►
Description
The Ortelius Journey with GitOps and Event Driven CI/CD - Brad McCoy, Moula
From early on the Ortelius team had seen value in events and how they can solve many of our problems that we face in our DevOps processes, particularly around Interoperability.
As they embarked on their journey implementing GitOps, they then realized that there were a lot of moving parts and tools in their CI/CD process so decided to adopt Keptn as the control plane to handle events. In this session we will explore how Ortelius, Keptn and GitOps form a new kind of pipeline that incorporates a central catalog for tracking the DevOps data in an event driven CD Pipeline.
For more Continuous Delivery Foundation content, check out our blog: https://cd.foundation/blog/
A
To
get
to
give
us
more
things
like
you
know:
testing
integration,
testing,
load,
testing,
quality,
gates
and
I'll
talk,
sort
of
high
level
about
the
concepts,
but
we
use
tools
like
argo
captain
and
and
as
at
the
end
I'll
sort
of
I'll
show
you
how
you
can
also
use
other
tools
like
flux,
for
example,
as
well,
so
a
bit
about
me
from
new
zealand
currently
living
in
australia.
At
the
moment,
I'm
a
cdf
ambassador
as
well
so
involved
quite
a
lot
in
the
community
and
I
contribute
to
open
source
as
well,
but
more.
A
A
They
had
some
sort
of
people
adopting
on
tls
that
had
like
30
qa
environments.
So
I
wanted
to
see
how
you
can
handle
the
microservice
spool
over
30
environments
and
different
dev
production.
So
that
was
quite
interesting
to
me,
so
it
was
also
quite
a
good,
easy
community
to
join
and
and
start
contributing.
It
wasn't
really
didn't,
have
a
lot
of
big
companies
in
it
as
well.
That
was
sort
of
taking
all
the
work.
So
it
was
quite
good
and
yeah
the
problem
statement
I
talked
just
about
then.
B
A
You
have
lots
of
environments
and
how
can
you
track
all
of
your
different
components
and
applications
so
before
we
started
the
the
get
ops
journey?
You
know
your
traditional
ci
cd,
some
some,
probably
a
lot
of
companies
that
you
have
worse
than
that.
So
essentially
it
was
both
ci
and
cd
together
and
then
that
would
just
be
all
in
one
script
pretty
much
and
go
to
the
the
dev.
The
test,
uat
prod.
A
A
lot
of
people
only
have
two
environments
well,
where
they
have.
Development
branch
goes
to
staging
and
prod.
So
we
we
wanted
more
and
that's
why
we
wanted
to
adopt,
get
ops
and
then
that's
where
we
brought
in
argo
cd.
So
we
had
the
choice
of
choosing
argo
city
and
flux,
but
because
it's
argo
was
had
the
ui
and
it
was
a
little
bit.
A
A
So
we
use
what's
called
customize
home
and
then
that
would
give
us
overlays
of
each
environment
as
well.
So
you
can
see
here,
that's
just
an
example
of
an
application
set
and
you
can
have
applications
set
for
each
environment
and
then,
when
you
need
to
update
your
your
images
or
your
manifest
files,
you
can
just
update
them
in
the
environment.
A
So
yeah
that
that
worked
quite
well
and
then
we
wanted
to
make
that
better
and
then
we
wanted
to
have
different
deployment
strategies
like
blue-green
canary
deployment.
So
we
then
started
to
introduce
argo
rollouts
as
well.
So
you
can
see
here
that
we
put
in
argo,
rollouts
and
prometheus
as
well,
so
using
argo
roll,
it's.
What
they
enabled
us
to
do
is
to
have
canary
deployments
and
there's
a
concept
called
analysis
run.
So
what
analysis
run
does
is
if
you
want
to
deploy
25
of
your
new
app
into
the
environment
analysis
run.
A
A
So
then
it
will
start
introducing
50
of
that
that
new
version
and
then
75
100,
for
example-
and
it
was
quite
fun
playing
with
argo
rollouts
because
they
have
this
really
cool
application,
called
agar,
all
that's
demo,
and
then
you
can
actually
test
these
concepts
working.
So
you
can
see
here
that
I
had
the
the
blue
app
there
and
then
I
would.
A
This
is
setting
25
weight
to
the
new
yellow
tag,
so
just
simple,
blue
and
and
yellow,
and
then
you
can
see
here
that
in
the
era
that
you
can,
you
can
give
it
errors,
and
then
you
can
see
it
rolling
back
as
well,
so
that
that
was
quite
a
cool
thing
to
do
because
norm.
Well,
I
always
work
in
devops
and
sometimes
when
the
developers
complain
that
something's
wrong-
it's
actually
their
fault.
A
So
this
sort
of
safeguards
them
coming
to
us
saying
that
there's
something
wrong
with
the
infrastructure,
because
you
can
see
it's
wrong
with
that
app
so
that
that
was
a
really
cool
project
that
we
enjoyed
playing
around
with
the
reason
we
chose
argo
rollouts
is
that
was
we
want
to
go
to
service
mission
instead
of
other
rollouts,
but
argo
rollouts
was
a
again
easier
to
learn
to
prove
these
concepts
of
what
we
wanted,
the
in-picture
to
look
like
and
that
application
called
argo
rollouts
as
well.
So
that's
you
can
find
that
on
github.
A
So
now
that
we
did
other
rollouts,
one
thing
that
we
found
pretty
easy
is
that
when
your
analysis
run
is
working
and
you're
you're
talking
to
prometheus,
if
you
don't
have
load,
testing
or
real
traffic
going
through,
it's
always
going
to
say
yes,
this
app
is
fine,
so
we
then
realized
well
what
happens
if
we're
doing
the
deployment
late
at
night?
There's
no
traffic.
A
A
So
this
is
what
a
rollout
specification
looks
like.
You
can
see
here
that
we
have
replicas
five
and
then
you
specify
your
your
strategy
canary
and
then
you
have
analysis
template
so
that
will
that's
how
you
specify
the
roller
it's
using
a
replica
set,
and
then
this
is
what
your
analysis
template
will
look
like
as
well,
so
you
can
see
you
can
you
can
run
any
prometheus
query.
A
We
have
here
just
a
simple
one
saying:
if
we
get
any
four
to
five
hundred
errors,
then
we're
going
to
say
this
and
you
don't
get
a
success
rate
of
0.95.
Then
it's
going
to
fail
and
then
roll
back
to
the
the
last
app
that
we
had
or
the
last
version
of
the
app
so
very
cool,
but
we
wanted
more
as
well
and
you
can
do
lots
of
more
things
with
it,
but
yeah.
So
here
you
can
see.
A
This
is
version
one
and
then
you'll
see
the
canary
working
and
the
analysis
runs
coming
up
here.
So
it
was
quite
a
cool
visual
to
see
argo
and
all
of
your
you
know,
visualizing.
What's
actually
the
canary
deployment,
which
is
very
cool
and
yeah,
so
that
was
great
and
then
we
we
had
that
working
and
then
this
is
where
we
wanted
to
go
another
step
further
and
introduce
captain.
So
another
argo
app
from
the
family
is
called
argon
notification.
So
that's
just
been
into
the
the
argo
cd
repro
now
as
well.
A
So
it's
shipped
now
with
argo
cd
and
what
the
argo
notifications
will
do.
Is
it
will
it
will?
Let
you
on
sync
I'll
show
you
soon
here,
so
it
will
have
something
called
triggers,
so
you
can
have
a
trigger
on
app
created,
app
deleted,
etc.
So
we
also
wanted
to
so.
First
of
all
I'll
say
what
is
captain
so
essentially
it's
quite
hard
to
understand
what
captain
is
when
you
first
look
at
it.
So
it's
a
event
orchestrator
that
has
many
different
capabilities
of
it.
So
it's
essentially
you
can.
A
A
So
you
can
also
do
things
like
remediation
as
well,
so
it
will
send
an
event,
a
problem
event
like
through
dynatrace,
for
example,
and
then
that
will
trigger
a
sequence
which
we'll
talk
about
soon,
so
that
that
was
a
big.
You
can
use
argon
notifications
for
for
slack
teams
they
have,
they
have
the
concepts
of
service
template
and
trigger,
so
the
the
trigger
will
come
in
from
argo
cd
and
then
it
will
whatever
based
on
your
template.
A
So
we
use
ones
like
out
of
sync
is
quite
good,
because
you
can
once
argo
becomes
out
of
sync,
then
we
can
get
captain
to
go,
say,
okay,
go
see
if
this
is
going
to
work,
and
this
here
is
yeah,
that's
a
essentially
a
specification
of
argo
notifications.
So
you
can
see
here
that
the
template
and
really
that's
just
a
payload
of
a
cloud
event
and
then
within
the
cloud
event,
is
the
the
data
that
we
need
for
captain
to
to
go
through
its
sequence.
A
Even
you
could
do
chaos,
engineering
and
some
really
cool
stuff.
So
that's
how
it
looks
when
it
comes
in
and
then
that
will
orchestrate
through
the
steps,
so
in
captain
there's
a
concept
of
a
shipyard
file.
So
it's
quite
hard
to
to
understand
what
it
is
the
first
time,
but
essentially
a
shipyard
file
is
three
things,
so
you
can
see
here
it
has
a
stage,
a
sequence
and
a
task.
A
So
I
can
send
the
cloud
event
and
say:
okay
go
trigger
this
stage
and
then
it
will
go
through
my
tasks
and
my
services
will
be
subscribed
to
them,
which
we'll
see
soon
as
well,
so
a
little
bit
a
lot
to
take
on
straight
away.
But
essentially
it's
a
it's
a
declarative
way
of
saying.
I
want
to
do
things
so
I'm
not
I'm
not
locked
into
things
like
okay.
A
I
want
to
use
j
meter
for
for
load
testing
and
if
I
want
to
use
k6,
I
can
just
flip
that
out
and
subscribe
to
that
event
as
well.
So
with
captain,
when
you
go
through
a
task,
you'll
have
three
events:
it
will
be
essentially
started,
I'm
doing
things
and
then
I'm
finished
or
I've
failed,
and
you
can
see
it
here
as
well.
So
this
is
what
it
looks
like.
So
essentially
here
you
have.
A
These
are
the
events
over
here
on
the
side,
and
it's
saying
that
I
want
to
subscribe
to
these
events
and
then
these
services,
so
so,
when
that
event
kicks
in
that
you
subscribe
to,
it
will
then
go
use
the
services.
So
there's
many
services
in
kept
in
as
well,
so
we
have
sandbox.
You
can
build
your
own,
so
so
at
the
bottom
here
we're
building
the
autelius
one
and
that's
on
this
event
here
when
the
catalog
is
triggered.
A
So
that's
where
we're
working
on
at
the
moment
where
we're
gonna
have
to
catalog
the
the
changes
and
the
components
of
the
microservice,
so
you
can
also
use
the
webhook
service
as
well,
which
is
quite
generic
and
easy
to
use
so
yeah
that
hopefully
that
makes
sense.
But
essentially,
if
we
go
back
to
the
shipyard
file,
you
could
subscribe
to
the
the
sequence
in
the
in
the
stage
here
and
and
then
you
can
see
here
that
you
know
these
are
all
the
events.
A
So
it's
it's
very
powerful
because
obviously
it
especially
big
enterprises
when
you
get
a
new
boss,
come
over
he's
like
okay.
Let's
change
this,
then
a
new
one
comes
okay!
Let's
go
back
to
this,
so
it
allows
you
to
take.
You
know
your
devops
tools
in
and
out,
which
is
quite
handy
and
for
me
for
me:
it's
really
powerful,
because
it's
you
can.
A
If
you
have
a
ci
tool
or
a
ci
cd
tool
that
you
you
need
to
pay
for
the
runners,
etc,
that
can
clock
up
lots
of
hours
waiting
for
things
synchronously
and-
and
this
is
just
a
really
nice
way
to
event-based
change
and
then
then
the
captain
tim
came
along
and
then
said:
hey.
We
want
to
work
on
this
as
well,
so
it
wasn't
just
the
autelia,
so
we
sort
of
joined
forces
and
now
we're
working
on
this
integration
here.
So
it's
in
the
sandbox,
the
the
captain,
argo
city
integration.
A
A
It
took
me
probably
half
a
year
on
this
project,
so
there's
a
lot
of
learning
to
be
involved,
so
we're
trying
to
make
it
easier
for
the
next
person
that
comes
along
as
well,
because
it's
quite
a
steep
learning
curve
to
learn.
You
know,
argo
captain
get
ups,
there's
a
lot
of
things
so
and
that's
the
the
next
iteration
again
of
this,
so
we're
continuously
improving
on
this
and
just
taking
it
step
by
step.
So
this
is
what
thomas
came
up
with.
A
We
don't
want
to
rely
on
argo
rollouts,
because
some
clients,
or
or
folks
don't
want
to
necessarily
use
rollout.
So
therefore,
where
we
want
to
make
it
more,
you
can
bring
anything
you
want,
so
it
could
be
just
a
deployment
or
automatic
whatever
you
have
and
then
now
we're
saying
that
okay,
so
when
it's
out
of
sync,
the
argo
will
not
notice
the
change
and
then
it
will
kick
off
the
deployment,
and
this
is
going
back
to
what
I
was
saying
before
about
how
kitten
works.
A
And
then
you
have
your
deployment
that
says
start
it
says:
yes,
I'm
I'm
doing
it
and
then
it
will
come
back
and
say
I've
finished,
so
you
could
in
that
deployment
tooling.
You
could
actually
put
anything
that
you
want
so
the
great
thing
about
captain
is
you
bring
your
own
tools?
You
can
use
anything
to
deploy,
there's
pretty
much.
Every
tool
now
has
its
own
form
of
deployment
as
well.
A
It
doesn't
mean
it's
better
or
worse,
but
you
might
just
have
a
personal
preference
of
using
your
own
tooling
and
then
you'll
see
that
you
can
also
run
tests
as
well.
So,
whatever
tools
you
want
to
run
testing,
you
can
use
them
and
you
can
use
one
or
many
you
can
use
as
many
as
you
want
you
just
in
the
shipyard
file
that
we
said
you
just
put
that
into
the
the
tasks
in
that
sequence.
A
So
this
is
where
it
gets
really
cool
because
you
can
start,
you
can
do
things
like
integration
testing
and
you
know
we
want
to
start
to
do
chaos
engineering
in
there
as
well
and
and
just
have
a
really
solid
event-driven
pipeline,
and
then
the
thing
that
sort
of
first
attracted
me
to
captain
before
I
knew
what
it
really
did
was
the
quality
gates.
But
that's
just
a
small
part
of
it.
So
you'll
see
here
that
you
have
evaluate-
and
that
will
go
around
in
here-
it's
down
trace,
but
that
could
be.
A
A
It's
pretty
much
there,
but
we're
still
doing
a
lot
of
testing
and
and
getting
it
right
and
then,
essentially,
you
can
subscribe
to
any
event
that
you
want
and-
and
you
can
make
this,
however,
generic
that
you
want
and
and
leverage
other
people's
work
from
the
marketplace
as
well
or
the
sandbox.
Should
I
say
so.
Thank
you.
I
hope
that
made
sense.
It's
quite
heavy
to
learn.
A
Quite
you
know
it's
two
quite
big
topics,
but
if
you
do
want
to
join
the
community,
we
have
ortillius
community
and
a
captain,
one
so
join
us
like
and
say
hello,
everyone's
friendly
and
we're
more
than
happy
to
help
you
adopt
this.
If
you
want
or
if
you
want
to
learn,
if
you
want
to
contribute,
it's
a
nice
place
to
do
that
so
yeah.
Thank
do
you
have
any
questions.