►
Description
Now that we've looked at other continuous integration (CI) tools, let's see how Tekton stacks up. Tekton is a young yet powerful CI tool for Kubernetes and OpenShift that's optimized for building and deploying cloud-native, microservice-based applications. When it comes to CI, there are lots of options, each with a slightly different set of features, so we'll look at why you'd want to use it and dive into the features Tekton offers.
About Tekton:
Tekton is a powerful and flexible open-source framework for creating CI/CD systems, allowing developers to build, test, and deploy across cloud providers and on-premise systems.
https://tekton.dev/
Follow us:
Christian Hernandez https://twitter.com/christianh814
Chris Short https://twitter.com/ChrisShort
What is GitOps Guide to the Galaxy?
Every other Thursday at 3am ET hosts Christian Hernandez and Chris Short sit down to discuss everything in the GitOps universe, from end-to-end CICD pipelines to creating Git workflows.
Subscribe to Red Hat's YouTube channel: https://www.youtube.com/redhat/?sub_confirmation=1
About OpenShift:
Red Hat OpenShift is an open source container application platform based on the Kubernetes container orchestrator for enterprise application development and deployment.
Openshift.com https://red.ht/33Fh6uV
OpenShift on github https://github.com/openshift/
OpenShift on redhat.com https://red.ht/2o4ccDk
#RedHat #GitOps #Kubernetes
A
B
It's
kind
of
like
without
ci,
really
you're
just
doing
infrastructure
as
code
with
get
up
so
so
you're
not
even
really
doing
get
ups.
If
you're,
if
you
don't
have
your
ci
process,
so
I
think
this
this
would
be
a
good
show
to
introduce
techton
as
sci
right
kind
of
explored
a
little
bit.
I'm
going
to
be
doing
a
lot
of
hacking,
I'm
going
to
be.
I
have
a
presentation.
I've
been
doing
a
lot
of
hacking,
so
everyone
sit
tight.
B
Yeah
yeah,
so
you
know
but
sit
tight.
I
might
go
a
little
long
today.
So
for
those
who
are,
you
know
who
have
a
who
have
a
heart?
Stop
just
rew.
You
know
you
can
rewatch
the
the
the
recording
after
but
just
fair
warning.
I
might
go
along
because
I'm
gonna
kind
of
go
over
tech
down
a
little
bit.
Then
I'm
gonna
hack
at
it.
So
I'm
gonna
go
from
zero
to
one
and
then
I'm
gonna
go
from
one
to
ten
right.
So
this
isn't
gonna
be
not
even
a
crash
course.
B
B
B
B
Yeah
so
yeah,
somewhere,
yeah
he'll
he'll
put
that
in
the
chat
somewhere
and
so
I'll.
I'm
gonna
go
over
like
what
techton
is
right.
So
like
ci,
I
I
hope
some
most
of
you
know
is
you
know
it's
something
that
everyone
knows
right
already
so
ci,
I'm
not
gonna,
go
too
deep.
What
ci
is
I'm
pretty
sure
someone
experienced
ci?
Oh
also,
hillary
was
supposed
to
be
on
chat.
She
said
she's
gonna,
braided
silver
by
the
way
and
that
that's
like
her
her
background
to
ci.
So
she'll.
A
B
A
B
Table
here
I'm
not
going
to
read:
I'm
not
going
to
read
off
the
tape
this
table
here.
There's
the
links
in
the
chat,
so
you
guys
can
take
a
look
at
that
presentation,
but
mainly
is
the
traditional.
What
we
call
traditional
ci
cd
was
built
before
cloud
native
architecture,
so
it
has.
It
has
a
lot
of
tech
debt
behind
it
right
and
that's
you
know,
I'm
not
saying
that
as
a
to
be
graduate
or
anything
it's
just.
That's
just
the
state.
The
state.
B
And
you
know
it
just
happens
right
and
so
and
and
the
why
behind
the
cloud
native
cicd
is
essentially
it's
like.
We
don't
want
to
bring
over
all
that
debt
attack
debt
onto
a
cloud-native
platform
right.
So
some
things
like
the
cia,
traditional
city
is
really
designed
for
virtual
machines
and
cloud
native
cicd
is
designed
for
containers
and
really,
I
think,
really
that's
the
core
of
it
and
then
everything
else
kind
of
kind
of
falls
into
place
after,
if
you
keep
that
in
mind
right-
and
so
you.
B
Yeah,
that's
that's
right
exactly
this
is
the
peanut
butter
and
jelly
aspect
of
it
right
is
you
know,
cloud
native
ci
process
right
and
r
and
really,
let's,
let's,
let's,
let's
not
beat
around
the
bushes,
but
I'm
talking
jenkins
versus
tank
ton
right
and
so
and
not
not
to
say
anything
bad
about
jenkins,
but
even
jenkins
itself.
They
know
this
limitation
because
there's
projects
like
jake
and
x,
right
yeah,
so
which
is
cloud
native
jenkins
right.
B
A
B
B
B
B
We're
gonna
hear
me
talk
about
steps
right,
so
steps
is
essentially
like
running
commands
in
a
container
right,
like
git,
clone
maven
build
that
sort
of
thing
right,
a
docker
push
and
a
task
is
essentially
a
list
of
steps.
So
so
task
is
nothing
but
a
collection
of
steps
to
take
right.
So
a
task
can
be
like.
B
Like
the
previous
example,
I
said
like,
let's
say
build,
you
know,
build
application,
you
know,
get
clone
right
and
then
task
will
do
a
git
clone
and
the
step
will
do
the
git
clone.
But
the
task
wraps
that
into
a
a
workable
unit
here
and
then
a
pipeline
is
nothing
but
a
collection
of
tasks.
So
you
have
a
pipeline
we'll
run.
One
task
test
run
on
the
other
task,
run
the
other
tasks
and
they
could
run.
B
You
know
those
tasks
in
parallel
or
serially
right,
so
it
kind
it's
kind
of
like
it
builds
on
each
other
right.
You
can
think
of
those
little
russian
dolls
right.
A
step
is
nothing
but
a
collection
of
tasks,
tasks.
Oh
sorry,
a
task
is
nothing,
but
a
collection
of
steps
pipeline
is
a
collection
of
tasks
pipeline
resource,
it's
actually
deprecated.
We
don't
use
that
anymore.
B
A
task
task
run,
and
this
will
make
more
sense
once
I
once
I
start
drilling
down,
but
a
task
run
is
a
invocation
of
a
task
right.
So
a
task
is
nothing,
but
a
definition
and
task
run
is
incessant.
Instantiation
of
that
definition,
so
it's
kind
of
a
difference
between
an
image
and
a
container
right
container
is
a
running
version
of
the
image
right.
So
it's
kind
of
that
sort
of
paradigm
and
the
same
thing
with
pipeline
run
right
pipeline
run,
is
an
instantiation
of
a
pipeline
right.
A
B
Especially
today
has
been
so
so
steps
right.
So
here
when
I
talk
about
steps
like
maven
build
is
a
step.
You
know,
you
know
parsley.
Ammo
is
a
step.
Git
clone
is
a
step
right,
so
they're
container
specific.
So
when
you
are
providing
environment
variables,
volumes
can
big
map
secrets
right.
You
are
you're
providing
that
to
the
the
step
right.
So
that's
what
what
a
step
is.
So
a
task
is
a
collection
of
steps
right.
B
B
So,
for
example,
here
you
know
task
maven,
right,
mdm
build
right
and
you
can
parameter
and
like
you
and
then
basically,
these
are
supposed
to
be.
I
guess
agnostic
right,
like
you,
your
your
steps
are
specific
and
then
your
tasks
are
agnostic
because,
like
here
you
can
say
here
here
are
my
parameters.
B
So
when
I
do
a
maven
build,
I
do
maven
build
and
then
something
right
and
that's
something
that's
passed
via
task
via
parameters
like
git
clone
is
a
task.
I'm
sorry,
git
clone
is
a
step.
Git
clone
a
specific
repo.
That's
a
task
because
you're
passing
that
repo
as
a
parameter
so
and
then
a
task
run
is
basically
like.
I
said
before
an
instantiation
of
that.
You
know
you
have
your
tasks
set
up
and
a
task
run
is
when
you,
when
you
run
the
task,
it'll
ask
you
okay.
B
Well
what
you
know
like
my
previous
example
of
git
clone
right:
okay,
well,
repo!
You
want
me
to
clone
and
then
once
it
clones
it,
that's
a
task
run
right
at
the
very
end,
it's
like
the
end
result
of
all
that
right,
so
in
a
task
run,
runs
in
a
in
a
pod.
Essentially,
so
each
task
runs
in
a
pod,
and
so
a
pipeline
is
a
collection
of
tasks
right.
So
you
can
kind
of
see
how
I'm
building
this
already
is.
I
have
there's
steps.
B
A
task
is
a
collection
of
steps
and
a
pipeline
is
a
collection
of
tasks
and
then
basically
defines
the
task
execution
right
in
the
order.
So
you
can
do
things
like
serially
right,
like
one
two,
three
four,
you
can
have
things
run
in
parallel,
right
things
that
are
that
you
want
to
run
at
the
same
time
that
don't
really
depend
on
each
other,
and
then
you
can
put
dependencies
in
right.
I
want
you
to
run
task
one
after
test.
Sorry,
I
want
you
to
write
task
two
after
task.
B
One
task,
four
doesn't
matter
when
you
run
it,
you
can
run
it
just
as
long
as
you
run
it
after
task
two,
but
in
parallel
to
test
three
right,
you
can
get
really
really
crazy.
With
this,
as
I
as
I've
learned
right
and
so
just
just
kind
of
a
heads
up,
I
I
learned
techton
by
just
beating
at
it
for
a
while.
A
B
It
just
was,
it
wasn't
making
sense
to
me,
but
then,
once
it
made
sense,
I
go
okay
yeah
I
could.
I
could
see
the
power
in
in
in
tecton
and
I'll
and
I'll
call.
I
guys
I'll
show
you
some
good
resources
after
this
right,
so
kind
of
powering
through
this
right.
So
if
every
task
runs
in
a
pod,
you
know
how
am
I
supposed
to
chain
these
tasks
together
in
a
pipeline
right?
B
B
You
know
all
kinds
of
nodes
in
your
cluster,
so
you
need
to
use
a
persistent
volume
right
so
and,
if
kind
of
kind
of
a
fun
fun
fact,
if
you
use
a
persistent
volume
that
has
block
storage,
all
your
all
your
tasks
run
on
the
same
node.
Just
because
that's
just
like
you
can't
because
it's
block
storage,
you
can't
share
it
between
nodes,
since
so
it's
kind
of
a
fun
secret
that
I
found
out
that
may
or
may
not
matter
to
you,
but
just.
B
Well,
I
was
overloading,
like
my
node,
wasn't
scheduling
anything
because
I
was
like
overloaded.
I
was
like
oh
I'm,
using
block
storage
yeah.
That
makes
sense,
like
all
the
pods
want
to
jump
on
the
on
this
one
node,
because
so,
if
you're,
so
just
you
know,
read
wide
read,
write
once
versus
read,
write
many
use
rewrite
many.
If
you
can
so
that
way,
your
pods
schedule
in
different
nodes,
so
pipeline
run
kind
of
like
a
task
run
is
basically
an
of
your
pipeline
right.
B
You
define
your
pipeline
and
when
you
say
pipeline,
run
it'll
it'll
ask
you
the
series
of
questions
right.
The
parameters
whatever
you
put
in,
and
you
know
when,
when
I
when
I
start
hacking
away
it'll
make
more
sense.
But
that's
what
a
pipeline
running
is
right
triggers
are
pretty
cool
triggers
is
something
I'm
going
to
show
as
well.
It's
basically
something
that
well
the
way.
The
reason
I'm
the
way
I'm
using
it
is
that
I'm
using
it
get
web
hooks
right.
When
I
make
a
commit
it'll.
B
B
Definitely
definitely,
and
so
so
yeah
so
you
have
a
the
event.
Listener
has
a
trigger
template,
meaning
that
when
you
do
a
pipeline
run
it'll
ask
you
for
the
parameters,
but
you
can
pre-predetermine
that
parameter
right.
So,
when
someone
hits
this
web
hook
pass
these
parameters
to
this
pipeline-
and
you
know
you
can
you
can
already
see
how
things
are
starting
to
chain
together
so
so
tasks
as
you'll
see,
you
can
pretty
much
write.
Anything
is
like
you
can
do.
B
B
B
B
A
B
B
Yeah
yeah,
which
which
you'll
see
some
of
the
stuff
that
I've
done
here,
so
this
is
actually
pretty
easy
right
with
the
pipeline
architecture.
You
define
a
pipeline,
you
know,
it'll,
run
many
tasks,
and
once
you
have
those,
then
you
can
reuse
it
right
and
for
many
things
right.
So
technically
speaking,
you
can
define
up
one
pipeline,
but
use
it
for
many
pipeline
runs.
B
So
this
is
kind
of
a
paradigm
that
the
jenkins
people,
you
know
those
those
who
are
really
into
jenkins.
This
will
mess
you
up
because
it's
usually
a
one-to-one
relationship.
Right
like
this
is
my
project
fubar
pipeline.
This
is
my
project
baz
pipeline
and
as
their
two
distinct
pipelines.
Actually
in
tectonics
is
one
pipeline
right
like
it's
generic
enough
to,
whereas
one
pipeline
the
pipeline
run
is
what
differentiates
those
pipeline
runs,
and
so.
B
You
could
you
could
essentially
pipeline
as
a
template
and
then
you
can
use
that
over
and
over
and
over
and
over
again.
So
this
will.
This
messed
me
up
a
little
bit
and
I
still
haven't
quite
figured
out
how
to
make
a
pipeline
template,
but
that's
the
idea,
so
I'm
just
kind
of
just
giving
you
the
the
nirvana
concept.
B
B
B
B
B
A
A
B
B
B
B
Come
on
come
on
cluster.
There
you
go,
do
it,
you
can
do
it
and
so
I'm
actually
like.
Hopefully
they
don't
delete
this
cluster
as
I'm
using
it.
So
we'll
see
it
should
it's
less
than
24
hours.
So
I
guess
it's
good.
So
now,
if
I
do
oc
get
tasks,
oops
trasks,
it
says
there,
but
actually
there's
a
techton
cli
I
can
do.
I
think
it's
task
list
right.
It
essentially
uses
the
same
thing
so
nice.
So
now
I
have
the
task
loaded
up.
Let's,
let's
run
it
right.
B
Oh
no
yeah,
no
she's
she's
talking
about
something.
No,
I'm
not
I'm
not
using
that
that
that
cluster
I'm
using
something
else,
yeah
I'm
using
something
else
yeah.
So
here,
techton
task
start
right.
You
see
it
says
hello,
world,
pretty
simple!
Essentially
right,
if
you
say
hey
run
this
task.
If
I
do
oc
get
pods,
I
think
yeah.
A
B
Completed
right
it,
iran
actually
ran
a
pod
there
and
I
just
ran
that
and
it
exited.
So
if
I
do
oc
get
task
run
right,
it's
essentially
there.
So
if
I,
if
I
wanted
to
run
it
again,
right
run
it
again,
it'll
right,
hello,
world
beautiful.
If
I
do
ot
get
a
task
list
right,
it'll
have
one
task.
I
do
get
tasks
right,
one
task,
but
if
I
do
task,
run
right,
there's
multiple
tasks,
so
this
is
kind
of
like
this
will
show
you
it'll
say
every
time
it
runs.
B
It
essentially
creates
a
new
cloud
and
runs
that's
a
new
instantiation
of
that
so
cool.
So
that's
like
the
simple
hello
world.
Let's
add
some
parameters
right.
So
let's
go
parameters
so
this
file
here,
so
it
says
so.
I
want
to
be
able
to
reuse
this
right.
I
don't
want
to
just
say
hello
world.
I
want
to
say.
A
B
And
then
someone's
a
name,
so
here
I'm
going
to
parameterize
it
right,
I'm
going
to
say
I'm
going
to
create
a
parameter
called
person,
and
this
is
kind
of
description
name,
a
person
to
greet-
and
this
is
it's
a
type
of
string
right,
so
you
can
actually
have
types.
You
have
string.
You
have
booleans,
you
know
you
have
ins
whatever
right
and
then
the
default
is
world.
So
if
no
one
specifies
anything,
I'm
just
going
to
say
hello,
world
and
essentially
here.
This
is
how
you
specify
the
parameters.
B
B
B
A
B
A
B
A
B
B
He
he
he
he
always
keeps
you
engaged
even
in
writing.
So
it's
always
fun
right.
So
it's
something
to
say
right.
So
what
should
I
say,
the
defaults
hello?
How
long?
I
should
wait
right.
So
it's
kind
of
giving
me
a
something
to
sleep
on
right
because
right
step
here
it
says
sleep,
param,
pause,
duration
and
then
say
whatever
you're
going
to
say
right,
so
I
kind
of
chained
those
together.
That's.
B
Yeah,
so
you
can
kind
of
just
see.
You
know
asleep,
however
long
then
echo
whatever
you're
going
to
say
so
in
the
pipeline.
Let's
come
down
here
now.
I
have
two
tasks
so
task
one.
This
is
kind
of
hard
to
see,
so
I'm
gonna
highlight
it
a
little
bit.
So
this
is
task
one
from
29
to
36,
so
task,
one
I'm
gonna
pass.
I
want
you
to
sleep
for
two
seconds
and
then
say
hello.
This
is
the
first
task
in
the
task
ref
meaning
like.
B
B
A
B
B
B
B
A
B
B
This
pi
this
next
pipeline
is
also
I'm
going
to
pause
for
two
seconds
and
then,
but
if
you
notice
here
lines,
23
and
24,
I'm
creating
a
dependency
here.
I
want
to
say
run
this
second
task
right
after
the
first
one,
so
in
in
the
previous
example-
and
we
we
you
couldn't
see
it
because,
just
because
of
the
nature
of
the
thing,
but
both
of
those
pods
fired
off
at
the
same
time,
because
we
didn't
build
a
dependency.
B
B
I
have
also
a
dependency
here
on
the
third
third
task
run
after
the
first
task,
but
I'm
not
putting
a
dependency
on
the
second
task
right
it
I
see,
I
have
no
dependency
on
the
second
task,
only
the
first
task,
so
this
so
this
kind
of
says
the
second
task
and
the
third
task
will
run.
At
the
same
time,
however,
I
want
you
to
note.
The
value
of
the
third
task
for
to
sleep
is
one
and
the
second
one
is
two,
so
this
will
produce
an
interesting
result
and
then
the
fourth
task.
B
A
B
A
B
B
B
B
And
so
so
I
have
a
pipeline
here
and
it's
doing
a
bunch
of
things
right
and
so
first
thing
I'm
doing
is
I'm
cloning
the
rainbow
and
that's
like
one
task
right.
I
clone
the
repo
and
then
I
set
an
image
tag
right,
so
I
define
so
I'm
not
a
big
fan
of
floating
tags
right,
so
floating
tags
for
those
you
don't
know,
is
like
dev,
prod
stage
right
yeah.
So
like.
B
Yeah
yeah,
I'm
not
I'm,
I'm
not
I'm
not
a
big
fan
of
those
tags,
I'm
actually
more
of
a
fan
of
like
a
version
like
v
2.1.3
or
what
I'm
doing
here
is
I'm
just
taking
the
I'm
chopping
off
the
last
part
of
the
hash.
In
my
repository,
my
code
repository
that
specific
git
commit
I'm
taking
that
tag,
I'm
taking
that
hash
and
I'm
just
using
it
as
an
image
tag.
So
I
know
what
version
is
what
and
so
that's
task
two
test.
Three
is:
has
many
steps
right
so
here
these.
B
B
I
so
here
I
have
a
parallel
task
right.
I
have
two
tasks
that
are
independent
of
each
other,
so
I
run
them
in
parallel,
so
I
do
a
scopio
copy
of
the
of
the
task,
the
of
the
the
tag
that
I'm
using
to
latest
right.
So
my
my
dev
version
is
always
latest,
so
I
just
do
a
scopiocopy
making
sure
the
image
is
updated
and
then
I'm
cloning,
my
deploy
repo.
B
B
Yeah
this
this
is
my
code
repo.
This
is
nothing
but
my
source
code
right,
and
this
is
what
this
is,
what
I'm
building
here.
I
actually
have
a
deployment
repo,
so
I
have
two
repos
for
my
application.
I
have
one
for
the
actual
source
code
and
one
for
the
actual
get
ops
stuff
right,
which
makes
perfect
sense.
Yeah.
You
want
to
keep
those
separate,
a
big
proponent,
trying
to
do
it
all
in
one.
Repo
is
heading
so.
B
Well,
and
also
they
have
different
life
cycles
right.
My
application
is
going
to
be
changing
a
lot
versus
my
infrastructure,
not
as
often
not
as
often
as
you
think,
and
so
so.
I'm
cloning
that
deployment
repo
right
the
deployment
repo,
where
I
defined
a
lot
of
these
like
right,
I
deploy
this.
I
have
an
overlay
for
a
dev
overlay
for
production
right
and
I'll
I'll
show
you
how
that
works
in
a
little
bit
so.
B
And
so
I
have
you
know
I
cloned
the
the
deployment
repo
and
I
do
that
in
parallel
with
scorpio
copy,
because
they're
they're
not
related
so
now,
I'm
using
customize
to
patch
the
image
inside
my
deployment
repo.
So
I
use
customize
set
image
and
that'll
update
my
customized
file
in
the
git
repo
and
then
I
actually
commit
to
dev.
So
since
it's
my
dev
environment,
I
just
screw
it
right.
Just
push
it
straight
to
dave.
I
want
to
see
this
could
be
sandbox.
B
It
could
be
whatever
environment
you
want,
but
in
my
case
it's
dev.
So
I
just
want
to
do
a
get
push
of
that
update
that
I
made
next
step.
Is
I'm
going
to
patch
prod,
so
I
want
to
say,
hey
production
I
want
to.
I
want
to
also
update
the
image
for
production,
but
instead
of
committing,
I
want
to
create
a
branch
and
then
the
last
step
is
create
a
pr
right
and
so
notice.
All
of
the
steps
here
in
tecton
there's
absolutely
nothing
to
do
with
argo
right.
B
A
B
Right
all
argo's
doing,
and
here
here's
my
dev
environment,
all
argo's
doing
is
looking
at
my
git
repo,
so
techton
right
does
the
ci
argo.
Does
the
cd
part
and
git
is
the
center
of
everything
right,
and
so
you
know
chris.
We
always
say
that
git
ops,
isn't
a
tool
to
practice
right,
but
if
get
ops
was
to
is
is
a
tool,
it's
not
argo.
It's
not
flux.
It's
not
acm!
It's
actually
get
right.
B
B
B
You
know
the
difference
between
production
and
development
or
like
a
dev
and
sandbox
is
actually
not
a
lot
right.
What's
what's
really
different
is
maybe
the
image
you're
using
and
the
secrets
environment
variables,
but
the
the
base
is,
and
now
I'm
kind
of
talking
about
customized
the
base
is
usually
the
same
deployment
like
deployment
service
route.
B
A
B
Yeah
we'll
talk
here
and
so
yeah,
so
my
deployment
here
is
this
is
kind
of
where
the
magic
had
this
is
the
workflow
here
is
kind
of
what
glues
it
together.
So
right,
let's
make
a
change
right,
so
I
have
my
development
here.
It
says
blue
in
my.
Let
me
move
this
thing
stupid
there
we
go.
My
production
says
blue
as
well,
so
they're
the
same
currently,
so,
let's
make
a
change:
let's
go
to
visual
studio
code.
B
B
A
B
B
It
triggered
that
right
and
so
so
as
that's
running
I'll,
show
you
where
that's
set
up
so
here,
let
me
go
back
real,
quick,
it's
gonna
go
through
all
those
steps
right
and
then
my
pipeline
runs.
You
can
see
the
history
and
you
can
see
the
tasks
are
starting
to
come
up
and
so
I'll
keep
this
up
here.
But
I
set
up
a
my
settings
so
since
this
is
a
cooking
show,
I
did
this
beforehand
web
hooks
right.
I
set
up
a
web
hook
here.
B
B
My
pipeline
and
so
yeah
here
so
here
I
cloned
the
repo
right.
Does
it
get
cloned
kind
of
straightforward
set
image
tag?
Essentially
I
take
the
hash
of
the
commit
and
then
I
am
providing
it
as
the
tag
for
clay,
so
quite
io
welcome,
app
and
then
just
a
hash
right.
So
I
can
identify
a
specific
commit
to
a
specific
image.
A
A
B
You
don't
have
like
a
platform,
I
mean
nothing,
there's
anything
wrong
with
that.
It's
just
a
different
way
of
doing
it,
but
you
don't
have
a
platform
like
jenkins
or
like
cloudbees,
you
doing
everything
in
cluster
you're
leveraging
your
own
cluster
to
do
a
build
because
everything's
containers
right.
So
you
know-
and
you
don't
have
to
have
things
running
all
the
time
you
just
kind
of
serve
this
type
of
thing
where
it
spins
up.
B
B
Yeah,
so
that's
that's!
That's
a
good
question,
so
I'm
doing
a
clone
repo
and
doing
set
image
tag
and
you're
like
how.
How
are
you
doing
all
that
if
you
go
to
storage
and
persistent
volume
claims
right,
I
have
a
volume,
okay
and
you
can
set
up
your
pipeline
to
use
the
same
volume
so
right
now,
the
way
I
I
have
it
set
up.
I
it
just
creates
a
new
volume.
Each
time.
A
B
B
B
B
B
And
then-
and
I
do
get
so-
I
have
a
have
a
token
and
an
ssh
key
and
you
just
load
it
up
in
so
this
is
kind
of
like
the
black
magic.
I
kind
of
wish
we
did
this
with
jafar,
but
this
is
like
middle
of
the
night
with
jafar.
When
you
add
a
git
secret.
Oh
that's
right!
So
you
add
a
now.
I
remember
so
you
added
the
your
secrets
and
when
you
are
calling
them
in
the
task,
no
in
the
the
task
runs
as
a
specific
service
account.
A
B
A
B
B
Yeah,
so
I
have
you
know
we
left
off
here
build
and
push
right
so
step
one
build
it
step,
two
push
it
step.
Three,
you
know
cut
the
information.
I
need
and
store
it
until
variable,
because
I'm
using
it
later
so
I
clone.
So
these
two
run
at
the
same
time,
clone
deployment
repo
tagged
to
latest
right.
So
I'm
doing
scopiocopy
right,
I'm
gonna,
say
quay.
B
A
B
I
patch
prod
right
same
same
it's
actually
patch,
prod
and
patch.
Dev
are
the
same
tasks
except
what
repo
I
I
target.
What
directory
I
target?
I'm
sorry
branch,
the
prod
right,
I'm
creating
a
branch,
and
then
I
do
a
pr
right.
So
how
does
this
look
like
argos
should
have
already
yeah.
So
argo,
if
I
refresh
dev,
it
says
green
but
prod
does
not
say
green.
B
I
have
a
process
right.
I
go
to
my
pr
and
looking
to
see
here.
Magic
fm
right.
Magic
here
is
the
p.
Is
my
ci
process
created
this
pull
request
and
this
pull
request
is
basically
I
go
to
files
change
right
and
if
you
notice
here
it
says,
overlays
prod
customization.
Let's
look
at
the
full
file
right
in
this.
Basically
the
section
and
the
customize
edit
that
you
saw
on
there
essentially
updates
this.
This
field
here,
if
I
go
back
so
you
can
see
here
so
here
you
know
you
cannot
you
can.
B
I
can
say
this
is
dumb
right
like
you
can
go
through
your
you
know,
start
the
review
start
the
process
that
sort
of
thing
right
so
kind
of
just
to
go
back
here.
So
what
I'm
doing
is
I'm
I
I
commit
to
dev
directly,
but
the
way
I'm
doing
the
gating
is
essentially
I'm
doing
it
through
a
pr
and
using
get
using
your
workflows
right.
My
workflow
is
pretty
simple,
but
your
workflows
will
be
a
little
more
complex
right.
There's
gonna
be
code
review.
B
A
B
I
don't
need
this.
I
don't
need
this
branch
anymore,
so
I'll
just
delete
it
and
then
the
code
is
updated,
and
so,
if
I
go
to
do
a
hard
refresh
on
you
see
that
argo
says:
oh,
hey,
there's
a
new
image,
updated
and
argo
automatically
creates
a
new
pod,
and
so,
if
I
go
to
my
application,
it
should
say
green
now
the
route
stale
route
come
on
there.
We
go
green.
So
now
my
dev
matches
production
because
I
just
approved
that
pr.
So
there
is
a
question
here.
B
So
it
looks
like
braided
silver
also
says
it
depends
as
well,
but
you
your
your
pipeline,
it
just
depends
where,
where
you'd
be
running
it
right,
so
in
my
pipeline,
I
would
do
it
all
in
that
same
cluster.
The
deployment
of
that
pipeline,
though,
is
different
right.
The
deployment
argo
cd
can
deploy
to
many
clusters
right.
Argo
city
can
deploy
to
one
cluster
right
this.
B
Pipeline
right,
like
that's,
not
the
idea
of
a
tecton
is
you
won't
have
like
a
fleet
of
tectons
running.
You
would
have
one
instance
of
tecton,
but
I
can
see
why
you
would
want
that
because
you're
coming
from
the
the
jenkins
world
right
where
it's
like
you'll,
have
because
jenkins
has
like
what
one
of
those
jenkins
agents
or
whatever
that
they
deploy
and
do
their
own
work.
This
is
like
another
way
of
thinking
of
doing
that.
Yeah
pipelines
could
spin
up
a
test
cluster
as
well.
A
B
B
B
Yeah
yeah
I'm
a
fan,
and
I
I've
said
this
to
a
lot
of
people.
They
know
I'm
a
fan
of
clusters
as
cattle
right
like
if
you're
gonna
deploy
you're
a
new
application,
you're
deploying
a
new
version
of
openshift.
But
that's
just
me
I
I
know
I'm
talking
crazy,
there's
gonna
be
a
lot
of
people
that
says
I'm
talking
crazy
andrew
says:
I'm
talking
crazy.
I
understand
that.
B
B
The
build
on
one
cluster
and
be
on
the
other,
that's
actually
that's
actually
tecton's
job.
That's
that
that
that
relies
on
the
lap
of
of
tecton
tecton
will
coordinate
that
for
you.
So
argo
cd
in
this
scenario
in
ci
cd
is
actually
has
a
very
small
role
to
play.
I
would
say
it's
actually
a
a
simple
but
a
simple,
slash,
powerful
design
of
argo
is
like
hey,
I'm
just
a
reflection
of
what
you
want
to
get
right.
It
has
other
tool
sets
that
make
it
actually
pretty
neat.
B
B
A
A
B
Anything
else,
cd
is
really
you've
done.
All
your
testing
you've
done
everything
in
in
your
ci
process
when
you're
finally
doing
cd
like
it,
it
should
be
like
it
should
be,
like
the
safe
thing
to
do.
I
know
we're
on
where
I
know
a
lot
of
people.
Aren't
there
right
we're
not
all
there
as
an
industry,
but
the
idea
is
cds
shouldn't
be
scary,
because
the
scary
stuff
happened
in
ci
right,
it
failed
in
ci.
That's
that's
why
you
wanted
to
have
scaled
nci.
A
A
B
Yeah,
so
this
is
also
the
a
paradigm
shift
as
well.
Is
that
if
you
want
notifications,
you
have
to
build
in
the
notification
right
and
it's
not
it's
not
like
out
of
the
box
right
like
like
hey
after,
if
step
one
and
step
two
finished,
I'm
gonna
curl
this
you
know
you
can
be
as
as
as
as
coarse
as
that.
There
might
be
something
on
the
tecton
hub.
Let's
just
look
real
quick.
There
might
be
a
notification
task.
Let's
look
for,
let's
look
slack
send
to
slack
channel.
Look.
B
A
A
B
B
Yeah
so
some
hooks
hook
to
argo,
tell
the
devs
to
go
to
argo
and
shut
up
so
so
there's
you
can
set
up
hooks
to
argo
as
well,
so
post
post,
sync
hooks
right.
So
there
is
multiple
notifications
you
can
set
so
the
the
argo
sorry,
the
the
tecton
right
in
your
ci.
You
can
have
hooks
going
in
to
send
notifications
and
argo
actually
has
a
postsync
hook
right.
So
like
once
everything
is
finished,
you
can
say:
hey
the
sync
was
successful.
B
B
A
A
A
B
Correct
correct
yeah
you
could-
and
this
is
because
I'm
using
a
get
ops,
workflow
right,
you
can
use
event
driven
right.
So
what's
really
cool
about
argo
cd
is
that
you
can?
You
can
turn
off
the
automatic
sync
and
just
use
hooks,
and
then
it
turns
into
like
an
event
driven,
so
you
can
actually
still
do
event
driven
architecture
with
cloud
native
tools
right
so
and
so
yeah
also
pipelines
aren't
linear
as
well
right.
So
it
doesn't
have
to
pause
and
wait
for
things.
B
A
A
A
B
A
B
B
A
B
A
Right
we
both
worked
at
duke
and
duke
university,
and
he
was
on
the
the
duke
I
t
side
and
I
was
working
on
the
research
side
and
we
were
both
doing
kind
of
similar
things
at
the
same
time.
So
we're
you
know,
feeding
off
each
other.
So
the
idea
now
is
that,
like
he's
been
an
esri
for
a
while
he's
all
these
different
ways
of
doing
dedicated
now
and
let's
bring
that
to
the
channel,
so
it'd
be
cool
to
get
a
bunch
of
srvs
from
throughout
that
org.
B
B
Yeah,
I
know
this
was
a
crash
course
because
I
went
from
like
hello
world
to
full
pipeline,
but
I
kind
of
wanted
to
you
know:
go
over
the.
You
know
how
it
fits
in
with
get
ops
and
how
argo
cd
fits
in
and
how
really
your
git
workflow
will
detect
all
that.
That's
the
center
of
everything
so
yeah.
So
unless
there's
any
other
questions,
I
think
I
think
we're
good
for
this.
Oh
one
housekeeping
totally
forgot.
B
B
So
I
have
yeah,
I
have
a.
I
have
a
thing
I
tweeted
it
out
on
kubecon
right.
I
actually
you
guys
kind
of
kind
of
got
a
sneak
preview,
because
I
kind
of
go
over
this
demo
in
that
in
that
talk
and
two
we're
having
get
ops
con
right.
So
if
you
haven't
sign
up
for
get
ops,
con
cfps
were
submitted.
We
we
emailed
out
the
congratulations.
It's
happening
so
I'll,
be
at
get
ops
con
I'll,
be
at
kubecon.
So
we're
taking
a
break
that
week
because
there's
going
to
be
a.