►
From YouTube: Chris Timberlake on Kubernetes Configuration
Description
Recording with the Ed Services team on a way to configure GitLab and Kubernetes to deploy a static site to a cluster.
A
A
A
A
B
The
class
was
having
some
trouble
getting
get
set
up
and
especially
getting
the
ssh
keys
set
up
and
he
recommended
hey.
Why
don't
you
just
have
all
the
students
run
a
doctor
image
with
git
and
you
can
they'll
all
use
the
same
key?
I
I'm
not
sure
about
the
security.
I
need
to
figure
this
out,
but
anyway
he
was
suggesting.
B
If
we
just
have
all
the
students
use
a
get
docker
container.
That
would
make
everything
run
a
lot
more
smoothly.
On
the
other
hand,
then
the
students
have
to
have
the
docker
engine
installed,
which
doesn't
work
for
everyone,
but
anyway,
I'll
make
an
issue
just
want
to
give
you
a
heads
up
on
that.
Okay.
C
B
A
C
B
A
C
C
A
A
C
Okay,
all
right
yeah,
if
you
want
me
to
do
that,
I
can
absolutely
do
that.
But
you're
gonna
have
to
give
me
just
a
second
here
to
find
the
right
doc,
because
there's
a
series
of
permissions
that
I
have
to
get
to
and
then
we
can
just.
I
can
just
walk
through
the
whole
process
start
finish.
If
you
want
go
ahead.
B
C
C
No,
no!
No.
I
wish
all
right,
so
I'm
gonna
start
sharing
with
safari,
so
here's
safari
right
so
the
first
place
is,
is
I
use
my
personal
vulture
account
for
a
bunch
of
stuff,
so
I'm
going
to
go
ahead
and
add
a
cluster
here
and
just
start
setting
up
a
cluster.
Now
I
have
a
personal
project
that
I
have
to
get
the
setup
for
so
I'm
going
to
go
ahead
and
set
this
up
against
that
anyways.
C
C
C
Okay,
yeah,
it
is
yep
yep
and
we're
gonna
go
ahead
and
deploy
oh,
we'll
just
call
it
nodes.
I
don't
even
know
what
else
to
call
it's
called
nodes,
so
this
is
going
to
take
a
minute
or
two
to
spin
up
in
the
meantime,
what
I'm
going
to
do
is
this
is
my
project
I'm
going
to
go
into
the
kubernetes
section,
real,
quick,
our
kubernetes
integration,
stuff
is
in
flux
and
it's
constantly
changing
they're
constantly
updating
it.
C
So
this
is
going
to
be
a
little
bit
bumpy
and
I
apologize
in
advance
just
kind
of
follow
along
as
much
as
possible
and
I'll
recap
at
the
end.
But
effectively
it's
going
to
be
it's
going
to
be
a
bit
bumpy.
Okay,
you
can
set
it
up
easily
easier
with
gitlab
using
eks
and
gke,
but
I
don't
use
either
amazon
or
google.
For
my
cloud
I
use
vultr
right,
and
so
this
is
going
to
get
spun
up
should
take
a
minute
or
two
to
get
spun
up
yeah.
C
I
think
it's
almost
up,
so
we're
just
going
to
call
it
this
right
and
we're
going
to
set
the
environment
scope
to
star.
That
means.
It's
a
development
environment
right
with
that.
How
do
I
put
this
with
that
environment
star?
This
is
really
important,
because
if
you
said
it's
like
dev
or
prod,
depending
on
the
environment,
that
you
have
in
your
pipeline,
that
you
assign
the
environment
in
your
pipeline
right
because
you
can
make
environments
and
you
assign
them
to
your
pipelines,
but
the
environment
you
use
in
your
pipeline.
A
C
C
C
C
B
C
C
I
have
no
idea
I
actually
just
picked.
I
just
changed
my
background
like
yesterday,
because
I
was
like
oh
look,
I
don't
have
a.
I
have
a
normal
crappy
background,
so
so
this
is
the.
This
is
the
file
that
you
get
from
volteer
and
a
lot
of
cloud
providers
will
provide
this
file.
This
is
like
your
your
auth
stuff
that
you
need
for
the
cert.
C
C
B
C
A
C
C
I
prefer
the
openshift
way
of
doing
this,
because
the
open
shift
way
is
like
light
years,
light
years
easier
because
you
can
just
do
like
oc
get
token,
and
it's
done
so.
This
service
account
right.
A
service
account
token
scope
to
cube
system
with
cluster
admin.
That's
what
it
was
there's
a
specific.
C
It
gives
it
complete
and
utter
access
to
everything
you
can.
You
can
have
cluster
roles,
which
is
a
role
against
the
entire
cluster,
and
then
you
can
have
a
normal
role
binding,
which
is
a
role.
That's
that's
locked
to
that
specific
name
space
so
like
if
I
did
the
same
command,
but
I
did
like
roll
binding,
it
would
only
give
the
gitlab
admin
service
account
access
to
admin
access
inside
of
that
namespace.
It
wouldn't
give
it
admin
access
outside
of
the
namespace.
C
Gitlab
is
sorry
everything
inside
of
kubernetes
is
usually
name
spaced,
so
this
usually
involves
some
kind
of
name
space
so
like
even
now,
when
I
get
when
I'm
doing
this
command.
To
get
the
token,
the
login
token
of
that
service
account
what
it's.
What
it's
doing
is
it's
pulling
the
admin
it's
pulling
the
token
from
inside
that
name,
space.
C
C
C
This
is
super
important
for
auto
devops,
for
a
gitlab
managed
cluster,
allow
gitlab
to
manage
namespaces
and
service
accounts
for
the
cluster.
This
is
important
for
being
able
to
create
namespaces
and
service
accounts
for
individual
projects
and
then
there's
a
namespace
per
environment,
so
deploy
each
environment
to
its
own
namespace
right
so
like
if
you
have
a
dev
or
a
dev
and
prod,
and
all
those
environments
go
to
their
own
own
sections,
so
integrations
yeah.
Okay,.
A
C
C
C
B
C
C
C
I'm
going
to
use
this
I'm
going
to
use
my
personal
email
for
this.
Just
because
again
this
is
my
personal
project.
I
had
to
get
the
setup
with
anyway,
so
so
cert
manager
and
keep
in
mind.
This
is
just
the
bare
basics:
to
get
installed
a
a
basic
application
on
kubernetes
with
git
lab
right,
because
you
need
the
ingress
router
to
be
able
to
direct
traffic
inside
the
cluster.
C
C
C
B
B
B
C
Have
to
yeah
once
you
get
into
kubernetes
like
this
is
the
you
know.
This
is
the
shortest
path
there
yeah.
Okay,
thanks
so
I
mean
I'm
trying
to
trust
me.
I'm
trying
to
give
you
the
shortest
path,
but
that's
it
is
what
it
is.
That's
true.
This
is
not.
I
mean
I
think
it's
over
complicated,
but
it
you
know
it's.
That's
not
my
not
my
ball
game.
C
C
C
This
does
a
weird
ci
cd
process
like
it's
like.
Does
it
on
one
of
the
docker
runners
and
then
then
it
pushes
everything
into
your
cluster.
So
there's
like
this
russian
nesting
doll
process
that
happens.
The
git
lab
a
git
lab
a
normal
git
lab
runner
is
going
to
spin
up
it's
going
to
conduct
this
process
and
execute
everything
against
the
cluster
spin
up
the
runner
on
the
cluster
and
spin
up
the
ingress
controller
and
everything
there.
So
a
lot
of
folks
can
will
do
this.
C
These
steps
manually
right
because
the
ingress
controller,
the
gitlab,
runner
and
stuff
like
when
you
get
to
this
part
of
it
installing
the
gitlab
managed
apps.
A
lot
of
folks
may
do
that
part
manually.
But
if
they
do
that
manually,
you
can't
take
advantage
of
the
full
functionality
of
kubernetes
inside
gitlab.
That's
the
that's
the
catch
you
just
you're,
not
able
to.
C
It's
like
when
I
say
full
functionality.
What
I'm
talking
about
is
like
all
of
the
like
all
of
the
monitoring
stuff
right.
You
can't
see
all
the
monitoring
stuff.
You
can't
see
all
of
the
security
and
firewall
based
protections
that
we
offer
you
can't.
Let's
see
what
else
is
there?
So
it's
like
you
have
to
install
you
know
century
right
to
be
able
to
which
one's
a
for
monitoring,
yeah,
so
century
does
monitoring,
and
I
think
post
hog,
yeah
postdog
does
product
analytics
and
then
there's
prometheus
right
like
so.
C
C
So
there's
a
lot
of
a
lot
of
like
things
that
get
tied
back
into
gitlab
and
unless
you
go
through
this
whole
process
and
connect
kubernetes
as
a
as
a
cluster
to
gitlab
and
provision
the
cluster
with
gitlab,
where
git
lab
basically
controls
the
cluster.
A
lot
of
these
function
out,
like
that
a
lot
of
the
security
and
compliance
stuff
and
deployment
stuff
is
in
monitoring
stuff
related
to
kubernetes.
C
Doesn't
work.
That's
like
my
number
one
complaint
about
git
lab
and
the
kubernetes
integration
is.
Is
that
for
all
these
things
that
we
build
with
kubernetes,
we
expect
gitlab
to
control
the
kubernetes
cluster,
the
majority
of
our
enterprise
customers,
don't
really
allow
that
they
don't
allow
for
that
to
happen.
C
C
C
C
So
this
ci
pipeline
is
like
special
in
that
it
has
like
special
credentials,
and
then
this
template
right
here
right
it.
It
has
a
bunch
of
logic
inside
of
it
that
reads:
this
managed
apps,
folder
and
based
on.
What's
in
the
managed,
apps
folder,
it
will
go
out
and
it
will
provision
everything
inside
of
kubernetes.
C
C
A
C
If
you
look
over
here,
you'll
notice
on
my
name,
spaces,
there's
two
new
namespaces,
there's
git
lab,
managed,
apps
and
there's
get
lab
provisioning
right.
So
now,
if
I
do,
if
I
do,
you
know,
get
live,
managed,
apps
and
I
do
get
pod.
It
shows
me
the
pods
are
in
here,
so
I've
got
cert
manager,
I've
got
ingress
and
I've
got
the
runner
all
set
up
and
configured
inside.
That
group.
A
A
C
So
it's
got
a
deployment,
it's
got
a
scaling
replica
set
set
to
one
right
now
and
then
it's
got
some
other
information
here.
The
concurrency
set
to
one
executor's
kubernetes
yeah-
and
this
has
all
of
my
all
my
items
in
it
in
terms
of
the
registration
and
all
that
you
can
modify
these
as
needed,
and
you
can
also
set
up
a
custom
kubernetes
helm.
You
can
set
up
a
custom,
helm
config
for
these
yeah.
A
C
So
where's
the
thing
I
was
working
on
web
bounties
all
right,
so
this
one
this
is
a
node.js
application.
It's
got
a
docker
file.
It's
got
everything
configured
here,
yeah,
it's!
You
know
it's
gonna,
build
the
application,
set
the
application
up
and
everything
and
I'm
not
going
to
build
a
pipeline
for
this.
I
don't
think
because
it'll
just
drive
me
nuts,
but
what
I
am
going
to
do
is
I'm
going
to
turn
on
auto
devops
and
we'll
use
a
continuous
deployment
to
production.
C
A
A
C
C
C
Yeah
so
you
have
to,
I
have
to
set
a
domain
here
and
then
I
have
to
redirect
the
domain
to
my
kubernetes
cluster.
So
like
here's,
I'm
setting
k8s
abstractos.org
right.
So
I'm
going
to
set
that
as
that
and
then
with
this
temp
project
I
just
had
which
I
lost,
because
I'm
bouncing
between
too
many
things
with
this
base
project.
If
I
go
to
test
project,
if
I
go
here
and
look
at
my
deployment,
my
ccd,
so
let
me
turn
on
auto
devops
real
quick.
C
The
base
domain
is
the
domain
from
which
all
the
pods
like
derive
their
url.
You'll,
see
it
in
a
minute.
So
I'm
turning
the
strategy
on
the
warning
went
away.
I
go
into
pipelines,
I
run
a
pipeline
now
this
pipeline
is
going
to
run
and
it's
going
to
build
the
application
and
you'll
see
that
the
items
here
should
show
up
for
a
deployment.
C
C
Cluster
subnet
ip
address
all
right,
so
here
comes
the
fun
part
too.
So
I
need
to
do
that.
I
need
to
figure
out
my
ingress
right,
so
I've
got
the
ingress
controller
running,
but
it's
only
on
one
node
at
a
time,
which
means
so
I
need
to
add
a
domain
here.
I'm
going
to
stop
sharing
for
just
two
seconds.
While
I
log
into
my
domain
account
real
quick,
I
don't
want
you
to
see
all
the
crappy
domains
I've
reserved
over
the
years,
especially
in
a
recording,
but
give
me
a
second
here.
Let
me
login.
C
I
have
to
pay
immediately
okay,
but
I
mean
I
think
you
can
there's
a
referral
you
can.
I
can
try
to
set
you
up
at
the
trial,
but
it's
cheaper
than
aws
honestly
than
some
of
the
others.
I
don't
I
mean
so
you
could
argue
about
expensing
it.
I
just
don't
because
this
is
a
personal
project
of
mine,
so
I
have
to
set
this
up
anyways.
It's
a
personal
project
of
mine,
not
a
big
deal,
but
I
think
you
could
probably
argue
to
expense
it.
If
you
really
wanted
to.
A
It's
just
like
you
know
like
sending
up
a
cluster
or
whatever
on
g
gke.
It's
like
you
spend
so
much
time,
just
like
trying
to
figure
out
how
to
set
up
the
cluster
that
it's
almost
like.
If
there's
an
easier
way,
especially
with
you
know
like
what
is
it
like
account,
kind
of
stuff
and
authorization
kind
of
stuff,
you
know,
I
think,
that's
a
better.
I
think
that's
better
way
of
doing
it.
Even
with
the
you
know,
the
300,
whatever
hours
or
whatever
they
give
you.
C
Yeah,
no,
I
agree,
that's
why
I
use
it
partially.
It's
because
it's
super
easy
to
just
get
a
cluster
stirred
up
digitalocean
is
kind
of
the
same
way,
so
I'm
just
adding
this
ka8s
with
this
ip
address
to
my
to
my
domain,
if
it'll,
let
me
it's
not
saving
it,
I'm
just
adding
it
to
that
domain
so
that
that
subnet
or
whatever
will
work
am
I
still
here.
C
A
C
C
C
It
makes
a
lot
of
assumptions
for
you.
It's
really
easy
to
get
set
up
with
our
templates
and
if
you
have
a
project
that
meets
auto
devops
requirements.
Like
perfectly
you
know,
it's
super
simple:
to
get
a
pipeline
set
up,
but
when
you
start
getting
anywhere
underneath
the
hood
anything
complicated,
which
is
like
what
a
lot
of
our
enterprise
customers
have
when
you
hit
that
point,
auto
devops
just
goes
out
the
door
like
it
just
becomes
terrible.
A
C
C
C
C
C
A
C
I
did
not
make
the
endpoint
I
I
installed
the
ingress
controller.
The
ingress
controller
is
what
what
what
sets
up
the
accepts
the
internal
communication,
so
whenever
any
traffic
hits
that
ip
address
so
I
hit
the
domain,
I
go
to
blah
blah
dot,
k8s
dot,
abstract
os
dot.
That
traffic
goes
to
the
cluster
ip
that
was
given
and
then
because
the
ingress
controller.
C
C
C
A
C
So,
to
recap:
the
traffic
pattern
right
so
the
end
user
types
in
blah,
blah
blah
dot,
k
a
abstract,
and
that
goes
to
the
dns.
The
dns
says:
okay,
here's
the
ip
address
so
establishes
a
connection
at
the
front
of
the
kubernetes
cluster.
The
kubernetes
cluster
then
says:
okay,
we
have
a
service,
the
ingress
controller,
that's
on
port
80.
C
C
A
C
Yeah,
there's
actually
more
than
that,
but
yes,
because
there's
this
whole
thing.
That
also
happens
where
it's
like.
It
also
has
to
figure
out-
and
this
is
something
I'm
worried
about
and
we'll
see
if
it
actually
works
or
fails,
but
like
the
ingress
controller
right,
what
will
happen
is
sometimes
when
you
hit
that
cluster,
that
cluster
is
a
one
ip
address,
but
there's
three
computers
behind
that
cluster.
So
when
it
hits
that
cluster,
sometimes
what
it
does
is
it
round
robins
between
the
nodes
and
so
our
engine.
C
C
So
there's
a
lot
of
networking
stuff,
that's
happening
behind
the
scenes
and
that's
the
part
you
know
that's
the
product
kind
of
cut
out
from
that
it
just
you
know
it
hits
the
for
all
intents
and
purposes
for
the,
for
you
know
this
video
and
for
people
who
are
working
with
it.
The
connectivity
hits
the
kubernetes
cluster,
the
kubernetes
cluster,
directs
it
to
the
ingress
controller
through
its
bizarre
mechanisms
of
networking
and
then
the
ingress
controller
checks
to
see
hey
this
domain.
Does
it
does
it
correlate
to
a
pod?
A
B
A
B
A
C
So
once
good
quality
is
finished,
production
will
fire
up.
Hopefully
it
will
not
deploy
postgres
again
and
then
we'll
have
a
deployment.
We
have
the
namespace
ready
to
go
so
we'll
have
a
deployment
and
then
we'll
check
it
out
by
the
way
the
the
url
that's
going
to
happen
by
the
way
it's
going
to
be
like
this.
C
C
A
C
B
A
C
C
C
A
A
C
A
B
A
A
C
C
B
A
A
A
C
C
I'm
perfect,
thank
you.
I
have
a
load
bouncer
right
so
going
through
the
docs
and
everything
I
set
up
the
ip
address
for
that
address
right,
but
it
doesn't
actually
hit
anything.
I
think
part
of
the
reason
why
that
is
is
because
I
think
my
ip
address
is
wrong.
I
was
using
the
wrong
ip
when
voter
sets
it
up,
there's
a
specific
external
ip
that
goes
into
the
load,
balancer
and
I'm
pretty
sure,
that's
the
ip
we
want
to
use,
not
the
other
one.
C
C
C
C
And
everything
I
don't
care,
what
anybody
tells
you
or
sells
you
kubernetes
clusters
on
each
of
the
cloud
providers
is
wild,
are
wildly
different,
but
for
volterr
what
happens?
Is
they
set
up
a
separate
load
balancer
on
their
infrastructure
and
that
load
balancer
connects
directly
to
each
of
the
kubernetes
nodes.
B
C
B
C
A
A
A
C
A
C
C
C
C
B
A
B
A
C
Yeah
and
it's
kind
of
an
evolving
like
I
said
it's
an
evolving
target.
This
method
that
we're
doing
right
now
is
technically
considered
deprecated
because
they're
going
towards
the
kubernetes
agent
way,
but
it's
the
way
that
it
works
the
best
right
now.
Currently,
you
know,
and
later
on,
you
know
we'll
be
switching
to
the
this.
This
method,
which
involves
the
which
is
like
migrate,
it's
cluster.