►
From YouTube: Agones Community Meeting - October 2021
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
B
Welcome
everyone
to
the
october
28th
edition
of
the
aegonius
community
meeting
we're
going
to
go
ahead
and
kick
it
off
from
the
top
of
the
agenda.
I
put
a
couple
things
on
the
agenda
this
morning.
Please
feel
free
to
stop.
B
If
you
want
to
talk
about
it,
the
first
couple
things
are
sort
of
like
just
quick
announcements.
The
release
was
between
the
last
meeting,
and
now
I
just
wanted
to
highlight
that
it
has
a
number
of
enhancements.
B
It's
got
improved,
grpc
support,
so
if
you
have
been
running
in
terrible
with
the
grp
finance
alligator
allocator
service,
that
has
some
fixes
now
has
a
new
feature:
promotion
to
stable
and
a
new
feature
in
alpha
all
which
are
definitely
worth
checking
out.
I
also
wanted
to
point
out
that
one
night
coming
soon,
but
now
in
the
next
meeting
I
think
and
it's
gonna-
have
support
for
kubernetes
1.1,
so
we're
upgrading,
kubernetes
again,
which
leads
me
to
the
second
item
on
the
region,
which
is
cuban
121,
which
seems
to
me
our
continued
duration.
B
It's
a
little
bit
stable
and
we're
seeing
more
flakes
than
we
were
before
we're
still
going
to
work.
Our
way,
through
those
marks
of
the
history
to
fix
one
of
them,
our
main
concern
here
is
whether
or
not
these
new
flickers
are
exposing
bug
in
our
actual,
like
controller
code
or
whether
they
are
just
really
sort
of
like
you
know,
tests
that
are
in
our
test
furnace
that
are
not
written
to
be
reliable
or
like
good
enough
to
run
against
the
cluster,
as
the
underlying
cluster
changes
a
little
bit.
B
In
this
case,
the
firm
looks
like
it
was
being
caused,
because
the
underlying
gt
cluster
was
using
a
little
bit
more
system
level.
Resources
like
there
were
some
like
things
that
there
were
some
pods
running
on
every
node
that
were
now
taking
a
little
bit
more
space,
and
so
that
made
one
of
the
tests
start
to
fail
so
working
through
those.
B
So
far,
we
haven't
seen
any
indication
that
it's
the
agonizing
controller,
that's
causing
problems
as
opposed
to
us
not
being
resilient
enough,
but
certainly,
if
we
do
find
it,
this
fall
would
give
us
a
chance
to
fix
those
substantial
props,
probably
also
the
race
conditions,
or
something
things
that
were
problems
before
that
we
just
had
seen
that.
B
So
I
think
that's
hopefully
going
to
turn
out
pretty
well,
and
the
next
thing
I
wanted
to
announce
is
that
the
this
is
also
part
of
the
118
release,
but
the
rolling
update
on
ready
feature
beta
to
ga.
We
have
two
other
features
that
I
think
I
found
something
up
which
will
be
promoted
during
this
next
release
and
the
last
one
I
wanted
was
from
support
118.
I
want
to
update
it
to
1.0.
B
I
found
a
couple
places
where
I
had
to
just
like
clobber
my
existing,
like
look
tf
state
file
incompatible
between
all
the
new
versions.
I
think
you
run
into
that
the
other
day.
B
A
A
A
Maybe
showing
up,
why
that's
being
a
problem,
but
anyway
I
just
wanted
to.
Let
have
a
look
at
that.
C
I
guess
maybe
it's
not
so
I
mean
it's
bad
having
flaky
tests,
but
at
least
now
we
know
that
they're
flaky
and
if
we
discover
a
bug
with
the
controller,
then
that's
good
too
right.
So
it's
kind
of
just
something
to
get
through
just
yeah
might
frustrating
yeah
yeah.
Exactly.
Let's
see
I
mean
yeah,
I
mean
I'm
not
going
to
fix
these.
C
So
it's
easy
for
me
to
say,
but
if
you
you
find
a
a
broken
node.js
one,
I'm
happy
and
then
yeah
I
might
say
I
I
did
that
pr
and
of
course
I
did
at
the
published
version
which,
which
makes
sense.
I
I
guess
I
guess
I
got
another
question
I
didn't
write
down.
It
was
that
what
is
that?
Because,
when
the
pr
is
merged,
the
documentation
gets
published
automatically?
C
A
C
I
see
yeah
okay,
so
there's
a
benefit
to
that
yeah
because
it's,
I
guess
it's
inconsistent
with
the
releases,
but
then
yeah,
that's
how
we
and
then
so
yeah,
actually
the
published
version,
but
then
does
that.
I
think
there
were
some
formatting
issues
when
I
did
the
polish
version.
Well
so,
like
the
syntax
is
really
horrible
as
well,
but.
A
That's
just
yeah.
This
is
templating
yeah.
But
yes,
yes,
formatting
is
a
thing.
What
should
we
call
it?
A
A
If
you
ever
want
to
go
see,
one
of
the
pr's,
like
one
of
the
things
we
do
is
we'll
go
and
delete
the
old
blocks
and
just
leave
in
the
new
one.
So
it
all
gets
cleaned
up.
In
retrospect,.
C
Okay,
do
you
have
to
search
like
the
manual
thing,
okay
and
then,
but
then
what
about
so
is
our
documentation
is
not
versioned
like
if
or
or
is
it
like,
because
you
know,
we've
got
that
version
drop
down
that
you
pointed
out
last
time.
A
A
Slightly
ahead,
but
yes,
the
documentation
is
urgent.
C
But
then,
if
it's,
I
wonder
if
it's
versioned
couldn't
we
like
have
a
dev,
you
know.
Basically,
what
you
want
in
maine
is
to
be
published
straight
away.
Could
that
go
to
or
you
had
an
edge
version
or
whatever
you
want
to
call
it
on
the
documentation
and
then
what.
B
A
Well,
we
have
a
development
version,
but
that
will
show
stuff
that
has
been
released
that
isn't
in
an
actual
release.
So
we
can't
like
there
is
a
if
you
look
in
the
drop
down.
There's
a
development,
so
that'll
show
the
stuff
for
the
next
release,
which
may
include
features
that
don't
exist
in
the
current
one.
C
A
C
Have
dated
or
anything
no,
I
wonder
if
there's
a
way,
then
to
say
like
instead
of
putting
all
these
published
version
tags
throughout
everywhere,
you
could
just
say
well
that
doesn't
whatever's
on
publish
there
doesn't
get
published
to
the
the
latest
release
it
just
gets
published
into
the
dev
lots
of
different.
Then
you
copy
you.
Well,
then
you
don't
have
to
add
published
version
tags,
so
everything
in
main
gets
automatically
added
to
the
to
dev.
A
A
D
C
A
Very
strange,
I
went
and
did
some
hunting
on
that,
and
I
couldn't
see
anything
in
the
lobs,
those
particular
fleets
like
nothing
without.
I
actually
wondered
if
extra
logging
in
the
most
recent
garage,
if
to
track
and
debug
and
controller,
if
you
can
see
specifically
why
a
game
server
might
shift
over
unhealthy
for
any
particular
reason,
and
I
was
wondering
if
the
game
temperatures
are
somehow
dropping
into
healthy
for
some
reason,
but
that
doesn't
seem
to
be
a
thing.
A
A
B
I
was
gonna
say
so
sometimes,
when
we're
waiting
for
replica
either
means
it
can
was
like
falling
behind
like
reconcile
fast
enough,
it
could
mean
that
we're
out
of
capacity.
So
I
think
the
graph
you
shared
with
me
yesterday
makes
it
look
capacity,
and
so,
if
there's
something
we
can
monitor
on
the
controller
to
make
sure
that,
like
it's
keeping
up
like
is
the
controller
cue
length
getting
really
long,
where's
the
controller
like
what
problems?
Does
it
get
to?
What's
calling
this
number?
A
So
just
so
actually
just
just.
B
B
We
could
do
is
yeah
when
these
things,
when
they
time
out
if
we
could
like
dump
the
state
of
the
cluster,
basically
like,
like
here's,
the
pods
here,
the
fleets,
here's
what's
playing
on
each
node,
because
then
we
come
back
and
say
like
we
have,
because
I
sometimes
write
that
manually
like
if
the
test
is
running
like.
Can
you
catch
it
at
the
right
time?
You
can
go,
look
at
it
and
see
what's
wrong.
Yeah.
B
I
have
yeah.
E
Yeah
I
got
access
to
the
logs.
Where
are
we
just
to
show
you
the
dashboard?
Because
it's
interesting,
I
think
so
I
built
this
building.
B
A
So
one
one
thing
that
I've
always
wondered
was
whether
or
not
we
were
maxing
out
our
end-to-end
cluster
in
terms
of
like
utilization,
so
I
created
dashboard
incline
monitoring
which
I've
never
actually
done
before,
and
it
was
fun
and
here
comes
my
dog,
and
so
we
can
actually
see
and
then,
if
we
need
to,
we
can
do
alerting,
but
we
can
actually
see
like
some
these.
A
This
shows
some
nodes,
so
some
nodes
get
quite
full
so
when
it
hits
one
that
means
like
a
node
is
full
on
cpu
allocation,
but
we
can
see.
We've
definitely
got
some
nodes
that
that
are
still
have
room
in
them.
These
are
around
like
0.5,
so
these
are
fine,
so
I
just
did
a
a
mean
across
all
of
them,
and
these
are
all
this
is
all
max.
So
we
can
see
okay,
so
we
usually
get
to
about.
You
know
protest
about
0.4,
so
we've
still
got
some
room,
it
doesn't
seem.
A
It
doesn't
seem
like
we're
running
out
of
room
in
our
cluster.
I
did
the
same
for
memory,
but
memory
is
like
0.05.
So,
like
that's
whatever
not
a
big
deal,
but
at
least
now
we
can
actually
see
this,
which
is
super
nice,
quick
question
mark.
A
So
if
you
can
see
the
number
of
parts
that
are
created,
that's
a
good
question.
No
yeah,
that's
legit!
All
right!
That's
that's!
Another!
Okay,
cool
cool
cool
where's!
My
pen,
I
should
probably
write
this
down
somewhere.
I
could
probably
write
an
issue
that
would
probably,
but
that
is-
you
are
a
smart
person.
D
A
A
Two
action
items
add
to
monitoring.
A
D
Yeah
now,
then,
when
you
look
at
the,
if
you
get
graphene
probabilities
matrix,
you
can
see
how
many
parts
per
node
is
being
allocated.
So
it
can
tell
you
if
you
are
hitting
the
110
liters.
A
D
Think
the
initials
eventually,
if
it's
keep
up
to
date,
but
I
believe
on
the
test.
This
is
just
that
on
the
different
clusters,
we
were
using
some
stress
test,
one
for
the
releases,
if
that's
the
one,
they
should
have
the
private
use.
D
B
A
A
I
think
driver
is
disabled
on
here,
but
really
all
I
need
to
check
on
this
is
pods
per
node.
I
don't
need
to
do
anything
about
game
servers,
so
I'm
not
doing
anything.
D
E
I
like
that
a
lot
I
mean
I
can
do
that
now
for
sitting
here.
If
we
want.
A
B
A
I'm
happy
to
sit
here
and
poke
at
this,
but
that's
up
to
you
so
where.
D
E
D
D
E
A
E
D
D
B
F
A
A
F
Because
what
do
I?
Because
the
dash
all
right
so
how.