►
Description
Join us for Kubernetes Forums Seoul, Sydney, Bengaluru and Delhi - learn more at kubecon.io
Don't miss KubeCon + CloudNativeCon 2020 events in Amsterdam March 30 - April 2, Shanghai July 28-30 and Boston November 17-20! Learn more at kubecon.io. The conference features presentations from developers and end users of Kubernetes, Prometheus, Envoy, and all of the other CNCF-hosted projects
CNCF Cross-Cloud CI - Adding Support for New Platforms - Andrew Kutz, VMware
The CNCF cross-cloud project deploys Kubernetes and popular cloud native applications to multiple platforms on a nightly basis. The results are available each day on the CNCF cross-cloud dashboard, https://cncf.ci. The dashboard helps visitors decide which platform to use when scheduling cloud native workloads. In this presentation, Andrew Kutz will review how to add support for a new platform to the cross-cloud testing project, common challenges faced when doing so, and tips for debugging failed deployments.
To learn more: https://sched.co/GrSk
A
All
right,
everybody
I,
think
we're
gonna
go
ahead
and
get
started.
It's
a
quarter
till
and
I
tend
to
ramble.
You
know
so
probably
need
all
the
time
I
have
and
twice
that
my
name
is
Andrew
Coates
and
I
work
for
VMware
and
in
case
you
are
not
where
you
think
you
should
be.
This
is
a
talk
on
adding
support
for
new
platforms
such
as
AWS,
vSphere,
GCE
to
CNCs,
cross
cloud
dashboard
and
with
that
we'll
go
and
get
started.
A
So
this
is
our
agenda.
For
today
it's
pretty
straightforward.
Don't
have
a
lot
of
time
to
add
the
humor
I.
Normally.
Would
that's?
Okay,
you
probably
saved
from
that.
So,
first
of
all,
what
is
the
CNC
F
cross
cloud?
Ci
I'm,
going
to
assume
you
all
know
what
the
CNC
F
is
and
just
move
past
that
the
CNC
F
cross
cloud
CI
ultimately
culminates
in
something
like
this.
A
This
dashboard,
they
have
a
project
called
cross
cloud
that
builds
kubernetes
and
provisions
it
to
multiple
platforms
again
such
as
GCE
AWS
packet,
vSphere,
and
then
it
builds
and
deploys
multiple
CN
CF
projects,
and
you
can
see
on
there
some
of
those
linker
D
fluent
the
core
DNS,
and
it
ensures
that
different
versions
of
kubernetes
that
always
builds
the
last
GA
release
and
then
head
can
coexist
harmoniously
with
several
well-known
CN
CF
projects.
It
doesn't
run
the
kubernetes
e
to
e
conformance
tests,
but
it
does
run
some
tests
for
the
projects
you
see
up
there.
A
Cross
cloud
is
really
three
distinct
components,
build
cross
cloud
and
cross
project.
Today,
we're
going
to
be
focusing
on
the
cross
cloud
component
and
it's
what
enables
the
multi-platform
support,
and
that
is
a
pretty
up-to-date
view
of
what
platforms
are
currently
supports
in
case
you
can't
see
it.
It's
AWS,
microsoft,
azure,
GCE,
IBM
cloud
packet,
openstack,
vmware,
vsphere
and
OCI
or
oracle
cloud
infrastructure,
and
today's
talk
is
to
how
to
help
you
put
your
platform
on
the
dashboard.
A
The
cross
cloud
project
actually
leverages
terraform
and
when
we'll
get
into
that
in
a
bit.
But
when
you
want
to
add
support
for
a
new
platform,
you
have
to
basically
either
create
or
live
leverage
an
existing
terraform
provider
for
your
platform
directories
at
the
root
of
the
project
mapped
to
those
platforms
again
same
list,
I've
been
repeating,
and
the
other
directories
are
terraform
modules
that
the
platform
projects
then
import
so
modules
for
doing
things
like
deploying
DNS,
deploying
kubernetes,
creating
your
cube,
config
file.
What
so
long!
A
So
how
do
we
actually
add
support
for
a
new
project?
Well,
it's
easy
as
like
I
say
one
two,
three
and
minus
one:
it's
several
steps,
but
it's
pretty
straightforward.
There
are
some
hiccups
along
the
way,
and
hopefully
this
talk
will
help.
You
handle
those
with
the
aplomb
that
I
did
not.
So
before
you
begin.
What
are
the
requirements?
A
Is
there
a
terraform
provider
for
your
platform?
Luckily,
when
I
began
work
on
this
and
myself-
and
we
who
is
here
today,
you
know
I'm
the
one
on
the
stage,
but
honestly
he
did
most
of
the
work
and
if
I
can
get
that
arrangement
for
all
my
projects
going
forward,
then
and
then
it'll
be
boring,
but
at
least
that
I
can
put
the
blame
on
him.
So
is
there
a
provider?
If
there's
not,
you
will
need
to
create
one.
A
That's
not
in
the
scope
of
this
talk,
but
there
is
a
link
from
the
presentation
that
takes
you
to
the
list
of
terraform
providers
and
you
can
see
if
there
is
a
provider
for
your
platform.
You'd
be
surprised,
there
are
providers
for
dozens
and
dozens
is
docker
installed
locally.
Is
the
next
requirement?
Why
do
you
need
docker
locally
to
do
testing
to
test
that
your
changes
are
working
and
we're
going
to
use
that
in
a
bit
to
actually
deploy
kubernetes
with
the
cross
cloud
provisioner
from
my
laptop
to
a
public
vSphere
instance?
A
And
finally,
it's
not
required,
but
it
would
be
really
good
to
have
an
IDE
that
has
support
for
Tara
forms
syntax
so
that
when
you
look
at
a
file,
it's
just
not
black
and
white,
and
it's
HCl.
It's
hash,
II
core
language,
but
yeah.
It's
it's
easy
to
read,
but
the
syntax
highlighting
is
not
as
nice
and
finally,
if
you've
not
used
Tara
form
before
there
is
gonna
be
a
learning
curve.
It
was
new
to
me.
I
might
use
vagrant,
but
vagrants
files
are
based
on
Ruby.
Tara
form
is
not.
A
Without
that
experience
there
could
be
a
slight
learning
curve,
but
again
I
was
brand
new
to
it
as
well,
and
we
went
from
not
being
on
the
dashboard
to
being
on
the
dashboard
in
about
a
month
and
a
half
completely
net
new
to
it.
I
was
also
brand
new
to
kubernetes
this
year,
so
I
don't
know
why
they
let
me
come
up
and
give
a
talk,
but
if
nothing
else
you
all
look
way
more
capable
than
me
and
I
was
brand
new
to
kubernetes
in
April,
so
not
new
to
CI.
A
So
if
you're
new
to
both
maybe
they'll
be
a
bit
more
of
a
learning
curve,
but
so
the
first
step
to
creating
a
new
platform,
adding
support
for
new
platform
is
forking.
The
cross
cloud
project
on
github
I'm
not
going
to
demonstrate
that
because
I've
already
forked
it
but
I
think
we
all
know
how
to
port
projects
on
github.
Actually
I,
don't
sorry
I
don't
have.
A
Let
me
show
you
that
project,
because
it
just
now
occurs
to
me
that
I
don't
have
it
up.
So
this
is
the
cross
cloud
project
on
github.
There's,
actually
a
link
to
this.
In
the
presentation,
it's
just
cross
clouds
and
then
cross
cloud
and
again
there's
a
link,
but
I
just
wanted
to
show
you
what
it
looks
like
in
github.
So
you
fork
that
I
always
find
it
helpful
to
go
ahead
and
add
a
get
remote
as
upstream.
So
when
I
clone
my
fork,
I
immediately
do
a
git,
remote,
add
upstream.
A
A
This
next
step
this
this
actually
used
to
be
a
lot
more
difficult.
Very
recently,
we
created
a
small
tool
that
can
help
you
create
a
skeleton
for
adding
a
new
provider
and
I'll
demonstrate
that
here.
In
a
moment,
the
skeleton
includes
initial
documentation
like
the
readme
bare-bones
terraform
files,
as
well
as
some
helper
scripts
and
a
hack
directory
to
make
standing
up
tearing
down
clusters
a
little
easier.
So
we
can
actually
show
that
off
right
here,
I'm
just
going
to
copy
this,
as
is.
A
Hey
it's
created,
so
it
kind
of
tells
you
the
steps
that
you
need
to
do
next
now
this
tool
relies
on
some
markers
and
two
files,
a
docker
file
and
the
provision
script
that
aren't
merged
yet
into
master.
If
that,
if
that
was
the
case
since
they're
not
yet
merged,
it
still
can
work
without
it.
If
we
do
this.
A
A
So,
for
example,
this
just
ensures
that
the
version
is
that
as
equal
or
greater
to
the
specific
version
of
the
provider
and
it
configures
it
with
its
credentials
and
allows
unverified
SSL
what-have-you.
But
so
you
would
need
to
configure
your
version
of
that
and
I
thought
I'd
put
a
link
to
the
list
of
providers.
It's
a
go.
It's
actually
a
fine
time
to
just
go
ahead
and
show
that
terraform
providers.
A
A
Okay,
next
step
after
you've
got
that
configured.
You
might
have
noticed
that
there
are
certain
things
using
these:
what
they
look
like
variables
right,
they
have
the
dollar
sign
than
the
VAR
dot.
What
are
they
well?
They
are
variables
they
actually
reference,
terraform
input
variables
and
they
are
defined
in
this
file
called
input
TF,
and
these
are
all
of
the
configuration
properties
that
your
project
can
use.
For
example,
like
the
name
of
the
cluster
or
the
directory
inside
the
container
where
the
data
is
stored,
and
then
beyond
that
is
just
common
data.
A
These
are
actually
that's
not
even
true.
This
is
the
only.
This
is
the
only
piece
of
information
even
specific
to
this
project.
The
rest
of
this
is
just
common
data.
That
is
true
for
all
of
the
platform
providers.
Now,
why
do
we
define
them
here
and
not
at
the
root
of
the
cross
cloud
project?
As
I
said,
every
platform
directory
is
a
terraform
project.
There
were
two
ways
that
the
CNC
F
could
have
built
this
and
they
chose
B
instead
of
a
either
both
are
valid
version.
A
Instead,
they
run
terraform
inside
of
the
platform
directories
and
then
load
all
the
other
directories
that
install
kubernetes
create
the
cube
config
all
that
as
modules,
because
of
that
your
directory
controls
the
input
for
the
project
and
because
some
of
those
other
modules
have
expected
inputs
themselves,
you
have
to
define
them
at
the
root
of
your
project,
because
any
one
of
these
at
the
root
of
your
project
can
then
be
the
set
via
a
configuration
file,
but
also
via
the
command
line.
So
TF
VAR
name
equals
any
configuration
parameter.
Any
variable
define
an
input.
A
A
A
The
other
modules
that
this
file
loads
are
the
modules
responsible
for
configuring,
the
common
components.
So
master
templates,
you
can
see
it
loads
it
from
source
dot,
slash
master
templates
1.10.
This
is
what
loads
or
deploys
kubernetes
and
all
of
its
dependencies
to
the
master
net
or
the
control
plane.
Nodes
module.
Worker
templates
is
responsible
for
the
worker
nodes
and
then
load
balancer
loads.
Our
load
balancer
dns
deploys
dns
TLS.
This
generates
all
of
the
certificates
for
that
are
necessary
to
deploy
kubernetes
and
finally
cube.
A
So
right
now,
the
cross
cloud
project
does
expect
your
platform
to
have
a
valid
cloud
provider.
I
have
a
branch
out
there
for
which
I
don't
think
yet.
I've
opened
a
PR,
but
I've
updated
their
their
templates.
To
give
you
the
ability
to
not
need
the
cloud
provider,
you
can't
just
give
it
an
empty
input,
because
the
value
the
name
of
the
cloud
provider
is
also
used
by
them,
for
other
things
just
adds
a
boolean
disabled
that
defaults
to
false,
but
right
now
it
does.
A
It
does
expect
you
to
have
a
cloud
provider
and
the
scaffolding.
Just
creates
a
template
file
that
loads.
This
thing
called
cloud
conf
and
then
it
interpolates
the
variables
in
there
with
these
input
values
from
input
TF.
Where
is
that
used?
Well,
you
can
see
it's
template
file,
Cloud
kampf,
when
the
module
gets
loaded,
a
cloud
config
it
loads
it
in
here
it
looks
up
the
template
data
unit,
just
interpolates
that
data
with
the
input
variables
that
you
provided
it.
A
The
modules
that
we
create
again
are
the
master
and
worker
nodes,
and
so
they've
got
very
similar
files.
You've
got
your
input
file
the
number
of
nodes.
You
want
the
name,
the
cloud
in
it,
data
the
number
of
CPUs
amount
of
memory
and
Meg's.
That's
your
input,
but
you
can
also
have
output,
both
the
control,
plane,
nodes
and
worker
nodes
output,
the
IP
addresses
of
the
machines
they
create,
and
that
is
because,
in
the
parent
project,
the
route
of
our
project
here,
you
will
see.
A
Where
is
it,
it's
gonna
be
at
the
load:
balancer
I
think
module
dot,
master
master
ip's
that
map's
back
to
this,
its
module
master
master
ip's,
so
the
modules
can
share
information
with
each
other
via
puts,
and
finally,
we've
got
an
O
in
the
control,
plane
and
worker
modules.
We
actually
then
define
the
resources
that
get
created
and
again
this
type
of
resource
is
going
to
be
specific
to
your
terraform
provider.
So
I,
the
scaffolding
can't
assume
what
it
is
it
does.
A
In
fact,
you
can
see
what
this
does
when
it
outputs
the
master.
Ip
addresses
that
star
says:
I
want
the
basically
it
joins
this
array,
and
this
array
is
a
list
of
the
IP
addresses
from
each
machines
and
machine
nodes
default.
Ip,
address
value
and
again,
the
load
balancer
is
the
scaffolding
that
it
created.
This
load
balancer,
the
scaffolding
creates,
is
very
dumb.
It
just
says
give
me
a
list
of
the
target
IPs
and
it
just
outputs
the
first
value
from
that
list.
A
There
actually
is
there
actually
are
more
concrete
implementations
of
it.
This
is
V
spheres
and
actually
leverages
AWS
for
the
load
balancer,
so
it
can
be
as
complex
or
as
simple
as
you
want
it,
but
the
scaffolding
doesn't
make
any
assumptions
and
it
just
creates
something:
that's
pretty
dumb
and
that's
pretty
much
everything
that
oh
finally,
the
scaffolding
does
creates.
A
Some
helpful
commands
here
deployed
SH
and
basically,
you
could
just
run
hack,
slash
deployed
SH
and
give
it
the
name
of
a
cluster
to
deploy
and
then
as
well
as
destroyed
at
SH,
and
that
just
calls
this
and
you'll
see
an
example
in
a
minute
war
actually
and
invoke
something
very
similar
to
that.
It's
just
I
use
those
files
or
versions
of
those
files.
A
Whenever
I'm
doing,
testing
that
way,
I
don't
have
to
type
out
the
whole
docker
run
command
every
single
time
and
those
are
referenced
and
they
read
me
as
well
yeah,
let's
go
ahead
and
see
Oh
last
two
files
that
you
have
to
update
the
scaffolding
did
or
the
yeah.
The
script
did
update
those
because
it
has
a
like
I
said
it
has
markers.
So
if
the
you
ran
that
script
on
a
version
of
the
cross
cloud
project
that
didn't
have
these
markers,
you
would
need
to
remember
to
add
that
copy
cube,
conciencia,
slash,
cube
con.
A
A
A
A
A
You
know
the
wireless
worked
great
on
Monday
before
everybody
arrived
and
then
the
show
floor
open
all
right,
so
yeah
we
would
we'd
want
to
rebuild
the
the
docker
image
and
actually
we
can
go
ahead
and
give
it
a
quick
demo
of
that.
What
I'm
going
to
do,
though,
is
going
to
remove
those
remove
cube
con
docker
boom
G.
A
A
Technically,
there's
no
reason
it
couldn't
be
pushed
pushed
to
a
you
know,
a
registry,
but
you
know
when
you're
doing
development
you're
not
going
to
be
using
the
the
public
one
anyway.
So
so
it
updated
the
image
and
we're
gonna
deploy
a
new
kubernetes
cluster.
Now
this
is
just
an
example
of
what
you
would
do
you
do.
The
docker
run
make
sure
that
image
is
removed
or
the
containers
removed
after
it
quits.
A
A
A
These
DNS
servers
are
important.
We'll
tell
you
why,
in
a
minute
we're
mapping
the
data
directory
in
the
folder
I'm
currently
in
to
CN
CF,
slash
CN
CF
data.
They
use
that
directory
in
their
image
to
store
the
terraform
data.
This
isn't
necessarily
required
this
just
Maps
the
dot
terraform
directory
in
the
working
directory
from
the
container
back
to
the
current
directories,
terraform
directory.
Why
is
that
nice?
It's
the
first
time
you
run
this.
A
It's
gonna
have
to
download
a
bunch
of
terraform
plugins
which
can
take
a
while,
and
if
you
map
it
back
to
your
local
directory,
you
don't
have
to
download
them
again.
These
eflags
are
passing
environment
variables
in
to
configure
certain
things.
These
are
the
ones
specific
to
cross
cloud.
These
up
here
specific
to
our
provider
cloud,
tells
it
the
type
of
platform
to
which
it's
deploying
back-end,
remember
I
mentioned
they
use
s3
files,
also
a
valid
candidate
name,
we're
gonna,
deploy
a
cluster
called
cube
con
and,
finally,
the
command
to
execute
deploy.
A
The
other
option
is
destroy,
so
let's
go
ahead
and
execute
this
and
hope
it
works
and
again,
if
it
doesn't,
I've
got
a
video
of
it
working,
I
kind
of
hope.
It
does
because
I
want
to
be
able
to
show
it
off
in
on
an
actual
V
Center
environment.
So
it's
actually
going
on
here
is:
it
is
creating
certificates
for
all
of
the
kubernetes
services,
including,
like
the
provision
service,
account
certificates
that
needs
a
CA
for
that
right.
What
is
that
giant
blob?
A
You
see
anyone
take
a
guess:
texts,
yeah,
it's
text,
it's
a
base64,
encoded
version
of
the
cloud
data
user
data,
so
the
way
that
they
deploy.
This
is
through
cloud
in
it.
In
fact,
while
this
is
doing
this,
we
can
actually
look
at
that.
They
use
cloud
init
to
deploy
against
everything.
So
you
can
see
these
different
versions.
What's
the
difference
between
1.1
know
and
the
same
one
at
bun
2?
If
we
look
here,
this
is
non
abun
2.
It
expects
it
to
be
core
OS.
A
This
is
just
a
cloud
learn
it
template
and
the
abun
two
version.
It's
called
a
bun
2,
but
it's
not
specific
to
a
bun
2
cloud
in
it,
even
though
it
was
created
by
canonical
became
very
popular
with
Korres,
and
so
you
know,
you'll
often
see
cloud
lean.
It
files
with
this
at
the
top
of
it,
but
that's
specific
to
core
OS
and
so
they've
got
one
that
works
without
core
OS
as
well.
A
The
longest
step
here,
at
least
for
us,
is
this
one
right
here:
creating
the
AWS
load.
Balancer
I
tried
to
figure
it
out
because
I
could
create
a
load
balancer
in
about
two
seconds.
But
if
you
go
look
at
the
source
code,
they
actually
will
not
move
past
this
until
the
load
balancers
listeners
are
responding,
but
that
blocks
everything
else.
So
it
takes
a
bit
before
this.
You
know
takes
about
two
minutes,
usually
before
it
decides
just
move
on,
because
I
can't
the
listeners
can't
respond
until
those
machines
are
up.
A
A
Once
doing
that,
I
actually
will
do
the
resolving
common
issues.
So
remember,
I
kept
saying:
hey
notice.
These
DNS
entries
crawl
squad
uses
a
shared
DNS
server
when
it's
standing
up
the
environment,
one
of
the
things
that
it
does
and
in
fact
we
can
see
that
if
I
had
thought
to
do
it,
although
actually
doesn't
have
yeah
here
it
is
it's
creating
these
DNS
entries
and
their
shared
DNS
server.
A
You
can
see
it
right
there
and
that's
basically
making
it
easy
for
them
internally
to
the
kubernetes
cluster,
to
reference
the
nodes
without
needing
to
know
necessarily
know
the
IP
addresses
ahead
of
time.
The
nodes
are
just
responsible
for
or
terraformers
responsible
for
entering
that
data
into
the
shared
DNS
server.
If
you
run
the
provisioning
container
without
that
DNS
one
whatever
it
is
one,
four,
seven,
seven,
five,
six
nine
two:
three!
A
No
one
you're
not
going
to
know
how
to
access
your
cluster,
because
the
cube
kuddle
that
the
come
sorry
the
cube
config
they're
right
out
has
the
DNS
name
has
the
fqdn.
It
doesn't
have
the
IP
address.
So
your
operating
environment
needs
to
be
able
to
resolve
names
against
their
DNS
server,
and
you
can
add
it
to
your
host
or
you
can
just
do
docker
run
and
include
the
DNS
there.
A
A
So
it
actually
brings
it
up
in
about
three
minutes
and
then
it's
adding
a
bunch
of
manifests
to
ensure
that
everything's
there
and
happy
now
it's
going
to
wait
for
the
nodes
to
show
up.
So
what's
the
next
common
issue
that
we
see
yeah
the
cross
cloud
image
should
be
built
from
the
the
project
or
give
it
a
dot
dot
slash
if
you're
building
it
from
the
platform
already
mentioned,
make
sure
you
add
the
dash
T
flag,
or
else
you
can't
cancel
out
of
a
deployment
and
here's
one.
A
We
had
to
figure
out
the
hard
way,
even
though
you
can
immediately
add
entries
to
the
shared
DNS
server
until
the
CNC
F
team
white
lists,
your
domain
that
cloud
name
there.
They
are
sky,
DNS
server,
won't
respond
to
queries
for
your
domain,
so
you
need
to
make
sure
they
add
that,
and
they
will
add
that
you
know
super
quick,
alright,
so
it's
online
and
we
can
verify
that
real,
quick
by
just
using
our
helpful
wrapper
here
and
say,
gets
all
and
yep
see
how
many
nodes
it
has.
A
It's
got
one
node
and
then
see
what
pods
it's
got
in
the
cube
system.
Namespace
should
have
like
the
cube
DNS
you
proxy
EEP
stir,
and
what
does
that
actually
look
like
over
in
vSphere?
Well,
it
deployed
three
control,
plane
nodes
and
one
worker
node.
Now
why
one
worker
to
three
control
plane
nodes,
not
positive,
but
that's
their
default
configuration,
although
you
can
modify
that
as
you
wish
to
do,
and
then
the
load
balancer
it
creates
is
over
here
at
AWS.
A
Now
you
may
not
use
the
aw
AWS
load
balancer,
you
may
use
one
of
your
own
or
you
may
not
use
one
at
all,
doesn't
really
matter
yeah.
Here
we
go.
That's
the
load
balancer
from
yeah
a
few
moments
ago.
So
let
me
get
back
to
the
presentation
and
then
finally
just
destroying
it.
It's
pretty
easy.
You
just
run
in
fact
we'll
just
use
the
helper
command
now
and
remember.
A
Remember
it.
The
scaffolding
will
create
the
same
helper
commands
that
I'm
using
here
for
you,
so
that
whole
docker
command
we
use
to
deploy
it.
You
could
have
just
done
hack,
slash,
deployed,
sh,
cube
con,
be
a
lot
easier
to
remember
that
and
have
to
type
out
everything,
but
it's
good
to
know
how
it
works
from
the
bottom
up
before
you
start
using
the
shortcuts.
A
So,
while
it's
destroying
the
cluster
and
by
the
way
it
destroys
all
the
resources
associated
with
it
as
well,
the
load
balancer,
the
TLS
certificates,
the
cube
config
the
machines.
Well,
it's
while
the
cluster
is
being
destroyed.
That
kind
of
want
to
open
it
up
to
questions
or
feedback.
Does
anyone
have
any
questions
about
something
I
didn't
cover?
Maybe
something
I
missed,
or
that
I
should
have
spent
a
little
extra
time
on.
A
Yeah,
okay!
Well,
let
me
see,
while
we're
doing
still
destroying
again
that
load
balancers
it
takes
forever.
So
alright,
so
it
looks
like
it
should
be
destroyed.
Let's
verify
that
it's
gone
yep,
we
don't
see
cube
Khan
there
anymore
and
that
load
balancer
usually
takes
a
moment
to
go
away.
Yep
it's
gone
so
that
was
it
and
again
just
using
the
scaffolding
command
that
is
up
here
should
get.
A
You
started
to
creating
one
of
your
own,
adding
support
for
your
platform
pretty
quickly,
and
if
you
have
any
questions
about
how
to
do
that,
or
just
want
to
suggest
some
ideas
that
could
make
this
even
you
know
easier,
smoother,
please
feel
free
to
contact
me
or
my
colleague
we
and,
like
I,
said
the
link
to
the
cross
cloud
repository
is
in
the
presentation.
Probably
should
have
been
sooner
and
if
you
actually
want
to
see
the
working
version
of
one
of
these,
these
are
all
the
pull
requests
for
the
VMware
platform
provider.