►
From YouTube: Kubernetes Deployment Options
Description
This is a demo of GitLab's Kubernetes deployment options as well as a discussion of the implications for the Defend roadmap.
Notes: https://docs.google.com/document/d/1OTzDtRV1EOesU_dyrNAUpjibyk0gKUWP5bVrTzTmTw8/edit?usp=sharing
A
B
A
So,
for
example,
if
I
was
a
sre
at
a
company
are
probably
installing
these
applications
we
find
with
other
other
methods,
such
as
using
git,
ops
or
some
sort
of
some
sort
of
script
that
I'll
do
it
so
that
it's
repeatable
rather
than
clicking
through
a
lot
of
things
which
is
what's
happening
here.
So
that's
the
first
one
I
have
you
seen
the
screen
before?
Yes,.
B
A
For
example,
yesterday's
there's
one
customization
option
here,
but
every
single
one
of
this
takes
a
lot
of
code
and
then
having
to
upgrade
number
between
versions
is,
is
just
nightmare-inducing
a
little
bit
so
kind
of
moving
we're
kind
of
shifting
responsibility
from
having
gitlab
do
this
to
having
a
CI,
CD
job
duties,
which
also
then
allows
people
to
view
the
errormsgs.
It
happened
a
lot,
so
the
thinking
is
that
we
will
keep
kind
of
facade
but
behind.
A
A
You
can
see
its
current
in
alpha
because
we're
still
building
the
building
blocks
we're
building
the
CIC
the
pipeline
for
it,
which
is
almost
done.
The
only
thing
the
things
that
are
not
done
is
hooking
up
this
facade
through
this,
the
pipelines
right.
So
we
need
to
establish
a
link
to
you,
know
the
user
clicking
it
and
if
there's
an
error
or
success,
we
need
to
show
here's
a
pipeline
that
actually
did
things
we're
trying
to
do
here.
So
this
is
one
step
for
the
evolution
right.
A
A
So
let's
come
explore
the
other
two
options
as
a
name.
We
can
come
back
into
the
future
of
these
things,
okay,
so
the
other
two
options
is
examples
of
how
other
parts
of
gillip
dustings
so,
for
example,
Prometheus,
which
you
can
use
outside
of
Cuban,
is
or
inside
get
my
notice
and
in
fact,
in
credits,
Cuba
notice
and
just
two
ways
are
doing
things.
So
the
first
way
of
doing
things
is
for
you
to
one
click,
install
cominius
and
it
will
install
with
some
default
custom
parameters.
A
The
other
way
of
doing
things
is
where
you
can
install
images
yourself
and
it's
up
to
you,
how
you
want
to
do
it.
You
can
follow
the
officer
ducks
in
you
know,
there's
a
lot
of
customization
options
as
well,
and
this
is
what's
probably
most
people
do
rather
than
switching
one
clicked
and
then
then
you
can
go
into
your
project
and
that.
B
A
There's
a
lot
of
reasons.
The
first
one
is
customization
customizability,
so
the
current
way
is
when
we
install
prometheus
this
installs
it
to
a
static
place
and
it
installs
it
with
a
bunch
of
static
things,
static
options
right
that
we
define
and
metrics
and
so
on
and
so
forth.
So
if
you
want
the
any
ability
to
have
your
own
options,
then
this
is
definitely
not
the
option
for
you
right.
A
B
A
A
Will
click
install
oh
I?
Will
click
install
no
kind
of
testing
and
look
into
it
a
lot,
and
it
will
take
a
long
time
so
I
would
say
that
most
people
wouldn't
do
that
I
mean
on
serious
production.
Great
clusters
I
mean
if
I'm
playing
around
you
know.
For
me
this
this
is
very
convenient
for
playin
or
demoing
or
demoing.
A
Mr.,
this
is
great,
so
how
I
would
imagine
that
people
would
do
it
is
that
they
were
exposed
it
that
way
and
they
they
want
to
do
the
menu
installation
so
that
they
can
use
Prometheus
for
your
own
needs,
but
they
want
the
ability
to
surface
some
of
those
things
in
gala,
and
it
should
be
as
simple
as
this
right.
Just
just
saying:
here's
how
you
can
access
my
prometheus
endpoint
right.
C
A
B
A
Know
yeah
the
other
things
permissions
as
well,
so
the
other
reason
why
people
would
not
do
this
there's
permissions.
It's
we
require
quite
high
permissions
here,
whereas
some
clusters,
your
operator,
might
have
only
limited
permissions
to
only
a
certain
namespace,
and
that
means
that
outer
box.
This
just
fails
to
install
because
you
don't
have
the
permissions
and
then
then
years
ago,
back
in
sick
admin
permissions
and
do
that
dance
and
it's
sometimes
not
justified
under
perspective.
A
So
this
two
involves
a
little
bit
our
UI
and
it's
still
a
little
bit
unclear
what
this
does.
But
I
think
this
has
been
in
Gila
long
enough
that
people
sort
of
understand
what
it
does
now,
which
makes
that
and
the
other
option
that
people
do
at
the
other
kind
of
application
that
people
use.
Is
you
get
up
run
so
again,
you
can
install
get
a
runner
here
and
it
will
install
Gilliam
run
it
for
you
in
your
queue,
bananas,
cluster,
but
again
it's
fixed.
So
it
only
gives
you
four
concurrent
clusters.
A
If
your
hosta
is
under
or
powered
for
it,
and
you
want
to
change
it,
not
you
candolyn
or
if
you
want
auto
scaling,
like
you
know,
having
telling
the
Gallup
random
manager
to
automatically
increase
or
decrease
the
number
run
is
based
on
demand.
It's
not
enable
here.
No,
we
can't,
because
we
need
to
configure
extra
settings
to
to
give
that
runner
the
data
to
scale
up
and
down.
So
most
people
would
probably
just
install
the
Elm
start
directly.
I
think
is,
if
you
can
do
it.
A
A
B
A
B
I
was
thinking
there
was
like
a
totally
different
way
in
gitlab,
but
this
actually
makes
way
more
sense
to
me,
then
obviously
I
was
just
getting
confused
by
that
thread,
so
this
makes
way
more
sense.
It's
just
either
you
go
into
kubernetes
and
you
install
directly
or
you
use
one
of
two
options
and
pilots
get
laughs.
Yep.
A
A
Lots
and
lots
of
ingress
versions
installed
already
and
having
to
kind
of
figure
a
way
to
upgrade
to
the
latest
versions
means
that
we
have
a
lot
of
different
deltas,
so
we
have
ingress,
say
1.2
1.3,
1.4
1.5,
and
we
have
to
support
all
of
that
to
have
the
ability
to
upgrade
to
the
latest
ingress.
We
should
say
English
1.16
or
something
like
that.
So
at
some
point
we
have
to
kind
of
drop
support
for
people
to
upgrade
those
things.
Oh
maintain
some
sort
of
dogs
to
say,
click
a
unit
upgrade
to
this
version.
A
First,
and
this
version
7
is
version
first,
which
we
don't
even
have
the
ability.
Yet
so
that's
that's
a
huge
problem
and
not
to
mention
some
applications
like
still
manager
have
breaking
changes,
almost
every
single
version
and
that's
as
possible.
There
is
no
way
to
automate
that
if
it's
a
breaking
change
for
example,
so
it's
our.
B
A
It's
the
same
from
them,
but
with
CI
city
apps.
We
at
least
have
the
ability
to
screen
things
a
little
bit
and
we
can
detect
that
and
try
and
try
and
work
around
things
all
have
the
ability
to
maybe
pin
those
things
and
pin
versions
down
and
say
yeah
if
you're
really
ready
to
upgrade.
We
can
do
that
because
the
separate
founder,
Gillett
version
we
can
sort
of
maintain
of
great
powers
like
that.
A
C
A
A
Right,
yes,
yes,
that's
right,
yeah
I
agree,
but
is
you
a
fraud
exercise
say,
for
example,
we
saw
it
manager
where
we
can
we'll
probably
have
device
users
on
every
single
upgrade.
Here's
a
breaking
change
and
here's
some
manual
steps
to
do
it
every
single
time.
We
do
this,
so
it
kind
of
defeats
the
purpose
of
having
a
managed
service
where
you
have
the
we
throw
back
to
the
user
you're
having
to
do
many
steps
every
single
time
we
upgrade.
A
Yeah,
so
that
is
why
increasingly
I'm
very
tempted
like
this
is
the
realization
that
just
popped
into
my
mind-
maybe
a
few
weeks-
oh
but
I'm,
increasingly
tempted
to
shift
the
responsibility
of
install
and
I'm
grates
to
the
user.
So
they
can
control,
however
much
they
like,
and
that
means
that
we
don't
have
to
be
tools
like
pinning.
We
don't
appear
tools
like
upgrade
windows,
for
example.
A
This
is
especially
important
for
ingress,
because
whenever
we
are
going
grows,
it
creates
a
availability
hole
because
that's
that's
where
their
website
comes
in
right,
ingress.
If
you,
if
your
English
is
upgrading,
you
can
search
traffic
for
example,
and
then,
if
it
breaks
then
you're
done
so
we
don't
need
to
do
upgrade
windows
or
manage
upgrades,
debugging
error,
stuff
and
show
it
back
to
the
user
to
install
manage
themselves
then,
and
then
we
provide,
then
what
we're
providing
is.
Instead,
the
ability
to
integrate
what
you
have
installed
into
get
lab.
A
Which
is
exactly
what's
happening
with
something
like
gillip
runner?
Yes,
we
manage,
we,
we
manage
the
runner
code,
but
it's
up
to
you
when
to
I'm
going
and
how
to
I'm
going
to
get
LeBron
and
then
we
can
even
have
breaking
changes
and
it's
fine,
because
it's
people
can
follow
the
steps
to
get
to
where
they
need
to
get
to.
B
C
A
Are
learning
things
out
or
they
are
interested
in?
Oh,
this
is
in
support
for
outer
DevOps,
where
other
develops
promises
you,
the
ability
to
have
below
local
pipeline,
just
testing
different
ways
straight
to
keep
analyst,
and
this
there's
a
few
applications
here
tree.
Actually,
in
fact,
this
top
tree
support
alto,
DevOps
and
I'm.
Thinking
that
you
know
why
not
have
a
button
to
say,
install
that
you
need
found
to
DevOps
and
then
the
rest
here
in
sort
of
really
just
applications
that
we
like
to
integrate
with,
but
they're,
not
necessarily
needed.
A
Yeah
I
mean
this.
This
method
is
this
method
of
buttons
and
installing
it's
not
necessarily
needed
I
think
it'd
be
equally
valid
to
have
instructions
to
say
which
you
like
integration
with
Jupiter
hub.
Here's
two
lines
of
his
two
steps
step.
One
is
install
it.
However,
you
like
and
in
step
two
is
somehow
registry
with
Caleb
or
integrator.
A
B
So
right
now
we're
wanting
to
add
things
to
the
UI.
We're
managed
apps
live
one
to
make
it
easier
for
users,
but
also
it's
important
to
us
for
discoverability.
So
in
twelve
eight
we
released
we
integrated
with
so
iam,
where
so
iam
can
be
deployed
into
the
cluster,
and
it
provides
container
network
policies
to
help
filter
the
traffic
between
clubs.
B
The
challenge
with
that
is
right.
Now
it's
not
very
discoverable
at
all,
so
it's
two
murders
go
and
read
our
documentation.
They
can
see
how
to
set
it
up
and
they
can
know
it's
there
and
understand
how
it
works,
but
we
don't
have
anything
in
the
UI
that
says
Julie
I'm
right
this
says:
hey.
We
have
a
cilium
integration
if
you're
just
using
the
UI
and
not
reading
our
Docs
or
calling
our
release
notes
you're,
not
going
to
know
that
it
it
exists
as
of
right
now.
So
yes,
we
want
to
make
it
discoverable.
B
We
also
want
to
make
it
very
easy.
The
other
problem
to
be
solved
is
with
cilium
it's
a
little
bit
different
from
other
things
and
get
loud
because,
like
the
ingress,
you
know
we're
actually
getting
in
line
with
customers,
traffic
and
so
suppose,
for
some
reason
that
you're,
you
know
any
number
of
reasons
write.
A
bad
policy
set
a
bad
upgrade.
Maybe
it's
just
bad
compatibility
with
some
other
tool
in
the
cluster
I,
don't
know,
but
suppose,
there's
a
problem
with
cilium.
B
If
there's
a
problem,
you
probably
want
to
turn
psyllium
off
as
fast
as
possible,
and
so
that's
the
other
reason
why
we
want
to
have
a
UI.
You
know
to
easily
turn
stoolie
them
on
and
off,
and
so
that
when
there's
a
problem,
you
can
kind
of
hit
that
panic
button
and
say
you
know,
turn
it
off
right
now.
So
right
now
we're
deploying
Trillium
through
the
CIC
be
managed
apps,
but
we
want
to
add
it.
B
A
A
A
But
again
it
has
its
risks
where,
as
you're
saying,
is
it's
very
much
in
line
with
the
customers,
data
and
everything?
So
if
I
do
this
in
a
new
Saviour,
what
has
happens
is
that
this
in
line
upgrades
attempts
to
update
their
ingress,
and
if
you
haven't
configured
your
ingress
to
be
more
than
one
replica
having
redundancy,
then
it
brings
down
the
ingress,
potentially
all
if
it
breaks
or
something
like
that,
there
goes
the
ingress
as
well.
So
again,
I
feel
like
this
is
okay,
but
I.
A
Think
if,
if
I'm
a
serious,
if
I'm
running
a
production,
great
cluster
I
will
not
use
this
at
all,
I
would
be
more
comfortable
with
a
two-step
instruction
like
how
runner
does
it,
which
is
and
run
so
imagine
we
have
something
like
this
section
here,
where
it
tells
you
to
go,
install
it.
However,
you
like,
and
the
subtext
is
that
you
can
upgrade
it
whenever
you
like,
and
then
there's
some
instructions
here
on
how
to
say,
tell
get
lab.
Yes,
you
have
installed
Salem
and
know
you
have
an
install
serum
or
something
like
that.
A
A
It's
it's
a
huge
thing:
I
wouldn't
want
the
code
in
to
get
laughs
like
the
so
many
edge
cases,
I
mean
sure
we
can
try,
but
it's
it's
I
feels
to
me
like
we
were
playing
whack-a-mole
with
all
the
edge
cases
that
will
appear
and
then
the
more
I
know
we'll
be
adding
more
and
more
code
to
handle
all
those
cases.
I.
B
A
A
We
should
kind
of
strip
all
this
down
to
its
bare
minimum,
and
this
is
really
for
convenience
sake,
and
we
should
have
something
in
here
that
there
looks
like
this
Rhino
section
in
state,
where
you
have
little
snippets
of
instructions
that
tells
you
what
to
do
to
install
Salim
and
it
can
be
maintained
by
ourselves
as
well.
That's
fine
and
also
to
enable
or
disable
see,
Liam
again
is
also
instructions
as
well.
I
think
that
we'd
the
way
to
go
for
myself.
B
B
B
A
C
C
Now
it
works
roughly
as
you
described
so
user
has
an
ability
to
enable
cilium
for
the
cost
applications,
but
we
have
a
large
red
notice
area
in
documentation
that
say
that
it's
not
ending
there.
You
have
to
restart
your
parts
and
billing
directly
to
the
documentation
of
the
cilium,
where
it
says
the
same,
that
you
have
to
restart
supports
to
move
ports
on.
C
That
sounded
like
that's
for
users
to
do.
There
is
an
option
in
helm,
Chad
for
cilium
to
do
that
for
you,
but
it
only
does
it
for
the
cube
system,
namespace
and
again
for
a
serious
deployment
with
what's
located,
not
the
namespaces.
It's
not
going
to
work
so
cerium
in
most
cases
will
require
user
intervention
for
unstuck
cell
installs.
B
C
B
C
Clouds
filled
with
cube
net,
so
this
means
that
there
will
be
no
civility
of
communication
between
ports.
They
won't
be
able
to
see
each
other
and
obviously
related
checks
will
all
fail,
like
probes
and
stuff
this
hand
recommended
to
restart
the
pods
that
were
started
before
a
cilium
deployment.
For
this
case,
of
course,
yeah.
A
B
Well,
so,
right
now,
right
now
is
your
lab:
isn't
really
adding
anything
that
you
couldn't
just
go
to
play
solium
on
your
own
yeah
right,
so
we
don't
have
value-add
yet
because
we
just
came
out
in
12.8
with
it.
So
you
know
whether
you
install
it
via
get
lab
or
whether
you
just
install
it
by
yourself.
It's
not
really,
there's
no
difference
right
now,
but.
C
There
is
a
sound
halyard,
Katyn
editing
like
some
features
already
merged,
and
this
specifically,
is
that
so
we
already
have
a
dashboards
on
top
of
the
Salam
and
Salam
installation
actually
involves
installation
of
additional
diamond
sets.
That
will
be
related
to
starts
collection
in
parameters
integration.
So
there
is
a
certain
battle.
C
They
merge
direction
in
12.9,
even
dashboards.
It's
just
under
feature
flag
and
in
1210
feature
flag
has
been
removed.
Okay,
so
certain
value
like
in
quest
applications
right
now
for
user.
Because
of
the
bias
it
will
involve
:
and
get
familiar
with
to
projects
and
intricacies
of
some
of
the
configuration
values
for
the
sillim.
C
It
would
not
lift
the
requirements
to
get
familiarity
with
kubernetes
networking
and
just
architectural
silly,
but
there
is
a
certain
man
up
there
and
again,
we
already
had
quite
introducing
dashboards
and
I
feel
like
what
a
develop
support
is
also
up
there
for
psyllium.
One
thing:
I
also
wanted
to
mention
that
salam
is
not
hard
requirements
for
network
policies.
We
technically
can,
and
they
even
mentioned
that
you
can
install
something
else
like
on
google
quad.
You
can
enable
calico
bass
need
for
choruses
during
the
cost
of
deployment.
B
A
B
A
B
A
Awesome
yeah
what
I'm
trying
to
get
it
is
that
I
think
we
should
get
out
of
the
game
of
providing
value
to
facilitate
installation
because
so
context
dependent,
and
we
should
instead
focus
on
you
know
what's
facilitated
by
this.
Is
that
integration
part?
So
what
allows
you?
What
kind
of
value
that
you
get
when
you
integrate
or
register
that
application
into
Caleb,
so
in
foreign,
is
really
clear?
Suddenly,
the
runner
and
the
runner
now
has
the
ability
to
pick
up
jobs
from
from
the
scallop
instance
for
prometheus.
A
C
I
completely
agree,
but
at
the
same
time,
situation
based
parameters
that
is
maintained
by
management
team
and
some
of
the
our
products
that
we
integrated
is
a
bit
more
complicated.
It's
not
like,
let's
say
parameters.
Installation
involves
also
deployment,
an
additional
Cuban
ettus
plugin.
That
is
hard
requirement
for
our
iteration
metrics
and
dashboards,
and
on
top
of
that
there
is
a
whole
list
of
default,
metrics
defined,
like
by
parameters
team.
It's
also
goes
in
parallel
with
deployment
like
this
one-click
deployment,
so.
B
C
That,
let's
go
I
can
ask
users
to
install
and
configure
all
that
is
a
bit
of
a
stretch,
and
permit
is
probably
the
extreme
example
of
that,
but
even
for
us,
this
sillim
the
features
we're
adding.
On
top
of
that,
like
it's,
not
just
go,
install
and
salome,
and
then
installation
is
different
for
areas
in
google
quad
and
then
on
top
of
that
our
dashboards
have
a
requirement
and
additional
packages
provided
by
Silom
team.
So
it
might
be
complicated
in
some
cases.
A
Yes,
but
I
I
kind
of
I
agree
entirely
from
media's,
and
thanks
for
pointing
that
out
when
it
gets
to
psyllium,
if
we
have
all
those
things
does
I
mean
that
decision
is
completely
up
to
you,
your
team,
your
group,
actually
I,
just
wanted
to
kind
of
ask
some
questions
like
does
this
complicate
things?
If
we
transform
all
that
complexity
into
code
on
Gillett
or
calm,
or
would
it
be
better
to
I.
C
C
Approach,
but
the
main
purpose
of
me
being
on
this
call,
is
try
to
see
a
perspective,
and
so
I'd
really
helpful
to
me,
because
I
think
what
the
Stratton
is
right
now
in
tourmaline
in
inside
the
team
is
that
engineering
team
is
moral
support
in
support
of
the
quad
based
approach,
but
from
the
project
management.
We're
getting
a
lot
of
requirements
for
the
UI,
and
we
just
don't
know
right
now
how
to
perform
those
tasks
and
which.
C
Go
and
again
to
answer
a
question
ice
is
the
quite
cost
application
being
just
right
now,
a
great
like
solution
for
our
problems
in
terms
of
just
hiding
a
bit
of
complexity
of
sillim
installation,
but
still
providing
all
the
flexibility
our
customers
might
need
during
this
process.
So
I
would
say,
classifications
is
where
I
see
see
him.
He'll
be
yeah.
A
Yeah,
okay,
that
sounds
good
to
me
in
terms
of
where
I
see
I
see
faster
applications,
as
potentially
one
piece
of
the
puzzle,
I
think,
eventually
long
term,
we
will
have
to
focus
more
on
the
value
that
we
provide
when
deprecation
is
integrated,
regardless
of
where
the
P
use
faster
applications
or
even
one-click,
install
if
they
exist
or
if
they
install
it
themselves.
I
think
the
value
is
increasingly
more
and
more
on
what
happens
when
they
included,
especially
for
production
great
clusters.
A
Okay,
cool
thanks,
I
hope
that
provides
good
in
our
context.
Just
another
point
is
that
I
will
not
be
opposed.
If
we
include
you
know
a
different
method
here
and
yes
is
embarrassingly
inconsistent
and
looks
incomplete,
but
I
feel
like
this.
One
click
is
it
and
long
term,
and
it
would
be
great
if
we
can
start
moving
in
the
direction
of
cost
applications,
or
you
know
the
ability
to
have
to
shift
the
burden
of
install
today
a
context
that
makes
sense
for
the
users
context
yeah,
so
I
guess
step.
B
C
But
from
how
internal
team
discussions
were
always
he
was
in
in
discussion
much
for
like
knowledge
level
during
which
implementation
so
I
was
just
curious.
I
know
they
have
a
personas
on
the
issues
and
one
of
those
is
sree
like
persona.
So
my
question
how
our
PM's
should
handle
like
this
particular
issue,
who
hole
should
be
a
targeted
persona
for
our
features,
how
we
should
what
about
aim
knowledge
level
of
the
end
user?
B
Least
for
for
the
ones
that
I
put
out
right,
like
often
up,
we
target
Sam
who's,
the
security
analyst
right
now,
the
personas,
the
team
that
builds
out
the
personas
I
would
say,
doesn't
have
the
security
team
very
well
fleshed
out
the
Sam
security.
Analyst
persona
is
a
little
bit
of
a
catch-all
for
a
bunch
of
different
personas
that
exist
within
the
security
organization.
B
As
far
as
what
level
of
like
experience
like
have,
technical
users
are
generally
the
feedback
that
we've
gotten
at
least
so
far,
is
that
on
average
they
are
fairly
technical,
which
is
fine.
You
know,
I,
don't
know
that
we
necessarily
need
to
have
UI
built
out
for
everything.
In
fact,
some
users
do
prefer
to
keep
things
in
code
versus
UI,
at
least
for
this
particular
one,
though,
when
it
comes
to
turning
it
on
and
off,
I
feel,
like
the
two
goals
that
I
put
there
in.
B
The
notes,
though,
are
valid,
regardless
of
the
experience
level
right
one.
We
want
to
make
it
discoverable
so
that
they
can
actually
know
about
sodium
without
having
to
read
our
Docs
and
then
two
making
it
easy
to
turn
things
off
when
something
goes
wrong.
I
don't
know
that
it
necessarily
has
to
be
a
button.
It
could
be
a
set
of
instructions
right
there,
and
you
know
that
may
be
enough,
but
it's
got
to
be
something
where
you
know
they
can.
There's
got
to
be
some
way
that
have
a
panic
button.
B
A
It
makes
perfect
sense
to
me
I'll
point
you
to
an
issue
that
configures
thinking
about
as
well.
I
feel
like
that
applicable
to
almost
a
whole
ops
section,
primarily
because,
like
we
share
some
of
these
issues
as
well
like
with
monitor,
especially
and
maybe
a
little
bit
of
defend
too
so
we
have
three
different
personas
here
that
we're
talking
about
one
is
developer
and
which
I
think
is
relevant
or
to
DevOps.
But
then
then,
with
just
two
personas
that
we're
trying
to
flesh
out
so
I
think
like
this
could
be
useful
food.
A
So
I
think
in
my
my
thought
is
that
the
application
operator
would
already
have
knowledge
or
if
they
don't,
then
they
they
can
gain
that
knowledge
of
trying
things
out,
whereas
they
will
not
be
able
to
gain
knowledge
by
trying
to
understand
what
this
install
button
does
in
fact,
they're
probably
put
off
by
it,
because
that
means
that
they
would
have
to
go
digging
into
the
gillip
source
code,
which
is
probably
an
order,
magnitude
task
harder
than
say.
Reading
the
helm,
dogs
for
ingress
ingress
to
you.
B
B
C
B
A
C
B
A
C
B
So
right
now
all
we
have
is
just
these
wireframes
that
I
dropped
into
the
notes
here.
I
guess:
I'm,
not
100%,
clear.
All
right,
sir,
where
your
mind
is
on
on
going
from
here
like
should
we
have
Andy,
continue
to
flesh
out
the
mocks
with
an
install
button,
or
are
you
proposing
that
we
do
something
else
here,
they've.