►
From YouTube: Minikube Intro – Dan Lorenc, Google (Any Skill Level)
Description
Want to view more sessions and keep the conversations going? Join us for KubeCon + CloudNativeCon North America in Seattle, December 11–13, 2018 (http://bit.ly/KCCNCNA18) or in Shanghai, November 14–15 (http://bit.ly/kccncchina18).
Minikube Intro – Dan Lorenc, Google (Any Skill Level)
This session is dedicated to providing users with an introduction to the Minikube project. Attendees will learn the basics of Minikube's architecture, how to use and configure it, and some "power-tips" for using Minikube to setup an excellent development environment.
About Dan
He is a software engineer at Google.
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
A
Some
of
the
features
that
supports
the
original
design
goals
you're
going
over
the
history
of
it
and
keep
has
been
around
for
a
few
years
now
and
then
we'll
get
into
some
advanced
kind
of
tips
and
features
you
might
not
know
about
if
you're,
a
mini
cube
user
to
hopefully
improve
your
development
environment
and,
finally,
we'll
go
over
some
details
on
how
to
contribute
to
mini
cube.
It's
a
great
way
to
get
involved
in
the
kubernetes
community.
A
This
talk
is
meant
to
be
interactive,
mini
cube
is
a
tool.
Basically
everybody
in
the
kubernetes
community's
been
something
familiar
with.
Hopefully
so
stop
and
ask
me
questions
at
any
point
as
we
go
on
and
we'll
stop
at
the
end
too,
for
some
more
questions
to
start
off
with,
though,
and
help
me
get
a
feel
for
the
audience.
I
have
a
couple
of
questions.
So
could
you
put
your
hands
up
if
you
used
mini
Cuba
before
a
lot
of
people?
Has
anybody
contributed
to
mini
cube.
A
A
It's
a
little
bit
background
on
myself,
I'm
a
PLM
of
a
container
tools
team
at
Google
in
San,
Francisco
I've
been
a
mini,
cube
maintainer
since
the
start
of
the
project,
I
worked
with
some
other
people
to
get
the
project
started
and
have
been
running
it
and
helping
the
community.
Ever
since
recently,
our
team
has
worked
on
some
other
kubernetes
and
container
features
like
scaffold
rules
for
building
containers
with
basil,
some
distro
list,
container
images
and
a
whole
bunch
of
other
things
like
that
little
plug.
We
are
hiring.
A
So
if
this
stuff
sounds
interesting,
you
can
email
me
after
the
talk,
so
the
original
golden
mini
cube
was
to
give
people
away
get
started
with
kubernetes
and
2-1-1
command
to
start
up
a
cluster
on
your
laptop
or
on
a
desktop
before
mini
cube.
There
was
no
real
way
to
do
that.
You
had
to
set
up
a
cloud
provider
enter
a
credit
card.
A
Do
everything
on
VM,
somewhere
mini
Kuban,
was
meant
to
be
a
way
for
people
to
kick
the
tires
of
kubernetes
start
to
get
familiar
with
it
and
to
be
make
it
more
accessible
to
people.
It
was
also
meant
to
be
used
day
to
day
by
people
that
are
developing
applications
for
kubernetes,
so
many
queue
provides
a
lot
of
ways
to
streamline
your
development
and
to
do
image
builds
inside
of
that.
So
it's
a
lot
faster
than
having
to
do
builds
against
a
remote
cluster
where
you
have
to
upload
your
source.
A
Every
single
build
and
the
goal
is
to
support
most
of
the
kubernetes
features
that
make
sense
locally.
Obviously,
not
everything
is
going
to
make
sense
on
a
little
laptop
compared
to
a
giant
cluster
in
the
cloud,
but
the
goal
is
to
make
it
as
close
as
possible.
So
you
shouldn't
have
to
change
your
application
source
code
to
run
it
locally
versus
in
a
real
cluster
somewhere
else.
Things
like
multi,
node,
scheduling
and
affinity,
don't
really
make
sense
in
a
single
node
environment,
but
everything
else.
We
try
to
support
storage,
networking,
even
some
basic,
auto
scaling.
A
So
a
history
of
the
project,
it
was
one
of
the
first
sub
projects
created
in
the
kubernetes
org
and
it's
been
a
great
experience
to
be
part
of
the
community.
Since
the
very
beginning,
we
wrote
a
document
as
part
of
the
kubernetes
1.3
release
approved
at
one
point
10.
Now,
it's
kind
of
crazy
to
go
through
that
many
releases,
so
fast,
around
April
of
2016
work
started
immediately.
A
There
was
a
start-up
at
the
time,
read
spread
that
had
done
a
whole
bunch
of
work
in
the
space
with
the
tool
called
local
queue,
which
was
a
single
go
executable
that
ran
every
single
kubernetes
component,
that
you
would
need
inside
the
same
process
as
different
go
routines.
It
stood
up
XD
the
API
server.
Everything
like
that,
so
you
could
just
run
one
command
and
start
everything
up.
A
A
After
a
couple
months,
we
finished
up
the
first
couple:
mini
queue
releases
and
launched
with
kubernetes
1.3.
There
was
a
blog
post,
then
we
got
tons
of
usage
and
tons
of
feedback
from
the
community
right
away
and
tons
of
help
from
the
community
as
well
they're.
Only
a
few
of
us
really
working
on
it
at
the
time.
There's
something
like
300
contributors,
too
many
cubes
since
so
it's
definitely
something
that
could
have
happened
without
the
scale
of
the
communities
community.
A
A
So
the
next
one
to
talk
about
the
mini
cube
architecture,
which
kind
of
explains
how
how
it
works
and
how
it's
been
designed.
There
are
two
main
mental
models
that
I've
heard
a
lot
of
people
use
to
explain,
mini
cube,
I
kind
of
like
to
think
of
it
as
a
funnel
mini
cube,
is
supposed
to
run
a
kubernetes
cluster
on
a
laptop.
But
everybody
here
at
probably
has
a
different
laptop.
A
different
operating
system
installed
different
tools
on
it,
so
it's
kind
of
hard
to
support.
A
The
other
model
that
I
hear
a
lot
is
that
mini
cume
is
like
a
turducken
you're,
not
familiar
with
the
turducken.
It's
a
it
is
a
bird,
that's
popular
in
America
or
it's
a
turkey
with
a
duck
and
a
chicken
inside
of
that
and
by
that
uh-huh
and
mini
food
has
a
lot
of
layers
inside
of
it,
so
it
can
be
complicated
to
deal
with,
but
most
of
these
are
pretty
necessary.
A
What
this
looks
like
usually
is
you
have
a
laptop
and
then
you
start
a
mini
cube,
and
then
it
starts
up
a
virtual
machine.
A
couple
things
run
on
the
virtual
machine
like
at
Stian
and
the
couplet.
They
can't
really
run
inside
of
containers,
but
then
we
started
with
a
container
runtime
and
then
inside
of
that
container
on
time
we
run
containers
with
the
rest
of
the
kubernetes
components
in
them.
We
use
cout
Badman
to
do
that
setup,
so
it's
actually
somewhat
self-hosting,
where
kubernetes
is
actually
running
the
kubernetes
components
inside
of
itself.
A
Those
are
things
like
the
api
server,
the
scheduler,
all
the
other
kubernetes
server
processes.
So
actually,
if
you
start
up
a
mini
cube
cluster
locally,
that
should
that
one
here
and
you
do
a
get
pauses
before
you've
deployed
anything.
You
won't
see
anything
running,
but
if
you
do
get
pods
with
all
namespaces
flag,
you
can
actually
see
the
kubernetes
server
components
that
are
running
here
in
the
VM,
and
so
these
are
all
the
components,
the
kubernetes
and
managing
itself
inside
of
itself
inside
of
this
VM.
A
A
There
is
also
no
VM
driver
for
mini
cube,
which
runs
everything
directly
on
the
VM.
That
can
be
a
little
bit
harder
to
manage,
though,
because
you
need
the
exact
same
versions
of
all
the
components
that
kubernetes
requires
and
if
you
don't
have
a
way
to
set
all
that
up,
it
can
be
a
lot
of
work
for
you.
A
A
That's
correct
yeah,
so
mini
Cuba
when
you
first
start
it
up:
it
installs
the
couplet.
It
downloads
a
version
of
the
couplet
that
Kubb
ATM
uses
and
installs
that
on
the
host,
it
runs
it
inside
of
system
D
a
system
to
unit,
and
then
it
uses
crew,
Bay
DM
to
bring
up
the
rest
of
the
components
inside
of
containers.
A
The
main
way
to
talk
to
things
that
are
running
inside
of
the
VM
are
using
node
ports,
which
is
one
of
the
bigger
differences
between
mini
cube
and
running
in
a
full
cluster.
Today,
the
mini
qvm
has
one
IP
address
for
the
whole
vm,
and
if
you
want
to
address
a
service,
we
can't
really
set
up
a
load
balancer
on
your
laptop,
so
the
recommended
way
is
to
use
in
node
ports,
so
your
service
gets
a
specific
port,
and
then
you
can
talk
to
that
over
the
mini
cube
IP
address
on
that
port.
A
A
A
A
Another
less
commonly
known
feature
I
talked
about
a
little
bit
earlier,
is
that
you
can
switch
out
the
container
runtime
use
inside
a
mini
cube
today
it
supports
docker
rocket
and
cryo
using
the
CRI
spec.
So
all
these
are
installed
in
mini
cube.
You
can
only
use
one
of
them
at
a
time,
so
you
have
to
tear
down
your
cluster
and
start
another
one
up.
If
you
want
to
switch
these,
but
that's
something
that
sig
note
is
working
on
improving
and
you
cube
also
uses
the
cni
for
network
interfaces.
A
Just
like
a
cloud
provider
would
and
an
upcoming
feature.
I'll
talk
about
a
little
bit
later
is
load
balancer
support,
so
we've
been
working
on
proposals
for
a
while
and
how
to
better
fake
the
load.
Balancer
cloud
experience
to
make
it
so
users
don't
have
to
change
their
configurations
when
running
and
mini
cube,
vs.
running
inside
of
a
larger
cloud
cluster.
A
So
once
you've
got
your
cluster
up
and
you
played
around
with
kubernetes
a
bit
and
you're
trying
to
set
up
mini
cube
in
your
day
to
day
development
for
your
kubernetes
application.
There
are
some
features:
we've
added
to
Minich
you
to
make
that
faster
and
easier
for
you,
depending
on
which
operating
system
you're
on
mentioned
before
you
can
mount
volumes
from
the
host
into
the
virtual
machine
and
use
this
to
sync
data
or
sync
code.
A
So
you
can
build
your
images
inside
of
the
same
docker
daemon
that
will
be
used
to
run
them,
so
you
can
skip
pushing
and
pulling
those
images
without
this
you'd
have
to
do
a
bill
to
push
it
to
a
registry
summer
and
then
have
your
cluster
or
your
laptop
pull
the
image
back
down
this
way,
you
can
do
a
bill
and
just
run
the
new
container
immediately
and
depending
on
your
operating
system.
There's
a
command
like
this
one
to
set
up
those
environment
variables
so
that
your
docker
seal
I
can
talk
directly
to
mini
cube.
A
You
can
also
run
a
whole
bunch
of
custom
add-ons
that
we
built
in
the
mini
cube.
Some
of
these
are
enabled
by
default,
like
the
dashboard
I
showed,
but
there
are
quite
a
few
inside
a
mini
cube
that
you
can
play
with
that
have
been
contributed
by
the
community,
so
you
can
see
a
full
list
of
these
with
mini
cube.
A
Mini
Cube
add-ons
list
and
you
can
see
which
ones
are
enabled
by
default,
the
the
add-on
manager
itself
is
an
add-on.
The
storage
class
is
what
sets
up
default
host
path,
provisioning
and
the
min
to
keep
dashboard
are
all
enabled
by
default,
and
you
can
turn
on
some
of
these
other
features
at
core
dns
ingress
is
a
very
common
one
keeps
tier
if
you
want
to
experiment
with
auto
scaling
or
monitoring
your
containers.
A
The
other
comment
feature
that
isn't
quite
known
is
that
many
cube
is
heavily
configurable
and
this
configuration
can
be
saved
and
shared
with
people
across
your
team.
So
all
these
things,
like
the
add-ons
you've,
enabled
the
special
flags
for
docker
special
flags
for
the
kubernetes
components
can
be
saved
in
a
mini.
Keep
the
configuration
file
written
yourself
into
JSON,
or
you
can
use
it,
set
these
with
mini
cube,
config
yet
and
then
keep
the
config
set
and
it'll
be
persisted
across
different
clusters.
A
A
A
A
This
extra
one
is
pretty
powerful
and
enables
you
to
use
the
configuration
pretty
well.
You
can
mount
in
special
folders
from
your
host
into
the
mini
qvm
at
startup
time,
so
they
can
configure
the
docker
demon
in
a
lot
of
ways.
You
need
to
setup
configuration
files,
they
can't
be
done
with
flags
and
they
need
to
be
there
when
the
cluster
first
starts
up
and
so
there's
a
mini
cube,
special
mini
cube
directory
in
your
home
directory
that
gets
copied
into
the
virtual
machine
mounted
from
the
root
directory.
A
So
if
you
need
to
override
a
certain
file
inside
of
the
unit
startup
time
to
configure
the
API
server
or
the
docker
daemon,
you
can
just
put
that
file
into
your
dot
min
you
keep
directory
with
the
right
path
and
it'll
get
copied
in
before
any
clustered
components
get
started
up
some
pretty
common
ones
here
to
put
a
script
in
the
boot
local
directory,
then
we'll
run
to
do
any
further
customization.
You
can
download
other
tools
set
those
up
even
swap
out
some
of
them
and
you
keep
the
components.
A
Another
very
common
feature:
that's
used,
a
lot
of
mini
cube.
Users
is
to
change
out
the
cluster
version.
Kubernetes
releases,
new
cluster
versions,
major
versions
every
three
months
and
you
might
have
versions
constantly.
You
typically
want
any
keeping
to
match
the
exact
version
you're
running
in
prod,
which
you
probably
don't
upgrade
that
much
it's
almost
a
full-time
job
to
keep
upgrading
kubernetes
that
fast.
So
you
can
pin
mini
cube
to
a
specific
kubernetes
version
using
the
Caze
version.
A
And
so
before
that
it
was
a
bunch
of
work
for
the
mini
cube,
maintainer
x'
to
set
up
new
builds
for
every
kubernetes
released.
It
would
come
out
so
I'm
gonna
keep
it
tended
to
lag
behind,
but
now
as
soon
as
a
cuvette
DM
build
is
published
even
once
from
ahead,
you
can
switch
to
those
in
your
cluster.
A
This
final
one
at
the
bottom,
is
about
configuring
all
of
the
kubernetes
components:
dynamically
there's
about
a
million
switches
to
all
the
different
kubernetes
components
like
the
API
server,
the
proxy
and
everything
that
people
need
to
change,
and
it
would
be
basically
impossible
to
add
flags
and
configuration
for
all
of
that
directly
to
many
cute
because
they
constantly
change.
So
there's
a
configurator
feature
in
mini
queue.
It's
the
extra
config
flag
and
you
can
set
any
flag
to
any
kubernetes
component
using
this
one.
A
So
in
this
example
of
passing
extra
config
equals
and
we're
setting
a
flag
on
the
kulit.
The
food
bar
fly
equals
bad,
and
so
you
can
look
up
the
cuvette
Dion
configuration
for
all
of
your
components
and
anything
that's
edible
via
the
cuvette
niem-uml
you
can
set
with
configuration,
flies
mini
cube
here.
A
A
Want
to
talk
a
little
bit
about
contributing
to
mini
queue,
it's
a
great
way
to
get
started
in
the
community
like
I
mentioned
before.
We
have
something
like
300,
different
contributors
that
have
used
this
as
a
way
to
jump
into
larger
kubernetes
projects.
A
common
first
PR
that
people
send
us
is
around
the
add
ons
that
we
support
they're,
constantly
getting
updated
and
people
like
to
add
new
ones
and
update
the
existing
ones.
So
there's
some
great
documentation
on
the
kerbin
ideas
web
page
about
how
to
do
that.
A
A
A
Yeah,
there
are
definitely
some
examples.
I
wasn't
aware
that
Travis
had
installed
a
mini
cube
by
default,
but
I've
seen
a
bunch
of
blog
posts
and
tutorials
on
how
to
set
that
up
inside
of
Travis.
A
lot
of
people
use
the
no
VM
driver
for
that
scenario,
and
so
they
have
pre
submit
tests
that
deploy
configurations
and
make
sure
that
everything
comes
off
correctly
inside
of
a
CI
environment.
That's
definitely
something
that
we
use
not
for
mini
cube
itself,
but
for
other
projects,
and
it
is
useful.
A
Great
question,
testing
mini
cube
is
very
difficult
because
it
requires
virtual
machines
and
not,
although
there
was
support
running
inside
of
other
virtual
machines,
where
tests
typically
run.
We
do
have
a
whole
bunch
of
physical
machines
that
are
typically
under
people's
desk
at
work
running
things
like
Windows,
and
we
have
a
bunch
of
back
menus
for
testing
Mac,
which
is
hard
to
virtualize
as
well,
and
so
we
do
a
full
set
of
intend
tests
across
all
of
these
different
drivers
and
operating
systems
on
physical
machines.
A
It's
hard
to
keep
them
passing
all
the
time,
because
these
machines
need
to
be
rebooted
and
logged
in
to
to
keep
them
up,
but
yeah.
We
do
try
very
hard
to
test
many
keep
across
all
of
these
platforms.
I
think
we're
the
only
kubernetes
project
that
has
tests
on
Mac
OSX
today,
because
it's
so
difficult
to
do.
A
A
So
that
the
question
was
regarding
the
difference
between
the
configuration
file
and
flags
for
configuring
mini
cube,
not
all
flags
are
supported
in
the
configuration
file
yeah.
We
our
goal
is
to
try
to
support
all
of
them
in
the
configuration
file,
but
the
configuration
file
is
JSON
and
some
of
these
flags
have
very
strange
and
difficult
types
to
marshal
into
JSON,
so
we
typically
have
two
special
case
flags
that
people
need
that
we
can't
just
turn
into
JSON
directly.
If
there's
a
specific
one
you're
interested
in,
we
can
add
support
for
that.
A
The
question
was:
do
we
know
what
the
mini
keep
users
are
using
it
for?
Is
it
local
development
or
CI?
Was
it
used
before
git
push
or
after
I
think
the
answer
is
both
people
use
it
for
all
sorts
of
things?
There
was
just
a
very
large
community
survey
done
by
say,
gaps
where
we
added
a
whole
bunch
of
mini
cubed
questions
and
got
some
of
the
best
feedback.
We've
gotten
so
far
on
how
users
are
using.
It.
I
think
that
was
one
of
the
questions.
A
A
A
The
very
first
thing
that
happens
is
a
fixed
version
of
mini
key
was
installed,
mini
cube,
does
new
releases
and,
as
a
new
kubernetes
releases
come
out,
we
switch
the
default
version,
so
one
common
pitfall
I
see
is
that
people
use
the
latest
version
of
mini
cube
and
then
one
day
the
kubernetes
version
will
change
and
some
of
their
things
will
break.
So
you
should
app
into
a
specific
version
of
mini
cube,
and
that
holds
true
for
anything
you're
installing
in
the
CI
system.
A
A
A
Yes,
so
the
question
was:
if
you
use
the
no
VM
driver,
are
there
any
extra
steps
you
have
to
follow
and
yeah?
Basically,
the
no
VM
driver
works
by
running
kubernetes
on
whatever
VM
or
physical
machine
you're
trying
to
use
it
on,
and
so
you
have
to
make
sure
that
all
the
kubernetes
dependencies
are
present.
Kubernetes
isn't
fully
self-contained.
It
depends
on
a
whole
bunch
of
special
versions
of
Linux
tools
that
are
installed
on
the
VM
itself.
A
That's
basically
it
so
you
have
to
make
sure
you
have
the
correct
versions
of
things
like
eath
tool
and
IP
tables,
and
so
cat
and
things
like
that.
The
proxy
and
the
other
kubernetes
components
will
use,
which
is
why
it's
so
much
easier
for
many
people
to
test
itself
inside
of
a
VM
where
we
can
control
the
versions
of
everything.
But
if
you're
doing
it
yourself
on
your
own
machine,
then
it's
up
to
you
basically.
A
A
So
the
question
was:
is
there
a
way
to
fake
resources
inside
of
mini
cubes?
So
you
can
pretend
your
laptop
has
four
CPUs.
If
it
only
has
two
for
that
specific
example,
I'm
not
sure
if
it
would
be
possible,
mini
cube,
does
support
flags
to
configure
the
CPUs
and
RAM
that
are
available
inside
of
the
cluster,
but
I
think
pretty
sure
you
can
only
configure
those
to
be
things
that
are
less
than
what
your
machine
actually
has
so
I
think
the
default
is
for
CPUs
and
for
gigabytes
of
RAM,
or
something
like
that.