►
Description
Full notes for this episode: https://github.com/heptio/tgik/tree/master/episodes/083
- 00:00:00 - Welcome to TGIK!
- 00:04:29 - Week in Review
- 00:17:42 - Putting Kubebuilder in context
- 00:22:56 - Minecraft image we'll be using
- 00:26:07 - Running Minecraft on k8s manually
- 00:34:10 - Ingress for Minecraft?
- 00:39:17 - Starting with kubebuilder
- 00:47:46 - Adding our first API type
- 00:53:43 - Aside: Operators vs. Controllers
- 01:02:18 - First run of controller!
- 01:04:11 - Leaders election and HA in k8s controllers
- 01:14:32 - Starting our reconciler
- 01:16:00 - RBAC requirements via annotations
- 01:29:47 - Aside: `generatedName` and its history
- 01:52:33 - Bug: reversed objects in `SetControllerReference` (James Munnelly caught it but I missed it in the chat!)
- 01:58:00 - First run with reconciler!
- 01:59:44 - Debugging first errors
- 02:08:20 - Almost there! CrashLoopBackoff
- 02:13:48 - Success!
- 02:14:00 - Wrapping up
Come hang out with Joe Beda as he does a bit of hands on hacking of Kubernetes and related topics. Some of this will be Joe talking about the things he knows. Some of this will be Joe exploring something new with the audience. Come join the fun, ask questions, comment, and participate in the live chat!
This week I'll be doing some programming! We'll crack open kubebuilder and see what it takes to build a Minecraft controller. I don't know how far we'll get but it'll be fun!
A
Alright,
hello,
everybody
welcome
to
TGI
kubernetes.
My
name
is
Joe
Beda
and
I
am
your
host
today.
For
those
who
don't
know,
TGI
Cooper
Nettie's
is
a
weekly
livestream
that
me
or
somebody
else
on
the
vmware
team
here
does
where
we
take
a
piece
of
kubernetes
technology.
Sometimes
we
know
well,
sometimes
we
don't.
We
pick
it
up.
Look
at
it
figure
out
how
it
works
and
do
that
all
live.
Where
you
can
ask
questions,
we
can
all
learn
together
and
it's
a
it's
a
good
old
time.
So
joining
with
me
here
today
is
Duffy.
A
Who's
gonna
be
manning
the
chat
he
also
sometimes
hosts,
and
so
sometimes
it's
helpful
to
have
somebody
else
in
the
chat
that
can
actually
help
do
stuff.
You
know
there's
a
little
bit
of
lag
for
the
chat.
So
if
you
want
to
tell
me
something,
it'll
take
a
few
seconds
before
I
can
go
ahead
and
see
it.
So
I
like
to
start
out
by
saying
hello
to
everybody
and
I,
there
were
a
few
folks.
Who's
been
who've
been
online
for
a
while.
A
So
if
lumati
here
Martin,
you
know
from
the
Netherlands
Lou
Matty
good
to
see
you
again
Suresh
from
Hamburg
Filipe
from
Paris
Friday
night
yeah,
it's
Friday
night
in
Europe.
This
is
a
the
poor
timing
that
is
tjk
Rory
from
Scotland
good,
to
see
good,
to
see
a
Jamie
Jo
from
Virginia
James
how's
it
going
man
Steve
James
from
the
UK,
so
James
works
for
for
jet
stack,
does
the
cert
manager
stuff
and
then
Steve
is
another
VMware
employee
good
to
see
ya
reinhard
from
germany
federico
from
gothenburg.
A
He
doesn't
say
it
there,
but
I
know
he
taught
me
how
to
say
it
once
philippe
from
germany,
st
yak
from
bangladesh,
arun
good
to
see
ya
mark
is
gonna,
make
fun
of
me.
So
here's
the
deal
so
when
I
joined,
VMware
I
came
in
as
a
principal
engineer
and
and
they're
like
you
know,
you're
gonna
have
to
write
code.
They
actually
interviewed
me.
You're
gonna
have
to
write
code
and
I'm
like
yeah.
No,
that
sounds
good
I
really
want
to,
but
like
to
myself.
A
I'm
thinking
like
my
job
oftentimes
involves
a
lot
of
coordination,
a
lot
of
architectural
review.
I.
Don't
get
a
lot
of
time
on
day-to-day
basis
to
write
code
and
mark
was
one
of
the
folks
that
actually
interviewed
me
coming
in,
and
so
this
is
actually
gonna
be
a
little
bit
scary
because,
like
he
may
decide
that
he
made
a
mistake,
as
he
watches
me
fumble
around
here,
but
that's
gonna
be
part
of
the
fun.
A
Let's
see
Joe
from
London
good
to
see
you
send
deep
from
Poland
and
we
have
another
Joe
from
Atlanta
good
to
see
y'all
all
right.
So
today's
episode,
what
I'm
gonna
do
is
I'm
gonna,
actually
try
and
write
some
code
and
we'll
see
how
far
we're
gonna
get
I'm,
probably
not
going
to
be
doing
everything
I
want
to
do,
but
I'm
gonna
be
playing
with
this
tool.
Kit
called
cube
builder,
that
is
essentially
a
starter
kit
for
building
kubernetes
controllers
and
the
controller
that
I
want
to
build,
is
to
launch
and
manage
minecraft
servers.
A
So
the
idea
here
would
be
that
using
a
kubernetes
custom
resource
definition,
we
can
create
a
new
kubernetes
object,
which
is
a
minecraft
server,
and
it's
going
to
be
this
controller
that
it
will
take
care
of
actually
launching
and
managing
that
thing
and
we're
gonna
start
with
something
very
simple.
Just
like
you
create
a
CRD,
it
goes
ahead
and
creates
the
server
and
then
we'll
talk
about.
A
A
So
yeah
yeah,
everyone
say
mark
can't
fire
me
now
so
Bismarck
from
Hoboken
I
don't
have
a
beer.
I
still
have
work
to
do
after
this
I
could
grab
a
little
bit
of
scotch
or
you
know
or
whiskey,
but
I
think
I'm
gonna
pass
right
now
and
then
mark
nice
to
see
ya
all
right:
oh
yeah,
Demark,
hot,
okay,
so
and
Sayid
from
London
alright.
A
So
the
other
thing
we
like
to
do
is,
as
we
start
out,
is
we
have
a
bunch
of
sort
of
interesting
things
that
are
going
on
around
the
community
and
I
want
to
run
through
some
of
that?
What
we
have
here
switch
to
my
screen,
and
let
me
know
if
you
all
can
read
this
I
can
I
can
make
it
a
little
bit
bigger.
So
what
we
have
here
is
we
have
what's
called
an
a
hack
MD,
which
is
essentially
a
collaborative
markdown
document.
A
A
Let's
see
so
let
me
go
through
some
of
the
stuff
that
that
we've
pulled
together
here.
So
the
first
thing
that
I
think
folks
want
to
be
aware
of
is
that
there's
a
big
changing
a
big
change
coming
in
in
kubernetes
1.16,
and
that
is
we
deprecated
like
these,
like
as
kubernetes
api
is
evolved.
Well
at
some
point,
we
had
a
core
set
of
api's
and
we're
like
hey.
These
are
getting
really
big.
We
need
to
move
stuff
out
so
that
they
can
start
grouping
them.
A
We
came
up
with
this
idea
of
having
the
the
sort
of
grouping
mechanism
as
part
of
our
API
s,
and
the
initial
group
was
this
thing
called
extensions,
but
at
some
point
we're
like
hey,
that's
gonna,
be
really
really
nasty,
and
so,
and
so
we
deprecated
those
a
while
ago
and
in
1/16
they're,
actually
gonna
get
removed.
So
these
things
all
got
moved
to
a
different
sort
of
API
group
and
and-
and
so
specifically,
these
things
are
moving
from.
A
Like
extensions
v1
beta
one
to
like,
say,
networking
cat
Co
for
network
policy
or
policy
v1
beta
one
for
pod
network
pod
security
policy,
so
like
there's
different
mappings
as
these
things
got
moved
around,
but
the
old
extensions
hosting
of
this
is
actually
going
away.
These
things
have
been
deprecated
for
a
while.
This
is
gonna,
be
a
little
bit
of
a
painful
change
for
folks.
There's
gonna
be
a
lot
of
folks
who
are
gonna,
have
to
go
and
update
some
of
their
deployments,
some
of
their
scripts.
A
There's
gonna
be
a
lot
of
broken
samples
out
there.
Duffy
is
going
to
do
this
next
week
on
TGI
K,
where
he's
really
gonna
dig
into
this
a
little
bit
and
play
with
it.
So
that's
I
think
that's
the
plan
right
Duffy
is
that
we're
going
to
be
doing
next
time.
So
that
is
something
that
folks
should
be
aware
of
and
start
getting
start
getting
ready
for
now,
and
this
post
was
was
written
by
Valerie
and
she's
working
at
lyft
I
like
Valerie
a
lot.
So
that's
what
we
got
going
on
there.
A
The
next
thing
is
that
cube
con
San
Diego
CF
peas
have
closed
so
good
luck
to
everybody
who
was
submitted.
Hopefully
we'll
see
a
lot
of
you
there
and
a
lot
of
good
talks
about
what
folks
are
doing
with
with
kubernetes
and
beyond:
hey
olive
how's.
It
going
so
all
of
his
another
VMware
employee
she's,
one
of
the
field,
engineers
that
we
now
called
katz,
kubernetes,
architect
team
and
that's
the
same
team
that
duffy
and
josh
are
on
also
and
we're
trying
to
twist
olives
arm
and
have
her
do
a
TGI
kay
sometime
soon
also.
B
A
Right
so
the
next
thing
is
that
we
have
this
article
here.
This
is
really
cool,
assisting
pull
together.
Oh
that's,
my
sort
of
a
real
list
of
like
a
whole
bunch
of
different
security
tools
as
an
overview
and
I
think
this
is
really
really
really
interesting.
There's
a
lot
here.
One
of
the
things
that
I
did
this
week
is
I
met
with
I
met
with
the
folks
from
the
CN
CF
security
SIG's.
A
So
these
are
folks
that
work
with
and
advise
the
the
TOC
and
the
CN
CF
talking
about
how
they
can
actually
help
promote
sort
of
good
security
tools,
a
breakdown
of
security,
around
projects
and
what
they
do
and
what
different
things
are
good
for
what
across
the
entire
cloud
native
space
and
so
I
think
there's
some
interesting
stuff
going
on
there
and
hopefully
we'll
see
more
stuff
like
this
pull
together
in
a
curated
way.
But
this
is
actually
a
great
place
to
start
right
now.
Oh
Craig
actually
grabbed
the
beer.
For
me
here
want.
A
A
A
Us
if
you
have
questions
going
through
them,
so
cool
and
Christophe
from
Corsica
good
to
see
ya
all
right
and
then
there's
a
new
special
interest
group
that's
joined
and
this
is
called
sig
usability
and,
let's
see
you
can
read
the
Charter
here.
Did
this
get
merged
I
think
this
got
merged,
which
means
that
it's
real,
which
is
very
exciting,
and
the
idea
is
that
core
end-user
usability,
and
so
this
is
actually
going
to
cover
things
like
experience,
but
also
accessibility,
localization
internationalization
stuff,
like
that.
A
So
that
is
that's
pretty
cool
and
you
can
see
how
the
sausage
is
made
in
terms
of
this
actually
getting
merged,
but
but
that's
actually
pretty
cool.
It's
that
sig
usability
is
now
here,
and
so,
if
you
want
to
get
involved
in
that,
the
details
are
going
to
be
here,
sort
of
like
ok,
they're,
gonna
work
and
find
the
right
time
for
everybody
to
start
meeting.
A
Ok-
and
let
me
close
some
of
this
stuff
because
we're
not
there
yet
and
then
there's
a
lot
of
release
stuff
that
I
think
I'd
like
folks
to
be
aware
of
so.
The
first
thing
is
that,
with
this
latest
release,
we
actually
have
a
new
website,
which
is
rel
note,
skate
Co,
which
is
like
really
really
really
cool
and
the
person
who
wrote
this
I
don't
have
it
in
the
notes
here.
I
would
definitely
want
to
give
credit
here,
because
ok
do
we
have
it
about
here.
A
Oh
man
who
signed
their
work,
I
saw
this
on
Twitter
and
I
definitely
wanted
to
remember
to
call
out
the
person
who
did
it,
but
this
is
really
really
cool
because
it
essentially
like
we
do
have
a
process
where
every
time
you
actually
file
a
PR
there's
a
section
where
you
can
say
whether
this
should
be
part
of
the
release,
notes
or
not
and
then
like
and
but
that
was
essentially
something
that
could
be
scraped,
and
there
was
a
manual
process
for
curating
these
things.
But
now
we
actually
have
this
thing
that
automatically
scrapes
it.
A
So
you
can
actually
read
all
the
release.
Notes
break
it
down
by,
like
you
know,
the
area
that
you're
working
with
whether
it's
API
server
or
security,
whether
it's
a
bug
or
a
feature,
very,
very,
very
cool.
To
actually
start
seeing
this,
and
it
does
give
you
sort
of
a
a
scope
of
the
understanding
around
like
how
much
is
really
going.
Ok,
so,
oh,
it
was
Jessica
that
actually
did
the
the
new
the
new
really
release
notes.
So
thank
you
so
much
Jeff
for
doing
that
super
cool
stuff.
Let
me
do
that.
A
If
you
want
to
do
that,
so
we
got
oh
yeah
that
there's
Steven
and
then
also
a
cig
release
also
presented
to
the
community
meeting,
and
so,
if
you
want
to
actually
check
that
out
also,
this
is
one
of
those
things
that
where
we
really
it's
a
very
human
intensive
process
right
now
to
do
a
kubernetes
release,
and
so
there's
a
lot
of
ongoing
work
to
make
this
stuff
be
more
automated,
and
so
there's
a
lot
of
room
to
get
involved.
Learn
that
process,
but
then
also
help
to
to
make
it
be
more
automated.
A
A
They
they
announced
this,
like
trio
of.
Let
me
find
a
good
link,
for
it
will
do
the
linux
com1
sure
no
lame,
ok
source
there
we
go
so
I
haven't
had
a
chance
to
decode
everything-
that's
going
on
here,
so
this
is
come
and
arrow.
Ok,
so
they
really
T
is
through
note
three
new
projects:
there's
code,
wind,
Kevin,
arrow
and
app
city.
A
It
looks
like
there's
a
lot
here,
but
I
honestly
haven't
had
a
chance
to
really
dig
in
with
everything
that's
going
on
here,
but
I
think.
The
idea
is
that
this
is
sort
of
at
the
application
level
in
terms
of
the
development
workflow,
a
standard
set
of
templates
that
you
can
follow
so
I'm
going
to
put
a
leak,
a
link
in
there,
because
that
was
something
that
I
meant
to
mention
forgot.
To
put
it
in
the
notes
here,
IBM
releases,
three
new
projects.
A
A
This
Katherine's
yeah,
so
I'm
sorry
I,
was
confused
because
I
know
that,
like
you
know,
there
was
some
other
stuff
that
they
did
also
the
I'm
torn
on
it.
I
got
to
be
honest.
On
the
one
hand,
I
think
it's
really
interesting.
This
idea
of,
like
hey,
let's
lend
this
stuff
down.
You
know
because
it's
been
around
for
a
while.
A
The
things
that
I'm
really
conflicted
about
is
that
it
essentially
is
in
some
ways
a
fork
of
kubernetes
because
doing
things
like
like
slipping
in
sequel
light
as
the
engine
is
like
workable,
especially
when
you
have
a
single
node
master
for
this
thing,
but
it's
it's
something
that
is
a
significant
divergence
from
what's
happening
upstream.
So
I
really
wish
that
the
Rancher
folks
had
worked
with
had
worked
with
upstream.
A
So
that's
that's
kind
of
my
take
on
this,
but
but
definitely
a
lot
of
really
interesting
ideas
there.
Alright,
let's
go
ahead
and
get
started
here
all
right,
so
the
tool
set
that
we're
going
to
be
using
as
we
do
this
it's
going
to
be.
Okay.
So,
first
of
all
the
goal
I
want
to
write
a
new
controller
for
kubernetes,
and
now
I
actually
did
a
live.
A
Three-Part
sort
of
T
gik,
where
we
did
this
in
the
past,
but
we
started
from
scratch
with
client
go
number
one
client
go
has
changed
dramatically
since
then,
and
number
two.
We
now
have
some
tool
sets
that
actually
make
it
a
lot
easier
to
get
started
in
building
code
for
kubernetes.
Specifically,
there
are
things
like
queue
builder,
but
also
the
the
operator
framework
from
Red
Hat.
So
the
one
that
we're
going
to
be
looking
at
is
is
queue
builder,
which
is
really
a
sort
of
you
know.
The
goal
is
to
be
a
curated
easy-to-use.
A
A
Or
there's
a
version
2
that
I
think
is
in
beta
right
now:
okay,
so
we're
in
beta
right
now
for
4,
and
so
oh
wow.
This
was
released
an
hour
ago,
so
I
don't
know
which
version
I
have
so
anyway,
so
that
was
released
an
hour
ago
with
a
bunch
of
with
a
bunch
of
bug,
fixes
I,
wonder
if
that
was
the
TGI
K
effect,
or
hopefully
they
were
just
doing
that
stuff
anyway,
but
anyways.
A
I
think
if
we
can
get
that
up
and
running
sort
of
you
know
today,
that'll
be
good,
but
then
we
can
start
talking
about
sort
of
like,
as
we
turn
this
into
a
more
generic,
a
more
feature
filled
operator
for
minecraft.
What
are
some
of
the
things
that
want
to
do?
What
are
some
of
the
future
directions?
That
might
be
fun?
A
A
Like
you
know,
one
of
the
one
of
the
folks
working
super
hard
in
the
API
machinery
cig
talking
about
this
idea
of
structural
schemas,
because
there's
this
problem
right
now
with
CR
DS,
where,
if
you
typo
something
or
if
you
put
data
in
there,
that's
not
part
of
the
schema,
the
CRD
just
carries
it
forward
and
then
at
some
point
like
you,
don't
get
good
feedback
of
like
hey.
Your
thing
doesn't
conform
to
the
schema,
whereas
this
is
a
change
where
it's
like
hey.
A
If
this
is
not
defined
as
part
of
the
schema,
we're
not
going
to
go
ahead
and
accept
it.
So
it's
a
further
sort
of
evolution
of
CR
DS.
That
makes
them
more
and
more
like
the
built-in
objects.
So
that's
exciting.
I
don't
know
if
this
is
something
if
we're
gonna
actually
go
through
and
lock
things
down
to
this
level.
Maybe
this
is
something
that
we
can
play
with
as
we
go,
but
but
definitely
something
to
be
aware
of.
A
If
you're
doing
this
type
of
development
by
the
time
CR
DS
hit
1.0
the
goal
is
that
this
stuff
will
be
on
by
default.
So
this
was
one
of
the
sort
of
lessons
over
the
beta.
The
long
long
beta
period
of
CR
DS
well
links
get
xed
out,
so
we
should
be
like.
If
you
put
a
link
in
there,
there
should
be
a
way
that
we
can
approve
it
not
sure.
A
What's
going
on
just
ok
and
then
the
other
thing
I'm
excited
about
this
is
Stefan
and
and
Michael
house
Blas
are
also
they
have
this
book.
That
this
will
Riley
book
on
programming,
kubernetes
that
looks
really
interesting
and
I
started.
Reading
the
pre-release
one,
if
you're,
if
you're
a
Riley
Safari
remember
you
can
go
ahead
and
do
that
and-
and
this
looks
really
topical
for
this
type
of
thing,
so
I
think
it's
great
that
we're
starting
to
see
some
content.
A
Some
books
come
out
that
really
talk
about,
not
just
how
do
you
use
kubernetes,
but
how
do
you
actually
start
start
programming
it
not
out
yet
I.
Think
like,
let's
see
if
I
can
like
okay
I'm
logged
in
well,
you
can
they
have
like
the
first
chapter
here
and
then,
if
you
want
to
go
further,
I
think
if
you're
so
far,
you
can
do
but
like
that
you
know
this
again
talks
about
like
hey
look:
we
have
a
CR
D,
a
custom
controller
and
then
it's
going
to
manage
a
bunch
of
objects.
A
B
A
The
first
thing
so
resources
that
we're
gonna
be
using
as
we
go
is
there
is
a
oh
wait
wrong
thing:
there
is
this
minecraft
image
and
I'm
actually
using
this
with
my
kid
at
home,
not
not
under
kubernetes.
Just
like
you
know,
half-assing
it
with
a
host
and
docker
run,
which
is
essentially
so
this
person
I
t
zg,
who
is
this?
This
is
what's
his
name.
A
Jeff
born
who's
in
Texas
seems
like
a
good
guy,
so
he
actually
maintains
some
docker
files,
and
one
of
them
is
kind
of
like
the
minecraft
server
docker
file
and
oh
and
here's
old
somebody
did
include
a
link
to
something.
That's
already
so,
basically,
okay,
here's
one
using
the
operator
SDK,
alright,
so
the
different
different
approach
to
this
using
the
operator
SK
will
be
doing
the
same
thing
with
with
cubed
learn
alright.
So
anyway,.
B
A
This
is
kind
of
like
I,
think
kind
of
the
the
the
the
de-facto
container
image
for
Minecraft,
and
it
turns
out
that
minecraft
is
actually
a
little
bit
tricky
to
containerize.
There's
some
really
weird
things
around
it,
such
as.
If
you
just
kill
the
process,
you
may
lose
data,
so
you
have
to
actually
from
the
outside
gracefully
shut
it
down,
but
it
turns
out
that's
hard
to
do
because
you
can't
just
sort
of
send
it
a
cig,
hop
or
something
like
that.
A
A
B
A
There's
a
lot
of
sort
of
like
gunk
that
you
have
to
do
to
run
Minecraft
that's
taken
care
of
with
this
with
this
image,
so
we're
gonna
be
using
that
and
then
and
then
queue
builder
itself
has
a
essentially
a
book,
and
this
is
actually
what's
what's
really
interesting.
This
covers
queue
builder
version
2
that
essentially
provides
a
quick
start
in
a
guided
tour
and
so
we'll
essentially
go
through
the
book.
A
But
instead
of
doing
the
examples
there
we're
gonna
go
through
and
try
and
do
our
own
thing
we'll
use
the
book
kind
of
as
a
guide.
Now
I've
never
done
this
before
so.
I
want
to
be
stumbling
through.
This
we'll
see
again
how
far
how
far
we
can
get
okay.
So
the
first
thing
that
first
things
first,
what
we
have
here
is
Cube
control,
get
no
nodes.
We
have
like
a
two
node
plus
a
master
cluster
running
in
AWS.
We
can
do
Cube
control
version.
What
is
it
it's?
It's
1.13
all
right.
A
We
haven't
updated
this
in
a
while,
so
it's
getting
a
little
long
in
the
tooth,
so
it
probably
doesn't
have
support
for
the
structural
CRD
stuff
that
we
were
talking
about
and
and
then
one
of
the
things
I
did
right
before
we
actually
went
on
as
I
was
trying
to
get
a
version
up
and
running
just
a
test
that
I
could
connect
to
it.
So
what
I
have
here
is
I
have
a
Yama
file
with
a
deployment
for
Minecraft.
A
That's
running
this
thing,
we're
running
the
paper
type
server,
we're
saying
that
you
we're
exposing
that
over
a
port
with
type
equals
load,
balancer
TCP!
So
theoretically,
if
all
goes
right,
this
should
actually
be
a
mine
craft
server
running
at
this
particular
address
and
and
I
might
be
missing.
Something
but
like
get
I
didn't
have
I
actually
have
a
chance
to
verify
it.
So
there's
the
service
that
we're
running.
Let
me
go
ahead
and
launch
minecraft
and
we'll
see
how
far
off
we
are
because
I
think
the
first
things.
A
B
A
The
link
to
the
minecraft
server
there's
the
one
though
they
link
it
there
yeah
I
mean
at
some
point
like
that
kind
of
is
not
the
point
of
the
game
for
us
to
like
interactively
play,
minecraft
right
Oh,
so
it
says,
can't
connect
a
server.
So,
let's
see
if
we
can
figure
out
what's
going
wrong
here,
so
something
is
actually
not
happening
here
and
so
I
didn't
have
a
chance
to
debug
this
before
we
actually
go
ahead
and
went
ahead
and
got
started
so
you
cute
control,
it
should
be
okay.
B
A
A
A
Yeah
we'll
look
at
the
endpoints
just
to
make
sure
next,
okay,
I'm
gonna
show
folks
what
I
did
and
I
took
the
the
ELB
here.
Let
me
turn
this
down
a
bit
and,
and
so
now,
if
I
go
to
minecraft,
that's
the
server
address,
and
this
should
theoretically,
but
it
says,
can't
connect
a
server
and
so
I'm
gonna
actually
see.
What's
going
on
here,
cube
control
get
endpoints.
A
Minecraft,
okay,
so
that's
not
working
we're
not
actually
having
things
hooked
up.
Oh
you
know
what
I
I
know
what
I
did
wrong
I.
Actually,
the
selector
is
at
I
called
one
MC
in
the
other
minecraft,
so
I'll
go
through
and
I'll
do
that
I'm
gonna
change
this
to
the
selector
for
the
service
is
there
so
now
I
can
do
apply
again.
B
A
A
A
A
A
Yeah,
as
I
was
doing
this
one
of
the
things
that
I
did
is
I
actually
didn't
quote
this,
and
then
this
turned
blue,
and
so
this
would
actually
not
be
a
string
pasty
kubernetes,
and
so
you
have
to
remember
to
quote
this
stuff.
So
that's
like
the
you
know,
the
the
typical
gotcha
that
we
get
going
on
there,
yeah
I,
think
it'll
default.
I.
Think
we're
good.
Now,
though,
well
Washington
does
have
some
of
the
best
best
city
names,
so
we
have
Mukilteo.
A
All
right,
so,
let's
look
at
you
control
logs,
all
right,
so
it's
so
okay!
So
now
what
this
actually
does
under
the
covers-
and
this
is
part
of
the
scripts
here-
is
that
it
actually
downloads
the
server
version
dynamically
and
then
boots
it
up
and
has
it
running
and
so
now
where'd
my
minecraft
server
go
there
we
go.
Oh.
A
A
Well,
it
says
encrypting,
don't
whoohoo,
yet
we're
still
working
on
this
okay.
So
one
of
the
things
that
I
think
is
that
I
heard
was
a
problem
with
minecraft.
Is
that
when
you
proxy
things?
Oh
there,
it
goes.
Okay,
uh-huh
other
people
are
actually
joining
all
right.
So
we
are,
you
were
definitely
in
and
now
I'm
asking
to
be
owned,
all
right,
so
whoever
it
was
actually
sort
of
copied
that
thing
off
the
screen.
Good
job
I'm,
proud
of
you
all
right,
so
we
have
something
up
and
running.
A
A
One
of
the
advanced
things
that
I
would
love
to
do
over
time
is
actually
write,
ingress
for
Minecraft,
and
so
it
turns
out-
and
this
is
like
I'm
just
like
trying
to
think
ahead
here-
we're
not
going
to
have
a
way
to
get
it
to
get
to
this
today.
Is
that
yeah
two
to
three
minutes
and
there's
also
a
cost
associated
with
it
right
and
so
like.
If
we
were
able
to
do
ingress
for
Minecraft,
we
could
actually
have
one
ELB
and
then
actually
do
this
sort
of
the
multiplexing.
On
the
other
side,.
A
And
so
so
there
is
like
looking
at
the
Minecraft
protocol.
It
doesn't
look
that
bad
and
there's
actually
a
handshake
where
it
actually
sends
the
hostname
that
you
actually
were
looking
at.
So
we
could
do
something
like
a
wildcard
DNS.
Have
everybody
connect
to
that
and
then,
if
we
can
actually
sniff
the
handshake
of
the
Minecraft
protocol,
we
can
dispatch
based
on
that,
and
so
we
can
essentially
make
ingress
for
for
for
minecraft,
though,
we
can't
use
alb,
because
it's
not
HTTP,
it's
a
TCP
protocol
right.
A
So
at
the
end
of
the
day,
if
we
want
to
do
something
special
like
that,
we're
gonna
have
to
write
our
own
sort
of
reverse
proxy
server
yeah.
So
you
know
fun
fun
things
when
I
say
ingress
I
mean
not
HTTP
ingress,
something
in
an
ingress
that
speaks
the
minecraft
protocol
natively
and
is
able
to
actually
pull
out
this
this
string,
so
that
would
actually
be
a
fun
fun
project
at
some
point,
but
not
today,
probably
okay,
so,
okay!
A
So
now
we
have
a
proof
of
life
that
we
can
actually
connect,
which
I
think
is
really
exciting.
I
think
I'm
gonna
leave
this
running
in
the
background
just
for
the
music,
and
you
know
to
warm
myself
from
my
laptop,
but
let's
look
at
like
okay,
I'll,
you
know,
and
and
as
I
was
doing
this
just
to
be
sure
like
there's.
A
couple
of
that
Duffy
was
leading
me
through
is
that
you
can
do
cute
control.
If
we
look
create,
you
can
do
something
like
this
cute
control
create
service.
A
If
you,
you
guys
have
all
seen,
you
know,
I'll
probably
have
seen
they
like.
If
you
do,
the
oh
yeah,
mall
and
then
dry
run
this
essentially.
What
this
does
is
this
actually
gives
you
sort
of
a
starting
point
for
the
ammo
for
stuff
and
then
and
then,
and
then
you
can
copy
and
paste
this
in
and
then
like.
A
You
know
using
this
hostname
this
path,
these
other
sort
of
keys,
which
set
of
backends,
do
I,
redirect
that
there
are
people
who
do
switching
on
TCP
but
you're
either
doing
that
based
on
the
port
like
well.
If
it
comes
on
this
port,
I
want
to
send
it
to
these
things
or
you're,
doing
it
based
on
some
other
protocol
type
stuff,
and
so,
if
you're
doing
TLS,
you
can
do
it
based
on
like
s
and
I,
which
is
sort
of
the
the
the
way
that
you
can
actually
say.
A
Hey
I
want
to
do
a
TLS
connection.
Here's
the
here's,
the
parameters
and
the
host
that
I'm
looking
for
that
works.
For
now
with
s
and
I,
there's
actually
work
to
start
the
encryption
and
then
do
the
selection
of
the
certificate.
So
now
it'll
be
a
little
bit
difficult
more
difficult
to
to
do
that
in
the
future.
A
But
but
when
I
say
you
press
for
Minecraft,
what
we're
actually
saying
is,
instead
of
looking
at
an
HTTP
header,
we're
gonna
have
to
look
at
something
specific
to
the
minecraft
protocol,
not
the
HTTP
protocol,
but
that's
that's
yeah.
You
can
also
switch
based
on
based
on
on
cookies.
To
is
one
way
to
do,
but
that's
probably
not
something
that
we're
gonna
get
to
today.
A
Okay,
so
cool
all
right.
So
let's
actually
get
started
with
cube
builder.
Now,
because
we
have,
we
have
some
proof
of
life,
so
the
first
thing
you
need
to
do
is
install
cube
builder
and
so
that,
basically
you
curl
some
stuff
you
on
tar.
It
put
it
in
a
directory
and
put
it
in
your
path,
and
so
I've
actually
gone
through
and
already
done
this,
and
then
you
also
have
to
have
customized
along
also
and
and
if
we
look
at
this
I
put
this
in
my
home
directory.
A
And
then
there's
a
bin
and
it
actually
is
queue
builder,
but
also
there's
a
version
of
cube
control.
The
API
server
a
net
CD
here
also
I,
think
this
is
for
doing
local
unit
tests
is
why
this
stuff
is
included
here,
but
I
think
you
know
at
least
initially
we're
just
gonna
be
dealing
with
with
cube
builder.
A
So
I
can
do
you
know,
cube
Builder
version
or
okay
I,
guess
that
doesn't
work,
but
because
it
alright,
okay.
So
it's
it's
expecting
a
certain
amount
of
environment
to
be
set
up
for
it
first,
but
I
do
want
to
like
make
sure
that,
because,
like
just
an
hour
ago,
the
latest
version
was
released.
So
let's
make
sure
that
we're
actually
using
the
latest
version.
A
A
What
I'm
gonna
do
is
I'm
gonna,
create
a
new
go,
go
path
here,
so
make
dirt
go
export,
go
path,
equals
go
okay,
so
we
have
a
new
go
path.
Now
before
you
used
to
go
modules,
one
of
the
things
that
you
pretty
much
anytime,
you
want
to
do
anything
and
go.
You
had
to
have
it
be
someplace
in
your
go
path,
but
with
go
modules.
There's
actually
now
a
way
where
you
can
have
your
sort
of
your
route
program
be
someplace.
A
Besides
your
go
path,
and
so
you
don't
have
to
sort
of
CD
deep
into
your
go
path
to
be
able
to
do
stuff,
and
so
that's
what
we're
gonna
actually
be
doing
today
to
get
this
stuff
set
up,
so
we're
gonna
call
our
project.
What,
if
you
call
our
project
and
call
it
like
kind
craft?
That's
bad,
but
I'm
gonna
go
with
it.
So
we
need
to
go
into
kind
craft
and
then
what
you
can
do
is
so
here
it
says
you
can
do
a
go
mod
in
it.
A
So
it
says:
if
you're
not
in
go
path,
you'll
need
to
run,
go
mod
in
it
to
be
able
to
tell
you
cube
builder
and
go
the
base
import
path
of
your
module,
so
we're
gonna
do
that
and
we're
gonna
go
with
github
comm,
slash,
JB
de
slash
kind
craft
I
haven't
created
that
repo
yet,
but
we're
actually
good
to
go
so
now
this.
What
this
does
is
it
just
creates
a
go
mod
file.
This
go
mod
file,
says
hey,
you
know.
This
thing
was
done
with
with
go
1.12,
and
here
is
the
module
name.
A
A
Of
looking
at
the
current
directory
relative
to
go
path,
it
just
looks
at
go
mod
to
figure
this
stuff
out
and
then
cube.
Api
server
and
Etsy
D
are
built
a
sort
of
Darwin
maka
executables
right,
which
generally
those
things
aren't
built
and
distributed,
and
so
like.
Maybe
we
should
for
testing
I,
don't
know
okay,
and
then
we
can
do
a
queue
builder
in
it.
My
do
this
is
actually
gonna,
be
the
base
domain
of
the
types
for
our
CR
DS.
A
Now,
if
you
look
at
a
lot
of
sort
of
see,
RVs
that
are
part
of
the
kubernetes
project,
you'll
actually
see
them
they're
part
of
Kate's
dot.
Io,
don't
do
this
unless
you're
part
of
the
kubernetes
project,
you
shouldn't
use
Kate
style,
and
even
if
you
are
part
of
the
kubernetes
project,
you
probably
shouldn't
use
Kate's
do
unless
you're
signing
up
to
actually
go
through
the
full
API
review
process
in
kubernetes
insig
architecture.
A
Just
this
last
week,
we're
actually
for
four
things
that
are
sort
of
like
sig
related,
not
part
of
the
core
to
some
definition
of
core.
We
actually
introduced
a
new
domain,
X
Katie,
oh
and
so
I
registered
that
last
week,
and
so
we're
going
to
be
using
that
now
sort
of
the
sort
of
sort
of
adjacent
projects
as
part
of
the
kubernetes
projects.
But
you
shouldn't
even
do
that.
Instead,
you
should
think
your
own
domain
and
so
I'm
gonna
do
TGI
K
dot
IO,
because
that's
something
that
we
own
okay.
A
So
what
this
is
doing
here
is
it's
going
through
and
adding
a
whole
bunch
of
dependencies
and
building
stuff
and
setting
up
essentially
a
bunch
of
scaffolding.
So
this
is
in
some
ways
like
if
you've
ever
done,
Ruby
rail
Ruby
on
Rails
type
of
development.
This
is
kind
of
like
a
Rails
in
it.
It's
like
get
me
a
project,
that's
up
and
running,
and
that's
just
sort
of
scaffold
up
and
so.
B
A
Alright,
so
this
is
now
up
and
running,
and
what
we
can
look
at
here
is
if
we
look
inside
of
kind
crafts,
and
so
let
me
actually
do
code
this
and
let
me
close
this
other
code
window
because
I
don't
want
us
to
get
confused
here,
alright,
so
there
we
go.
So
what
do
we
have
here?
So
we
have
a
project
file
which
actually
this
is
like
metadata
about
our
project,
so
that
next
we're
running
when
we're
running
cube
builder.
A
It
actually
knows
about
this
stuff,
there's
a
make
file
to
make
a
whole
bunch
of
stuff
which
is
actually
kind
of
fun,
including
the
the
the
the
image
name
that
we're
gonna
be
building
and
pushing
where
we
can
find
things
like.
Go
bin
and,
let's
say
tests,
build
a
manager,
I,
don't
know
we're
gonna
learn
about
a
bunch
of
this
stuff.
We
do
controller
Jen,
which
is.
A
Yeah
this
gets
defined
here
where
we're
actually
building
this
thing
called
controller
tools.
So
there's
a
whole
bunch
of
sort
of
tooling
that
this
brings
in
and
stuff
like
that,
we
have
a
main
file
here.
Oh
I
need
to
actually
install
some
extensions
and
stuff,
so
we
have
a
main
file
here.
I
had
just
done
this
I
wonder
why
it's
like
getting
mad
at
me.
A
With
like
a
bunch
of
stuff
like
hey,
look,
we
actually
get
metrics
and
stuff
like
that
out
the
gate.
Some
go
mod
stuff
because
we
added
a
whole
bunch
of
things.
We
have
API
machinery,
Clank,
OH,
controller,
runtime
and
controller
tools,
these
things,
and
so
there
are
some
locks
here
in
terms
of
the
versions
we
want
to
get
a
docker
file
for
actually
building
our
thing
and
interesting
stuff,
where
this
is
actually
a
multi-stage,
build
that
we
have
going
on
here
and
then.
A
Let's
see
some
boilerplate
that
oh
wow
there's
a
whole
bunch
of
fun
stuff
going
on
here,
so
this
actually
gets
put
assumes
that
you
know
at
the
top
of
every
file
that
they
create
essentially
a
config
in
terms
of
like
how
do
we
actually
get
things
up
and
running
with
customized
and
includes
cert
manager,
which
is
fascinating.
Our
we're
gonna
have
to
dig
into
what's
going
on
there,
that's
pretty
opinionated,
the
manager
binary
actually
gets
built
locally
and
then
there's.
Oh,
that's
my
BS
code
setting.
B
A
And
we're
getting
some
completion
cool.
The
green
squigglies
are
spelling
correction
things.
It's
a
it's
a
mod
that
I
have
installed
alright,
so
let's
actually
figure
out
what's
going
on
here
so
now.
The
next
thing
is
that,
like
we
create
a
new
API
group
called
web,
app
v1
with
the
kind
being
guestbook,
and
so
let's
actually
copy
this,
but
instead
of
a
guestbook,
we're
gonna
do
what
are
we
gonna?
Do
so
cube,
build
or
create
okay,
ctrl
shift
P
ends
up
being
the
reload
window.
Okay,
thanks
boggin
create
API
Oh.
A
A
A
Alright,
so
there's
a
couple
tools
here
that
are
actually
being
downloaded
yeah.
It's
just
an
excuse
for
me
to
play
minecraft
and
drink
beer.
Exactly
so
there's
this
there's
this
binary
here
called
controller
gen
and
we
actually
have
sort
of
the
boilerplate.
That's
gonna
put
at
the
top
of
it
sort
of
saying
that
we're
doing
Apache
stuff
and
then
the
past
that
I
wanted
to
do.
And
then
this
is
actually
going
and
we're
running
a
couple
of
tools,
go
mega
and
ginkgo
and
I
think
it's
worth
and
the
it's
it's
running
these
things
for
us.
A
Automatically,
let's
see
so
this
is
I'll.
Get
this
all
test
stuff
go.
Megan
and
ginkgo
are
our
test
stuff.
There's
other
tools
and
I
think
those
are
probably
we
already
brought
those
in
that
essentially
take
proto
files.
Our
go
files
turn
them
into
things
like
like
proto's
and
stuff,
like
that
I'm,
not
sure
if
we're
actually
using
any
of
those
yet
but
alright.
So
what
we
have
then
is
under
API.
We
have
a
v1
alpha
one
and
under
that
we
actually
have
our
group
inversion.
A
Okay,
so
the
way
that
resources
are
identified
in
kubernetes
is
that
you
have
a
group
and
a
version.
So
the
group
is
minecraft
eg,
I
Kao.
The
version
is
v1,
alpha,
1
and
then
and
then
there's
a
whole
bunch
of
schema
management
stuff
for
being
able
to
actually
translate
stuff
back
and
forth
between
different
schemas.
A
Here's
the
type
that
we
have
here
so
it
essentially
we
have
a
server
here
and
it
allows
you
to
do
sort
of
lists
of
server
and
then
a
server
has
essentially
aspect
in
a
status,
and
so
this
is
where
we
actually
put
in
the
the
fields
that
are
going
to
help
define
our
CR
D
and
then
there's
generated
code.
That
knows
how
to
do
a
deep
copy
of
this
stuff,
and
so
in
the
kubernetes
world.
A
Our
generated
code
actually
has
a
ZZ
underscore
in
front
of
it,
and
then
you
end
up
with
things
like
get
ignore
that
says:
I
think
if
we're
right
here,
yeah
everything,
that's
easy,
that's
not
as
part
of
vendor.
We
actually
ignore
something
like
that.
I
think
is.
It
is
what
we
end
up
doing
or
no
maybe
I'm
wrong.
A
Skip
generated
files
except
for
vendored
files.
That's
what
it's
supposed
to
do!
I'm,
not
sure
that
that
does
what
it's
supposed
to
do.
Okay,
maybe
there's
a
bug
there?
Okay,
so
we
did
that
we're
not
gonna
edit
anything.
Yet,
let's
see
so
you'll
need
a
kubernetes
cluster
running
again
see
you
can
use
kind
to
get
a
local
cluster
for
testing
or
running
against
a
remote
cluster.
Your
controller
will
automatically
use
the
current
context.
Your
cube
config
file.
A
Maybe
we
should
do
it
again
sky.
Well,
let's
just
do
this
and
see
see
what
happens
here?
I,
don't
should
we
do
kind?
Do
you
think
we
should
do
I?
Have
my
cluster
up
and
running
we'll
actually
go
ahead
and
do
that
now?
What
does
make
install
gonna
actually
do
if
I
go
through
and
I
look
at
make
file
make
file
install
is
going
to
do
queue,
control,
apply,
eff,
config
CRD
basis.
So
if
I
look,
it's
config
CRD.
A
Oh
yeah,
so,
okay
you're
right,
it
also
generates
controller,
and
so
here's
like
we
have
a
test
suite
for
the
controller
server
ginkgo
and
go
mega
comes
in,
and
then
here's
this,
this
sort
of
the
boilerplate
for
actually
doing
the
server,
and
so
the
big
thing
that
we're
gonna
actually
adding
logic
to
here
is
this
reconcile
loop.
So
that's
the
interesting
stuff.
Oh
it
first
runs,
manifests
okay,
cool
thanks.
A
It's
gonna
generate
our
base,
manifests
for
the
CR
DS,
and
so
it's
essentially
going
to
read
the
goat
code
and
read
the
and
from
that
extract
out.
Essentially
the
schemas
that
we
need
for
the
schemas
that
we
need
for
installing
the
CR
D,
so
Philippe
says,
is
that
it
also
generates
operator
sources
in
controllers
yeah,
it's
okay,
so
operator
versus
controller.
This
is
a
very
confusing
thing
and
I
don't
think.
As
an
industry,
we've
actually
decided
upon
this
in
my
mind,
and
so
this
is
my
definition.
Other
people
will
disagree
with
me.
A
A
controller
is
a
generic
piece
of
logic
in
kubernetes
that
looks
at
some
sort
of
desired
State
and
tries
to
make
something
happen,
either
in
kubernetes
or
elsewhere,
and
then
writes
the
status
back
into
status
right.
So
it
often
pairs
with
a
CR
D,
and
so
it's
like
read
some
state.
Do
some
stuff
write
some
status
back?
Do
that
in
a
loop
until
things
settle
down
an
operator
is
a
type
of
controller
again.
A
My
definition
that
has
domain-specific
knowledge
about
managing
a
specific
system,
and
so
something
like
replica
set
is
pretty
generic
I
call
that
a
controller,
something
like
what
we're
doing
here
is
very
specific
to
minecraft.
In
my
case,
then
I
think
that
then
becomes
an
operator,
but
the
the
relationship
we
need
controller,
an
operator
and
like
when
you
use
one
versus
the
other,
not
super
well
defined
in
our
world.
A
A
A
Do
you
agree
to
the
EULA
and
we
can
do
you
bool
and
so
we'll
just
pass
these
things
so
we'll
take
a
bunch
of
the
environment
variables.
We
will
hoist
those
things
up
into
into
this
thing
here.
That's
probably,
let's
do
want
another
thing,
that's
fun
that
will
actually
be
able
to
see
here
because
I
think
that
those
things
are
boring
and
this
this
documentation
here
is
not
easy
to
find
stuff
in.
B
A
A
A
B
A
B
A
A
A
A
B
A
When
we
do
run
here,
it's
actually
just
doing
a
go
run
on
main,
and
so
what
happens
here
is
that
if
you
have
your
cube
config
setup
correctly,
you
can
actually
run
the
controller
as
you,
and
so
you
don't
have
to
build
a
docker
image
and
deploy
it
to
be
able
to
run
a
controller.
The
controller
can
actually
operate
from
within
the
cluster,
using
a
service
account
or
it
can
operate
from
outside
the
cluster,
while
you're
doing
development,
and
so
that
ends
up
being
really
interesting.
So
we
can
do
a
make
run.
B
A
And
you
can
see
it's
actually
doing
a
bunch
of
the
stuff
that
it
needs
to
do
to
be
able
to
actually
run
this
thing.
We
don't
we're
not
doing
anything
yet,
but
at
least
we
have
the
scaffolding
now.
So
it
turns
out
that,
like
we
haven't
done
anything,
but
we've
done
a
lot
here,
because
I
think
a
lot
of
folks.
It's
it's
not
easy
to
go
from
zero
to
actually
having
a
client
up
and
running
with
client
go
and
have
it
connect
to
a
server.
A
It
turns
out
that
that's
actually
a
fairly
complicated
thing
to
get
up
and
running,
and
so
it's
cool
that
we
actually
have
this
stuff
up
and
running
so
easily,
and
so
we've
already
sort
of
saved
a
lot
of
time.
So
now
the
the
scaffolding
that
we
have
here
for
for
the
main,
let's
actually
down,
dig
into
this
a
little
bit
and
see
what
we
have.
Oh
no
I,
don't
want
to
close
that
altogether.
Well,
maybe
maybe
I
do
I'll
close
this
altogether
for
now.
A
Okay,
so
we're
importing
a
whole
bunch
of
stuff,
including
I,
like
our
API
stuff
and
our
controllers,
and
our
knit
function
where
essentially,
what
we're
doing
is
we're
adding
a
new
scheme
and
then
we're
adding
to
that.
The
the
scheme
is
essentially
I.
Believe
now
correct
me.
If
I'm
wrong,
those
who
know
client
go
better,
it's
essentially
the
set
of
sort
of
like
versions
that
we
know
about.
A
Client
go
scheme,
add
to
scheme,
and
so
it
sort
of
collects
these
things
together
and
then
we
have
runtime,
which
is
part
of
the
API.
So
this
is
essentially
just
like,
essentially
the
the
directory
of
schemas
that
we
know
how
deal
with.
We
have
a
metric
server,
so
you
know
if
you
want
to
actually
have
Prometheus
track.
The
number
of
Minecraft
servers
that
we're
running.
We
can
go
ahead
and
do
that
and
whether
or
not
we
need
to
enable
a
leader
election.
A
Is
it
actually
uses
the
kubernetes
api
itself,
and
so
it
actually
writes
into
an
object
and
there's
a
race
to
see
which,
if
your
things
can
write
into
the
object,
the
first
one
to
write
in
the
object
wins
they
have
to
update
it
every
so
often
if
they
don't
update
it.
The
other
ones
notice
that
they
haven't,
that
it
hasn't
been
updated
and
then
they
go
through
and
race
to
see
which
of
them
can
actually
then
become
claim
leader,
and
it
used
to
be
that
people
synchronize
on
the
endpoint
object
in
kubernetes.
A
Don't
know
we'll
have
to
see
what's
going
on
here,
I,
don't
know
if
they're
still
using
the
endpoint
or
not
for
the
leader
election
stuff.
I'd
be
something
to
be
interested
in
looking
into
okay
and
then
we're
setting
up
a
logger
and
we're
using
zap,
which
is
a
logging
thing
that
I
actually
not
aware
of
packaged
app
contains
helper
for
setting
up
a
new
log,
our
logger
okay,
so
it
seems
like
everybody
is
like
log.
R
is
the
way
to
go.
That's
what
all
the
cool
kids
are
using
now.
A
A
A
No
wonder
why
I'm
like
what
it's
all
that
crap?
Okay,
so,
let's
see
so
we're
setting
up
the
loggers,
a
blog
or
okay,
so
we're
doing
that
and
then
we
set
that
logger
into
the
controller.
What
is
the
true
here
mean?
Oh,
whether
it's
a
development
thing,
I,
guess
I,
don't
know
and
then
we're
creating
the
controller.
Here
we
have
the
the
controller
runtime
we're
creating
a
new
controller
manager.
A
A
A
Alright,
so
there's
a
lot
going
on
here,
but
but
this
is
good
stuff.
So
essentially,
this
is
a
helper
manager
that
manages
a
lot
of
the
gunk
around
sort
of
bringing
up
in
managing
controllers
and
then
we're
passing
into
it.
A
scheme
which
is
the
thing
that
we
added
sort
of
essentially
all
the
types
that
we're
dealing
with,
that
we
know
about
it,
publishes
metrics
and
it
actually
takes
care
of
the
leader
election
stuff
and
so
like.
This
is
like.
If
we
look
at
leader.
A
A
And
so
this
is
the
thing
that
actually
goes
through,
and
this
is
our
set
of
controllers,
our
business
logic,
and
we
say
we
want
to
initialize
our
business
logic
with
our
client,
which
is
essentially
our
connection
to
the
kubernetes
api,
server
and
logging
information
and
then
and
then
we're
gonna
say,
set
yourself
up
using
a
manager
and
if
things
don't
work,
we
exit
out
and
then
we
go
ahead
and
you
start
the
manager
and
it
does
things
like
handle,
signals
and
stuff
like
that,
and
so
I
think.
The
interesting
thing
then
ends
up
being.
A
The
controller
we
have
set
up
with
manager-
and
we
can
essentially
do
you-
know
we're
calling
it
the
controller,
runtime
new
controller
managed
by
this,
for
this
is
actually
the
the
the
object
that
this
controller
is
watching
and
and
we
want
to
complete
it
and
actually
pass
it.
The
reconciler
and
so
the
reconciler
ends
up
being.
A
For
the
server
right,
okay
and
so
the
reconciler
is
the
place
where
we
can
actually
go
through,
and
we
have
a
context
that
we
can
work
with,
and
then
we
have
our
log
with
values,
server
the
the
namespace
that
we're
actually
operating
with
here
also,
and
then
we
can
do
whatever
logic
we
we
need
to
be
able
to
do
there,
okay,
so
that
I
think
is.
That
is
the
this
sort
of
the
rough
lay
of
the
land
here
that
we're
working
with.
A
A
This
is
a
place
to
put
it
here
which
is
useful
and
then
here's
how
you
go
through
and
actually
build
the
image
and
go
ahead
and
deploy
it,
and
you
may
need
to
do
some,
our
Bakke
stuff,
and
it's
like
you,
may
need
to
grant
yourself
cluster
admin
privileges,
let's
not,
but
we'll
get
there.
Okay,
all
right!
So
that's
the
quick
start
here,
but
let's
actually
get
a
little
bit
deeper
and
okay
so
launch
configuration
manager
in
our
back
okay,
so
there's
config
stuff
that
we're
talking
about.
A
We
have
go
mod,
make
file
project
we've
already
looked
at
all
that
every
program
has
a
main
look.
It
just
told
you
all
the
stuff
that
I
talked
about
so
the
controller
is,
it
will
actually
use,
go
routines
to
actually
run
a
bunch
of
controllers
in
the
background,
okay
groups
and
versions.
We
talked
about
this
a
little
bit.
We
have
kinds
and
resources,
so
kinds
are
essentially
types
and
then
resources
are
where
they
show
up
in
the
URL
path,
and
so
typically
these
things
are
one-to-one
where
it's
like.
A
A
A
A
Init
schema
builder
designing
all
right,
so
this
is
building
a
cron
job,
spec
they're,
going
deeper
on
this
stuff,
where
they
essentially
have
a
job
template
spec
spec
that
they're
actually
pulling
in
out.
So
this
is
a
way
that
you
can
reuse
like
a
pod
spec
or
something
like
that
to
be
able
to
do
this,
which
is
pretty
cool.
We
didn't
go
nearly
as
deep
here
and
then
we
define
a
custom
type
to
hold
our
concurrency
policy.
Oh
wow,
so
they're
going
pretty
deep
here,
okay,
we
don't
need
to
do
all
that.
A
A
A
A
Okay,
this
is
actually
really
cool.
So
one
of
the
things
that
you
can
do
is
you
add
annotations
here
in
terms
of
what
our
back
permissions
you
need
and
then
that'll
actually
help
build
the
install
scripts
that
you
need
for
this
stuff,
and
so
here
what
we're
saying
is
that
for
queue
builder,
our
back
groups,
we
need
the
server's
resource.
We
want
to
be
able
to
get
list,
watch,
update
patch
delete
and
then
for
status.
A
For
minecraft,
we'll
just
create
pods
directly
right,
and
so
if
the
pod
doesn't
exist,
we'll
go
ahead
and
recreate
it
because
we
don't
want
to
like
you
know,
there's
no
reason
for
us
to
use
replica
set
or
anything
like
that
under
the
cover.
So
our
example
we
use
deployment,
but
this
is
just
fine,
also
so
we'll
go
through
and
we'll
do
queue
builder.
Our
back
groups,
equal,
oh
and
then
go
explain,
pods,
which
is
gonna,
be
just
kinda
pod.
A
All
right,
let's
see
if
that
works,
you
know
what
we
don't
need.
We
can
do
the
up.
I'm
gonna,
do
a
to-do
list,
our
back
stuff
right
just
so
that,
like
since
we're
gonna,
be
testing
out
as
us
right
now.
We
don't
need
to
be
able
to
do
that.
Okay,
but
that's
actually
really
cool
that
we're
using
annotations
so
that
you're
not
going
through
and
doing
all
the
our
back
stuff
yourself.
You
can
do
it
right
here,
keep
everything
together
and
have
the
essentially
the
controller
declare
which
are
back
permissions.
It
needs
out
the
gate.
A
So
next
we
get
to
the
heart
of
the
controller,
the
reconciler
logic
so
scheduled
time.
Annotation
is
batch.
Tutorial
queue
builder
scheduled
death,
so
we
actually
have
a
annotation.
So
this
is
just
a
constant
string
and
then
reconcile
we're
already
there.
Okay.
So
now
we
need
to
load
the
thing
up.
So
now
we're
gonna
actually
go
through
and
we're
gonna
do.
Var
MC
server
equals
a.
A
Okay-
and
this
is
called
child
k
batch,
so
this
batch
here
I'm
confused.
What
is
this?
This
is
okay.
This
is
our
type
right
because
we're
doing
this
thing
where
it's
like
we're
wrapping
the
other
batch
thing.
So
we
have
like
batch,
which
is
our
batch
versus
okay,
but
whatever
okay,
MC
server
and
then
for
us.
This
is
going
to
be.
A
Okay,
so
Fernandez
that
for
the
are
back
annotations,
its
group
okay,
so
if
it's
part
of
the
core
group,
you
just
put
a
comma
there
to
essentially
say
that
the
core
troop
is
that
the
group
has
left
blank
thanks
for
an
M.
That's
that's
super
interesting
okay,
so
we
have
context
here
the
name
space
coming
in
and
then
we
go
through
and
we're
returning
result
and
then
the
ignore
not
found
what
this
is
essentially
saying,
like
look
if
we're
trying
to
like
if
we're
trying
to
load
something-
and
it's
not
there.
A
A
A
A
Okay,
Sally
says
that
we
can
also
do
group
equals
Corps,
Sally
I
didn't
know
you're
here,
Sally
is
you'll,
see
his
comments
in
this
stuff.
Sally
is
actually
working
on
this
stuff,
so
Sally.
If
you
see
me
doing
something
particularly
stupid,
please
let
me
know
here
because
I
haven't
programmed
this
stuff
in
forever
and
it's
all
changed
since
that
last
I
looked
at
it
and
I'm
learning
a
lot
here.
Okay,
so
the
get
next
schedule.
Does
this
little
sort
of
inline
function
here,
that's
going
to
be
returning.
A
A
Yeah,
this
is
a
usability
test
for
you
with
somebody
who
knows
a
very
odd
slice
of
what's
going
on
here.
Okay,
so
we
actually
go
through
we're.
Gonna
get
next
scheduled,
there's
a
next
run
here
and
then
and
let's
see
if
we
have
an
error
and
then
what
we're
actually
doing.
This
is
actually
rakul.
We're
actually
gonna
actually
write
the
next
run
in
there
we're
on
a
new
job.
If
it's
on
schedule
not
pass
a
deadline,
not
blocked
by
our
concurrency
policy,.
A
And
then
we'll
actually,
finally,
we'll
return
the
results
which
are
above
it
says
if
we
want
to
riku
when
our
next
run
would
need
to
occur,
this
taken
as
a
maximum
deadline.
So
when
we
return
scheduled
controller
results,
so
I
want
to
actually
see
what
this
type
is,
which
is
a
reconcile
result.
Oh
this.
A
You've
got
client
ignore
not
fun
as
a
pre-written
helper,
so
I
found
a
dock.
Well,
it's
not
clear
sort
of
which
of
those
things
and
we
have
to
copy
and
paste
or
not.
So
what
you're
saying
is
that
I
may
actually
get
a
get
a
bug
as
we
actually
go
here,
but
that's
okay!
Well,
we'll
figure
that
out
okay.
A
A
A
I'm
gonna
actually
break
this
out
into
a
separate,
separate
function
because
I
think
that's
what
we
want
to
do
funk
just
to
avoid
a
weird
Dhokla:
okay,
gotcha,
okay,
because
you're
not
actually
capturing
stuff
off
the
stack
instruct
pod,
okay
and
what
we're
gonna
be
taking
here
is
a
minecraft
server.
What
that
server
s
and
what
else
do
we
need
really
nothing?
And
here
so
now,
what
we're
gonna
do
is
we're.
Gonna
do
pod
equals
o
and
then
so.
This
is
okay.
This
is
one
of
the
like
crappiest
things
of
programming
with
kubernetes.
B
A
And
I
like
and
and
and
I,
have
to
find
the
right
sort
of
namespace
for
this,
so
we
actually
have
core.
V1
is
probably
what
I
want
for
the
pod,
so
you
have
to
know
where
this
stuff
is
and
we're
gonna
call
this
thing:
we're
gonna
call
this
core
and
I
believe
now.
I
can
go
through
corridor
pod
right
and
now.
A
Where
were
we
sorry,
okay,
cute
control,
API
resource
and
cute
control,
API
versions?
Okay?
What's
a
cube
control,
API
resources,
okay,
this
lists
them
all
and
then
API
version
so
yeah,
but
I
always
forget
where
the
path
is
like
translating
this
into
like
the
path
for
the
go.
Import
ends
up
being
a
little
bit
wacky
and
end
a
little
bit
tricky
all
the
time.
You
know
it
would
be
cool
if
there
was
a
if
there
was
code
to
say,
like
and
I'm
sure.
This
would
be
a
relatively
easy
tool
to
write.
A
A
A
So
yeah
yeah,
Mel
and
Jason
I
mean
kind
of
Jason
to
go
yeah.
You
could
do
Jason
to
go,
and
then
that
does
that
stuff
but
like
it
still
has
to
put
the
right
types
in
there
and
actually
give
you
all
the
imports
and
stuff.
So
that
might
be
a
way
to
do
it,
but
all
right.
So,
let's
actually
figure
out
the
object.
A
Meta
is
actually
going
to
be
very
similar,
so
we're
doing
labels
and
annotations
or
fine
name
we're
gonna
do
name
equals
format
print
you
know,
do
MC
and
you
probably
make
these
things
be
configurable
over
time.
What's
going
around
on
here
expected?
Oh,
you
know
this
is
me
like
being
stupid,
that
that's
what
I
need
right?
Yes,
okay
and
then
we
also
have
scheduled
time
eunuchs
and
will
actually
go
through
and
we're
gonna
do
a.
A
A
A
A
A
A
So
early
on,
with
GCE
from
the
get-go
I
decided
that
I
wanted
to
be
more
a
more
friendly
API
than
the
ec2
API
and
if
you've
ever
dealt
with
AWS
and
ec2
every
time
you
do
anything
you
get
this
opaque
identifier
and
it's
a
total
pain
in
the
butt
and
back
in
the
time
when
we
were
designing
the
GCE
API.
This
was
before
there
was
wide
tag
support,
so
there
wasn't
even
a
way
for
you
to
actually
give
a
friendly
name.
A
A
But
that
didn't
exist
in
ec2
at
the
time
and
I
wanted
to
be
like
friendly
names
are
the
way
to
go,
because
these
opaque
identifier
are
a
total
pain
in
the
butt
and
and
so
the
and
so
one
of
the
things
that
so
we
built
that
the
the
GC
API
we
gave
it
to
some
folks
to
start
using,
and
specifically
this
with
a
we
were
partnering
with
the
folks
at
write
scale.
At
the
time
we
gave
them
early
access
to
GCE
and
Thorsten.
Who
was
the
CTO?
A
There
was
like
what
the
hell
I
need:
unique
identifiers,
because
when
you're
starting
to
treat
VMs
as
cattle
having
a
unique,
unique
identifier
is
actually
a
really
really
useful
thing
right
because,
like
you
could
have
like
a
VM
called
foo,
and
then
you
delete
it
and
you
create
another
VM
called
foo
and
it's
like
it
could
be
the
same.
Vm
called
foo,
and
so
that's
very,
very
confusing
for
for
users,
and
so
my
advice
to
Thorson
is
like
pick
a
random
number
and
use
it
write
it,
which
is
like
totally
a
reasonable
thing
to
do.
A
A
The
idea
being
that
like,
if
you
wanted
to
say,
hey
I,
need
to
create
this
I,
don't
care
what
it's
called,
but
I
wanted
to
like
you
to
actually
pick
the
name
you
can
use
generated
name
and
the
generated
name
becomes
the
prefix
that
gets
used
in
the
eventual
name.
As
this
thing
actually
gets
created,
so
that's
a
little
bit
of
sort
of
kubernetes
history.
There.
A
A
What
we
want
to
do
is
we
want
to
go
through
and
generate
something,
but
like
one
of
the
things
that
we're
gonna
have
to
do
is
list
out
the
pods
blowing
the
server
we're
gonna
have
to
do
this
before
we
launch
it
either.
Otherwise,
every
time
we
reconcile
will
actually
be
do.
This
will
will
actually
create
another
one.
A
B
B
B
A
A
So
Joe
here
is
saying:
I
think
generating
name
only
happens
once
it's
persisted,
which
means
the
object
won't
have
a
name
when
it
goes
through
admission.
Control
I
feel
like
that's
an
issue
that
I've
hit
previously
that's
an
entry.
That's
an
interesting
thing
right.
It
probably
is
like
a
defaulting
layer
thing,
and
so,
if,
like
and
I
wonder
what
this
means
with
respect
to
audit
logs
interesting
stuff,
ok
Filipe
says
that
he's
using
control
click
on
on
code
on
pods
back
to
get
the
Deaf
bot
this
drunk.
Ok.
So
how
do
I
do?
A
A
A
A
B
A
A
A
A
B
A
B
A
B
A
A
A
B
B
A
B
A
Well
doing
string
manipulation
and
go
a
lot
of
times
ends
up
being
a
hell
of
a
lot
more
verbose
than
you
actually
wanted
to
be
okay.
So
this
is
what
we
need
to
do
here.
So
one
of
the
things
that
I'm
gonna
do
just
to
make
things
a
little
bit
easier,
is
I'm
gonna,
move
from
name
to
generated,
name
or
generate
a
name
to
name
right
just
to
like.
A
Cuz
other
okay,
I'll
tell
you
why
I'm
doing
this
is
because
what
happens?
It
will
get
an
error
when
we
actually
recreate
and
every
time
we're
in
the
reconcile
will
try
and
recreate.
The
pond
will
already
exist,
no
harm,
no
foul
and
then
and
then
Oh
Tim
peppers
asking
me:
what
do
you
think
about
c3s
the
kubernetes
ring
and
see
well,
string,
manipulation
and
see
is
more
of
a
tire
fire,
that's
during
the
manipulation
and
go,
and
so
that
way
we
like
we
like
like.
A
A
A
A
B
A
B
A
A
A
Here
you
had
us
logging
after
we
created
we're
using
the
run
the
we're
using
the
wrong
logging
here,
we're
doing
log
error
well,
where's
log,
coming
from
where
we're
pulling
in,
like
the
Prometheus
log
for
some
reason.
So
how
do
I
make
sure
I
get
the
right
log?
Oh
I,
you
have
to
do
log
here,
so
that
was
my
bad
done.
Okay,
so,
hopefully
that'll.
B
A
Decree
pods
in
a
loop,
if
you
know
protect
by
testing
the
pot
already
exist,
yes
and
that's
why
I
actually
went
through
and
I
actually
got
rid
of
generated
name.
We
can
add
this
back
when
we
test
for
existence,
so
we're
gonna,
try
and
recreate
a
pod
with
the
same
name.
It's
gonna
are
out,
and
so
that's
a
shortcut
just
so
that
we
can
get
up
to
the
point
where
we
actually
see
proof
of
life.
Here.
A
All
right,
so
we
got
a
controller
up
and
running.
Yes,
go
imports
likes
to
take
wild
guesses
for
sure,
okay.
So
now
the
next
thing
we're
gonna
do
here
is
we're
going
to
go
into
our
samples
here
and
we're
gonna
go
through
and
we
have
a
sample
server
and
so
I'm
going
to
change
this
to
my
server
and
then
the
fields
that
we
need
here
are
going
to
be
EULA.
A
A
A
B
A
A
A
B
A
Alright,
so
this
is
where
we're
actually
doing
it
so
I
think
not
supported
the
name
of
the
pot
sport
spaces.
Oh
no
though,
the
name
of
the
pot
is
that's.
Actually
who's
gonna
turn
into
an
environment,
variable
I
think
let's
actually
go
and
we'll
look
at
it
so
like
this
is
actually
going
to
be
the
server
name,
but
that's
gonna
be
a
different
name
right.
A
This
is
the
server
name
that
actually
so
when
we
actually
go
to
minecraft-
and
it
actually
says
here
so
here
it
says
minecraft
server
and
it
overwrites
that
so
that's
the
environment
variable
that
we're
actually
going
to
see
okay.
So
so
the
question
here
is
that
if
I
go
through
and
I
probably
need
to,
may
not
go
in
the
scheme
here,
so
he
have
Kline
go
scheme
this.
This
should
add
all
the
stuff
that
we
need.
Doesn't
it
so
check
if
you
populate
the
scheme
on
your
reconciler?
Okay,
maybe
I
missed
that.
A
A
B
A
A
A
I
mean
okay,
so,
oh
you
said
something
about
Maine:
okay,
so
I'll
have
to
get
there,
okay,
so
the
biggest
thing:
okay.
So
this
code
here
so
right
now,
okay,
let
me
explain
what's
going
on
here
with
the
setup
with
manager
right
now,
our
controller
will
get
kicked
every
time
we
actually
see
a
new
every
time
we
actually
see
a
new
minecraft
server
object
coming
on
board.
A
The
other
thing
we
want
to
do
is
we
want
to
actually
set
that
thing
up
so
that
it
actually
gets
kicked
every
time
we
see
a
change
in
any
of
the
pods
that
were
managing
so
I'm
gonna
actually
go
through
we're
gonna
skip
this
for
now,
because
we
want
to
go
through
if
a
pod
dies
or
gets
deleted.
We
want
to
recreate
it
right
and
so
I'm
gonna
put
it
to
do
here
to
do.
A
A
A
A
Oh
yeah,
no,
so
so,
okay,
so
what
I
did
before
is
I
launched
it
I
launched
the
program
was
running
in
the
background.
I
did
the
apply,
so
the
the
the
actual
CR
D
is
instance,
is
now
actually
sitting
in
kubernetes,
but
the
controller
crashed
and
so
now
I
did
have
the
controller
running
in
the
background
before,
but
now
I'm
actually
running
in
the
foreground
as
actually
debugging,
and
so
now
we're
going
through
a
scheme,
but
actually
who
is
actually
setting
the
scheme
up.
A
A
A
B
A
B
A
A
Well,
my
keyboard
is
kind
of
messed
up
here.
Okay,
so
now
we
can
do
at
the
apply
of
that
all
right.
Now,
it's
mad
what
it's
a
man?
Okay,
so
we
crashed.
Oh
we're,
actually,
okay,
we're
not
hard
crash
and
we're
getting
an
error.
Now,
reconciliation
error.
It
already
exists.
Okay,
so
now
you
have
cute
control.
So
this
is
the
type
of
thing
where
it's
like:
hey,
look
the
pod
turning
but
I
changed
parameters.
A
A
And
then
you
have
to
go
through.
You
have
to
think
about
all
the
different
sort
of
error
cases
that
you're
that
you're
working
on
all
right,
okay
created
pod,
so
we
have
the
pod
right
now
get
pods
all
right.
Now
we
have
an
crash
loop
back
off,
okay,
cute
control
logs
and
see
my
server
Oh
invalid
type
law.
So
we
still
have
the
the
pod.
A
The
third
way
we
could
do
it
is
that
we
could
just
opted
not
to
set
the
environment
variable
if
the
thing
is
is
blank.
Actually,
let's
do
that?
Let's
do
that
so
now.
What
we're
gonna
do
here
is
we'll.
Do
it
the
easy
way,
but
the
next
thing
we're
gonna
I,
will
put
a
note
in
there
here,
it's
like.
If
these
values
are
blank,
we
should
just
not
set
B
and
variable.
A
A
A
Craigs
asked
me
if
I
want
another
beer
and
the
truth
is
that
I
I
do
but
I
have
other
stuff
I.
Do
okay,
I
actually
have
the
thing
running
in
the
background,
so
I'll
just
foreground
it,
but
we
got
to
finish
up
I'm
hurrying.
Okay.
So
now
we
actually,
we
created
the
pod.
So
now,
if
I
do
cube,
controlled,
get
pods,
oh
we're
still
crashing.
Oh
it
didn't
like
it
get.
A
A
All
right,
so
that's
a
foregrounded
again
it
created
again
I'm
gonna
get
the
thing.
Did
we
actually
do
it
the
right
way?
This
time,
oh
look
type
equals
vanilla,
sweet
and
it's
running
yes.
Okay,
now
we
got
the
thing
running:
okay,
sweet
all
right.
So
what
I'm
gonna
do
here
just
to
actually
take
some
notes
here
is
I'm
going
to
create
a
new
file
here.
We're
gonna
call
this
thing.
A
Create
and
manage
a
service
along
with
each
pod.
Oh,
the
question
of
why
I'm
using
pod
versus
a
replica
set
is
the
replica
set
will
recreate
stuff,
but
we
don't
want
to
have
two
of
the
same
server
running
at
the
same
time
and
in
like,
if
we
have
split
brains,
I
mean
maybe
we
could
use
a
replica
set
just
for
it's
sort
of
like
worth,
or
a
replica
of
one
I
mean
that
might
be
a
reasonable
way
to
work
on
this
up.
To
think
about.
I
think
a
replica
said
may
be
a
good
solution.
Here.
A
Definitely
don't
need
to
use
a
deployment
because
we
don't
need
the
upgrade
stuff
and
the
replica
set
is
smart
enough.
That,
like
one
of
the
things
we
might
want
to
do
is
actually
allocate
a
persistent
volume
here
under
the
covers
behind
and
make
it
stateful,
we
can
make
it
a
stateful
set.
Also
one
of
the
things
I
worry
about,
that
is
that
the
state
will
set,
you
know,
ends
up
being
a
little
bit
slower
and
like
stateful
sets
really
come
to
their
own.
A
A
A
A
A
You
know
the
amount
of
time
that
it
takes
to
go
from
like
zero
to
something
where
you
can
actually
get
into
your
business.
Logic
is
really
really
really
hard.
You
know
right
now.
Client
go
is
not
an
easy
thing
to
get
set
up.
It
turns
out
that
there's
a
bunch
of
sort
of
like
fussing
around
with
with
you
know,
mod
and
in
module
versions
and
stuff
like
that.
A
So
the
fact
that
this
brings
that
in
it
has
you
all
the
scaffolding
to
be
able
to
use
the
the
the
controller
manager
it
automatically
does
things
like
set
up
logging,
like
that's
all
super
super
awesome,
and
it
even
takes
it
to
the
point
of
like
what
is
the
scaffolding
for
actually
being
able
to
deploy
and
manage
to
this
stuff?
I
love
the
annotations
around
our
back.
That's
super
super
cool.
A
Also
interesting
things
that
I
think
would
be
cool
would
be
like
the
the
yamo
versus
golang
sort
of
duality
is
a
total
pain
in
the
ass.
This
is
not
a
queue
builder
type
of
thing,
but
any
time
when
you're
doing
this
stuff-
and
you
know-
and
you
see
this
happen
a
lot
of
times
when
you
oh
yeah-
this
is
another
thing
to
add
to
the
readme.
A
That's
just
a
real
pain
right,
so
I
think
a
better
way
to
be
able
to
create
this,
and
this
is
because,
like
the
way
that
go
is
so
strongly
typed
the
way
that
it
actually
deals
with
being
super
explicit
about
typing,
like
the
fact
that
I
have
to
have
like
core
dot
container
here,
twice
total
pain
in
the
butt,
so
a
better
way
of
actually
constructing
these
objects
would
actually
be
pretty
sweet.
That's
a
separable
problem
from
queue
builder,
but
I.
A
Think
it's
just
like
doing
these
types
of
things
in
go
is
just
a
total
pain
yeah,
but
this
is
super
fun.
I'm,
going
to
check
this
in
on
my
on
my
my
personal
github,
and
you
all
can
take
a
look
at
it
and
see
all
stuff
that
I
did
I'm
really
excited
about
this
I
I
want
to
continue
doing
more
of
it,
I
think
it's
really
fun,
and
then
you
know
we
can
open
it
up.
You
can
all
create
Minecraft
servers,
how's
that
sound,
but.
A
Questions
before
I
sign
off
I'm
way
over
here,
I,
don't
know
if
there's
any
of
y'all
left,
but
thanks
for
sticking
with
me,
if
you're
still
here,
Oh
Timothy
is
asking
any
opinion
on
the
Q
language
for
configuration,
I
haven't
looked
at
it
deeply
I've
taken
a
glance
at
it
and
I'm
I
think
it
looks
interesting,
but
I
have
some
concerns
right
in
some
ways.
It
reminds
me
of
this
language
that
never
made
it
out
of
Google
called
bored
menthe
I
in
terms
of
the
way
the
parameterization
works.
A
There
is
this
idea
that
you'd
like
to
find
templates
and
you
can
parametrize
those
templates
I
think
that's
really
cool,
but
I
worried
that
in
reality
those
parameterizations
would
get
really
really
really
long.
And
so
my
question
is:
is
that
is
it
the
type
of
thing
that
works
well
in
the
small,
but
doesn't
work
well,
as
you
actually
get
bigger
and
more
complicated,
so
I
do
want
to
do
an
episode
on
Q
at
some
point
really
dig
into
it
I.
A
A
Did
this
stuff
and
I
think
it's
really
fascinating
to
watch
this
and
then,
if
you
go
and
watch
the
original
controller,
that
I
wrote
and
do
some
compare
and
contrast
the
the
client
go
and
especially
with
things
like
Q
builder,
it's
really
come
a
long
way
and
I
think
that's
super
super
cool.
So
thanks
everybody
for
sticking
with
me,
I'm
gonna
actually
be
on
vacation
for
several
weeks
and
like
I'm,
actually
coming
back,
but
I
can't
be
here
for
Friday.
A
A
Let's
see
we're
trying
to
strong-arm
olive
into
doing
it,
she's,
probably
asleep.
Now,
because
she's
in
London
and
and
then
a
Christian
is
is,
is
also
signed
on
to
do
some
more
so
we're
definitely
getting
a
lot
of
folks
joining
in
there
fun
with
it
and
I
think
everybody
brings
their
own
flavor
their
own
way
of
doing
stuff
to
it,
and
that's
really
really
exciting.
A
What
that
means,
how
to
deal
with
it,
sort
of
some
strategies
for
that,
and
so
that's
going
to
be
very
topical,
not
topical
yet
because
1.16
isn't
out,
but
it
will
be
topical
at
some
point
very
soon
in
the
future.
So
with
that
I'm
gonna
sign
up.
Thank
you.
Everybody
for
joining
me
and
I
will
see
you
gosh
I
have
to
look
at
the
calorie
things
like
August,
or
something
like
that.