►
From YouTube: Ops Section Product Walk Through - Kubernetes Agent
A
Hey
everybody:
my
name
is
kenny
johnston.
I'm
a
director
of
product
here
at
gitlab,
I'm
going
to
be
doing
a
walkthrough
of
our
new
installation
of
our
kubernetes
agent.
So
let
me
go
ahead
and
share
my
screen
this
one.
Okay,
so
just
walk
through.
I
was
expecting
to
be
able
to
set
up
the
agent
on
gitlab.com
did
have
to
have
my
project
authorized
to
be
able
to
create
agents,
or
at
least
agent
artifacts
in
the
database.
So
I
had
that
done
before
starting
this
discussion.
A
I
figured
I
would
just
start
with
googling
gitlab
kubernetes
agent
set
up
and
get
started
so
gitlab
kubernetes
agent
setup.
A
All
right,
so
obviously
I've
been
here
before
so
here's
gitlab
kubernetes
agent
can
be
good
for
integrating
more
secure,
pull
based
with
gitops
engine
all
right.
Okay,
here's
how
it
works.
Developer,
pushes
code
that
updates
the
repository
and
the
manifest,
and
that
manifests
the
agent
is
watching
and
pulling
what's
going
on
in
that
manifest.
And
then
the
configuration
of
the
agent
is
in
a
potentially
different
location,
getting
started,
install
the
agent
server.
A
I
have
already
done
that
because
I'm
using
gitlab.com
on
github.com
under
cast.gitlab,
okay,
so
I
can
skip
that
part.
I
don't
need
to
install
the
cast
okay,
I
need
to
define
a
configuration
repository,
so
I
created
this
example.
Project
called
quantization
to
test
and
what
did
it
say?
Oops?
I
already
have
this
hooked
up.
Let's
stay
over
here
define
a
configuration
browser.
Next,
you
need
a
gitlab
repository
to
contain
your
agent
configuration
minimal.
Repository,
looks
like
this
that
gitlab
agents.
Okay.
So
I'm
going
to
name
my
agent.
A
A
A
So
right
now,
so
this
is
saying
the
configuration
for
the
agent
when
it
gets
figured
is
hey.
Where
should
I
look
for
manifest
projects?
I'm
going
to
look
at
myself?
This
is
this
same
project,
so
got
that
part
done
back
to
instructions,
create
an
agent
record
in
gitlab.
Next,
create
a
gitlab
rails
agent
record
to
associate
it
with
configuration,
repository
project.
B
B
A
A
A
A
A
A
B
A
A
When
I
connect
to
it,
I'm
going
to
run
this
command.
That
will
basically
use
my
token
and
do
a
cubesatl
apply
of
a
specific
community
service
description
that
I
think
I'm
going
to
get
from
this
docker
image.
So
I'm
going
to
pull
a
docker
image,
that's
going
to
generate
an
agent
yaml
definition
for
my
kubernetes
cluster.
A
A
A
A
A
A
B
A
A
A
A
B
A
Clusters,
so
just
for
the
video
I
hit
refresh,
even
if
I'm
looking
at
agent
managed
clusters,
I
still
get
flipped
back
to.
A
Okay,
but
I
have
this
agent
record
and
a
token
for
that
agent
so
doing
I
was
going.
B
A
A
A
A
A
A
A
Okay,
let
me
see
what
else
so
now
that
I
have
this
set
up.
I
wonder
if
I
could
are
there
any
simple
adjustments?
I
understand
replicas,
let's
change
the
replicas
from
two.
B
A
A
I
guess
you
could
think
about
how,
if
you
have
this
process,
you
could
have.
I
wonder
if
you
can
have
your
agent
different
agents
looking
at
different
branches.
A
So
what
would
be
great
is
if
I
guess,
if
I
had
staging
in
production,
that
I
could
have
my
staging
look
at
a
staging
branch
and
then
the
agent
for
that
staging
cluster
always
pulls
from
that
from
the
specific
branch.
That's
in
staging.
Let's
just
add
a
note
about
that.
B
A
A
A
A
A
Manifest
into
it
and
using
the
one
replica,
and
then
I
would
have
this
pull
request
to
go
to
master,
which
would
update
master
if
merged.
You
could
also
think
about
that
from
yeah.
You
could
have
like.
A
It's
kind
of
like
similar
to
how
our.
A
Certificate
based
cluster
integration
works
today,
so
cool
I'm
going
to
stop
recording
fairly
straightforward,
that
one
error
that
I
think
is
likely
more
due
to
the
fact
that
I'd
had
some
of
those
resources
already
created
on
the
cluster,
but
maybe
not,
and
I'm
gonna
merge
this
and
then
shut
down
my
cluster.
A
Two
from
five
to
one:
it's
cool
all
right,
thanks.