►
From YouTube: Agones Community Meeting April 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).
B
B
B
C
B
Sdk
some
nice
new
validations,
if
you're,
if
you
create
a
game
server
in
a
namespace
that
we
haven't
supported,
yet
you
actually
have
a
nice
validation
in
the
event
stream
now,
which
is
nice
rather
than
it
just
being
stuck
just
also
super
nice
other
fun
things
more
configurable
log
levels
for
controllers,
so
like
better,
better
clean
up.
More
of
that,
so
we're
not
spamming
a
lot
of
stuff.
B
If
you're
using
super
tuck
Scott,
you
can
now
it's
now
using
a
nice
fork
that
allows
AI
connections,
the
AI,
bots
pharisee,
puck
Scott.
So
if
you're
looking
for
something
to
do
load
testing
with
an
existing
game
that
actually
runs
a
workload
that
could
actually
be
a
nice
way
to
do
it
to
to
do
to
do
memory,
resource
settings
on
the
sidecar
and
a
bunch
of
other
stuff
and
transient.
There's.
C
D
C
B
E
B
It
will
probably
be
just
the
go
and
rest
SDK,
so
I
expect
just
because
of
time,
but
we
can
we'll
see
how
far
I
get,
but
that
was
that
is.
That
is
my
goal.
For
this
release.
I
am
I'm,
gonna
be
hyper,
focused
I
want
to
see
some
major
features.
We
haven't
really
had
a
major
feature
in
the
Gunners
for
a
while,
so
I'd
like
to
see
some
some
more
rolling
at
that
forward.
So
that's
my
goal.
F
B
Also
had,
if
you
haven't
had
a
look,
we
did
do
some
changes
to
the
design
for
player
tracking
based
on
sort
of
the
the
work
that
we've
been
doing.
Nothing
too
drastic,
the
other
really
the
big
thing
there
is
actually
that
we
will
be
tracking
the
tokens
for
IDs
for
players
when
they
connect
and
disconnect
on
the
CR
D.
We
looked
at
that
and
decided
that
was
actually
gonna,
be
a
really
good
idea.
B
Did
I
pulled
away
some
other
functionality,
so
the
web
hook
functionalities
going
away
because
you
can
write
your
own
controllers,
so
yeah,
if
you
haven't,
had
a
look
or
you're
interested
in
you
want
to
make
sure
that
that's
gonna
line
up
with
what
you
want.
We've
got
something
census,
but
feedback
is
always
appreciated.
D
B
D
B
So
what
we're
doing
with
this
is
we'll
keep
two
layers,
so
one
on
the
SDK
server
we're
gonna
track
an
in-memory
state
of
what
the
connecting
disconnected
are.
So,
if
you're
connecting
to
it
from
the
game
server
itself,
you
always
have
that
most
valid
data
and
then,
basically,
whenever
you
do
a
player,
disconnect
or
connect,
it
actually
will
basically
chuck
something
in
the
queue
one
second
in
the
future,
so
that
we
can
then
batch
everything
all
in
one
request
and
keep
track
of
disconnects
in
connect.
B
So
we're
like
alright,
let's,
let's
set
it
up,
so
it
batches
those
connect
and
disconnect
counts
and
the
player
IDs,
and
it
does
it
all
in
one:
go
each
and
every
time,
yeah.
Also,
the
other
fun
trick
in
that
is
what
I'm
also
looking
at
Oh
Lord
I've
actually
done,
and
some
of
the
code
that
I
have
is
make
sure
that
we
don't
whenever
we're
setting
the
connected
IDs
make
sure
that
the
the
ID
a
list
stays
in
the
same
order.
So
there
it
doesn't
so
it
can.
B
G
B
The
same
and
stuff,
so
we
can
there's
there's
some
in
a
lot
of
places
with
stuff
you
can
climb
just
be
like
just
apply.
The
same
thing
doesn't
happen
that
often
but
I'm
putting
some
more
aggressive
checks
in
there
to
be
like
okay.
If
this
data
is
the
same,
then
don't
even
send
it
to
the
API,
because
we
don't
want
it
to
go
through
so
beings
like.
B
It's
like
any
API
like
it's
very
limited.
I
already,
don't
know,
but
if
you
would
agree
with
me,
but
I
would
say,
the
kubernetes
api
is
probably
not
a
hundred
cent
built
for
like
rapid
state
changes,
because
that's
just
not
like,
like
a
pods,
not
gonna,
change
that
often
and
there's
actually
stuff
coming
down
the
pipe
for
rate
limiting,
based
on
like
putting
your
own
api
limits
and
stuff
for
a
first,
I'm
sorry
but
you're
gonna
say
something:
oh,
they.
E
Say
like
the
underlying
data
store,
like
etsy
d,
was,
was
chosen
for
like
resiliency
in
terms
of
like
doing
quorum,
writes
and
reads
so
that
you
could
like
have
an
highly
available
control
plane.
It
was
not
chosen
for
throughput,
and
so
you
kind
of
generally
get
to
pick
one
or
the
other,
and
so
yeah
like
as
Mark
was
saying
like
most
of
the
state.
That's
expected
to
be
stored
in
kubernetes
is
not
expected
to
be
changing.
B
So
we
will
see
but
yeah,
that's
why
yeah
definitely
having
that
double
layer
so
that
you
know
if
you
need
to
get
out
from
the
SDK
and
know
what
your
player
accounts
are
from
within
your
game.
So
every
we
want
to
provide
that
and
also
make
it
if
it
essentially
eventually
consistent
to
the
CIO
d8
backing
it.
B
A
C
A
Right
he
moved
really
quickly,
so
anyway,
it's
basically
been
rewritten.
The
version
that
used
to
happen
is
going
away
soon.
It's
not
appreciated
yet,
but
no
more
features
will
be
coming
in
that
one.
So
the
we
have
the
link
to
the
blog
post
there.
It's
very
short
to
the
point,
but
one
of
the
things
was
was
that
we
had
this
weird
issue
that
was
involving
the
nodejs
stuff
with
a
weird
error
message,
and
so
there
was
the
idea
of
like
hey.
A
B
B
That
way,
or
something
else
needs
to
happen,
I
don't
know
or
maybe
I
genuinely
I'm,
not
sure
there
are
some
tweaks
on
the
SDK
server
side
that
I've
seen
in
some
tickets
around
like
keep
a
lot
of
times
and
things
like
that
that
I
could
try
but
I
kind
of
without
something
to
replicate
the
error.
It
kind
of
makes
me
a
little
scared
to
just
be
randomly
twiddling
knobs
and
see
what
happens.
B
B
D
B
B
D
A
E
So,
instead
of
forcing
people
if
they
want
to
create
a
new
cluster
and
run
a
supported
version
to
have
to
pick
the
latest
release
beginnings,
you
know,
if
they're
concerned
about
that
or
if
they
aren't
ready
to
upgrade
the
better.
We
can
do
in
terms
of
keeping
up
with
their
releases
a
little
bit
faster.
The
more
of
a
sort
of
window.
We're
in
don't
provide
people
in
terms
of
choosing
the
version
of
Donuts
that
they're
comfortable
running
having
a
version
of
kĆ¼bra,
it's
actually
available
on
their
cloud
provider.
E
E
E
B
A
E
Okay,
well,
I
think
the
backwards
looking
milestones
are
really
useful
and
accurate,
because
I
think
merge
marks
goes
through
and
like
make
sure
that,
maybe,
as
part
of
the
release
like
all
of
the
merged,
you
know
and
clothes
issues
get
tagged
with
the
milestone
for
the
release
that
has
been
tried.
The
future
looking
milestone
41.6
is
rather
empty
and
only
has
two
issues
in
it
which
are
sort
of
ongoing
maintenance
tasks
that
have
been
like
just
literally
bumped
forward.
Each
milestone
is
respective
existing,
so
it
seems
like
the
future.
E
Looking
at
milestones
like
we
should
either
think
about
not
using
them
and
just
saying
milestones
are
only
for
backwards
looking
stuff
or
we
should
think
about
a
process
where
we
can
for
as
each
release
gets
cut.
We
can
sort
of
talk
about
like
here's
the
stuff.
We
think
will
be
in
the
next
release,
so
a
good
example
is
mark
just
said,
like
I'm
working
on
player,
tracking
I
really
want
the
alpha
version
to
be
done
in
1.6,
but
player
tracking
is
not
part
of
the
1.6
milestone.
E
Right,
like
the
human
eye,
is
upgrading
to
one
that
15
and
I
just
talked
about
is
targeted
for
106,
but
that's
not
in
the
1.6
milestone.
So
if
anybody
is
trying
to
look
at
a
short-term
road
map
forgiveness,
they
don't
really
have
a
way
to
see
what
we're
looking
on
unless
they
happen
to
come
to
this
meeting
or
find
these
notes
and
not
really
know,
what's
coming
up
next,
so
we
can
decide.
E
That's
ok
and
we
should
just
like
keep
things
as
is,
and
if
people
want
to
know
what's
coming
up,
they
can
they
can
find
these
notes
or
come
talk
to
us
or
ask
questions
and
slack
or
if
people
think
it's
useful,
we
could
have
some
sort
of
process.
We
create
to
actually
use
milestones
for
things
that
we
are
working
on,
that
we're
targeting
and
that
can
be
as
simple
as
like.
E
You
know
if
I'm
working
on
the
Cure
Bernays
upgrade
like
I
should
add
that
milestone
or
if
marks
working
on
the
other,
think
you
should
add
that
milestone
or
if
you
don't
have
access
I,
think
everybody
should
be
able
to
do
it
through
through
the
prow
commands.
So
we
could
sort
of
encourage
people
to
do
that
in
the
same
way,
people
sort
of
assign
issues
themselves.
It's
there
working
on
them,
or
we
can
just
say
like
we
don't
care
about
milestones.
E
Looking
forward,
I
was
curious
if
anybody
thought
that
it
would
be
valuable
to
have
forward-looking
milestones,
especially
they
Steven
or
Dominic.
If
you're
like
looking
at
like
what
what
is
coming
next,
would
that
be
useful,
or
is
it
more
like
when
I'm
ready
to
push
a
new-build
like
I'll?
Look
at
the
current
version
make
sure
everything
is
there
that
I
want,
and
you
know
just
sort
of
request,
or
this
forum
or
somewhere
else
like
things
that
you're
interested
in
having
and
seeing
soon.
D
G
G
E
A
E
And
look
forward
every
time
and
we'll
we
should
just
maybe
write
a
short
note
that
says
like
milestones
or
backwards.
Looking
like
you
shouldn't,
try
to
figure
out
what's
coming
up,
you
can
tell
what
is
coming
in
actually
as
if
something
has
been
merchant
closed
that
people
have
proactively
tags
that
when.
B
E
E
E
B
B
E
C
E
A
E
Sorry,
go
ahead,
wrote
get
up
in
security
alerts,
occasionally
to
some
set
of
people,
apparently
not
mark
to.
Let
us
know
if
there,
if
it
has
detected
security
issues
within
our
project
right
in
the
way,
I
think
it
does
that
it
looks
at
your
dependencies
that
you're
pulling
in
and
and
matches
those
against
known
the
CVS
and
there's
been
a
CV
for
a
little
while
now
against
a
minimis
library
versions
that
are
older
than
0.2
0.1.
A
E
When
I
went
and
looked
at
the
CVE,
it
looks
like
it
probably
would
not
affect
us,
because
it's
about
some
parsing
and
input
validation
errors
and
I
think
the
way
we're
using
it.
We
don't
actually
use
it
for
any
parsing
or
validation.
So
it
seems
like
it's
pretty
low
risk,
but
it
would
certainly
be
nice
to
get
rid
of
the
warnings
from
github
and
the
warnings
from
the
other
people.
That
are
you,
notification
that
are
pestering
us
I.
D
E
D
E
E
D
D
Because
I
guess
in
our
case,
we
just
have
to
check
the
day
before
release,
or
you
know,
as
late
as
possible
or
where
we
can
check
when
the
the
security
alert
comes
and
see
if
there's
been
a
fix,
but
often
it
might
take
a
week
or
so
so
that
we
can
just
just
have
to
kind
of
keep
manually,
checking
until
it's
until
it's
fixed
and
then
another
one
may
turn
up.
So
probably.
E
E
A
E
Getting
threatened,
I've
tried
that
button
for
this
one
and
it's
spun
for
a
good
couple
hours
and
then
failed.
Just
don't
know
that
was
the
first
thing.
I
tried
I
was
like
oh
security
warning,
it
says,
it'll
fix
it
and
it
didn't
do
anything
so
and
how
it
because,
like
seems
like
it's,
it's
transitive
dependencies
at
the
time
like
this
was
I
think,
a
month
or
so
ago
the
versions
to
upgrade
out
of
the
range
weren't
there.
D
B
A
G
A
B
B
F
Yep,
so
more
just
gotta.
Try
it
to
see
how
many
bytes
we
can
pass
into
state
of
30
and
I
saw
that
when
we
extend,
for
example,
state
and
port
ports
in
in
this
state
for
games
error,
there
is
no
validation,
so
we
can
pass
whatever
we
can
with
the
cops
they'll
edit
comment.
So
do
we
need
to
validate
on
update
this
state
probably
could
be
reasonable
for
priorities,
checking
or
not.
H
B
F
There
is
a
question
about
their
form,
so
when
we
have
modules
so
we
have
two
of
them
helm
and
jockey,
for
example,
and
then
we
do
two
times
get
Cologne
of
hala
gonna.
Probably
we
can
split
out
two
separate
one
and
clone
after
mesh.
For
example,
two
main
agonist
reproduce
these
small
ones
if
it's
reasonable
to
save
some
time
disk
space,
because
currently
you
cannot
clone
only
one
folder
if
capital
go
files,
etc.
I.
I
B
F
D
B
F
F
Do
it
later,
probably
yeah
and
yeah
I
just
wanted
to
close
this
or
when
we
have
was
updating
all
versions
and
one
more
issue
that
I
see
probably
or
we
need
more
yeah
like
more
events
on
when
pod
could
not
be
created.
So
as
this
currently,
there
are
quite
a
lot
of
places
where
it
could
happens,
and
one
point
was
my
latest
bug,
which
was
it
wants.
How
to
resource
quotas
could
not
be
updated
because
of
the.
F
B
B
E
E
Well,
that's
kind
of
how
it
happened
is
like
we
didn't
go
into
an
updating
them
all
as
I've
been
changing,
all
of
that's
the
case
yeah
and
then
a
little
bit
later
I
went
back
and
was
like.
Oh,
it's
got
like
all
of
these
security
warnings
and
there
were
tons-
and
you
know
a
lot
of
more
just
like
low
severity
or
ya.
A
E
D
B
I'd
really
like
to
make
that
change,
I
think
I
think
it
we're
reaching
a
point
where
we
probably
should
I
posted
an
example
in
there,
where
you
can.
Actually
it
looks
like
from
from
what
people
are
saying
it
seems
like
you
can
use
helm
three
right
now.
You
just
need
to
change
the
command
a
bit
I
think
we
could
like
I
post
an
example
of
where
someone
else
has
done.
B
That
and
they're
like
peace
is
how
you
do
at
home
too,
and
this
is
how
you
do
it
in
three
and
that's
fine,
but
I
think
what
we
probably
need
to
do
on
our
end
is
start
to
change
over
a
tooling
to
move
it
to
home
three
and
properly
put
a
note,
no
dock,
saying
something
like
hey.
This
probably
works
on
home
too,
but
we're
testing
on
home
3.
B
B
You
delete
the
cluster
chair,
form
and
like
it,
doesn't
delete
the
helm
stuff
it
enough
time,
and
so
you
end
up
with
a
whole
bunch
of
stuff
left
over
in
gke
like
how
much
the
load
balances
stuff,
which
is
a
real
pain
and
it'll,
also
simplify
a
lot
of
the
terraform
stuff
as
well
and
bring
us
up
to
date
with
the
terraform
providers
and
that
kind
of
stuff
too
I
did
an
informal
poll
on
who's
using
one
and
it
looked
like
on
Twitter
terraform.
Is
that
clear,
sorry,
helm
3
is
a
clear
winner.
B
G
So
improbable
last
I
checked.
They
were
still
using
that
you
weren't
using
turn
they're
using
telephone,
but
not
to
create
the
Gunners
and
they're
using
helm,
but
not
for
the
deployment
begoƱa's
we're
hungry,
they
were
hand
rolling
it
with
yellow
files,
ok
and
customize
the
template
and
customize,
because
we
didn't
even
anything.
Okay.
C
D
F
F
F
H
B
C
B
B
B
B
A
A
C
C
G
G
G
A
Well,
cuz,
we
keep
saying
like
that,
everybody's
you
know
working
from
home
and
all
this
like,
we
need
to
find
a
game
that
we
can
all
play
together
and
we
tried
last
week
with
some
of
folks
on
the
team.
It
was
varied
results
so
something
that
everybody
could
do
on
their
own.
But,
like
still
be
you
know,
playing
something
together.
That
sounds
interesting.
That
could
be
good.
F
Yeah
theories,
google
steadiest
project
is
not
working
in
my
country,
so
I
try
to
I
have
tried
g-force
now
and
it's
like
interesting
experience
not
to
install
not
to
use
your
disk
space
etc.
So
it
was
fun
some
logs,
but
at
least
they
provide
free
also
months.
It
seemed
like
related
to
worrying
time.
Yeah.
A
And
that's
definitely
a
nice
benefit
without
like
every
time
my
husband
will
cuz.
My
office
is
right
next
to
the
room
that
has
the
TV
in
the
PlayStation
and
I'll,
see
him,
like.
You
know,
end
of
the
day
go
in
there
and
then
it'll
be
quiet
and
then,
like
30
minutes
later,
it's
like
well,
I
was
gonna
play
NBA
2k,
but
there
was
an
update
I'm
like
yeah,
welcome
to
the
world
of
video
games
like
that's
just
how
it
works.
Yeah.
You.
B
B
A
A
Okay,
well,
whatever
it's
downloaded,
I'm
sure
it's
fine
all
right!
Nobody
has
anything
else
exciting.
Well,
thanks!
Everybody
for
joining
will
get
their
recording
up,
we'll
still
be
in
touch
of
course,
I'm
thinking
next
months
we
switched
to
later
time-
and
today
we
talked
about
this
before
about
you,
know,
swapping
off
and.