►
From YouTube: Agones Community Meeting August 15, 2019
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
A
You
know
the
same
model
either
in
just
spend
some
requests
to
do
it
in
Europe
and
we're
also
going
to
try
to
do
something
at
coupon
in
November,
which
is
also
in
India,
because
that's
where
all
conferences
are
this
year,
but
if
you're
in
San
Diego
next
week,
even
if
you're
not
going
to
the
conference,
if
you
just
live
there
in
the
area,
we'd
love
to
see
you
have.
You
join
us
at
the
gaming
day
or
just
meet
up
at
some
point
during
that
week
to
tell
the
informations
in
the
working
Donkin.
B
Mention
of
people
that
12
have
been
released
has
a
whole
bunch
of
changes.
If
you've
got
configuration
without
11
you're
going
to
have
to
rewrite
bits
of
it.
4.12
all
the
example
have
been
updated.
I
mentioned
in
slack
how
to
update.
There
was
I
found
a
example
on
something
like
the
AWS
community
sites
which
which
was
outdated,
so
I
told
somebody
how
to
update
that,
and
but
if
you're,
you
know
running
your
own
game
servers,
you
know
have
to
rewrite
every
mo
between
11:00
and
12:00,
and
we
hope
that
this
is
the
last
time.
B
C
Yeah,
okay,
so
we're
we're
still
waiting
for
the
allocation
metrics,
but
we
have
a
PR
in
place
for
that
and
we
just
picked
visceral
to
get
back
from
holidays
so
concerned
about
that
the
missing
SDK
functionality
across
unity
and
unreal
I'm
kind
of
worried
about.
Since
we
don't
seem
to
have
people
in
the
community
to
jump
on
that-
or
at
least
no
one
who's
committed
to
doing
the
work
but
can
I
can
probably
tackle
or
maybe
between
us.
C
B
Could
give
it
a
shot?
It's
always
a
possibility.
How
do
we
feel
about
cutting
one
natto
with
some
of
the
SDK
is
not
being
fully
future
complete?
Does
their
way
begin
to
sort
of
denote
like
this?
Sdk
is
a
work
in
progress.
It's
not
fully
implemented,
and
here
are
some
that
actually
worked.
We're
still
working
on
this
one
it'll
come
in
at
future
release.
C
Yeah,
but
at
the
same
time,
if
it
doesn't
happen,
maybe
we
should
be
blocking,
given
that
that's
that's
a
thing
rather
be
saying:
hey
the
performance
on
this
is
great
rather
than
anything
else
there
is.
There
is
some
work
we've
got.
There
are
some
people
I
know
working
on
some
stuff,
I
need.
Actually
there
are
some
people
working
on
some
stuff
being
deliberately
vague,
who
are
using
unreal?
Who
may
be
able
to
do
it?
C
B
C
C
D
C
C
You
know
it
for
conformance.
I
was
actually
thinking
with
they
get
the
game
server
status.
We
do
you
have
a
swagger
Jason.
We
could
probably
generate
out
at
least
the
game,
server,
spec
format
or
the
game
server
format.
Out
of
that,
and
then
probably
port,
that,
across
out
of
like
a
c-sharp
generated
I,
don't
actually
remember
what
swagger
swagger
general.
C
B
And
I
think
the
other
point
there
was
like.
If
you
do
know
unity,
you
should
look
at
the
code
that
is
there
and
make
sure
it
seems
reasonable.
Even
if
the
simpler
stuff
is
done,
if
it
wasn't
done
well,
we
could
fix
that
Yeah
right,
I
think
I'm
using
the
right
words.
I,
don't
know
it.
It
runs
for
me
right,
like
I,
can
run
the
samples
and
it
seems
to
be
working
like
the
very
basic
example,
but
that
doesn't
really
mean
that
it's
production
already
yep.
C
I'd
also
love
to
see
the
unity
unreal
SDKs
to
have
like
a
similar
thing
to
like
what
we
have
connect
and
other
ones
where
in
like
it
weights,
it
blocks
until
the
SDK,
so
ever
comes
up
or
be
tries
at
least
or
something
like
that.
I
think
we
can
just
do
you
like
with
the
those
those
both
unity
and
unreal
use,
the
HTTP
endpoint.
C
B
I'm
I'm,
ok
with
that
I
mean
I.
Think,
like
you
said,
if,
if
those
are
the
ones
that
you'll
really
really
want,
then
it's
a
little
bit
unfortunate,
but
we
do
have
a
number
of
other
SDKs
that
work.
You
know
I,
think
Steve,
you
guys
are
using
the
nodejs
one
right,
and
so
we
have
ones
that
we
know
work
and
I
think
we
can
say
ago.
Is
it
so
Forks
and
if
you
really
need
these
missing
SDKs
like
please
come
help
us
to
finish
them.
Yeah.
C
F
A
C
I
I'm
just
saying
is
there
I
actually
think
most
of
all
I
thought
they
were
more
I
just
wanted
to
do
a
quick
pass
through
to
see
if
there
was
anything,
that's
a
bug
that
we
were
like
that
should
definitely
be
fixed
before
1.0,
but
now
I'm,
looking
through
it
I,
actually
someone's
literally
just
had
a
segfault
with
the
C++
SDK,
but
I'd
like
to
get
more
details
on
that.
So
yeah
I.
C
B
C
Cool,
alright,
okay,
so
it
wasn't,
doesn't
look
like
there's
anything
else
that
served
up
anyone
else,
which
is
good.
That
makes
me
happy.
The
other
thing
I
was
gonna
mention
just
cuz.
It's
fun
and,
and
April
mentioned
that
she
put.
Apparently
she
put
some
some
slack
emojis
that
we're
gonna
have
to
find
inside
there.
I
I
wanted
to
add
some
little
bot
commands
to
make
things
easier
when
like
troubleshooting
or
helping
people,
so
I
added
Bangla,
quote
version,
which
is
literally
like.
Can
you
please
provide
your
account?
C
C
And
Roberts
just
trying
that
out
right
now
and
I
also
want
to
add
some
put
some
some
links
to
add
to
the
troubleshooting
guide.
I
want
to
add
as
well.
If
there's
anything
else
that
you're
like
this
is
a
thing
that
I
keep
seeing
and
it
would
be
really
handy
if
we
had
a
bag
to
like
link
to
something
or
something
like
that.
Then,
like
just
thing
me
and
let
me
know,
if
there's
anything
else,
we
can
add
just
to
make
some
community
help
stuff
easier
and
more
reproducible.
A
C
C
A
A
F
E
C
C
C
B
So
my
plan
for
pro
is
to
try
and
set
it
up
as
sort
of
part
of
the
load,
testing,
stuff,
I
poked
ocher
a
week
or
so
ago,
to
figure
out
what
his
load
testing
setup
wasn't.
He
sent
me
some
work-in-progress
code,
which
I
had
three
base,
but
I
is
now
rebased
against
head,
so
I
can
try
to
run
his
load
testing
and
see
if
I
can
reproduce
it
at
which
one
I
can
send
a
PR,
and
we
can
try
to
get
that
checked
in
and
the
idea
what
I
would
be
like.
B
The
first
thing
to
do
would
probably
be
something
we're
not
already
doing
so
now.
Try
to
replace
functionality,
but
try
to
do
something
you
have
an
additive
is
the
thing
that
proud
can
do
that.
Cloudbuilt
can't
do
we'll
just
run
something
off
my
schedule:
that's
not
tied
to
pull
requests,
and
so
the
idea
would
be
to
try
to
run
the
load
testing
framework.
You
know
once
a
day
or
something
like
that.
Probably
unlike
a
different
cluster,
then
we
run
our
easy
ease
on
and
get
sort
of
a
new
signal
going
back
to
the
project
about.
B
So
how
we're
doing
you
know
on
a
slightly
slower
cadence
than
PRS,
but
still
something
consistent.
So
as
we
do
that
like
we
can
talk
about
what
other
little
like
prior
plugins,
we
might
want
to
run.
Just
like
I
think
there's
like
I,
don't
know
if
the
issue
closer
is
parked
per
hour.
That's
in
beta
bot,
I,
don't
know
about
to
run
against
your
stuff.
So
try
to
ask
Eric.
C
About
that
there
are
some
other
github
bots
that
will
do
similar
things
as
well,
like
90,
day
closes
and
stuff.
Like
that,
I
can't
remember
I,
remember
we
have
a
conversation
ages
ago
with
I'm,
just
thinking
about
like
long
term
and
like
being
able
to
do
like.
Ideally,
maybe
probably
roll
back
like
over
cloud,
build
stuff
and
move
it
into
prowl.
Does
the
prowl
cluster
run
inside
our
project
or
it
runs
inside
a
separate
project?
C
C
B
Think
there
are
two
ways
we
can
run
prowl
one
is:
we
can
run
the
whole
stack
ourselves
right.
We
print
havoc
units
cluster
that
runs
prowl.
If
we
do
that,
then
we
are
complete
control
over
like
what
the
URL
for
the
results
looks
like
and
everything
else.
We
also
have
to
maintain
prowl
yeah.
You
can
then
have
PowerPoint
its
own
cluster
to
run
the
jobs
or
a
different
cluster.
The
other
alternative,
which
is
what
I
proposed,
is
let's
let
the
kubernetes
and
prod
folks
run.
B
The
proud
control
plan
keep
it
up
to
date,
since
they
maintain
that
code,
and
we
want
people
worried
about
it
at
which
point
we
basically
bring
our
own
cluster.
So
we
say
that's
right
now
we
submit
jobs,
but
you
execute
them
in
a
our
cluster
and
that
cluster
can
be
in
our
project,
which
can
easily
furnish
to
RF
engine,
isn't
good,
and
then
we
can
do.
We
can
do.
C
I,
don't
think
art
is
gonna,
be
an
issue
anyway.
That's
fine,
but
though
that
makes
sense
thanks.
That's
awesome,
none
that
sounds
really
good.
That
sounds
really
good.
So
far
so
far,
my
string
of
bash
functions
and
cloud
functions
and
stuff
seems
to
be
fine,
but
yeah
one
day
it
will
break
it's
working
well
for
what
it
does.
C
Enterprise
e
I
don't
have
anything
else.
Does
anyone
else
have
anything
else,
all
right,
I'm.
B
Just
looking
through
the
open
bugs
to
see,
if
there's
anything,
it's
like
you're
cooking,
a
bug
in
January
about
packing
packing,
not
working
correctly,
which
I,
don't
think,
really
went
very
far,
and
there
was
a
bug
open
about
the
fleet.
Autoscaler
spawning
extra
game
servers
with
a
cone,
a
0.6
I,
don't
know
if
that's
worth
trying
to
reproduce
and.
C
C
B
Only
happens
if
they
have
no
pool
auto-scaling
term
problem,
yeah,
that's
a
little
eerie,
so
I
don't
know
pool.
If
that
would
be
something
it
would
be
probably
good
to
try
to
reproduce.
Yeah.
F
C
B
There
was
another
feature
request
that
seemed
relatively
easy
to
implement
and
might
might
help
people
in
production.
So
it's
kind
of
curious.
If
you
guys
have
any
feedback.
Somebody
asked
for
a
configurable
log
level
for
the
sidecar
because
they
said
that
normally
like
the
sidecar
is
too
verbose
like
they
just
want
to
see
warnings
and
errors,
and
not
all
the
infamous
which
is
it
show
up
all
the
time.
C
B
Yeah
how
how
widespread
of
a
feature
request?
It
is
if
it's,
if
it's
a
Colin
use
case
to
try
and
look
at
those
logs,
are
commonly
getting
in
the
way
or
if
it's
you
know
easy
enough
to
sort
of
work
around
them.
Using
your
blood
collection
pipeline
for
most
people
that
it's
not
worth
trying
to
have
to
work
on
yet.
B
The
log
levels
are
already
I,
think
annotated
in
the
output
shows
info
warning
errors
and
the
person
was
just
saying
like
or
so
many
ones
at
level
info
that
aren't
really
useful
for
debugging.
It's
making
it
hard
to
spot
the
error
and
warning
ones
and
I'm
guessing
there's
a
few
like
alpha
string,
yeah
most
of
the
health
check,
which
happens
really
frequently,
and
they
basically
just
like
a
configuration
option
to
say
like
for
this
fleet,
like
set
the
side
card,
adjust
warning
and
above
because
then
I'll
see
the
relevant
stuff.
C
B
H
Yeah
so
one
buck
which
I
currently
work
keen
on
his
container
crash
before
ready,
not
moved
to
should
restart
so
this
one
I
think
we
documented
and
it
should
be
prioritized,
but
it's
not
so
easy
to
like
prevent
in
the
so
I
will
have
a
discussion
later,
but
I
think
we
have
stated
like
four
strategies
for
health
check,
but
some
of
them
is
not
working
yet.
This
is
also
good
to
fix.
B
B
C
Yeah,
whereas
in
theory
what
should
actually
happen
and
it
would
have
hidden
that
particular
bug
as
well,
which
would
have
been
fun
and
a
whole
other
thing
is
if
it
like
comes
up
before
the
sidecar
and
crashes,
and
it
should
restart,
come
up
and
then
try
and
reconnect
and
then
finally
reconnect
and
continue
on
its
merry
way.
It
could
have
a
chance
to
become
healthy.
Yeah
should
have
a
chance
to
go
healthy
if
it
crashes
after
it
becomes
ready.
Then
you
know
that's
bad
and
like
dying,
go
away.
C
C
Because
you
have
an
initial
delay
as
well
in
your
health
check,
so
you
can
actually
control
how
many
times
like
how
long
you're
willing
to
wait
for
that
crash
loop
to
kind
of
happen
if
it's
happening.
Yes,
that
would
be.
That
would
be
the
ideal.
The
ideal
scenario
I
think
Alex
did
that
make
sense.
Yeah.
C
All
right
the
code
ages
ago
and
I
think
it
was
actually
coming.
Who
is
like
hey
if
I
delete
a
pilot,
the
game
server
doesn't
go
away
now,
it's
like
oh,
that's,
bad
and
then
I
didn't
actually
look
at
the
health,
the
health,
the
health
of
the
contract
that
we
had,
and
so
that's
kind
of
my
fault,
yeah
I.
H
C
H
B
There
that
might
be
a
worth
talking
about
as
long
as
we're
here
in
a
few
minutes
is
the
issue
that
hoon
opened
last
week
about
sort
of
release:
cadence
the
sixth
six
weeks,
cadence
the
time
between
release
candidate
and
final
cut
for
each
release
and
running
things
from
Esther
branch.
There
were
sort
of
multiple
meta
points:
that's
issued
number
989,
hoon,
I,
don't
know
if
you
want
to
jump
in
and
really
any
background
on
that
one
I.
C
That
are
I,
haven't
gone
through
them
or
I
think
I
usually
end
up
applying
labels
to
it.
So
it's
really
also
on
me
that
I
haven't
actually
looked
and
gone
to
see.
If
that's
the
thing
that's
going
through
or
even
thought
about
how
we
can
ask
where
we
should
ask
for
help.
I,
probably
like
I,
like
having
the
good
first
issues
at
the
very
least,
probably
more
than
anything
else.
C
C
I
I
But
if
we
have
intermediate
really
there,
then
somebody
can
basically
pick
up
the
change,
make
sure
please
their
change
Oryx
or
they
can
apply
their
change
to
the
right
start
using
their
change
versus
waiting
for
that
to
happen
after
six
weeks.
So
one
way
to
do
to
do
that,
we
can
have
a
deaf
branch
and
regularly
even
on
demand.
We
can
create
their
like
things.
I
That
needs
to
be
deployed
like
the
llamo
files
and
the
images,
and
then
people
can
try
on
that
one
and
then
it's
stable
than
marriage
and
the
master
so
there's
separation,
and
if
there
is
a
hot
pot
effects
that
goes
to
the
dev
branch
it.
Instead
of
getting
everything
from
the
dev
branch,
we
can
just
apply
that
hot
effects
and
then
release.
We
don't
have
a
good
story,
I
think
for
the
hotfix
already
or
I'm,
actually
missing
something.
So
those
are
the
issues
that
was
on
top
of
my
mind
when
I
opened
that
box.
C
So
it's
gonna
say
on
that:
we
do.
We
do
actually
do
build
images
for
each
master.
We
actually
build
images
for
each
PR
and
each
master
as
well.
P.
Ours
are
easier.
We
actually
have
a
few.
If
we,
if
we
look
at
PR,
you
will
probably
notice
that
at
the
bottom
of
a
comment,
there's
actually
a
a
line
to
install
that
particular
version
for
that
particular
PR,
and
that
we
keep
those
images
around.
So
that's
kind
of
their
I.
C
Don't
know
how
many
people
actually
use
it
to
foreach
build
of
a
PR
there's,
there's
there's
that
I
think
it
would
be
lovely
to
do
something
similar
from
master,
maybe
on
each
each
commit.
We
could
have
like
a
line
where
you're
like
just
get
check
out
this
commit
and
then
there's
the
helm,
install
line
and
you're
good
to
go.
C
That's
like
that's,
basically,
just
building
a
page
that
does
the
thing
which
is
not
terrible.
We
keep
those
images
for
like
30
days
remembering
so
we
could
definitely
do
that.
I
think
that's
kind
of
cool,
so
I
gonna,
say
building
out
dynamically,
creating
the
story
ammo.
We
don't
have
right
now,
but
we
could
do
in
theory.
C
B
C
C
Twist,
yes,
yeah
and
all
that
and
like
we,
we
release
on
that
actual
one.
So
that's
like
the
Helmand
store
for
a
PR
I'm,
looking
at
for
example,
so
we
do
something
similar
where
it's
like,
you
know,
get
fetch
get
such
this
particular
Shah.
For
this
particular
one
and
then
Helmand
stole
using
these
commands.
Maybe
that's
good
enough
or
we
could
have
a
repository
of
them.
We
could
stick
them
somewhere,
I,
don't
know
if
we
want
to
do
like
a
helmet
art
for
each
development
release
top
of
Lisl,
maybe
a
little
bit
overkill.
B
C
The
same
one
is
make
gen
install,
so
does
actually
doesn't
change.
So
if
we're
doing
like
the
1.0,
it's
always
gonna
be
one
point
over
three.
So
it's
basically
the
default
is
like
one
point
0.0,
and
then
we
just
tweak
it
for
like
that's
what
the
the
development
scripts
do.
Actually
they
do
exactly
that
step
and
they
just
pass
in
what
your
Shah
is
attached
to
the
base
base
revision
and
that's
it
tells
it
how
to
install
on
home
yeah
I,
like
I,
mean
I
like
the
idea
or
subversion
how
we
want
to
do
it.
I
So
we
need
to
have
to
have
a
boom.
We
already
tagged
them
by
a
release
attack
right,
so
this
should
be
fine,
but
I've
seen
in
the
github
repository.
There
are
a
lot
of
teams
that
use
a
dev
and
master
branch,
so
somebody
who
wants
to
for
both
brands
from
the
master.
They
know
that
that's
the
stable
one,
just
at
least
one
and
then
they
develop.
On
top
of
that,
then
monster
changes.
C
C
C
C
C
I
think,
actually
we
cut
the
we
we
cut
I'm,
trying
to
say
I'm
trying
to
remember
how
did
we
do
that
yeah?
We
had
to
create
a
release
branch
because
we
were
already
moved
past
that
point
so
we're
doing
this.
Basically,
we
did
that
fixer
we
can
get
stuff
upon
the
marketplace
for
GK
right
and
don't
I
feel
like
did.
I
B
B
Everything
in
for
master
and
that's
how
you
do
it
right.
You
start
from
the
release
branch,
make
your
fix
on
that
release,
branch
and
move
forward
and
oftentimes.
That's
a
cherry
pick
where
you
you
put
it
in
the
master
and
you
put
the
same
commitment
release
branch.
You
know
possibly
slightly
differently,
depend
on
how
far
they've
drifted
yeah,
but
sometimes
you
just
put
a
Kim
in
the
release
branch
that
doesn't
go
on
your
master
and
I.
B
Think
you
know
both
those
were
close
or
I
think
relatively
well
supported
with
the
master
being
the
development
branch
and
I
may
create
a
release
branch
every
time
that
there's
your
police
generally
those
release
perimeters
don't
go
anywhere
because
we've
create
the
next
release
instead
of
hot
fixing.
But
occasionally
you
know,
topics
in
the
police
branch
and.
I
They
remember
something
else,
something
that
triggers
is
another
thing
that
trigger
is,
if
we
stop
do
not
merge
your
changes,
I'm
steering
this
one
week
of
trial.
If
we
had
a
developed
branch,
those
changes
still
could
go
to
the
develop
branch.
Only
the
hotfix
would
end
up
to
the
master
branch,
and
it
would
help
it
see
the
lot
of
complex
issues
when
there
are
a
lot
of
like.
If
there
are
10
issues
waiting
for
a
week
sweeteners,
there
would
be
conflict
one
by
one.
I
B
I
don't
think
right
like
during
the
last
release
edit.
It
was
like
super
quiet,
you
know,
but
it
just
sort
of
waiting
for
the
weekend
mark
was
on
vacation
and
I
was
that
it
wasn't
a
lot
of
merging
going
on
anyway,
but
you
know
I.
Maybe
during
previous
you
don't
release
Canada
at
weeks
it's
dead
works,
yeah.
C
A
C
Mean
I'm
happy
with
the
five
one
I
think
it's
a
good
compromise
on
like
what
should
I
call
it.
I
I
totally
hear
you
I
was
like
people
may
not
have
time
within
a
week,
but
if
people
don't
have
time
in
a
week,
I'd
be
surprised
if
they
have
time
within
two.
But
that
seems
like
a
long
time
for
us
to
like
wait
on
stuff.
That's
the
best
is
my
opinion.
There
I.
C
Don't
know
how
you
feel
about
it,
but
it
sounds
like
it
sounds
like
having
a
set
finding
some
way
to
set
up
interim
like
basically
being
able
to
make
it
easy
to
set
up
hey.
This
is
your
build
if
you
want,
if
you
want
to
grab
the
latest
from
master,
here's
a
here's
a
nice
way
to
do.
It
would
probably
be
a
good.
A
good
happy
place
in
between
that
is
having
to
change
a
bunch
of
stuff.
C
E
C
A
C
I
mean
I
figure
is
actually
rather
than
doing
daily
I
mean
I
mean
we
could
do
daily,
which
I
think
it's
actually
easier,
because
the
the
build
chain
stuff's
already
exists,
therefore,
basically
creating
a
master
build
every
time,
because
it's
this
is
literally
cloud
build.
That
runs
on
everything.
So
the
the
contents
they're
assuming.
B
C
C
B
B
C
But
that's
actually
really
act
could
actually
I
kind
of
like
that
cloud.
Storage
idea,
just
because
we
could
put
that
as
a
step
in
in
the
cloud
build
like
if
it's
the
master
branch
I've
seen
some,
it
I'm
seen
some
hacky
stuff
in
cloud
build
where
it's
like.
You
write
a
bash
script
that
basically
says
if
it's,
if
it's,
if
the
environment
variable
equals
master,
then
do
this
thing,
then
we
could
just
get
up
in
cloud
storage
together
or
if
there
any
other,
clever
way
of
doing
that.
No
actually.
A
I
There
are
some
issues
that
I
opened
for
our
multi
cluster
allocation
policies
and
the
al-khair
service,
those
those
are
so
one,
the
biggest
issue
that
I
saw
and
I
haven't
filed
ticket
for
that.
Yet
we
marked
the
aisle
care
service
as
vivan.
That's
still
in
progress,
it's
not
documented,
so
I
think
it
should
be
fine
to
basically
roll
it
back
to
live
on
alpha
wine,
but
I'm
not
sure
if
this
is
a
good
idea
or
not,
and
there
are
changes
going,
that's
gonna
break
the
server
ideas
like
GRDC.
B
I
There's
a
locator
service
recently
changed
the
name
to
ago
nests
alligator.
That's,
basically,
a
load
balancer
service
that
accept
calls
with
the
certificate
from
external
external
to
the
cluster
like
from
a
matchmaker
outside
the
classroom,
and
that
also
is
used
for
talking
to
other
clusters.
But
we
are
not
like
the
API
is
not
finalized
and
it
shouldn't
be
marked,
as
we
whine
I'm.
C
B
C
C
B
E
A
All
right:
well,
we
did
record
so
we'll
get
this
up
on
the
YouTube
and
we'll
link
it
out
all
the
dots
as
well,
and
hopefully
we'll
see
some
of
you
next
week.
Maybe
otherwise
we'll
have
our
next
meeting
in
September.
At
some
point,
we'll
have
to
figure
out
the
exact
date
because
I
know
we
have
a
conflict
with
a
big
internal
conference.
I
think.