►
From YouTube: Kuma Community Call - June 23, 2021
Description
Kuma hosts official monthly community calls where users and contributors can discuss about any topic and demonstrate use-cases. Interested? You can register for the next Community Call: https://bit.ly/3A46EdD
A
Good,
okay!
Well
so
we
were
talking
about
the
the
release
last
week.
A
A
B
So
I
mean
I
think
most
of
the
people
on
this
saw
saw
the
demo.
I
thought
I
was
really
happy
with
how
it
all
played
out.
We
went
through
and
we
did.
B
We
stepped
through
the
layer,
seven
functionality
I
did
kind
of
I
don't
know
it
was.
It
was
somewhat
like
a
cheesy
demo,
not
not
the
best
one,
but
we
took
all
of
like
the
routes
that
I
had
set
up
in
engine
x
and
just
used
l7
to
handle
handle
service
routing.
Instead
it
showed
the
functionality
but
obviously
like
you
could
do
that
in
ingress
controller.
You
could
do
that
elsewhere,
but
it
did
show
the
functionality.
B
If
I
was
happy
with
that,
we
did
kind
of
a
campy
campy
little
demo,
where
we
we
took
marco's
demo
app
in
my
demo,
app
and
used
a
header
to
flip
between
them.
So
if
I
think
it
was
like
app
mode,
marco,
as
a
header
was
present,
it
would
use
marco's
app
and
it
switched
over
pretty
cleanly.
One
thing
I
I
really
enjoyed
was
firing
up
locust.
I
don't
know
if
anybody
else
has
used
locus
before,
but
I'm
a
big
fan
of
it.
B
For
doing
like
load
testing,
mostly
because
it's
written
largely
in
python,
so
I'm
a
fan
of
that
we
used
that
to
test
out
ilia's
rate
limiting.
So
we
thanks
for
giving
me
that
code
by
the
way
ilya
the
night
before
that's
kind
of
a
last
minute
rush
for
me,
but
it
worked
great.
There
was
no
honestly
that
that
one
had
the
least
amount
of
things
I
had
to
figure
out
in
order
to
get
it,
get
it
working
right.
So
that
made
me
happy
yeah,
but
the
demo
went
super
smooth.
We
didn't!
B
I
didn't.
I
don't
think
I
hit
any
hiccups
along
the
way.
I
don't
think
I
had
any
any
failure
moments
in
there,
yeah,
no,
no
issues
along
the
way,
so
it
was
really
it
was
a.
It
was
a
good
flow
and
we
had
good
feedback
on
it.
We,
you
know
we
got
a.
We
had
a
lot
of
people
watching
across
all
three
mediums,
so
we
had
it
going
on
linkedin,
twitter
and
youtube.
B
I
think
we
had
around
60
or
50
55
or
65
across
all
three
of
those
and
we've
had
a
ton
of
watches
since
then.
So
it's
not
even
just
like
the
live,
it's
the
people
going
back
and
watching
the
video
that
was
that
was
really
great,
so
yeah
it
was
a.
It
was
a
good
release
like
to
like
to
keep
doing
those
as
we
release
new
versions,
provided
we
have
like
good
features
and
stuff
to
show
off
right.
A
Okay,
sudipto.
We-
I
guess
that
we
were
chatting
about
this
before
you,
you
joined,
but
apparently
the
pr
that
you
you
that
we
chatted
about
the
last
time.
One
did
not
make
it
into
the
release,
but
as
jacob
was
saying
in
it,
we
are
willing
to
include
it
in
one
to
one.
C
Yeah,
anyway,
I'm
actually
following
up
so
I
happen
to
get
on
to
like
another
scenario,
wherein
let's
say
you
have
an
application
which
is
like
in.
D
C
An
overlay
network
and
you
want
to
scrape
scrape
the
matrixes
matrices
out
of
that
in
case
the
prometheus
needs
to
needs
to
scrape
that
endpoint.
We
need
to
let
prometheus
know
that
it
is
the
advertised
address
and
not
the
address
where
it
should
go
to
script.
C
So
to
do
that
prometheus
uses
a
function
called
get
get
ip
and,
and
that
also
would
need
to
be
like
changed.
But
then
I
see
that
the
gate
ip
is
used
at
three
places.
I'm
I
understand
the
users
of
gate.
Ip
are
two
two
of
them,
but
I
don't
understand
the
uses
of
get
ip
at
the
third
place
and
and
that's
the
suggestion
which
I
asked
jacob.
C
A
C
C
We
yeah,
so
the
l7
routing
would
be
something
which
would
be
like
immensely
helpful.
The
only
thing
that
I
want
to
add
on
top
of
that
is
to
have
a
little
bit
of
gateway
functionality.
If
I
can
have
that
will
really
help
a
lot
of
the
application
team
too.
I
mean
that
will
greatly
help
people
to
use
use
that
particular
feature.
A
That's,
that's
that's
an
interesting
topic.
So,
first
of
all
we
are.
We
are
working
on
something
around
the
gateway
functionality,
but
I
don't
remember
in
which
call
I
don't
think
that
that
it
was
here,
but
it's
in
some
call.
There
was
the
discussion
now
that
you
have
l7.
How
do
you
distinguish
what
you
do
on
the
gateway
and
what
you
do
on
the
service
mesh
side?
A
B
You
playing
playing
devil's
advocate
a
little
bit
like.
I
do
think
that
the
gateway
is
the
right
place
to
do
it,
but
I
think
that
they're,
you
know
we
talk
a
lot
about
mesh
and
creating
like
a
consistent
place
to
apply
policy
across
like
multiple
environments,
and
I
think,
like,
if
you're,
using
if
you're,
using
different
gateway
types.
There
are
times
where
the
way
you
apply
that
inbound
configuration
is
going
to
be
different
generally,
if
you're,
using
something
with
ingress
spec.
B
It
should
always
be
the
same
in
theory,
but
what,
if
they're,
using
straight
up
kong
gateway
they're
using
like
break
up
the
regular
gateway,
as
opposed
to
an
ingress
controller
or
if
they're,
using
some
other
method
of
getting
inbound,
then
you're
going
to
have
a
different
method
of
applying
kind
of
like
the
inbound,
the
inbound
connectivity
policy,
whereas
if
you're
using
just
the
l7
functionality,
it
would
be
consistent
regardless
of
what
inbound
gateway
they
were
using.
So
I
think
it's
like
that's
a
little
bit
of
like
an
edge
case.
B
I
think,
like
not
everyone's
going
to
have
that
path,
but
it
does
answer
a
problem
that
exists
around
like
consistent
ways
to
apply
traffic
configuration
across
environments
that
are
very
different.
So
I
think
there
is
a
use
case
for
not
using
that
config
on
the
gateway.
I
think
it's
just
a
matter
of
like
documenting
that
and
explaining
the
difference
and
explaining
when
someone
would
want
to
consume
one
versus
the
other
use
one
versus
the
other.
That's
my
just
my
two
cents.
I
could
be
totally
wrong
on
that,
but
just
a
thought.
C
No,
I
mean
I
I
agree
with
you,
I
mean
there
are
so
many
use
cases
right
and,
and
the
way
people
are
want
to
use
this
particular
gateway
are,
are
also
like
divided
into
so
many
personas,
so
a
few
of
them
would
like
to
use
gateway.
A
few
of
them
would
not
like
too.
Few
of
them
are
happy
with
india's
kind
of
functionality,
and
few
of
them
are
directly
routing
traffic
to
to,
and
they
want
to
basically
inject
the
traffic
directly
into
the
mesh
via
the
gateway.
C
B
C
Like
there
is
no
authorization
authentication
that
we
want
to
add
on
the
gateway,
but
at
least
if
we
can
add
some
kind
of
routing
functionality
we
can.
We
can
use
the
proxy
template
to
to
offload
some
of
the
authorization
on
on
the
gateway
listener
or
or
maybe
we
can
do
a
pass
through
and
then
send
it
to
the
back,
and
I
mean
there
are
so
many
permutation
combinations.
It's
very
difficult
to
talk,
talk,
talk,
taking
care
of
like
every
every
every
scenarios,
but
yeah.
A
I
think
that
last
time
we
we
we
opened
the
discussion
also
about
l7
did.
Did
anyone
had
the
chance
to
try
this
with
http
2
services,
because
at
least
looking
at
the
code,
it
should
be
applied
there?
It
should
be
working
by
default.
We
do
upgrade
http
connections,
although
they
get
terminated
so
should
be
working
from
what
I
can
tell.
But
it's
yeah
it's
interesting
to
try
that.
C
A
Okay,
great
so,
apart
from
from
the
traffic
route,
what
what
else
had
the
traffic
permissions?
This
also,
I
believe,
is
something
that
was
that
was
great,
so
zone
ingress.
Syria
put
a
lot
of
effort
into
that.
A
I
guess
that,
from
the
point
of
view
of
kubernetes,
that
probably
doesn't
doesn't
change
much
probably
for
universal.
That
would
be
mostly,
I
would
say
you
make
some
some
difference
in
the
way
that
the
resources
are
managed.
There
rate
limiting.
Maybe
it's
it's
http
only.
We
still
were
trying
to
figure
if
the
tcp
rate
limiting
makes
that
much
of
a
sense
and.
A
How
it
can
be,
I
would
say,
because
the
essentially
what
envoy
allows
is
not
rate
limiting,
for
example,
like
packets
per
second
and
things
like
that
on
the
tcp
level,
but
its
kind
of
connections
per
second.
So
if
you
have
a
long
running
tcp
connections,
I'm
not
sure
how
how
how
good
this
rate
limit
is
at
all
so
deep
in
your
environment.
C
And
and
services
are,
services
are
basically
are
waiting
on
on
saying
some
good
stories.
Like
I
mean
services
will
come
little
bit
later
on,
as
I
understand,
because,
because
it's
a
new
technology
and
they
they
are
basically
waiting
for
it
to
like
get
a
little
bit
of
matured
and
there's
a
lot
of
sla
and
a
lot
of
it
goes
on
top
of
these
services.
So
so
that
is
why
there
is
a
little
bit
of
lag
there,
but
application
teams
are
are
the
ones
who
who
who
will
basically,
who
are
basically
using
this
particular.
A
Okay,
good
with
with
the
interruption
that
we
have,
maybe
maybe
we
can-
we
can
have
like
10
minutes
more
so
are
there
any?
Are
there
any
questions
so
austin
on
youtube
that
you
want
to
bring
forward.
D
My
only
update
is
that
the
pr
for
prometheus
is
updated.
The
maintainer
said
he's
going
to
review
asap
going
from
there.
D
If
it
does
take
significantly
longer,
we
can
the
http
sd
was
released
in
the
last
version
of
prometheus.
So
that's
another
option.
We
could
expose
austin.
C
I
wanted
to
ask
you
something
so
a
short
question
right:
is
it
possible
for
us
to
like
export
it
as
stats
t,
instead
of
in
prometheus
format,.
D
It's
over
http
yeah,
I
mean
statsd
yeah,
the
you're
talking
about
the
format
of
the
scraped
metrics.
Are
you
talking
about
the?
What?
What
I'm
doing
with
prometheus
is
just
exposing
the
targets
to
scrape
I'm
not
sure
about
what
it
would
take
to
expose
a.
C
Statsd
format
yeah
on
on
y
can
actually
expose
in
in
two
formats.
One
is
in
prometheus
and
one
is
in
stats
d
format
in
star
c
format.
It
actually
pushes
it
to
the
to
the
ip
address
that
you
have
configured.
D
That's
interesting,
I
don't
know
if
the
top
of
my
head
would
be
needed:
okay,
but
yeah.
Sorry,
okay,.
A
Okay,
good,
okay,
then,
then,
then,
we'll
just
wrap
up
here.
Okay,
thank
you
see
you
bye.