the
Golden
Path
that
you
know,
yeah
ACB
cluster
fails
quorum
only
trying
to
get
it
back
in
back
in
flight
and
any
validation
hook.
Doesn't
let
you
touch
anything
in
that
in
in
there
and
now
I'm
debugging,
the
code
of
the
operator
did
get
it
back
like
well
anyway.
H
That's
that's
all
like
I.
Don't
think,
there's
anything
to
do
with
CF
this
time
testing.
That's
it
completely
completely
orthogonal
different
thing
about
the
overuse
of
operators.
I
think
I
want
to
write
like
a
paper
like
operate.
This
considered
harmful
I.
Give
you
either
like
to
use
a
staple
set
I
figure
out
how
to
use
it
table
set
unless
you're
like
one
of
five
people
yeah
anyway,
I
feel
that's
a
complete
perversion,
divergence
of
where
we
actually
are
like
I.
H
Think
two
CS,
like
the
real
question,
is:
is
there
a
value
in
having
a
CID
based
API,
one
underneath
Cappy,
and
maybe
one
between
Cappy
and
Erie
me,
like
maybe
they're
the
same?
Maybe
they're,
not
I.
Don't
know,
but
is
there
a
value
to
trying
to
create
C
IDs
for
those
which
users
could
use
if
they
wanted
to
or
go
directly
to
it,
that
via
the
API.
B
You
know
I
think
there
is,
there
is
volume
it
so
like
in
just
I
mean
you
already
had
the
value
proposition
in
your
question.
Right
people
can
use
this
as.
H
B
And
use
the
CF
CLI
or
they
can
decide
to
like
just
use,
cube,
CT
I'll
apply
to
do
their
thing
and
and
use
I,
don't
know
their
favorite
get
ups
operator
to
write.
Those
CRD
is
not
in
Iraq
with
this
year,
but
the
CLI
so
like
this
very
thing
specifically
is
alone
like
worth
looking
at
series,
at
least
to
me
right,
wean
copy
and
every
just
to
say,
like
hey.
This
is
optional
right.
B
C
C
On
the
aspect
of
good
idea
or
bad
idea,
but
what
I've
seen
in
the
kubernetes
community
is
definitely
this
thinking
around
like?
Can
we
strip
things
down
to
make
them
useful?
Also
standalone?
Yes,
like
you
see
that
in
K
native
versus
SCADA
and
like
what
Microsoft
is
doing
with
like
some
of
the
aspects
that
K
native
is
discovering
and
I'm,
not
sure
that
a.
H
I
think,
let's
see
like
the
CRV
would
make
both
erie
me
and
Cathy
independently.
So
in
a
way
I
mean
the
Cathy
was
something
that
was
just
here's
your
nice
easy
to
use.
Api
it's
your
hero,
clue!
You
know
here's
a
hurricane
and
for
your
kubernetes
right,
you
can
stand
this
up
and
it's
gonna
give
you
a
simplified
workflow
on
top
of
your
cuba
Nettie's
and
it
spits
out
these.
You
know
high-level
CR
DS,
which
you
can
map
on
to
the
back
end.
H
However,
you
like
right-
and
you
could
ship
that,
just
with
one
standard
implementation
of
that
CRV.
On
the
other
hand,
it
would
make
him
really
kind
of
repeatedly
useful
in
there.
It
really
just
becomes
like
a
top
level
thing
yeah,
an
axially,
essentially
that
can
spit
out
a
native
feeder,
whatever
you
like
from
CFR
definitions,
if
you
see
what
I
mean
I
think
I
think
the
immunity
wants
to
delete
ourselves
in
general
anyway,
can't
be
as
small
as
possible.
So
maybe
eventually
that
just
becomes
part
of
that
top
level.
H
H
H
You
know
in
it
was
what
we
were
saying
a
little
bit
earlier
and
I.
Don't
think
it
makes
sense
to
decouple
ourselves
from
kubernetes.
Excuse
me
a
stable
States,
but
I
do
think.
There's
some
value
in
decoupling
ourselves
from
what
particularly
animals
and
your
crazy,
including
at
ease,
because
there
is
still
quite
a
lot
of
you
know:
potential
innovation
wanna
be
able
to
take
advantage
of.
B
B
B
Like
what
what
does
this
mean
for
for
like
I,
don't
know
the
the
stuff
that
teams
like
Jules?
Does
it
change
anything
for
a
rainy
for
you
like?
Is
this
something
you
will
be
looking
at
soon
soonish?
My
my
feeling
is
without
like
work
done
and
copy,
whatever
you
will
be
doing
in
a
rainy,
won't
won't
help
budget
rent
copy
doesn't
start
writing
series.
B
B
H
B
H
Can
see
a
rescue
Chiara
is
happy,
but
I'll
be
honest.
It
seems
high
priority
as
long
as
it's
someone
else's
high
priority
and
I.
Don't
know
if
it's
our
highest
priority
if
our
highest
priority
is
still
probably
v3,
be
honest
and
say
I
think
people
do
need
to
be
three
API
to
work
between
the
two
more
than
then
you'd
see
all
these
cases.
H
D
H
It
does
seem
fair
to
say
that
no
one
in
this
fake,
which
I'm
going
to
take
these
in
the
ropes
and
stiff,
is
pushing
back
against
the
idea.
So
maybe
that's
an
interesting
data
point.
If
you
see
what
I
mean
they
were
all
kind
of
like
to
come
ther
E,
and
if
we
could
prioritize
it,
we
don't
just
like
the
idea
of
it
happening.
H
So
maybe
maybe
given
some
of
this
started
with
a
half
a
project,
the
puns
of
people
are
kind
of
motivated
and
when
it
dries
that
further,
you
see
what
I
mean
I
think
like
you
know,
there
is
a
clue,
so
it
can
be
done
and
maybe
maybe
about
something
similar
that
keeps
a
lung.
Maybe
there's
someone
interests,
they
didn't
have
to
try
and
push
it
forward
when
it
becomes
something
we
want
to
adopt.
H
That
might
be
tougher
and
more
than
that's
impossible,
but
that
might
be
tougher.
I
do
feel
like
a
cloud
Patrol,
it's
probably
the
one
bit
of
cloud
foundry
and
go.
Isn't
you
know
not
a
great
language
for
reviews
and
if,
if
you've
ever
and
I
know
you
have,
if
you've
ever
worked
with
the
kubernetes
go
clients,
you
probably
know
what
I
mean
go
is
not
a
language.
There
is
written
for
doing
that
stuff
and
it
I
mean
and
that's
wild
crazy.
H
A
H
I
A
H
A
B
I
was
I
was
about
to
say,
I,
think
there
was
a
project
like
a
few
I'm,
not
sure
like
when
it
exactly
happened,
but
I
think
it.
They
took
like
a
few
weeks
actually
to
try
to
rewrite
cloud
controller
in
golang
this.
This
happened
a
while
back
so
maybe
maybe
Shannon
or
jewels
or,
like
other
people,
remember
more
of
it
but
I.
Think,
like
the
whole
like
what
came
out
it
was.
It
doesn't
seem
like
at
this
point
in
time
with
a
reasonable
effort.
H
I
really
I
think
it's
a
different
thing
like
I.
Think
that's
like
one
thing
is:
should
you
rewrite
the
crowd,
control
lingo,
but
that's
one
side.
The
other
is
to
see.
Are
these
and
and
and
the
series
is
just
where
we
currently
have
this
sync
loop-
that
goes
in
pings
arena
for
each
thing
is
expecting.
We
can
just
create
a
CID
free
thing,
we're
expecting
a
like.
So
although
the
kubernetes
go
climb
looks
incredibly
scary
and
terrible,
it's
actually
a
perfectly
reasonable
REST
API
and
for
kubernetes,
which
is
I.
H
D
H
Yes
right
now,
you
hit
me
Renee
over
HTTP,
which
goes
and
creates
loads
of
Cuban
SEC
IDs.
Instead
of
that
cloud,
controller
crates,
the
top-level
CID
and
then
arena
is
like
does
asynch
lead
to
create
all
the
other.
You
know
deployments
and
services
and
whatnot,
but
yeah
it.
It
is
really
just
a
case
of
motion
that
in
and
he's
if
there
isn't
he's
a
massively
difficult
to
create
a
CID
and
from
Ruby
it's
really
just
and.
H
D
What
it's
worth
Marco
I,
do
remember
several
efforts
to
rewrite
the
cloud
controller
and
go
I.
Think
I
wasn't
on
those
teams,
but
I
recall
that
teams
were
spun
up
and
worked
on
it
for
a
couple
of
months
at
a
time
and
gave
up
I.
Don't
know
why
I
assume
it
can
be
done,
I,
don't
believe,
that's
impossible.
It
just
hasn't
been
done
yet.
D
H
I
mean
leaving
is
a
pretty
good
cloud,
like
just
I
mean
look
like
look
at
look
at
the
Cuban
fu
stuff,
like
there's
just
so
much
code
generation
that
it
has
to
do
because
it
doesn't
have
like
generics
and
when
you
like,
a
database.
Api
is
so
like
also
generated,
and
you
can
totally
doing
how
you
doing
though.
F
H
B
I'm
waiting
for
that
too,
and
get
a
repository
in
Cloud,
Foundry
incubator
at
some
point
in
time,
all
right,
I
think
we
are
on
the
top
four
of
our
our
leading,
so
I'm,
not
sure
if
we
really
get
to
some
closure
for
this
point.
But
I
am
feeling
that
we
had
a
good
conversation
on
the
holes
here,
the
stuff.
Now
it's
just
up
to
you
Jules
to
prioritize
all
the
stuff
and
then
we're
looking
forward
to
things
awesome.