►
From YouTube: Kuma Community Call - June 9, 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
Okay,
good
so
hello
and
welcome
to
the
bi-weekly
community
call
here
are
the
meeting
minutes.
They
should
be
in
the
invitation-
and
please
add
your
names
here
so
that
we
have
a
record
of
who's
visiting
and
for
today.
My
main
topic
that
I
have
prepared
well
is.
A
A
Adding
for
for
this
release-
and
I
will
start
with
that-
and
if
there
are
any
other
questions
or
questions
about
the
features
yeah
we
can,
we
can
discuss
and
chat
and
and
see,
see
how
we'll
go
okay.
So
essentially
the
plan
is
next
monday.
We
have
an
rc
one
version.
This
is
our
new.
Let's
say:
release
schedule
in
works.
We
are.
We
are
starting
to
do
this
or
before,
especially
on
the
major
releases.
A
We
are
going
to
do
an
rc
version
on
monday
and
then,
if
everything
goes,
fine,
no
major
issues
found
in
the
rc1.
We
are
going
to
tag
the
version,
the
real
release
on
wednesday,
it's
mostly
european
european
times
when
we
say
this
so.
A
Unfortunately,
jacob
is
not
he's
not
here,
but
he
was
the
domain
driver
and
the
person
that
actually
implemented
engineer
that
implemented
this
feature.
If
anyone
is,
I
would
say
curious
about
what
it
is.
We
have
a
preliminary
like
a
preview
of
the
documentation
site
with
everything
about
l7
at
least
integrated
here.
A
So
there
is
a
full
example.
Essentially
it's
a
modification
of
the
traffic
cloud
policy,
an
amendment
on
top
of
it.
It
extends
it.
We
have
a
new
http
session
section
and
it
has
all
these
fields
here
that
you
can
look.
This
is
the
the
full
the
for
example.
There
are
also
like
smaller
smaller
examples,
with
a
little
bit
more
explanation
that
that
people
can
follow
easily
when
they,
when
they
want
to
to
go
and
look
into
this.
A
So
there
is
a
seven
traffic
split,
there's
modifications
that
can
be
done
upon
routing
you
can
reroute
and
etc.
A
So
this
is
about
the
l7.
I
know
that,
like
when
browsing
documentation,
it
just
seems
a
couple
of
pages,
but
I
believe
that
this
is
moving
kumar
as
a
project
tool
to
a
whole
new
level.
Like
I
mean
it
opens
a
lot
of
new
use
cases
and
possibilities
with
it.
We
have
been
chatting
about
this
l7
for
a
long
time,
as
I
already
said,
and
I
I
think
that
everyone
will
be
happy
that
this
is
finally
coming
to
to
a
release.
A
The
next
two
things
are
about.
A
B
There
are
maybe
few
because,
as
we,
there
is
like
a
remote
control,
plane
is
used
inside
few
config
names,
so
some
properties
will
change
and
unfortunately
it
means
that
it's
not
backward
compatible
and
there
will
be
potential
problems
for
I'm
kind
of
working
on
right
now
on
a
guide
to
how
to
upgrade
it
using
him
and
like
our
cattle.
A
Yeah
and
we
are
providing
a
guide
to
actually
lead
you
to
through
this
process
upon
when,
when,
when
you
go
on
the
path
on
upgrading
this
and
the
the
other
one
is
zone
ingress
which,
like
we
had
a
lot
of
a
lot
of
conversations
and
discussions,
people
initially,
especially
in
the
kubernetes
world,
we
should
people
get
confused
about.
A
Okay,
is
this
the
kubernetes
ingress
or
something
else?
Is
it
a
data
plane?
Is
it
essentially
what
it
is
in
the
end,
and
it
takes
some
some
some
time
to
explain
and
we
we?
We
are
hoping
that,
with
this
with
this
move
that
we
do
now,
the
messaging
is
going
to
to
become
much
much
clearer
will
be
easier
to
maintain,
understand
and
again,
of
course,
maintaining
production,
etc.
A
So,
essentially,
it
is
not
only
about
renaming,
but
we
also
add
an
additional
zone,
ingress
resource
to
actually
describe
the
the
notion
of
having
zoning
in
the
end.
So
the
point
is
that
up
till
now
this
is
this
was
especially
visible
on
universal.
When
someone
deploys
a
new
zone,
they
had
to
deploy
this
special
data
plane
resource,
which
is
which
has
this
special
ingress
section
and
etc,
etc.
Now
all
this
is
going
away,
actually
it
stays
for
backward
compatibility,
but
we
are
also
adding
this
zone
ingress
resource.
A
C
A
B
A
No,
that's
that's
power
is
it.
We
are
also
facilitating
one
of
the
one
more
of
the.
A
Core
functionalities
of
envoy
this
is
the
rate
limiting.
So
we
are
we're
putting
a
new
policy
corporate
limit
policy.
We
try
to
simplify
a
little
bit
the
configuration
that
envoy
provides.
So
I
hope
this
will
be
useful.
I
know
that
some
people
in
the
community
have
tried
to
do
this
with
proxy
templates
and
we
have
seen
it
working
and
it's
it
is
working,
but
I
guess
that
with
this
this,
this
new
policy
that
essentially
is
building
upon
invoice
local
rate
limiting.
A
We
are
getting
like
people's
lives
better
and
much
easier
to
to
to
maintain
when
I
set
local
rate
limit.
Essentially
envoy
has
two
two
options:
two
to
do
rate
limiting
one
is
called
local
rate
limit,
which
essentially
is
kind
of
isolated
per
data
plane
or
per
ranvo
instance
rate
limiting
and
the
other
one
is
global.
A
We
have
looked
into
the
global
one,
which
essentially
means
that
your
control
plane
can
view
all
the
all
the
instances
of
the
particular
proxies
and
can
actually
control
the
the
rate
limiting
in
effect,
so
that
you
can
have
an
accumulated
rate
limiting
like.
If
you
set
100
requests
per
second,
it
will
distribute
you
to
all
the,
for
example.
A
If
you
have
three
instances
of
the
data
plan,
it
will
distribute
it
in
a
way
so
that
all
the
three
instances
will
effectively
handle
only
300
requests
now,
with
local
rate
limiting
is
not
so
so
what
we
are
applying
is
actually
a
policy
that
will
be
applied
per
instance.
A
So
if
you
have
a
single
service
with
three
instances
and
you
set
100
requests
per
per
minute-
let's
say
okay
per
second,
but
per
per
minute,
then
essentially,
each
of
the
instances
will
handle
300
requests
per
minute,
so
in
total
it
will
be
300
requests
per
minute
for
the
for
the
overall
service.
This
this
will
work,
I
believe,
from
from
our
researches
and
thoughts.
This
will
work
much
better
and
more,
it's
much
easier
to
explain
and
to
maintain,
especially
across.
A
So
that's
why,
for
the
time
being
we
we
we
put
global
rate
limiting
aside
and
we're
not
looking
into
it.
This
is
for
the
rate
limiting
proxy.
There
are
two
more
let's
say:
one
can
think
of
kind
of
custom
cosmetics,
but
you
know
they
can
upgrade.
They
can
break
your
upgrades.
So
it's
essential
to
also
also
be
noted.
These
these
things
will
lend
also
all
of
these
relent
in
the
change
log.
There
will
be
an
upgrade
md,
so
all
this
will
be
described
there,
but,
okay,
we
are
discussing
this
here.
A
So
one
of
the
one
of
the
other
things
is
we're
going
to
default
to
envoy
built
in
dns.
We
introduced
this
a
couple
of
minor
versions
back
as
kind
of
a
optional
feature.
A
Essentially,
it
allows
us
to
intercept
the
way
that
we
do
it
is
we
intercept
all
the
dns
requests
that
that
are
going
from
the
workload
directly
and
instead
of
letting
the
workload
to
consult
with
the
comma
cp
about
the
naming
of
the
services,
etc,
etc.
Everything
is
happening
locally,
so
kumo
cp
still
still
has
the
global
overview
of
all
the
mappings
of
names
to
ip
addresses,
but
the
actual
dns
serving
dns
server
execution
is
happening
on
the
data
plane
proxy
on
the
android
that
sits
next
to
to
your
workload.
A
This
allows
to,
for
example,
if
the
common
cp
is
having
some
outages-
or
you
know
is,
is
is
getting
carrier
start
your
workloads.
You
still
be
able
to
do
the
resolving
because
and
why
we
do
the
resolving
for
you
locally
and
also
it
allowed
us
why
we
introduced
it.
It
allowed
us
to
to
be
able
to
address
external
services
by
their
real
name.
A
So,
for
example,
if
you
want
to
consume
salesforce.com,
you
know
the
their
api
from
within
within
your
mesh
as
an
external
service,
you
can
still
call
it
cells,
salesforce.com
and
you
will
just
be
able
to
to
access
it.
So
we
are
making
this
the
the
default
behavior.
So
the
familiar
kumakato
install
dns,
that
was
that
causes
actually
many
troubles,
because
people
are
usually
kind
of
skipping
this.
This
step,
it's
just
a
single
line,
so
sometimes
people
don't
don't
see
it
and
it
caused
a
lot
of
pain.
A
So
we
believe
that
switching
to
this
new
new
mode,
which
will
be
automatically
enabled
once
you
deploy,
unless
you
you,
you
don't
want
it
and
you
you
will
exclude
it,
you
will
kind
of
disable
it,
but
with
this
automatic
motor,
everything
will
just
work
out
of
the
both
no
additional
steps
needed.
Also.
We
know
that
on
some
clouds,
like
asia,
we
had
problems
with
in
the
aks
modifying
the
core
dns
to
redirect
dns
requests.
A
A
A
It's
essentially
dropping
the
whole
support
for
ipv2,
so
you
cannot
do
api
v2
anymore.
We
are
dropping
it
in
in
this
version.
As
a
support
on
the
control
plane
side,
we
will
still
keep
our
envoy
117
for
now.
Maybe
we
upgrade
down
the
path,
and
why
is
this
important?
If
you
have
a
long
running
work,
workloads
data
planes
that
are
still
running
the
data
plane
from
1.0,
something
like
previous
to
1.1?
A
A
B
B
It
would
be
nice
to
get
in
for
the
release,
because,
if
not,
it
would
there's
a
configuration
option
that
I'd
like
to
make
breaking
before.
Okay
release.
A
B
Yeah,
okay,
but
I've
been
talking
with
the
right,
yeah
yeah,
it's
it's
very
close,
but
yeah.
If
it's
released
in
one
point
to
on
the
kuma
side,
then
we
can
write
a
blog
post
and
say
it's
supported
from
kuma
1.2
onwards.
Okay,
okay,
makes
sense,
makes
sense.
A
Yes,
I
see
this.
This
actually
gets
in
a
better
position
from
the
premier's
community,
so
we
can
say:
okay
now
you
have
a
release
that
actually
supports
this
yeah.
Would
you
integrate
our
patches?
Okay,.
B
Exactly
I
think,
there's
only
one
outstanding
question
I
have
with
the
envoy
community
about
how
to
specify
along
because
they
they
say
in
their
docs
that
long
polling
is
supported,
but
they
don't
have
any
spec
really
about
it
so
having
to
choose
how
the
client
sends
the
the
timeout
parameter.
They
haven't
answered
me.
So
I
don't
know
if
we
should
just
do
like
a
query,
parameter
or
a
header
or
something
like
that.
But
that's
in
the
pr.
B
A
B
A
Test
great
okay,
anything.
C
Else
all
the
updates
that
you
gave
were
like
amazing,
actually
very
happy
to
see
l7
which
got
merged
v2,
getting
deprecated,
dns
rate
limit,
all
all
great
zones,
even
the
max
pr
right.
They
were
all
like
pending
for
a
very
long
time
and
then
thanks
for
getting
them
done
actually.
A
Yeah,
if
you
want
to
compile
from
master
of
course,
otherwise
on
monday
we
have
rc1,
you
have
the
link
to
the
docs,
I
mean
it's
it's
already
merged.
I
don't
think
that
there's
going
to
be
any
any
changes,
at
least
we
don't
see
anything
now.
C
Okay,
super
I'll
I'll,
actually
pick
it
up
that
if
you're
cutting
your
rcm
india,
you
you
mentioned
that
I
have
a
very
small
pr
which
is
about
the
advertise
address.
If,
if
even
that
can
be
taken
care
of,
I
mean
if
there
are
any
comments,
maybe
I
can
take
care
of
that.
A
Six:
okay,
let's
play
use
of
it;
okay,
so
if
you.
A
A
Marco,
this
is
changing
the
data
plane
resource.
You
want
to
also
peek
at
this
and
hear
your
opinion
like,
at
least
from
the
configuration
point
of
view.
Just
essentially
is
adding
public
router
boundaries
for
the
data
plane.
It's
something
that,
like
we
have
in
the
zone,
ingress
now.
B
A
Yes,
this
is
this:
is
the
advertise
address
for
the
data
plane
itself
because
apparently,
sudeep2
needs
this.
C
I
can
explain
the
use
case,
so
there
are
use
cases
wherein
the
data
planes
are
running
behind
behind
an
at
or
behind
an
overlay
network,
so
because
of
which
other
database
cannot
reach
this.
C
These
data
planes,
which
are
running
in
the
world
in
network
and
and
it
needs
additional
ip
addresses,
so
that
so
that
when,
when
the
other
data
planes
are
trying
to
connect,
it
will
actually
connect
to
this
advertised
address
and
then
and
then
there
will
be
rooting
done
to
make
sure
that
it
actually
reaches
the
right
data
pane
behind.
B
A
Good
anything
else
for
the
release
or
any
other
topic.
We
have
five
minutes
left.
C
Do
you
also
have
any
plans
to
like
use
the
the
gateway
data
plane
as
as
a
gateway
as
in
it
should
not
without
having
a
I
mean
that
it,
the
gateway
can
can
can
receive
all
the
all
the
ingresses?
C
C
Two
quarters,
and
do
you
also
have
plans
to
terminate
tls
if
you
can,
or
at
least
do
some
kind
of
validation,
as
in
today
the
gateway
accepts
plain
text:
traffic
right?
Can
it
accept
tls
traffic.
C
C
So
essentially,
what
is
happening
is
the
gateway.
Data
plane
is
additional
additional
layer
which,
which
is
getting
introduced
into
into
applications
so
trying
to
see.
If,
if,
if
we
can,
I
mean
if
we
can
get
rid
of
that
and
and
if,
if
the
gateway
can
help
us
do
some
kind
of
traffic
traffic
steering
or
something
like
that,
it'll
be
really
good.
Yeah.
A
I
don't
know
I,
I
guess
we'll
see,
we'll
see
down
down
down
the
path
when
we
start
working
on
this.
But
that's
that's
a
good
good
feedback
and
yeah.
B
A
C
A
Put
any
final
questions:
suggestions:
okay,
there's!
No!
Let's
wrap
up
here,
be
excited
about
the
new
release
next
week
and
we'll
see
you
in
two
weeks,
bye.