►
Description
Delve into Helm: Advanced DevOps [I] - Lachlan Evenson & Adam Reese, Deis
Helm is not just for simple applications running in simple environments. In this talk, we delve into the depths of Helm, focusing on lifecycle management and continuous delivery (CI/CD) of Kubernetes-native applications in different environments. We show how to extend Helm’s capabilities with plugins and add-ons. We'll also see how sophisticated charts like OpenStack and Deis Workflow use these capabilities to model more complex deployments.
About Lachlan Evenson
Cloud evangelist and tire kicker. Pushing cloud to it's limits in a public/private environments. Passionate about infrastructure automation, architecture and cloud deployment strategy. Believer in open source and an active member of the community contributing to several projects. Spent the last year and a half working with Kubernetes and enabling Cloud Native Journeys. @lachlanevenson
About Adam Reese
Adam Reese is a core maintainer for the Kubernetes Helm project. As a Senior Engineer at Deis, he has contributed to many open source projects. Over his 15 years building software, Adam has built everything from distributed pipeline processors to embedded systems.
Join us for KubeCon + CloudNativeCon in Barcelona May 20 - 23, Shanghai June 24 - 26, and San Diego November 18 - 21! Learn more at https://kubecon.io. The conference features presentations from developers and end users of Kubernetes, Prometheus, Envoy and all of the other CNCF-hosted projects.
A
I'm
Rocco
and
Evensen
and
I'm
one
of
the
kubernetes
charts
maintain
us:
does
anybody
using
kubernetes
charts
a
stream
Lulu?
Thank
you.
Thank
you
for
contributing
and
making
it
a
great
place
for
people
to
get
started
on
on
kubernetes.
So
big,
thank
you
to
you
all,
and
over
here
I
have
my
esteemed
friend
and
colleague,
Adam
riess.
He
is
a
helm
core.
Maintainer
he's
been
there
from
day.
One
he's
been
fighting
the
good
fight
making
helm
what
it
is
so
I
think
between
us.
A
Both
we've
got
a
little
bit
experience
about
helm,
so
we
hope
to
take
you
on
a
30-minute
journey,
we're
going
to
have
a
lot
of
fun
and,
at
the
end,
there's
going
to
be
a
little
fun
demo
for
you
all.
So,
basically,
hang
in
for
that
stay,
awake,
otherwise,
I'll
wake
you
up
with
the
splash,
fantastic,
alright,
so
show
of
hands.
You
probably
tired
of
this
question.
Who's
running
set
up
kubernetes
in
production,
woohoo,
fantastic,
fantastic,
great
and
who's
running
helm.
A
Right!
Okay,
you
tell
me
what
helm
is.
If
I
don't
know,
kubernetes
package
manager
kay
helm
is
the
kuba
nice
but
correct.
Correct
come
on
down
come
on
down
and
when
you
prize
no,
it's
okay,
it's
okay!
You
say
that
say
that
I'm
going
to
make
you
do
a
backflip.
Do
a
circus
thing!
Okay,
can
you
really
do
the
hellmouth
thing,
the
helm,
anthem?
You
know
that
too.
So
how
long
is
a
package
manager
for
kubernetes?
Now
what
if
I
said
that
helm
is
not
just
a
package
manager,
helm
is
not
just
a
package
manager.
A
What
what
did
I
hear?
Matt
butcher
say
that
it
was
not
pre-approved.
Sorry
didn't
sign
off
on
this,
but
I
want
you
all
to
think
about,
and
what
we're
going
to
spend
30
minutes
doing
is
thinking
about
what
else
can
we
make
helm
do
so?
We
all
know
helm
is
a
package
manager,
but
how
can
we
stretch
helm
and
actually
make
it
to
some
other
things?
A
So
today
we're
going
to
I
want
to
challenge
you
all
come
with
an
open
mind
will
show
you
what
home
can
do
today,
but
let's
think
about
and
brainstorm
what
else
we
can
do
with
him.
Everybody,
okay
with
that
yeah
and
the
audience
said
fantastic.
Thank
you.
So
much
for
being
awake,
okay,
fantastic!
So
before
we
get
into
the
the
nitty-gritty
of
why
I'm
saying
helm
is
not
just
a
package
manager.
A
B
B
So
as
an
engineer
which
I'm
sure
many
of
you
in
here
are
and
can
relate
to
this,
when
I
see
a
new
piece
of
technology,
come
out
that
growing
as
fast
as
kubernetes
I
have
to
acquire
it
and
earn
the
Oracle
status
on
it.
So
it
it
starts
getting
built
up
a
little
bit
and
the
excitement
builds
and
you
want
to
implement
and
start
investigating
everything
about
it.
B
B
Thank
you:
go
ahead,
go
ahead.
Yeah
it
hurt
it
hurt.
So
there
was
a
vagrant
script
in
the
route
and
for
visiting
it.
You
had
to
make
sure
one
that
your
power
was
plugged
in
because
the
battery
would
dive
in
my
twenty
minutes.
After
doing
your
cube
up
and
then
about
the
fourth,
try
you
get
a
cluster
running
after
you
see
all
this
scroll
for
a
long
time
and
then
yes,
I
have
a
local
cluster.
C
B
B
B
C
B
B
Knowledge
of
these
different
resource
types-
you
know:
I,
want
to
build
a
proof
of
concept
if
I
have
an
existing
application.
I
want
to
build
that
out
in
a
manifest
and
deploy
it.
So
I
started
with
what
I
knew
and
grabbed
the
guestbook
application
and
started
plugging
different
stuff
in
which
ended
in
copy
and
paste
EML,
15,
dents
and
repeat
so,
if
I
made
it
this
far,
then.
B
So
that
was
a
present,
so
looking
at
Helm's
place
in
this
journey,
helm
up
gave
away
in
that
initial
stage
of
I
have
a
cluster
running
I
want
something.
That's
that's
real,
praying
sorry,
not
realize
and
say
that's
not
guestbook,
so
it
provided
ways
that
no
applications
that
you
knew.
If
you
wanted
Redis
running,
then
it
provided
a
real,
easy
way.
I
just
do
a
helmet
installed
label,
read
it
and
I
had
it
up
and
running
and
I
had
something
that
I
was
familiar
with.
B
B
B
It
the
package
manager,
you
know
we
can
install
stable
applications.
We
have
enough
configuration
options
that
we
can
tweak
some
knobs
and
get
it
running
exactly
how
we
need
to.
We
can
get
stuff
talking
together.
You
know
this
repository
that
we
have
is
backed
by
to
the
community,
so
we
can
leverage
the
knowledge
that
everybody
shares.
B
If
somebody
pulls
something
else
and
it
just
get
toast-
and
you
know
we
can
phone
all
their
friends
and
say:
hey
Joe,
do
this
and
you
know
it
serves
as
a
learning
ground
that
we
can
all
develop
some
best
practices,
because
it's
such
a
new
technology
like
and
there's
not
one
way
to
write
yamo,
so
we
can
figure
out
like
okay,
what's
the
best
approach
to
some
of
this
stuff,
what's
the
best
approach
to
labels?
What's
the
best
approach
to
naming
convention
all
right.
A
So
I
think
you
know
rounding
out
that
first
section
taking
a
look
at
what
helm
and
where
people
are
using
it
today,
I
think
that's
fairly
common
for
those
using
helm,
they're
using
it
to
you
know,
get
the
upstream
charts
and
start
playing
with
putting
resources
on
the
kubernetes.
So
you
know
if
we
take
a
look
at
that
thing
at
where
most
folks
are
now.
Let's
take
a
quick
look
at
what
the
core
components
of
helm
are
before
we
start
to
introduce
what
else
we
think
helm
can
do.
B
So
charts
can
be
installed
from
the
repository
which
we
were
talking
about
before
they
can
also
be
installed
from
a
local
file
path.
So
if
you
wanted
to
have
it
like
in
your
repo
for
your
application,
it
check
into
your
version
control
helm
would
support
that
you
could
just
run
helmet,
get
the
local
file
and
watch
this
on
the
right.
A
A
Who
he
has
used
cue
bats,
going
to
cube
at
home?
Do
you
like
it
give
it
a
shout-out?
So
that's
actually
the
work
of
another
project
called
monocular,
which
gives
this
kind
of
UI
on
top
of
Helmand
and
one
of
the
decor
you
are
done,
put
your
hands
together
for
art,
now
and
I
think
I
think
monocular
ancestor,
so
that
actually
makes
this
the
whole
front
end
quite
usable.
So
if
you
have
not
been
to
cube
apps,
calm,
I'd
suggest
you
go
there
and
it
may
be
the
first
time
you
install
something
on
kubernetes
successfully.
A
Who
knows
it's
like
this
one,
besides
guestbook,
so
looking
at
charts,
so
I
think
charts
being
an
integral
part
of
helm.
You
know,
as
container
images
are
two
containers
and
you
can
ship
those
units
everywhere,
let's
think
about
charts
in
the
context
of
helm
and
how
we
can
actually
use
helm
to
do
more
than
just
move
charts
around.
So
we
want
to
introduce
concepts
of
what
we
think
helms.
Other
hats
might
be
go
ahead,
switch
hat.
B
B
Although
repeatability,
if
we
say
state
so
if
your
cluster
gets
a
bad
way,
you're
going
to
be
able
to
pull
up
your
environment
again,
something
else
is
configuration
management.
So
here's
an
example
of
installing
into
a
a
specific
cluster,
which
is
the
cube
context
and
specific
values
which
is
development
yanil.
So
you
can
have
development
staging
and
production,
which
is
a
lot
like
the
rails
model,
which
I
hear
a
lot
of
people
like
and
inheritance,
so
helm
allows
you
to.
It
allows
code,
reuse
where
you
can
use
it
for
standardizing
certain
fields.
B
You
can
standardize
labels
and
have
that
be
templated
throughout.
You
know
like
if
you
want
to
enforce
that
a
certain
group
has
a
specific
label
with
their
group
in
your
org.
You
can
enforce
that
kind
of
sets
the
policy
enforcement
also,
if
you
have
a
monitoring
or
logging
system
that
is
used
for
your
company
across
all
applications.
Well,
that
could
be
included
in
a
chart
as
well,
and
you
could
just
have
it
out
of
the
box
every
time.
A
So
if
you
have
a
policy
wide
change,
the
thing
we
see
the
most
problematic
is
actually
between
versions,
tags
change,
alpha
notations,
beta
notation,
so
we'll
have
actually
a
standard
way
to
move
those
changes
through
without
everybody
having
to
update
all
of
their
chart.
So
that's
incredibly
valuable
yeah.
B
B
So
what's
the
time
when
you
do
a
deployment
with
microservices,
you
don't
just
want
like
your
one
micro
service
there.
You
know
it
needs
its
buddies
to
work
and
be
useful.
So
for
that
we
have
dependency
management,
so
charts
can
actually
be
included
in
other
charts
and
they
can
be
composed
together
to
build
out
your
your
application.
That
way,
and
we've
seen
a
few
big
examples
of
that,
one
of
them
being
OpenStack
and
also
Dave's
workflow
there
there's
the
charts
online.
You
can
go
and
look
at
them.
A
Yeah,
the
SATA
photos
and
the
AT&T
folks
has
OpenStack
charts
published
and
it
speaks
to
how
you
can
actually
compose
more
complex
applications
and
actually
have
them
work
and
run
on
top
of
kubernetes.
So
if
anybody's
interested
come
and
talk
to
us
after
this-
and
we
can
point
you
in
your
direction.
A
So,
ok,
crocks
and
more
so,
if
anybody
has
ever
seen
me
speak
before
you
know,
I'm
on
a
relentless
pursuit
to
actually
make
crocodiles
really
popular
in
the
world.
Does
anybody
like
crocodiles,
so
what
I'm
going
to
do?
Nobody
like
struck
it
off,
I,
don't
like
them
either
crocodiles
are
terrible.
You
do
not
want
it.
They
eat
dogs
and
there's
just
not
not
nice
animals
in
general.
A
But
what
I'm
going
to
do
is
actually
take
the
stuff
that
Adam
spoken
about
and
spend
a
lot
of
time
in
showing
you
how
we
can
actually
use
helm
in
different
contexts.
So
what
I
want
to
do
is
just
get
those
creative
juices
flowing
and
everybody's
brain
and
let's
see
what
we
can
do
with
this.
So
are
you
ready
to
come
on
a
demo
journey
who
likes
demos?
Ok,
all
right
now,
I'm
no
longer
lucky,
the
presenter
I
just
got
to
change
my
hat
I'm.
Now
lucky
the
app
developer
and
I'm
going
to
go.
A
I've
got
a
game
about
Crocs,
it's
going
to
change
the
world.
Some
of
you
may
see
in
it,
but
I
haven't
quite
made
the
billion
dollars
and
retired.
Yet
so,
let's
see
if
we
can
change
that
here
and
now
so
like
any
good
demo,
let's
start
with
an
IDE
but
I'm
an
app
developer
and
I
may
not
know
anything
about
kubernetes
and
I
may
not
know
anything
about
containers.
So,
let's
just
start
from
there
right,
so
I've
got
my
app
and
I
want
to
make
a
change
to
my
application.
So
it's
a
go.
A
Web
app
I
happen
to
know
that
I
have
my
source
code
here
and
all
I
need
to
do
is
actually
change.
One
little
thing:
I'm
changing
the
JavaScript
from
one
version
to
another.
So
let
me
pop
over
and
just
prove
to
you
that
there's
no
so
I've
made
this
change.
This
is
not
pre-recorded
I'm,
not
cleaning
typing,
but
let's
see
if
the
demo
gods
bite
me
for
saying
that.
A
So
let
me
just
check
that
in
and
I'm
going
to
bump
jas
version
now
before
I
actually
check
this
in
what
I
may
want
to
do
is
just
show
you
what
croc
hunter
is
because
if
you
you
may
die,
not
knowing
I,
don't
want
anybody
to
pass
out,
not
knowing
how
wonderful
it
is
so
croc
croc
hunter
is
here:
I've
got
to
make
a
change.
Pronto,
there's
a
great
game
in
certain
production,
but
my
boss,
just
told
me
that
you
can't
shoot
crocodiles
with
laser.
So
croc
hunter
is
back.
A
A
Alright.
So
what
I'll
now
do
is
go
over
and
take
a
look.
So
I
have
a
Jenkins
server
here.
What
Jenkins
should
do
is
actually
get
that
webhook.
You
can
see
that
it's
fired,
who
uses
Jenkins
here,
anybody
use,
Jenkins,
sorry,
I,
didn't
get
blue
ocean
or
whatever
it's
called
I'm
still
on
the
old
GUI.
Maybe
maybe
next
demo,
but
what's
happened,
is
I,
see
that
my
dev
branch
has
been
fired
inside
Jenkins.
You
can
see
the
builder
status
over
there
for
the
build
of
my
dev
branch.
Now.
A
So
under
my
source,
repo
here
and
let
me
just
increase
my
font
I
actually
have
my
charts
checked
into
my
source
repository.
So,
as
Adam
mentioned,
you
not
only
have
to
have,
you
can
have
jobs
and
repositories
or
you
can
have
charts
in
just
your
source
code
and
I.
Think
I
would
challenge
people.
Why
not
put
them
in
your
source
repository.
That's
where
you
put
everything
else.
If
changes
need
to
be
the
way
to
the
way
you
deploy
your
app,
it's
actually
a
common
place
to
put
them
in
there.
A
A
Okay,
great,
so
I've
got
a
green
now,
as
I
mentioned.
What
actually
happens
here
is
I'll
see
if
I
can
increase
the
font.
So
you
can
see
this,
but
as
part
of
my
build
steps,
I
actually
not
only
check
that
I
can
get
to
kubernetes
I
use,
helm
here
and
actually
check
that
my
chart
can
be
built
and
can
and
can
be
tested.
A
A
So
I'll
show
you
how
that
works
in
a
minute,
but
by
the
end
of
this
I'm,
going
to
publish
a
container
out
to
quad
I
do
that
I
should
be
able
to
reference
and
pull
in
so
here's
way
you
should
see
now
that
all
I've
done
is
get
push
and
I
actually
have
a
container
artifact
with
my
branch
and
my
get
char,
that's
pushed
out
and
available.
So
while
I
do
that,
let
me
just
prep
my
changes
for
production,
so
I
can
see
that
it's
gone
through
the
whole
build
pipeline.
A
All
right
so
here's
Paul
54,
so
I
can
actually
see
that
here's,
my
change
ready
to
go
I
can
also
see
that
Jenkins
has
picked
this
up
and
it's
actually
running
a
build
off
the
PR.
So
I'll
show
you
what's
happening
here.
Jenkins
is
configured
to
watch.
Watch
this
repo.
It's
seen
that
a
PR
is
gone
and
we're
actually
going
to
build
and
merge
on
the
base
master
branch
and
run
that
same
pipeline
again
so
again,
I
haven't
touched
kubernetes
and
I
haven't
touched
any
container
commands
this
is
all
being
triggered.
So
why
not?
A
While
we
wait
for
that
PR
to
be
built,
let's
go
and
pick
up
that
artifact
that
was
cut
off
dev
and
just
test
it
locally
and
make
sure
it
actually
works.
Okay,
so
before
I
go
over
to
that,
let's
just
pick
up
the
artifact
name,
please
sorry
to
switch
screens
around
doing
the
screen.
Dance
I'll
just
pick
up
this
artifact
name.
So
the
tag
on
the
container
here
is:
is
there
and
now
I
should
be
able
to
use
home
to
install
that?
Okay,
so
I'll
do
a
helm,
install.
A
Let
me
make
sure
I'm
in
the
right
directory
so
because
I'm
installing
from
my
git
repository
I,
have
the
chart
locally,
there
I
can
do
a
helm.
Install
I
can
give
it
the
charts
directory
in
the
croc
hunter
I'm,
going
to
set
the
image
tag
to
equal
the
image
tag
that
was
just
cut
off
my
branch
and
I'm
just
going
to
deploy
it
to
a
namespace.
A
So
I
can
test
on
this
kubernetes
cluster
called
Olivo
test
down
the
bottom
half
of
the
screen,
I'm
actually
just
running
a
watch
on
that
namespace,
so
we
get
wayfaring
orangutan.
Is
there
any
other
type
and
what
we
see
is
my
application
has
been
pushed
out
with
Helm's,
so
the
same
deployment
process
that
I've
just
step
through
and
Jenkins
I
can
now
replicate,
and
that
speaks
to
what
Adam
was
saying
about
repeatability.
A
B
A
Last
plugins,
so
helm
has
a
nice
plug
in
thing.
I
won't
go
into
it
today,
but
I'm
actually
doing
a
test
so
I'm
using
the
test
suite
actually
to
test
that
my
application
works
and
the
way
the
test
suite
works
in
helm
is
it
actually
goes
out
and
runs
commands
inside
kubernetes
and
you
get
a
red
green.
A
So
not
only
am
I
testing
that
my
app
can
be
installed
on
kubernetes,
I'm
also
testing
that
it
works,
so
I've
just
gone
and
replicated
a
test
on
a
completely
different
cluster
to
the
one
that
runs
my
production
workload
and
I
was
able
to
pick
that
up
and
run
it
again.
So
I
even
have
a
story
here
to
get
some
laptop
development
going
fairly
quickly
using
the
same
deployment
architecture
that
the
the
Jenkins
pipeline
is
using.
A
A
A
Alright,
so
for
the
final,
the
final
trick
here
Jenkins
will
again
pick
up
the
change
to
master,
but
this
time
that
will
actually
go
out
and
deploy
to
that
app.
That
I
showed
you
at
the
start.
So
it's
going
to
not
only
do
a
dry
run,
it's
going
to
do
a
test
and
push
it
out
once
it
passes
all
that,
so
you
can
see.
The
master
branch
has
already
come
up
here
now
for
bonus
points.
While
that's
building.
Let
me
just
introduce
you
to
another
little
trick
here.
A
This
Jenkins
cluster
that
I'm
using
here
was
actually
built
using
helm,
so
it
did
a
helm,
installer,
stabled,
Jenkins
and
I've
just
installed
Jenkins
from
the
upstream
chart,
repo
and
I.
Have
you
have
all
this
that
you
fingertips?
So
here's
actually
Jenkins
running
you
can
see
22
seconds
ago
the
agent
came
up
and
the
agent
is
performing
my
build
so
that
running
on
the
same
cluster,
that's
running
my
production
application,
so
I
think
another
persona,
that's
probably
maybe
not
as
well
represented
with
helm,
is
the
operator
level.
A
So
let's
go
back
and
make
sure
that
this
thing's
happening
okay.
So
at
the
moment
it's
actually
just
building
my
container
and
then
it
will
publish
that
container
out
to
to
quay.
At
this
point
you
know
I
want
to
round
out
and
get
you
start
thinking.
Is
there
any
questions,
while
we
wait
for
this
to
finish,
I
think
yeah
gentleman
up
the
front
yep.
A
Yes,
so
absolutely
the
question
was
I'm
running
it
on
Jenkins,
so
can
presumably
I
had
to
go
configure
all
the
pipeline.
So
I'll
show
you
what
I
had
to
configure
for
this,
and
it
was
just
this
there's.
No
other
configuration
I
just
had
to
tell
it
to
login
and
scan
it's
a
public
repo
via
get
limited,
so
I'm
using
some
credentials
and
go
and
grab
my
repo.
That
means
the
build
steps
are
actually
checked
in
to
my
source
repository.
A
So
all
Jenkins
does
is
read
this
file
in,
and
it
knows
exactly
what
to
do
and
I'm
using
home
for
the
last
mile
down
here.
So
if
branch
equals
master
go
ahead
and
do
a
helm,
deploy
and
I'm
going
to
deploy
that
chart
out
with
that
upgraded
artifact.
So
to
answer
your
question,
you
need
to
have
a
file
the
Jenkins
can
read
in,
but
you
can
programmatically
add
a
file.
Much
like
the
United
says,
provided
Travis
circle.
A
The
likes
of
those
folks
read
a
Yammer
file
out
of
your
repository,
so
I
didn't
do
anything
Jenkins
apart
from
Tillet
what
repo
to
look
at
here.
Okay,
so
we
should
be
okay
with
masters
gone
out
and
for
the
final
kicker
here,
let's
see
right
so
the
get
char
is
here
and
yeah.
Now
I
no
longer
shoot
laser,
as
I
actually
should
fish,
which
is
a
much
much
nicer
experience.
A
The
swamps
tape
again
so
fantastic
at
this
point.
I
will
open
it
up
for
questions
and,
if
you're
interested
in
the
slide
deck,
there's
actually
exactly
what
happened
as
a
diagram.
So,
if
you're
interested
in
just
understanding
exactly
what
I
did
in
that
flow
or
what
have
questions
about
it,
but
will
open
it
up
any
special
repos.
A
A
So
the
four
helm
test,
specifically,
you
actually
define
a
pod
manifest
and
it
executes
a
pod
manifest
inside
kubernetes,
so
it
will
actually
go
and
do
any
arbitrary
test.
So
what
I'm
actually
doing
in
my
test
just
to
peel
that
back
for
folks
is
show
you
here
so
because
I
have
my
chart
here.
I
can
I
can
go
and
show
you
where
the
tests
live,
and
over
here
in
templates,
I
have
tests
and
a
test.
A
This
test
is
actually
doing
a
double
you
get
against
a
service
name
inside
kubernetes
and
just
making
sure
I
can
pull
slash.
So
I'm
actually
testing
that
that
that
comes
up,
so
you
can
define
any
tests
and
any
type
of
containers.
If
you
had
my
sequel,
you
could
log
in
you
could
create
a
table.
You
could
drag
the
database
and
do
the
same
thing.
Thank
you
great
great
question.
Question.
B
A
C
B
B
That
defines
like
all
the
different
parameters-
parameters.
Ations
I-
can
talk
of
this
the
chart.
So
these
are
all
the
different
values
you
can't
configure
and
pass
in
and
secrets
are
handled
just
like
any
other
kubernetes
resource,
the
templating
happens
and
then
essentially
it
just
builds
the
mo
for
you.
So
you
don't
have
to
do
the
copy
and
paste
a
on
16
entropy
and
it
will
just
install
it
directly
into
kubernetes
from
there.
So
there's
there's
no
magic
happening
between
the
two.
C
A
Seen
the
reason
for
it
implemented
outside
of
helm,
but
there's
no
reason
that
it
couldn't
become
a
piece
what
I've
done
in
the
past
when
I
ran
ran.
This
was
you'd,
have
a
subset
and
helmed
in
reference
external
files
in
the
values.
So
you
can
say,
go
get
this
from
here
and
then
you
just
need
some
process
to
go
put
that
file
in
place,
and
you
just
read
from
that
file
and
pull
it
in
right.
So,
instead
of
having
hard
fix,
you
can
actually
say:
go
get
the
value
from
that
file.
A
So
I've
seen
people
going.
You
know,
go
and
grab
this
from
over
here
and
you
have
some
process
to
put
that
there
and
that's
just
the
result
of
pulling
it
from
a
key
valve
store
so
and
then
inside
something
like
in
the
context
of
that
app.
You
have
environment,
specific,
yeah,
more
files,
so
you
could
say
dead,
staging
prod,
and
you
just
say:
if
my
branch
is
this,
use
this
yeah
more
file,
which
says
point
to
these
files.
Here,
that's
what
I've
seen
most
commonly
used
for
environment,
specific
solutions,
yeah.
A
Yes,
so
yeah
I
probably
take
it
in
the
dryer
you
want
to.
So
how
long
has
two
components
of
the
server
side
component
called
tiller
until
I
has
a
G
RPC
interface?
So
you
can
either
talk
directly
to
that
and
somebody
in
the
community
members
implemented
a
HTTP
REST
gateway
as
well.
You
could
stick
a
rest
gateway
on
the
front.
A
B
A
B
B
Yeah,
so
the
question
was:
have
we
seen
the
project
landscaper
yeah,
that
the
guy
that
built
that
is
actually
very
active
with
with
us
in
the
project
and
yeah?
It's
an
awesome
project.
We
do
have
plans
for
something
similar
to
that,
not
not
the
the
composing,
the
many
charts
together
as
a
logical
unit.
Essentially
what
he
did
was
what
we
did
to
kubernetes
the
resources.
He
took
it
a
step
further
and
did
charts
on
that,
but
also
provided
a
lot
of
functionality
around
like
diff,
outputs
and
stuff,
like
that.