►
From YouTube: Agones Community Meeting October 2020
Description
No description was provided for this meeting.
If this is YOUR meeting, an easy way to fix this is to add a description to your video, wherever mtngs.io found it (probably YouTube).
A
I'll
put
a
link
to
the
working
doc
in
the
chat.
Just
so
everybody
can
see
it,
but
we've
got
some
stuff
from
dom
and
mark
on
the
list
and
dom
you're
first
up.
If
you
want
to
dive
into
it,.
B
Yep
so
hi
guys.
What
was
I
wanted
to
speak
about
this
month?
Oh
yeah,
so
my
seven
mark
had
gone
backwards
and
forwards
on
a
ticket
about
exposing
metrics
and
both
player
player.
Tracking
metrics
are
basically
where
we
started
and
then
arbitrary
metrics.
One
of
the
discussions
was
basically.
B
Where
does
the
line
end
between
the
gunners
and
what
kind
of
what
point
do
we
stop
with
this
there's
a
couple
of
other
projects
that
might
be
of
interest,
however,
the
I
think
the
consensus
that
me
and
mark
got
to
was
we'll
give
it
a
try,
we'll
build
out
and
we'll
flesh
out
the
api
and
what
it
might
do
to
change
the
sdk
and
we'll
have
a
look
at
that
put
it
forwards
as
a
proposal
and
see
if
people
like
that
or
not
and
see
how
much
of
a
burden
it
might
be
to
maintain.
B
So
that's
kind
of
kind
of
where
we
got
to
with
that
one
marketing.
If
you
want
to
jump
in
and
add
anything
to
that,
that's.
B
One
of
the
ones
I
have
been
looking
at
is
the
possibility
of
maybe,
if
it's
just
coming
from
the
crd,
using
cube
state
metrics
instead
and
try
to
write
a
build
of
cube
state
metrics,
which
could
do
a
lot.
But
if,
if
you
want
other
other
metrics
from
that
sidecar,
then
I
think
we're
best
just
implementing
the
open
census.
Metrics
endpoint
in
that.
B
Metrics
it's
a
project
which
will
look
at
metrics
based
on
any
deployed
thing
in
kubernetes,
so
it
has
things
like
number
of
deployments.
We
were
created
destroyed
all
that
sort
of
stuff.
So
I
kind
of
thought
it
could
be
a
good
one
to
add
in
an
extra
thing
here
for
game
servers
and
then
get
the,
but
I
think
it's
effectively
what
we're
doing
with
using
events
inside
the
controller.
B
So
it's
very
similar
to
that.
So
I
think,
like
we
said,
just
keep
keep
with
that
pattern
in
a
minute
or
expose
a
metrics
endpoint
on
the
sidecar.
I
think,
depending
on
what
the
the
api
looks
like,
we
might
get
more
information
at
that
point.
B
Yeah,
which
came
comes
on
to
the
next
kind
of
thing.
I
wanted
to
talk
about
the
minute
when
we
do
player
added
and
we
get
quite
a
a
small
event
which
is
just
set
to
one,
and
so
you
played
count
set
to
one
and
what
I
might
prefer
is
maybe
player
connected
or
player
player
left.
B
I
can't
remember
what
the
api
actually
calls
it
when
players
joining
players
leave,
but
giving
the
id
of
the
player
as
well,
seeing
as
though
we
have
it-
and
this
would
definitely
be
much
better
in
my
telemetry
systems,
because
I
try
and
fire
off
anything
around
game
service
to
a
product
called
honeycomb,
and
it
basically
allows
me
to
do
more
insightful
stuff.
With
my
events.
C
Yeah,
I
I
was
hesitant
to
put
that
in
there
just
because
we
were
already
shifting
around
so
many
bits
by
adding
player
tracking,
and
it
felt
that
count
was
like
a
good
happy
medium
to
just
be
like
okay,
what's
happening
with
my
game
server,
I
can
see.
If
you
I
mean,
if
you
need
that
data,
you
know
you
can
get
at
it
through
the
event
system
right,
because
you
can
see
the
diff.
C
Yeah
I
was
sort
of
like
if
you
know
a
thousand
players
join
at
once
within
a
single
slot
like
that
means
we're
adding
it
to
the
like
we're
dropping
that
data
into
the
api
twice
and
that
just
kind
of
felt
that
that's
good,
like
I
just
was
like.
I
don't
know
that
was
that
was
primarily
my
concern.
There.
C
B
B
D
A
Yeah
eric
it's
kubecon,
north
america,
which
is
november
17th
or
something.
A
Okay,
yeah,
I
was
gonna,
I
was
gonna
put
it
in
the
notes
because
that's
very
cool
and
I
haven't
registered
for
kubecon
yet,
but
I
guess
I
should
do.
A
Even
if
you
don't
attend
kubecon,
the
talks
are
obviously
recorded
and
they
always
make
it
available
afterwards,
so
you
can
catch
it
afterwards,
but
that's
very
cool
we're
very
excited
here.
Yeah
I
got
it
mark's
not
had
a
talk
except
that
could
at
coop
kanabatagones
for
a
while.
So.
C
Awesome
september
was
a
lot,
so
I
didn't
do
the
things
from
last
meeting.
So
this
is
mostly
me
reminding
myself
that
I
need
to
do
the
things
that
I
said
I
would
do
last
meeting,
which
is
create
tickets
for
sends
sdk,
send
game
server
on
watch
execution,
so
that
was
the.
When
you
connect
to
watch.
You
immediately
get
an
instance
of
the
game
server
we
had
to
go
through,
I
think
a
while
ago.
C
Actually
that
probably
should
go
to
beta
at
the
very
least,
but
I
need
to
write
that
ticket.
If
somebody
decides
they
want
to
write
these
tickets
before
I
get
to
them
like,
please,
don't
let
me
stop
you.
By
the
same
token,
there
should
be
some
discussion
around
player,
tracking
and
bringing
that
to
beta
as
well.
It
seems
like
there's
some
people
using
it,
which
is
good
and
so
far,
hopefully
nothing's
broken
so
sweet.
C
C
C
C
Dominic
seems
to
think
that
there's
going
to
be
possibly
an
issue
with
getting
unreal
to
do
the
watch
api,
but
I'm
like
well,
we
could
do
it
in
unity.
So,
like
I
feel
like
you
know,
if
you're
just
saying
that
unreal
is
just
not
as
good,
then
that's
you
know.
That's
on
you.
B
B
I
think
that
was
kind
of
where
I
got
to
in
my
head.
When
I
last
looked
at
it
yeah.
B
C
Yep,
what
was
I
going
to
say?
Yeah,
we,
I
think
in
all,
I
think,
in
unity
we
had
to
dig
into
some
of
the
middleware
stuff.
Some
of
that
was
a
bit
more
exposed
on
on
the
http
request.
Side
of
things
like
you
could
intercept
stuff
as
it
came
through
in
chunks
and
it
worked
out
reasonably
well.
I
have
no
idea
what
works
on
the
unreal
side,
but
cool.
C
We
can
talk
about
that,
but
yeah
most
of
that
that
watch
execution
stuff
is
about
more
about
the
functionality
of
watch.
Any
questions
on
that
no
I'll
move
on
from
there
cool
some
of
the
people,
who
probably
aren't
here.
Actually
some
customers
of
ours
discovered
that
there's
some
performance
issues
with
the
grpc
allocation
endpoint
that
would
have
been
highlighted.
I
I
will
I'm
getting
probably
unless
somebody
beats
me
to
it.
I
will
probably
jump
in
on
this.
Primarily.
C
The
big
issue
here
is:
if
you're
not
familiar
the
kubernetes
like
client
throttles
itself
to
about
four
requests,
a
second
by
default,
which
is
not
very
fast.
So
unless
you
specifically
override
it,
things
go
slow
if
you're
doing
stuff
through
the
cube,
the
kubernetes
like
go
client
or
any
like
crd
client,
and
so
when
we
built
the
allocator.
Unfortunately,
we
missed
that
one
because
we
have
it
fixed
in
the
controller,
but
we
missed
it
in
the
allocator
and
they're
like.
Why
is
we
only
getting
four
requested
seconds
per
allocation
so.
C
To
go
in
add
some
home
things
and
then
add
some
boost
performance.
We
we
did
some
internal
testing
and
a
little
kind
of
hacking
stuff
in
it.
Yeah
it
goes
up
to
like
100
plus
requests
a
second
once
as
soon
as
we
make
the
change,
which
is
pretty
nice.
C
We
also
there's
another
error
here
because
of
the
way
vacation
we
bin
pack
in
the
allocation
system,
because
we
have
multiple
pods,
they
can
try
and
bin
pack
exactly
the
same,
and
so
they
may
try
and
allocate
the
same
game
servers
at
the
same
time
a
lot
and
that
can
result
in
a
lot
of
contention
and
eventually
it'll
fall
over
at
particularly
high
rates.
I
think
we
could
probably
do
some
slight
randomization
with
some
like
configuration
options
to
maybe
get
some
more
throughput
on
that
as
well.
C
So
we
might
take
a
step,
but
I'm
going
to
try
and
get
maybe
those
both
done
before
the
next
release
in
a
couple
of
weeks,
but
we'll
see
how
my
timing
goes.
Did
you
have
anything
else
you
want
to
add
to
that?
I
know
you
were
looking
into
some
of
this
stuff
too.
C
Cool
just
wanted
to
make
sure
didn't
miss
anything.
So
yes,
more
performance
better
on
the
g,
that's
specifically
the
grpc
allocation
point.
E
And
perhaps
we
need
to
have
a
discussion
and
adding
the
performance
testing
before
google.
C
B
Mark
was
that
just
against
the
allocator,
or
is
that
against
any
kubernetes?
Yes,
apr.
C
I
would
say:
allocation
fleet
scale
up
scale
down.
Is
there
any
other
factories?
I
can
think
of
sorry.
B
I
mean:
was
the
book
just
against
the
geography
allocation
endpoint
or
is
it.
C
It's
just
the
grpc
allocation
endpoint,
the
the
one,
the
custom
resource
definition.
One
is
fine,
because
that
goes
through
the
controller.
B
And
you
don't
know
if
there's
any
other
throttles,
for
instance,
so
through
my
example,
I'm
using
select
against
labels
heavily
in
kubernetes
to
make
sure
I
get
the
right
game
servers
because
I
basically
just
reuse
allocated
game
servers
constantly
got
it.
I've
not
seen
any
performance
hit
so
far
so
generally
across
the
board.
C
No,
it
should
be
fine,
so
way
way
way
back
in
the
day
when
we
first
started
doing
stuff
with
gone
as
some
of
our
senior
engineers
who
work
on
other
stuff
now,
actually
we're
like.
How
fast
can
we
make
this
minimum
has
been
down?
We
realized
it
was
super
slow
and
the
biggest
thing
we
found
was
in
the
controller
for
agonized,
like
the
qps
was
still
set
to
four,
which
was
the
default,
and
so
we
bumped
that
up
by
a
bunch
and
then
did
some
other
work
as
well.
C
So
anything
that's
running
directly
through
the
controller
which
involves
all
the
crds
and
the
extension
apis,
which
is
also
like
a
game
server
allocation
or
anything
like
that.
That
has
a
nice
boost.
I
think
we
run
about
a
few
hundred
qps
per
second.
We
have
that
set
to,
but
because
the
allocator
is
actually
completely
separate,
binary
just
missed
it,
and
so
we
just
need
to
pass
that.
We
just
need
to
create
some
hallmarks
for
that
that
we
can
set
that
independently
and
then
pass
that
information
through
and
make
that
change.
F
C
Actually,
since
I've
seen
you
drop
in
here
ilker,
can
I
bug
you
publicly
to
get
a
prn
to
get
your
your
performance
testing
tool
thing
that
you
built
for
application.
C
C
F
Yeah,
this
is
what
I
did.
I
had
to
write
a
new
client
pretty
much
because
I
was
using
the
I
mean
when
I
was
testing
more
than
a
year
ago,
using
a
regular.
I
mean
gsa
allocation
to
the
client.
I
mean
now
it's
different,
so
not
that
different,
but
it's
a
little
bit
sweet.
C
Awesome,
I
think
the
only
other
item
I
have.
I
think
I've
asked
we've
gone
through
a
few
iterations
of
like
what
was
the
allocation
based
on
player
capacity.
It
now
seems
to
have
formed
into
a
I
think,
a
quite
nice
reallocation
policy
that
also
works
for
player
capacity,
but
in
theory,
could
also
be
used
for
multiple
sessions
per
game,
server
or
multi-tenancy
within
a
single
game
server
or
like
high-density
game
servers
or
probably
a
bunch
of
other
stuff.
So
if
you
haven't
checked
it
out,
please
do
and
provide
any
feedback.
C
I'd
like
to
get
started
on
that
soon.
I
think
I'm
pretty
excited
about
that.
I
think
that's
going
to
have
some
really
nice
nice
applicability
and
the
ability
to
be
able
to
do
things
like
give
me
a
game
server
that
has,
you
know,
give
me
a
game
server
that
has
room
for
two
people,
whether
it's
allocated
or
not,
or
I
prefer
to
have
that
and
then,
if
I
can't
find
it
give
me
one,
that's
just
ready
and
allocate
as
per
normal,
but
also
because
you
can
define
your
own
custom
labels.
C
You
could
be
like
this
game
server
tracks.
How
many
rooms
it
has
internally
and
we're
just
gonna,
keep
a
label
that's
free
until
like
we
shut
that
down
and
close
that
off
and
there's
a
couple
of
different
patterns,
you
could
do
to
do
that
too.
So
you
could
use
it
for
all
kinds
of
different
allocation
strategies
which
I
think
are
pretty
exciting.
I
quite
like
it.
So
if
you
haven't
checked
it,
please
do
provide.
Any
feedback
seems
like
most
people
have
looked
at
it
already,
which
is
nice.
G
C
I
would
say
released
independently,
keep
them
keep
them
separate.
I
think
that's
a
good
question
so.
G
C
Yeah,
I
don't
yeah.
I
think
I
think
my
the
concerns
I
personally
had
around
graduating
player
tracking
was
more
about
usage
and
creating
this
functionality
was
a
way
to
get
people
well,
hey.
I
wanted
to
get
a
functionality
in,
but
was
to
get
people
to
use
it
more.
But
if
people
are
using
it,
then
I'd
have
less
concerns.
C
Although
I
know
that
player
tracking
has
some
api
changes
stuff
that
we
want
to
get
in
there.
But
yes,
no
and
also
my
thought
on
this,
and
maybe
I
should
actually
document
it
was
that
it
would
sit
under
a
different
feature
flag,
and
so
you
could
independently
turn
each
piece
on
and
off
as
well.
G
Yeah
exactly
right,
because
the
the
other
flag
should
start
to
default
to
true
and
this
being
new
functionality
would
be
an
alpha
and
the
flag
would
default
to
false
right.
So
exactly
what
you'd
get
is,
if
you
wanted
both
you'd
have
to
flag
this
one
on
yourself,
but
the
other
one
should
already
be
turned
on.
Unless
you
really
don't
want
it
and
you
turn
it
off.
G
I
don't
know
if
there's
a
way
for
us
to
track,
but
it
would
be
really
useful
to
track
to
know
if
people
are
feature
flagging
things
off
for
any
region
reason
once
we
start
defaulting
them,
because
the
idea
is
when
they
graduate
to
ga.
We
eventually
remove
the
ability
to
turn
them
off.
So
I
think
that
would
be
a
really
good
signal
that
it's
not
ready
if
people
are
finding
that
they
have
to
turn
it
off
for
something.
B
I
think
the
comment
is
fully
up
today.
I
haven't
fully
read
it
through
yet,
but
they're
taking
the
comment.
That's
with
the
full
design
is
today.
B
I
fully
support
the
new
game
server
state
that
is
really
good,
I'm
just
doing
it
by
a
label
at
a
minute,
so
yeah
nice.
H
Yeah,
so
my
question
really
is
I
I
don't
know
what
else
I
can
do
on
that
one,
because
the
question
in
the
bug
is
basically
like.
Can
you
even
turn
this
on
elsewhere,
and
so
you
know
that
one
seems
kind
of
stuck.
H
This
is
the
owner
references
thing
where
in
newer
versions
of
kubernetes,
if
you
don't
have
the
owner
reference
defined
sorry,
if
there's
a
feature
enabled
in
cube,
if
you
don't
have
the
owner
references
defined,
it
complains.
This
was
the
issue
we
ran
into
on
the
live
stream,
and
so
the
bug
is
kind
of
an
rfe
like
hey.
Can
you
please
put
the
owner
references
in.
C
H
C
G
G
H
Installing
openshift
can
be
a
high
bar
and
this
may
be
somewhere
where
maybe
we
can
help
you
all
where
we
we
make
it
easy
or
available
to
you
to
just
deploy
one
somewhere,
because
we've
got
internal
systems
that
I
might
be
able
to
get
some
clearance
for
somebody
from
your
project.
C
D
G
You
can
also
have
automation
assigned
bugs
if
you
type
slash,
assign
and
then
at
your
your
github
handle.
It's
also
possible
that
issues
I
don't
think
they
can
be
assigned
to
people
that
aren't
part
of
the
org,
I
believe,
is
what
it
is.
So
it's
possible
that
it
can't
be
assigned
to
you,
no
matter
what
you
should
just
put.
C
H
C
I
think
robert,
I
think
it's
you,
ruby
cool.
G
Yeah
I
just
wanted
to
give
a
heads
up.
I
filed
an
issue
on
this,
but
I'm
not
sure
anybody
saw
it.
It
occurred
to
me
last
week
or
so
that
helm
v2
is
reaching
end
of
life
like
it's
been
deprecated
for
a
while
there's
been
lots
of
migration
guides
to
mp3,
we
moved
to
home
v3.
I
think
it
was
around
1.617.
G
So
that's
you
know
multiple
months
ago,
but
the
official
end
of
life
for
home
b2
is
happening
in
about
a
month
on
november
13th
of
this
year,
at
which
point
the
it's
not
clear,
exactly
what's
going
to
happen
with
home
b2,
but
it's
certainly
no
longer
going
to
be
supported
in
any
way.
They're
going
to
close
all
of
any
outstanding
issues,
they're
not
going
to
take
any
future
requests.
Stuff
may
or
may
not
continue
working.
G
I
know
there
was
some
discussion
about
whether,
like
the
images
and
that
kind
of
stuff
would
still
be
hosted.
I
think
that
microsoft
stepped
up
to
to
pay
for
hosting
of
stuff,
so
so
stuff
would
not
just
completely
break
on
that
day,
but
we
have
moved
to
helm.
V3,
there's
been
a
note
in
there
for
a
while
that
you
should
start
using
helm
v3,
so
I
sent
a
pr
yesterday
to
delete
our
home
v2
documentation
and
configuration
files
for
terraform,
so
I
just
want
to
give
people
a
heads
up.
G
We
think
we've
mentioned
recently
that
oh
google
host
says.
Okay,
we
mentioned
a
while
back
that
you
should
be
moving
to
helm
b3.
Hopefully,
people
have
had
a
chance
to
do
that
if
they're
using
helm,
if
not
now
would
be
a
great
time
to
do
that,
because
the
documentation
and
code
changes
should
be
live
with
1.10,
which,
coincidentally
releases
about
three
days
before
home
b2
is
into
play.
G
Alternatively,
if
anybody
is
really
still
using
home
b2
and
strongly
objects,
please
comment
on
the
pull
request.
We
can.
We
can
delay
deleting
it
because
it
theoretically
should
keep
working,
but
I
definitely
want
to
remove
the
documentation,
because
we
don't
want
to
encourage
people
to
keep
using
it.
C
Nice
and
I'm
also
just
looking
like
it,
looks
like
what
you've
done
here
is
you've
said:
okay,
we'll
remove
the
documentation
with
the
next
release,
you're,
not
just
like,
deleting
it
straight
away.
I
think
that's!
That's
nice.
G
C
H
I
made
a
comment
in
the
chat.
The
call
for
presentations
for
red
hat
summit
is
open,
so
if
anybody
wants
to
do
anything
on
agonize
or
just
games
in
general
on
on
kubernetes
and
openshift,
I'd
be
happy
to
help
be
a
co-presenter
and
and
ferret
through
any
of
those
things.
So
we
just
live
stream
some
stuff
again
eric
we
can.
We
can
do
that
too,
but
some,
if,
if
anybody's
itching
for
a
presentation
at
a
conference,
that
one
is
open
right
now
for
the.
C
Oh
there's,
some
there's
some
interesting
stuff.
Actually,
there
was
something
interesting
that
came
up
in
slack
a
while
ago.
I
seem
to
having
just
as
a
just
as
type
of
brain.
I've
had
a
few
conversations
recently
about
two
particular
things.
One
is
like
people
asking
about
how
people
are
mounting
assets
externally
and
populating
those
assets
which
we
have
a
ticket
on
and
we
sort
of
bounce
around
a
bit.
C
So
people
have
patents
that
they've
like
to
use
for
like
volume,
mounts
or
putting
assets
somewhere
on
the
internet
and
then
providing
them
to
particular
images
in
some
way,
shape
or
form
I'd
love
to
have
some
of
those
best
practices
documented
somewhere.
I
think
the
conversation
keeps
coming
up
a
lot
and
I
think
a
lot
of
people
are
reinventing
the
wheel,
the
other
one
that's
come
a
few
times
in
a
few
places
is
people
doing
like
crash
dog
crash
crash
dumps
and
finding
patterns
to
do
that.
C
I
think
we
had
a
conversation
in
slack
about
that,
and
there
were
a
few
different
things
that
came
out
of
there.
I
had
the
same
conversation
again
the
other
day
with
another
user
and
having
some
some
defined
patterns,
for
that,
I
think,
would
also
be
kind
of
cool
if
people
have
found
ways
of
doing
that.
I
think
be
really
good,
just
documenting
some
of
that.
C
C
That
would
be
useful
write,
some
notes.
B
I
I
definitely
have
thoughts
about
the
uploading.
I
think
I
was
on
the
slack
thread
when
we
talked
about
it
yeah
it's
definitely
doable.
I
demonstrated
the
way
I
would
go
about
it,
but
I've
not
had
to
do
it
myself
yet
so
I've
actually
built
it.
So
I've
been
leaning
on
what
other
people
are,
what
people
other
people
are
doing.
C
Yeah
the
the
things
that
were
interesting
at
the
follow-up
conversation
I
had
from
that.
Was
it
working
out
how
to
ensure
that
the
crash
dump
was
identifiable
as
to
which
game
server
it
came
from
and
so
like
making
sure
it
went
into
like
a
directory
with
the
game
server
name
in
it
or.
C
The
thought
I
actually
had
as
well
like
to
your
point
like
it
seems
like
you,
want
to
take
your
crash
dump.
It's
where
the
question
up
is
stored,
seems
to
be
signed
inside
like
a
config
file
inside
the
linux.
There's
a
linux
config,
but
it
looks
like
you
can
pass
it
to
a
secondary
program
and
since
you
could
pass
environment
variables
through,
the
secondary
program
could
be
like.
Oh
I'll,
take
this
file
and
just
rename
it
and
then
stick
it
in
the
appropriate
spot
that
it
should
go
or
something
like
that
interesting.
C
B
C
B
So
I
would,
I
would
have
gone
down
the
route
of
mount
mountain
volume
with
a
certain
name
and
drop
it
in
there
yeah.
But
how
do
you
know
what
that
name
is
yeah
yeah
yeah?
It's
a
good
point.
C
Yeah
that
was
the
this
little
bit
of
chicken
and
egg,
but
you
do
you
can
get
the
environment
variable
of
like
you
can
use
the
downward
api
and
k8
to
pass
the
name
and
the
name
space
through
so
and
the
pod
name
matches
the
game
server
name
for
a
lot
of
reasons.
So,
if
you
can
it
looked
like
in
that
config
you're,
like
crash
dump,
the
config
allows
you
to
pass
the
file
to
a
separate
program
either.
C
Maybe
as
a
thing,
so
you
could
like
do
stuff
like
uploading
or
whatever
you
want,
but
what
you're
actually
just
going
to
do
is
just
rename
it
or
stick
another
file
next
to
it.
Maybe
that
has
the
meta
information
or
something
in
that
sort
of
vein
when,
before
you
check
it
in
directory,
that
was
my
thought.
That
was
my
theory.
C
C
With
the
actually
to
your
actually
to
your
point,
dominic
when
you're
talking
about
an
unreal
crash
dump,
is
that
coming
from
unreal
itself,
or
is
that,
like
a
linux
crash,
dump.
B
I
think
I
was
unreal.
Okay.
B
With
them
in
a
minute,
maybe
literally
crushing
the
game
server
at
that.
B
B
I
mean
unreal
is
just
all
all
kinds
of
fun
if
you,
if
you
want
to
get
into
unreal
and
human
readable
logs,
that's
really
helpful
yep
wow.
This
is
helpful,
no
concept
of
tracing
metrics
or
anything
like
that.
So
that's
even.
C
D
E
Looks
like
on
all
the
configurations
that
we
introduced
to
the
home
are
not
introduced,
the
terraform
that
we
added
for
agonize.
How
are
you
maintaining
that.
I
Yep
previously,
I've
avoided
them
only
what
I
was
changing
a
lot,
but
we
can
make
a
one-to-one
correlation
between
what
we
have
in
the
here
and
there.
I
Okay,
I
mean
the
problem
here
that
we
are
introducing
all
these
for
any,
for
example,
g,
key
eks,
etc
so,
but
we
could
require
for
any
new
helm
configurations
to
pass
it
to
terraform.
I
think.
E
C
Oh,
but
you
can,
you
can
pass
the
value
you
can
pass
the
value
files
through
that
is.
That
is
an
option,
so
you
you
do
have
access
to
all
the
values
going
through
to
the
helm.
Configuration
that
way.
I
think
we've
just
pulled
some
stuff
out
where
people
are
like.
I
think
whenever
someone's
like.
Oh,
I
wish
I
had
this
thing
in
the
terraform.
They
go
and
put
an
arg
in
there.
I
Alex
yo,
what
is
the
camel
file
was
here,
it
has,
is
some
new
variable,
it's
just
to
use
local
one
or
local
defaults
or
use
what
is
predefined.
C
Oops,
no
I'm
in
the
wrong
spot,
terraform
helm3!
There
we
go
variables.
C
G
C
Was
literally
just
looking
at
that
all
right,
so
values
is
that
and.
C
G
C
G
C
I
By
the
way,
there
is
a
terraform
test,
so
you
can
run,
make
reform
tests
and
or
to
tweak
it
a
bit
to
test
what
you
expect
to
have
to
see.
Yeah.
That's
nice
actually.
I
Yeah,
it's
reform
apply
then
in
the
middle.
What
can
have
cube
client
in
it?
So
you
can
make
some
queries
and
after
that
do
perform
destroy
yep.
I
Because
say
it's
only
one
default
test
so
and
this
therefore
there's
this
teragrant
or
test
allows
you
to
do
lots
more
like
this
in
the
helm,
configuration
files-
and
there
is
a
ticket
for
that-
I
can
push
it
to
the.
J
C
A
C
D
A
Well,
actually,
I
guess,
for
november
our
next
meeting
would
be
on
thanksgiving
in
the
u.s.
So
that's
probably
not
going
to
happen
so
we'll
probably
move
it
up
a
week
to
be
on
the
18th.
C
No,
what
is
it
also
just
a
reminder?
What
is
it
next
week,
the
week
after
israelis,
candidate.
A
A
G
A
G
G
C
A
C
C
J
C
H
I
binged
the
forest
a
few
weeks
ago
and
then
started
on
night
of
the
dead
very,
very
similar,
but
very
different
experiences.
H
H
Haven't
I
I
never
played
two
so
okay,
I
played
one
like
I'm
a
weird
gamer
and
that
I
I
end
up
playing
stuff
like
way
after
the
curve
yeah,
I
understand
so
you
know
by
the
time
I
played
division.
Two
like
nobody
was
playing
it
anymore.
So
I
was
all
by
myself
all
the
time.
Oh
I've
been
actually
waiting
to
play
start
playing.
C
C
H
I'm
sorry
we'll
have
to
we'll
have
to
be
more
in
sync,
with
our
with
our
old
game
playing.
C
Yep
I've
been
playing
broke
force,
which
is
ridiculously
dumb
bro
force.
Yes,
it
is
yeah.
I
think
I
saw
that
one.
It
looks.
C
C
Yep
yep,
it's
dumb,
it's
really
dumb.
It's
a
side.
Scrolling
shoot
em
up
with
like
just
bro
everything,
it's
it's
dumb
and
very
silly,
and
when
did
it
come
out
like
2013?
Oh,
my
god,
it's
really
old,
but
it's
all
like
it's
pixelated.
It
doesn't
matter.
A
I
finished
a
spirit,
fair
and
now
I'm
back
in
the
stardew
valley,
speaking
of
pixelated,
graphics,
so
spirit
repair
is
good,
but
it
it
did
make
me
cry
at
least
a
couple
of
times,
so
I
can
see
that
it
has
feels
it
has
feels,
and
it
also
like,
is
really
engaging
up
front
and
then,
like
you,
reach
a
point
where,
like
all
right,
this
is
the
same
stuff
over
and
over
again.
So.
H
Yeah
my
issue
with
games
like
disco,
is
mostly
I
play
things
in
fits
and
spurts,
and
so
I
might,
I
might
binge
for,
like
you
know
three
hours
a
day
for
several
days
and
then
not
touch
anything
for
four
months
and
any
kind
of
like
rpg,
with
a
story
that
you
need
to
follow.
I
come
back
to
it.
I'm
like
I
have
no
idea
who
any
of
these
people
are,
and
then
I'm
just
clicking
my
way
until
then.
A
C
J
Recommend
I
find
myself
playing
a
lot
of
mobile
games,
not
just
not
just
because
of
the
job,
but
can't.
J
I'm
playing
this
it's
kind
of
weird,
it's
called
sandship.
You
basically
set
up
a
production
line
to
make
metals
and
like
rods
and
screws
and
things,
and
then
you
use
those
to
build.
A
A
A
To
run
and
yeah
as
long
as
there
was
a
starbucks
nearby
because
he
does
need
his
pumpkin
bread
franklin
is
my
dog.
He
gets
he's
obsessed
with
pumpkin
bread
like
I
have
my
starbucks
right
now:
pumpkin,
bread,
hey
franklin!
Oh
look,
you're
right
here
how
weird
yeah
he's
a
bean
he's
a
fiend
he's
a
cute
fiend.
A
Oh,
that's
there's
spell
bound
is
that
the
new.
A
It's
on
everything
and
it's
free.
I
think
it's
like
fortnight,
I
guess,
but
with
like
wizards.
C
A
Which,
theoretically,
I'm
always
about
you,
know
I
like
the
wizarding,
usually
because
you
don't
aim
as
well
as
you
do
with
a
gun,
but
I
haven't
really
played
it
yet,
but
the
it
seemed
pretty
straightforward
from
the
tutorial
so
and
it
is
a
multiplayer
cross
platform.
C
H
Well,
so
here's
the
thing
I
bought
like
a
ton
of
the
books
and
never
played
it,
because
the
only
two
kids
that
I
was
friendly
with
at
the
time
that
were
interested
in
that
stuff,
one
of
them
her,
the
the
mother,
was
like
this
super
religious,
evangelical,
christian
and
so
any
kind
of
role-playing
game
didn't
matter
that
there
was
no
magic
in
cyberpunk
like
it
was
devil
stuff,
so
he
couldn't
technically,
he
wasn't
allowed
to
play
and
then
the
other
kid
was
fine.
H
And
then,
by
the
time
I
found
a
group
of
kids
to
do
paper
and
pen.
We
did
the
star
wars,
d6
rpg,
which
was
awesome.
I
had
a
great
summer
playing
v6
star
wars,
rpg,
which
is
totally
so
cool
if
you're
into
star
wars,
so
just
never
got
a
chance
to
do
paper
and
pen
cyberpunk.
So
now
that
the
video
game's
coming
out,
I'm
like,
oh,
is
it
gonna
ruin
it
for
me,
like
I
don't
know.
H
A
C
C
C
A
A
J
A
And
how
it
would
be
a
thousand
times
better
if
it
was
running
under
gonna.
So
that's
what
we
expect
so.
H
We
are,
we
have
been
building
a
multiplayer
game
as
part
of
a
live
streaming
effort
with
the
openshift
stuff,
but
we're
we're
so
early
that,
like
we
don't
even
have
a
server.
Yet
it's
still
like
really
really
really
early.
I
don't
know
that
we'll
be
able
to
use
agonize
because
it's
like
a
different
paradigm,
but
we'll
we'll
see
once
we
get
to
a
point
where
there's
actually
a
server
to
be
run.
H
Yeah,
we
actually
have
like
a
bunch
of
games
at
red
hat
that
have
been
built
sort
of
unofficially
randomly
for
various
things.
There
are
a
lot
of
like
node,
no
jse
javascripty
browser
games,
but
we
were
like
well
what
would
it
take
to
build
something
proof
of
concept
wise
that
you
could
actually
like
scale
really
large,
using
openshift
and
kubernetes
and
other
pieces
of
the
red
hat
portfolio,
and
so
we're
just
kind
of
hacking
our
way
through
it?
You
know
like
how?
How
could
we,
how
could
we
use
messaging?
H
How
could
we
use
our
data
grid?
How
could
we,
you
know,
do
whatever
so
yeah,
we'll
we'll
see
when.
A
We
get
somewhere
because
you
should
put
a
link
to
that
in
the
working
doc,
so
folks
can
check
it
out.
D
A
All
right:
well,
I
guess
we
can
wrap
it
up
and
thanks
everybody
for
joining,
and
then
we
will
see
y'all
in
november
if
we
can
pull
ourselves
away
from
the
new
releases,
that
we
will
be
playing
them
and
I'll
get
the
recording
up
soon,
and
I
will
ping
in
slack
when
it
is
ready.