►
From YouTube: 14. #everyonecancontribute cafe: Kubernetes deployments to Hetzner Cloud with Max Rosin
Description
Hetzner Cloud: https://www.hetzner.com/cloud
Demo repository: https://gitlab.com/ekeih/k3s-demo
Direction: https://about.gitlab.com/direction/configure/infrastructure_as_code/
Features:
- Protected Terraform states (developer access): https://gitlab.com/gitlab-org/gitlab/-/issues/227108
- Terraform MR widget integration, next steps: https://gitlab.com/groups/gitlab-org/-/epics/3441
- 5 minute production app with Terraform in the background: https://about.gitlab.com/blog/2020/12/15/first-code-to-ci-cd-deployments-in-5-minutes/
- Terraform Registry: https://gitlab.com/gitlab-org/gitlab/-/merge_requests/18834
- Move Kubernetes cluster creation into Terraform: https://gitlab.com/gitlab-org/gitlab/-/issues/220201
A
A
B
Right,
yeah,
okay,
cool
then
welcome
everyone,
yeah,
I'm
max
from
germany
and
a
while
ago
I
was
talking
to
micah
on
twitter
and
we
are
talking
about
hetzner
and
kubernetes
and
that
it's
interesting
stuff-
and
he
invited
me
to
this
meeting,
and
so
I
attended
a
few
times
now
and
today
I
wanted
to
start
to
look
into
hetzner
and
into
kubernetes
with
all
of
you,
and
so
it's
it's
not
like
a
conference
talk.
It's
not
like
every
word
is
prepared.
B
I
prepared
some
demos
and
some
rough
agenda
what
we
are
going
to
do,
but
the
rest
of
it
is
improvised.
So
we
can
talk
about
it.
We
can
do
something
else.
If
we
realize
we
want
to
test
something,
it's
completely
free,
what
we're
actually
going
to
do
so
for
everyone
who
is
looking
at
the
live
stream
for
the
first
time
in
this
meeting
we
usually
just
do
whatever
we
want.
B
So
I
don't
think
that
we
will
go
through
everything
today,
so
we
will
start
today
with
part
of
it
and
then
maybe
next
week
and
the
week
afterwards,
depending
how
fast
we
go,
we
can
look
into
more
of
it,
but,
depending
on
how
fast
we
move
and
into
how
much
details,
we
want
to
look
yeah.
It
depends
a
little
bit
if
we
need
one
hour
or
five
hours
so
before
we
start
from
scratch.
I
want
to
look
a
little
bit.
B
B
At
first,
it
uses
terraform
to
provision
all
virtual
machines,
all
networking
stuff
that
is
required
if
this
is
done.
Oh,
this
is
a
bad
example.
There's
a
one
step
missing
here.
There
comes
the
second
step
where
it
uses
ansible
to
configure
those
virtual
machines.
We
will
see
this
later,
so
don't
be
confused
that
that's
not
here,
but
here's
a
step
missing
then,
when
this
is
done,
I'm
using
tanka,
which
is
another
tool
we
will
learn
about
if
we
want
to
which
creates
the
first
few
things
in
the
kubernetes
cluster
like
namespaces
and
stuff.
B
If
you
don't
know
about
kubernetes
namespaces,
don't
worry.
We
will
see
this
today
or
next
week
and
then
in
the
end,
I'm
using
help
file,
which
is
a
wrapper
tool
around
helm
and
helm.
You
can
imagine
that,
as
like
a
package
manager
for
kubernetes-
and
we
are
going
to
use
this
to
install
more
software
in
our
cluster
and
yeah,
each
of
those
steps
is
can
be
yeah.
We
can
look
at
it
at
a
like
just
an
overview
or
we
can
look
into
it
at
a
very
detailed
level.
B
B
B
D
B
B
B
I
have
to
open
my
own
notes
all
right,
so
let's
go
to
hetzner
first,
so
I'm
talking
about
it
like
everyone
just
knows
what
headset
is,
but
I'm
not
sure
if
everyone
knows
but
that's
nice,
so
I'm
going
to
show
it
for
a
second,
it's
a
cloud
provider
we
can
log
in
here.
Hopefully
I'm
already
logged
in-
and
I
am-
this
is
an
overview
of
my
projects,
and
here
I
have
an
empty
project.
Everyone
can
contribute.
We
will
use
this
one
today
and
yeah.
B
Someone
from
hetzner
was
so
nice
to
send
me
some
credits,
so
we
can
create
servers
and
networks
and
load
balancers
and
volumes
here
for
free
for
the
day
right.
The
only
thing
I
prepared
was
that
I
added
an
api
token
over
here,
so
I
don't
have
to
expose
the
secret
of
it.
While
I
show
you
my
screen
and
I
did
the
same
with
gitlab-
I
created
an
api
token
for
gitlab.
B
If
you're
on
gitlab.com
you
can
go
into
your
settings,
then
I'm
going
to
move
zoom
in
a
little
bit,
because
I
then
know
that
you
won't
see
any
secret
stuff
and
then
you
can
create
an
access
token,
and
I
did
that.
I
created
an
access.
Token
gave
me,
gave
it
api
permissions
and
then
created
it
it's
down
here
and
it
expires
in
seven
hours.
So
we
can't
do
this
session
long
other
than
seven
hours,
because
then
the
key
is
expired
and
yeah.
B
B
If
anything
is
too
small
or
not
something,
then
let
me
know
yeah
what
do
we?
What
do
we
need
for
a
kubernetes
cluster
kubernetes
cluster
usually
is
built
from
some
control
plane
and
some
worker
nodes,
and
I
think
in
the
past
it
was
like
called
the
masternodes
and
the
minions,
but
the
naming
changed
a
little
bit
more
to
server
and
agent,
at
least
in
k3s.
B
It's
always
called
your
server
and
your
agent
and
the
control
plane
is
running
on
the
server
node
and
your
actual
workload
you
want
to
run
in
your
cluster
is
running
on
those
worker
nodes
or
on
your
agents,
and
so,
let's
maybe
start
with
a
small
cluster.
We
say
we
create
one
vm
with
the
control
plane
and
two
vms
as
worker
nodes,
and
one
way
we
could
do.
This
would
be
to
go
into
the
headsnow
web
interface.
B
Let's
call
it
server,
then
we
could
click
create
and
buy,
and
then
we
do
it
two
more
times
for
the
workout
notes
over
the
agents
and
if
we
want
to
have
30
worker
notes,
then
we
do
it
30
times
and
that's
terrible
because
each
time
we
do
this,
we
could
do
a
mistake.
This
takes
a
lot
of
time
to
do
this.
If
we
want
to
delete
something
and
replace
it,
we
would
have
to
click
here
again.
We
don't
want
that.
So
it's
industry
best
practice
to
not
do
this
manually
in
the
web
interface.
B
B
B
This
is
maintained
by
hetzner
themselves,
which
is
quite
nice,
because
this
way
we
always
know
that
this
will
work
with
the
headset
cloud,
because
if
they
change
their
cloud
then
they
will
also
change
the
telephone
provider
because
they
they
want
people
to
use
it.
So
that's
very
good
for
us
to
have
this
kind
of
official
support
and
not
just
a
community
project.
B
B
B
B
B
B
So
first
thing
we
are
going
to
put
in
here
is
some
basic
configuration
we
have
in
here
that
we
are
telling
us.
We
have
some
variable,
that's
a
just,
a
arbitrary
name
I
choose,
and
we
are
telling
telephone
to
use
a
special
backend
via
http,
and
then
we
are
telling
it
that
we
want
to
use
a
provider
called
hcloud,
which
sounds
very
similar
to
this
one
right.
Our
dependency
is
hcloud,
so
we
want
and
we
we
want
to
configure
this
and
we
give
it
a
token.
B
It's
a
parameter
of
this
provider,
and
this
one
is
coming
from
this
variable
and
yeah.
Where
is
this
variable
coming
from?
In
our
case,
it's
coming
from
an
environment
variable
in
my
shell,
which
is
a
hcloud
which
is
something
like
this.
It's
called
a
terraform
var
h
cloud
token
and
then
terraform
will
automatically
use
it
for
this
value
and
how,
where
you
store
this
variable
depends
a
little
bit
on
your
preference.
D
B
And
define
this
and
then
you
could
put
a
git
ignore
in
your
repository
to
never
push
this
file
to
your
repository,
because
this
token
is
secret,
and
but
often
it's
used
like
this
if
you
are
working
locally,
but
I
didn't
want
to
put
it
in
here,
for
hopefully
obvious
reasons,
so
I'm
not
going
to
save
that
just
just
so,
you
know
where
this
variable
comes
from.
This
token
is
already
here
in
my
in
the
my
shared
environment,
all
right.
A
C
B
And
the
the
next
topic
is
a
little
bit
about
the
state,
because
terraform
has
a
state
because
it
it
needs
to
know
what
is
already
in
ex
what's
already
existing
in
your
cloud
provider
and
if
it's
managed
by
terraform
or
not
and
that's
by
default,
it
will
store
a
local
file
in
this
dot.
Terraform
directory
and
just
yeah
store
it
there.
And
then,
if
you
go
to
another
machine
and
want
to
use
terraform,
you
can't
because
the
state
fire
is
not
there.
B
B
You
could
use
an
s3
bucket
in
aws,
which
I
think
was
the
most
used
solution,
and
then
another
way
would
be
to
put
the
state
file
in
your
git
repository
and
that
maybe
works
if
you
are
working
alone
or
with
two
people,
but
it
definitely
does
not
work
with
a
team
and
also
you
are
putting
your
state
which
could
be
sensitive
into
a
git
repository
so
and
yeah.
I
think
the
last
possible
solution
that
existed
under
recently
was
to
run
your
own
postgres
database
and
use
this
as
a
state
backend.
B
B
B
B
Basically,
it
tells
terraform
to
use
those
endpoints
to
store
the
state
and
most
of
the
state
goes
here
and
then
we
have
also
some
other
options
where
it's
locking
the
state
that
ensures
that
only
one
instance,
one
person
is
using
the
terraform
state
at
the
same
time,
because
you
can't
write
into
it
from
two
different
terraform
instances,
then
we
use
my
gitlab
username
and
my
gitlab
token.
Now,
if
you
remember,
I
earlier
told
you
that
I
created
an
api
token
on
gitlab
for
me,
so
that's
this
one!
B
We're
going
to
the
terraform
directory
and
pasting
this
long
command.
Now
it's
doing
a
few
things.
One
of
it
is
that
it's
downloading
this
headsnap
provider,
which
we
default
defined
in
our
versions,
file
and
yeah.
Then
it
initializes
the
back
end
and
it
tells
us
that
everything
was
successful.
This
is
nice
and
now,
if
we
run
run
the
terraform
command
here
on
my
machine
and
one
of
you
would
initialize
terraform
with
the
same
back
end,
we
would
have
the
same
state
and
that's
really
really
nice.
B
B
So
we
could,
for
example,
look
into
the
provider
documentation
over
here,
headset
cloud
provider,
resources,
and
then
we
can
see
there
is
hcloud
ssh
key
and
for
each
resource.
That
is
available
on
hetzner.
There
is
an
api
or
there
is
a
there's,
an
object
here,
resource
we
can
use,
and
then
there
is
a
little
example
of
what
we
can
put
in
there
and
I
prepared
an
ssh
key
for
today.
B
And
then,
let's
run
terraform
plan
again
and
now
we
see
the
terraform
actually
wants
to
change
something
it
wants
to
create
this
ssh
key
cool.
Let's
do
this
start
from
a
play
and
it
gives
us
a
preview
again
and
we
sell.
And
yes,
let's
look
into
the
headsnow
web
interface
again
on
the
security
there's
now
an
ssh
key
that
wasn't
here
earlier,
when
I
showed
you
that
I
created
an
api
token,
not
sure
if
you
remember,
but
this
page
was
empty
all
right.
One
ssh
key
is
nice,
but
it's
not
helping
as
much
earlier.
B
B
Let's
make
it
hcloud
server,
let's
just
call
this
one
server.
The
last
part
is
the
name,
and
we
want
to
have
exactly
one
of
it.
It's
not
strictly
necessary
to
define
account
here.
If
you
just
want
to
have
one
instance,
I'm
doing
it
anyway,
so
it
will
be
called
server
0,
our
first
instance
so
in
the
future.
B
So
let's
look
and
do
this
we
are
giving
our
server
a
name,
and
so
this
name
is
a
name
for
this
terraform
resource,
and
this
name
is:
what
do
we
want
to
call
our
server
in
the
head
center
api,
like
what
name
should
it
display
when
we
look
at
our
server
and
what
we
want
to
call
it
server
dash
the
index
of
our
account?
So
that's
why
our
first
server
will
be
called
server.
0..
B
We
want
to
use
an
ubuntu
image24,
we
want
to
use
the
server
type
cx21
and
we
want
to
have
a
location
in
feinstein
nurnberg
one.
I
guess,
if
you
remember
what
we
looked
earlier,
that
here
when
we
could
set
at
server
the
options
are
quite
similar,
feinstein,
ubuntu,
24
and
down
here
cx21,
it's
two
cpus
and
four
gigabytes
of
memory.
B
B
So
let's
go
back
into
the
code.
I
just
said
he
says
here
is
our
ssh
key.
Let's
add
this
as
well,
just
like
this
and
here's
one
nice
part
of
terraform,
there
are
a
lot
of
nice
parts
about
terraform,
but
this
is
also
one
of
it.
We
don't
have
to
hard
code
this
one
here
like
to
put
this
one
here
we
can
reference
the
id
of
this
object
in
terraform,
so
when
terraform
applies
our
configuration
later
on,
it
will
make
sure
that
it
uses
exactly
this
key.
B
What
we
want
to
do
just
we
can
write
any
string
or
any
pair
of
strings
its
key
values,
and
we
are
going
to
use
those
later
to
decide
which
virtual
machines
are
servers
and
which
virtual
machines
are
agents,
and
that's
why
we
call
gives
it
a
quite
unique
name.
We
are
using
our
domain,
so
it's
really
unique.
Probably
no
other
people
would
use
this
prefix
until
it's
server,
true
and
agent
false.
B
B
So
when
you
are
doing
networking
stuff
in
headset
cloud,
you
usually
have
a
network
and
then
you
also
have
a
subnet
which
has
to
be
a
part
of
the
network.
It
can
be
the
same
or
it
has
to
be
smaller
and
their
own
documentation
says:
there's
no
much
use
of
this
right
now,
but
features
will
come
so
we
are
just
going
to
create.
One
network
create
a
rather
large
ip
range
here
and
then
we
are
create
creating
a
subnet
in
there
again
we
are
referencing.
B
B
So
this
way
we
are
making
sure
that
this
server
will
be
put
into
this
network,
and
in
a
few
minutes,
when
we
have
several
servers,
we
will
see
how
they
are
put
together
in
one
network
to
be
able
to
reach
each
other
with
ips
from
this
network
all
right
for
the
network.
We
need
to
add
a
little
dependency
here
to
tell
teller
form
that
this
object
can
only
be
created
after
this
one
is
created.
That's
why
this
dependency
is
down
here
so
cool.
B
B
B
B
B
B
That's
all
we
need
for
as
a
base
setup
to
run
kubernetes,
and
if
we
look
into
networks,
then
we
see
that
we
all
have
the
subnet
we
just
created
with
terraform,
and
we
have
all
three
servers
in
here
and
next
time,
not
today.
We
will
look
into
this
a
little
bit
more
because
we
will
add
a
bit
more
networking
stuff
to
actually
have
a
load
balancer
in
this
subnet
and
get
traffic
in
there
from
the
outside.
B
B
We
are
now
able
to
see
this
state
we
created.
This
is
the
state
file
that
I
mentioned,
that
is
by
default
on
your
local
disk,
and
it's
now
living
here
in
the
gitlab
api
and
under
a
few
days
ago
it
was
invisible
here
you,
the
only
way
to
figure
out
that
it
exists
was
to
talk
to
the
gitlab
api
manually,
and
now
this
is
quite
nice.
There
are
no
fancy
features
here.
You
can't
just
like
give
me
a
preview
of
my
state
or
something,
but
you
can
at
least
see
that
there
is
a
state
file.
B
You
can
download
it
and
then
look
into
it.
It's
a
json,
so
it's
machine,
readable
and
human
readable,
but
it's
not
so
great
human
readable,
mostly
for
machines,
and
but
you
can
also
delete
it
or
lock
it
or
unlock
it.
If
it's
locked,
for
example,
if
someone
else
or
you
are
using
terraform
and
it
locks
this
state
and
then
terraform
crashes
and
never
unlocks
it,
then
you
have
an
easy
way
now
to
unlock
it
again.
B
B
D
B
B
C
C
C
C
C
Yeah,
okay:
workspace
is
the
correct
command.
Okay,
okay,
but
I
can
explain
aside
a
little
bit
to
the
workspace
okay,
so
literally,
there
are
different
steps
you
can
do
to
have
in
terms
of
that,
we
want
to
separate
our
environments
from
the
infrastructure
as
code
base
and
literally
more
people
or
in
the
past.
I
use
this
also
for
myself.
C
I
copied
only
the
whole
state
file,
but
it's
not
needed
what
you
can
do
is
you
can
create
a
new
word
space.
It
will
be
managed
in
the
same
state
file
and
then
you
have
a
new,
clean,
terraform
environment,
where
you
can
pull
up
anything
you
want
to.
So
that
means
what
you
need
to
do.
You
can
first,
you
can
check
it
with
clear
for
workspace
list.
C
B
C
C
Mostly,
the
idea
behind
the
word
space
is
that
you
don't
need
to
change
your
code
for
the
environment,
so
that
means
you
have
a
deaf
environment,
a
production,
environment,
a
staging
environment,
and
you
have
only
one
state
file
with
separated
workspaces
and
you
can
test
them.
The
features
out
that
you
want
to
or
applying
these
configurations,
it's
bounded
to
the
word
space
and
you
can
name
the
word
space
like
you
want.
B
A
No
it
it
was
meant
for
as
an
additional
insight
into
what's
planned
or
what's
what
what
can
be
built,
what
we
want
to
add
is
gitlab
for
managing
terraform
or
making
the
integration
better.
One
of
them
is
a
tariff
on
package
registry
and
the
other
one
is
the
merch
request.
Widget
integration,
which
outputs
the
terraform
plan
currently
should
also
have
more
insights,
and
there
are
a
couple
of
more
things
to
come
and
watch
out
in
in
the
linked
epics.
A
C
A
Yeah
I
had
I
had
a
coffee
chat
with
victor
nagy
a
while
ago,
who's
the
the
product
manager
and,
I
think,
the
protected
state
files
and
also
the
metric
wrist
integration.
Now
the
ones
which
are
anticipated
next,
but
totally
up
to
you
with
influencing
the
the
world
map,
with
more
ideas,
more
comments
and
so
on.
Making
the
integration
easier.
A
I
think
the
knowledge
is
there,
I
could
exp.
So
my
terror
from
knowledge
is
not
that
deep,
but
changes
changes
hard
in
in
terms
of
not
breaking
things
but
yeah.
It's
totally
looking
forward
to
that,
because
also
for
the
the
five
minute
production
app
project
which
we
keep
running,
we're
also
using
terraform.
In
the
background,
I
would
say
to
provision
aws
ec2
instances
and
hiding
hiding
everything
away
and
actually
you're
just
deploying
your.
B
Okay,
all
right,
then,
the
next
part
of
it
would
be
the
ansible
setup
to
yeah
to
do
some
basic
configuration
on
those
notes
and
we
will
use
ansible
yeah
for
a
few
small
things
and
especially
we
will
use
it
to
actually
install
k3s
there
and
looking
at
the
time.
I
guess
the
k3s
part
will
come
next
week,
but
we
can
already
start
to
set
up
our
instable
environment
for
it.
B
Your
ansible
is
software
for
configuration
management
and
we
can
define
our
desired
state
in
yammer
files
and
then
apply
this
to
our
infrastructure,
and
then
ansible
will
connect
to
all
our
servers
per
ssh
and
apply
the
state
that
we
want
to
have,
and
in
this
case
what
we
mostly
do
is
install
some
packages,
render
some
configuration
templates
and
put
those
rendered
files
there
and
yeah.
We
will
see
what
we
will
actually
do
with
it,
but
we
before
we
can
use
it.
B
We
need
to
set
it
up
a
little
bit,
and
so
you
can
install
ansible
with
your
own
package
manager
like
if
you
are
using
ubuntu.
You
can
use
ipate
install
ansible,
something
like
that
that
never
really
worked
great
for
me
because
then
ansible
is
installed
as
a
global
package
and
usually
I
want
to
have
a
local
version
of
the
ansible
version
and
all
the
dependencies.
B
B
Let's
install
this
and
all
right
so
and
we
can
now
see
we
are
using
the
switch
environment
and
now,
if
we
do
any
python
things,
then
they
will
be
installed
in
a
local
directory
of
my
user
instead
of
the
of
my
whole
system.
So
if
we
look
at
where's
python
living
it's
now
in
home
marks
which
length
blah
blah
blah
all
right,
let's
create
a
new
directory
here
in
our
project.
Let's
call
it
ansible
because
there's
no
relation
between
our
telephone
code
and
our
ansible
code,
it's
completely
independent.
B
There
are
a
few
ways
how
to
combine
those.
So
you
could,
you
can
call
teraform
as
part
of
ansible
or
you
can
use
terraform
to
write
some
configuration
for
ansible,
that's
possible
and
that's
our
valid
approaches.
But
in
our
case
we
will
separate
those.
So
we
can
close
all
terraform
files
for
now
and
just
look
into
the
ansible
directory
and
before
I
do
something
wrong,
I'm
going
to
check
my
notes.
B
B
B
B
And
also
not
installing
all
our
dependencies.
This
can
take
a
few
seconds
because-
and
somebody
has
quite
a
few
dependencies
because
it
supports
so
much
different
things,
so
it
needs
a
few
more
packages
for
this.
While
this
is
running,
we
can
create
another
file
here.
Let's
call
it
again
requirements,
but
this
time,
let's
call
it
call
it
the
unify
requirements.
So
this
is
not
for
python,
but
it's
for
ansible
itself,
because
we
are
going
to
use
two
dependencies
again
like
it's
quite
nice
right.
We
we
are
often
defining
just
some
dependencies.
B
We
did
this
entire
form.
We
added
this
hcloud
provider
as
independency,
and
now
we
are
using
a
lot
of
packages
from
ansible
and
now
we
are
using
yeah
like
call
it
plugins
or
modules
for
ansible
and
one
of
them
or
they
call
it
collections
since
the
recent
release,
one
of
them
is
hetzner
cloud
and
one
of
them
is
deb
ops.
B
Okay,
yesterday,
the
link
worked,
I
think,
there's
one
in
the
end
is
wrong
this
one.
So
that's
a
very
similar
support,
wise,
like
the
provider
for
terraform,
it's
also
managed
by
hetzner
themselves.
So
if
they
extend
the
feature
set
of
the
cloud,
then
they
also
are
very
fast
at
updating
their
ansible
provider
or
their
ansible
collection.
B
C
C
B
C
B
B
D
B
I
think
they
are
in
somewhere
here
collections
and
then
we
can
see
that
there
is
this
library
and
everything
in
there.
So
that's
why
it's
already
here
on
my
machine,
okay,
cool.
So
I
said
dynamic,
inventory
and
that's
because
ansible
needs
an
inventory
and
the
inventory
tells
ansible
what
hosts
success
like
host
names
or
ip
addresses,
and
maybe
some
special
settings
for
those.
B
B
When
we
run
ansible,
then
it
will
look
for
this
file,
ansible,
config
and
I'll
just
copying
and
pasting
my
configuration
in
there.
So
the
important
part
is
that
we
call
we
tell
ansible
to
look
into
this
inventory
file.
We
tell
to
use
the
root
user
and
for
yeah
for
demo
purposes.
We
also
tell
it
to
ignore
if
the
host
key
is
not
matching
because
yeah,
if
you
run
this
in
production,
you
usually
want
to
have
this
on.
B
B
I
think
we
should
now
be
able
to
look
in
at
our
inventory
in
the
beginning
of
the
talk.
I
told
you
that
we
have
this,
that
I
have
this
hcloud
token
variable,
where
my
hcloud
api
token
start
and
the
ansible
module
is
able
to
use
the
same
variable.
So
I
don't
need
to
define
this
somewhere
in
the
ansible
config
right
now
I
can
say
ansible
inventory
list.
B
Hopefully
it
works.
Yes,
it
does
so
it's
some
nice
json
output.
So
if
you
want
to
parse
this
with
some
other
program,
that's
that's
easy
to
parse
yeah
over
here
we
have
now
our
all
three
of
our
hosts
and
we
can
see
that
we
have
all
this
information
available.
We
have
those
labels,
we
have
the
location,
we
have
the
type
we
have
the
id.
We
have
everything
here.
What
we
need
and
that's
nice.
B
What
we
also
have
is
some
groups.
One
group
is
called
hcloud
with
all
three
machines
in
there
and
then
one
group
all
it's
a
kind
of
a
default
ansible
group
where
it
puts
this
group.
So
all
children
of
the
h
cloud
group
are
also
part
of
all
and
if
we
had
any
other
hosts,
they
would
end
in
the
ungrouped
group,
but
there
are
no
other
hosts
right
now.
B
B
So
I
I
like
that
very
much
of
the
this
way
of
using
a
dynamic
inventory
for
ansible
all
right
next
thing.
We
want
to
do
yeah
next
thing.
We
actually
want
to
test
if
our
ansible
is
able
to
connect
to
our
machines,
and
for
this
we
deployed
an
ssh
key
earlier
on
teleform.
If
you
remember
this,
one
that
exists
in
hetzner
is
this
key
for
me,
and
I
want
to
tell
ansible
to
use
this
key
to
connect
to
my
machines,
so
I'm
going
to
create
a
directory
here
in
ansible,
it's
called
groupwise.
B
And
ansible
has
a
way
to
define
variables
for
hosts
or
for
groups
that
that
are
settings.
I
want
to
apply
to
those
machines
and
I'm
going
to
do
this
for
the
group
all
earlier.
We
saw
we
have
this
group
all
right
here
in
our
inventory,
where
all
our
servers
are
in
here
and
yeah
generally.
You
are
completely
free
on
what
to
put
in
there,
but
I'm
going
to
put
in
a
little
bit
of
answerable
configuration
to
tell
ansible
that
it
should
use
a
private
key
file
which
is
stored
in
my
home
directory
and.
D
A
B
B
B
If
you
build
a
production
setup,
don't
add
the
security
later,
but
maybe
if
you
just
want
to
showcase
something,
it's
fine
to
do
the
security
later
to
to
get
faster
to
the
point
where
we
can
yeah,
we
can
actually
see
our
kubernetes
cluster
right.
I
think
gary
is
leaving
so
gary
thanks
for
joining.
We
will
see
you
next
week.
B
A
I
I
think
it's
really
great.
So,
first
of
all,
thanks
for
preparing
everything,
I
learned
lots
of
things
and
probably
it's
best
to
tease
for
next
week
and
say:
hey
we're
pulling
this
together
and
actually
deploying
kubernetes,
so
q,
a
this
will
be
next
week.
Yeah
and
probably
I
don't
have
any
questions,
I'm
a
little
mind
blown
to
be
honest.
B
So
that's
a
little
bit
of
a
teaser
what
we
are
going
to
next,
what
we
are
going
to
do
next
week
and
yeah
in
my
experience
this
moment
feels
quite
awesome
all
the
way,
all
the
all
the
time,
if
you
at
this
point
where
stuff
is
running
and
you're,
seeing
everything
coming
together
like
ansible,
terraform,
k3s
and
suddenly
there
are
yeah
computers
doing
what
you
tell
them
to
do.
It's
nice
so
automatically.
B
We
can
escalate
this
a
little
bit
later
on
by
doing
everything
in
gitlab
ci,
which
would
be
a
little
bit
more
connected
to
gitlab.
Then
again,
we
could
try
out
the
new
autoscaler,
because
since
yeah
currently
hetzner
is
working
on
an
autoscaler
implementation
for
kubernetes,
which
is
nice,
it's
stuff
like
we
don't
have
to
do
it
like
spend
the
next
five
weeks
doing
kubernetes
stuff,
but
we
maybe
you
can
work
on
it
next
week
and
then
do
something
else
the
week
after
and
then
come
back
to
this.
Whatever
we
want
and.
B
B
Yeah
yeah
they
added
a
pull
request
to
the
upstream
kubernetes
repository
for
the
auto
scaler
and
then
yeah.
The
customer
asked
katzner
hey
this
pull
request
is
almost
ready,
but
to
be
honest,
I
don't
want
to
maintain
it.
Do
you
want
to
do
that
and
I
think
headset
just
said?
Oh,
this
is
great.
We
want
to
do
this
and
so
a
few
days
ago,
lucas
from
hetzner.
B
A
B
A
I
have
no
like
objection
to
say:
hey
we're
doing
this
for
the
next
two
months,
because
I
think
it's
a
real
great
way
to
learn
new
things
like
putting
lego
bricks
together,
and
you
know
I
love
lego,
so
I'm
I'm
totally
up
for
like
let's
try
next
week
and
then
we
have
another
topic
with
the
autoscaler
and
then
we
have
maybe
a
different
idea
of
trying
graphing
cloud,
for
example,
and
add
some
monitoring
to
it
and
then
integrate
it
in
gitlab,
ci
and
yeah.
I.
B
Think
yeah,
I
think
I
actually
wanted
to
put
this
here
like
deploy,
monitoring
stack,
because
that's
this
way.
We
can
also
talk
a
little
bit
about
external
dns,
to
configure
our
dns
dynamically,
to
get
ssl
certificates
with
cert
manager
and
to
get
traffic
actually
inside
of
our
cluster
within
ingers
controller
yeah.
I
guess
there
are
two
additional
sessions,
so
if
we
are
out
of
topics,
we
can
just
continue
to
build
this
cluster
forever.