►
From YouTube: Kuma Community Call - January 6, 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
A
Yeah,
so
I
think
here
john,
I
don't
think
that
I
have
seen
you
before.
Are
you
the
person
that
wanted
to
implement
bravely
the
l7
routing
thing.
B
No,
I
don't.
I
don't.
B
Me
no,
I
posted
a
question
there
just
before,
just
at
the
end
of
the
year
about
circuit
breakers
and
jacob
had
given
me
a
good
response
to
that.
So
I
just
thought:
I'd
join
in.
B
It
was
it
wasn't,
it
wasn't
an
issue.
It
was
more
of
a
question
that
I
raised,
but
yeah
it
wasn't
the
issues
yet,
but
so
I
thought
I'd
join
in.
I
think
it's
13
24.
A
A
A
B
A
A
Great
feedback
from
lots
of
people
like
we
have
a
lot
of
things
to
to
to
improve
people,
are
constantly
hitting
some
corner
cases
that
we
are
not
covering
in
our.
A
A
A
Okay,
so
essentially
the
last
thing
that
we
did
last
year
was
104.
I
was
just
just
going
to
the
things
that
we
have
accumulated
already
in
the
release
branch.
I
don't
think
that
we
are
going
to
release
anything
this
week.
A
We
already
see
some
some
novelties
there
and
I
think,
can
you
remind
me
today
we
were
comparing
the
between
victory
of
something.
Was
it
the
logging
of
the
tracing?
I
don't
remember,
but
I
mean
okay,
there
are.
There
are
lots
of
interesting
things
in
the
in
the
new
api
that
are,
of
course,
people
are
going
to
evolve
and
they
even
have
before
alpha
or
something
going
on.
So
I
mean
obviously
v2
is,
is
something
that
is
going
to
do
to
the
obsolete
okay.
It
should
be
absolutely
by
now
I
mean
the
next
versions.
A
I
think
one
seven
117
of
them
only
on
v3.
Why
I'm
saying
here
this
because
this
is
like
the
the
switch
between
v2
and
v3,
is
what
bounces
to
the
next?
Like?
I
don't
know
if
you
should
call
it
major
release,
but
let's
say
one
one:
1.1
release
of
comma,
so
this
is.
This
will
be
the
the
turning
point
where
I
actually
switch
switch
to
the
support
this
in
general,
from
what
I
can
tell
this
shouldn't
be
affecting.
A
A
Some
people,
like
a
lot
of
people,
would
just
like
to
do
this
like
this,
so
we
would
respect
this,
especially
like
specifically
at
this
point
like
this
point,
switching
the
apis
of
the
envoy
shouldn't
be
showing
that.
A
But
this
is.
This
is
essentially
like
the
cornerstone
that
we
have
put
for
the
new
for
the
new
release
of
milestone.
So
it's
better
wording.
So
once
we
finalize
this,
we'll
probably
start
looking
at
1.1
and
introduce
this
is
a
breaking
change.
A
As
I
said,
I'm
not
getting
this
to
be
so
much
of
a
breaking
chance.
Yet
switching
on
apis
is
big
enough
other
than
that
yeah
again,
we
are
continuing
our
strive
to
move
from
sandboxing
to
incubation
stage
in
the
cnc,
which
is
somehow
bound
to
having
successful
deployment.
So
we
are
looking
for
use
cases.
So
if
any
one
of
you
has
something
that
we
can,
we
can
like
draft
together
into
into
some
marco.
I
don't
remember
the
format
should
it
be
like.
A
B
Yeah
I
mean
I
would
be
happy
to
you
know
we're
I'm
still
evaluating
kuma
for
for
our
company,
but
everything
is
looking
really
good
so
far
that
I'd
be
happy
to
participate
in
that
once
we
we
start
the
rollout
for
that.
I
wouldn't
anticipate
us,
though,
being
probably
rolling
out
in
a
major
way,
probably
till
maybe
march
ish.
It
kind
of
depends
on
a
few
other
things,
but
as
we
get
there,
I'm
more
than
happy
to
contribute.
B
Probably
to
be
several
hundred
data
planes
or
more,
our
two
non-production
clusters
will
probably
be
in
the
realm
of
two
probably
three
to
five
hundred
data
planes
before
production,
and
then
a
production
will
be
five
to
ten
times
larger
than
that.
A
Yeah
I
mean
I,
I
think
that
even
even
like
staging
environments,
these
are
still
still
something
of
interest
for
such
use,
cases
that
we
can
present.
So
so
that's
good
enough
and
the
the
amount
that
I
mean,
the
numbers
that
you're
quoting
here
are
impressive.
You
know.
B
A
We're
having
a
kind
of
an
open,
open
discussion,
we
don't
have
any
specific
agenda
just
like
in
a
couple
of
days,
just
think
about
generic
things
around
comma.
What
are
our
plans.
C
Yeah,
I
was,
I
saw
the
new
updates
from
kuma,
I
mean
one
was
the
easter,
I
mean
on
the
vm,
you
have
you
have
something
like
what
is
it
called
non-intrusive
way
of
using
the
proxy
or
the
kuma
dp?
C
So
have
you
already
discussed
that
or
maybe
I
can
get
some
idea
from
you
guys
as
in
how
do
you,
what
are
the
things
that
are
to
be
done?
If
we
have
to
basically
introduce.
A
We
have
a
date
that
the
documentation
you
should
be
able
to
see
these
in
the
dedicated
section
here,
so
it
has
some
guidelines
of
what
you
can
and
you
want
to
do,
but
so
it
really
depends
on
the
the
environment,
so
this
has
been
tested
in,
let's
say
in
ubuntu
in
some
cases
and
it
works,
I
mean
it's,
it's
part
of
our
end-to-end
testing
in
our
cia,
so
it's
continuously
tested
so,
but
this
is
again
still
like
repeating
one
at
the
same
test.
A
So,
for
example,
I
haven't
checked
this
in
centos
or
something
like
this,
so
there
might
be
some
variations
here
and
there,
but
I
think
that
the
larger
part
of
this
is
this.
Here
is
what
what
it
is.
So
I
would
suggest
that
you
go
and
read
what
we
have
here.
Maybe
try
it
out
and
see
how
this
works
for
you
and
then
come
with
questions.
I'm
sure
there
will
be
some
some
things.
C
Yeah,
I
I
didn't-
I
should
have
seen
here.
First
actually,
sorry
yeah,
let
me
look
at
it
and
then
come
back,
but
it
would
need
some
privileges
right
because
it
needs
to
install
the
ib
table
related
stuff.
As
I
see.
A
Yes,
but
this
is,
this
is
the
one,
so
you
should
ruin
this
as
root
right.
Okay,
so
one
time
thing
like
once
you
once
you
run
this
on
the
hose
as
root,
then
then
the
host
mine,
like
the
the
rules,
are
set
there.
They
will
be,
you
know
they
are
saved
and
they
will
be
very
good,
so,
okay,
perfect
and
then,
of
course,
your
data
plane
proxy
is
like
instead
of
outbounds,
you
can
you
not
only
like
you
can
just
use
the
transparent
proximity
sections
now.
A
C
Got
it
thank
you
thanks,
and
I
have
a
specific
scenario
that
I
wanted
to
discuss.
C
So
let
me
talk
more
about
that.
So
what
what
I
was
trying
to
do
was
to
basically
have
a
front
proxy
and
have
couple
of
route
rules
to
route.
I
mean
the
front
proxy
will
behave
like
a
gateway.
B
C
Terminate
the
connection
and
it
will
use
the
route
rules
and
it
will
pass
it
to
the
back
end
application.
So
as
I
know
that
kumar
does
not
support
the
routing
rules.
So
what
what
I
did
along
with
my
colleague
I
mean
what
we
did
is.
Basically
we
we
used
the
proxy
template,
construct
and
and
added
all
the
route
rules
that
we
that
we
are
supposed
to
do.
But
after
that,
what
we
see
is
in
the
proxy
template.
We
are
using
the
service
name,
the
backend
service
name,
but
that
is
not
getting
resolved.
C
Whereas
if
I
use
the
same
configuration
in
a
static
mode,
then
then
then
the
back
end
gets
resolved.
Doesn't
then
there's
no
back
end?
Then
we
have
to
give
the
ip
address.
So
then
that
works.
But
let's
say
from
kuma
we
configured
the
gateway
kind
of
a
construct,
and
then
we
add
proxy
template
to
put
the
routes
so
that
it
can.
C
A
I
guess
I
guess
it
would
be
better
if,
if
you
post
on
this,
like
the
specific
or
at
least
so
you
just
see,
but
is
this,
is
this
universal
mode
or
kubernetes?
Yes,
it's
universal,
so
are.
Are
you
setting
like
domain
name
like
fully
fqdn,
like
I
mean
like
something
that
can
be
resolved
to
dns
or
what.
C
No,
no,
no
so
we
have
defined
the
services
as
we
defined
in
kuma,
so
we
have
a
service
called
backend.
Let's
say
back
and
one
back
and
two
back
and
forth.
C
And
what
we
are
doing
is
we
are
using
proxy
template
to
route
the
incoming
connection
to
all
of
these
backend
depending
on
the
path.
So,
let's
say
slash:
backend
one
should
go
to
service
backend
one.
So
when
we
configure
this
particular
configuration
via
proxy
template,
the
backend
one
back
in
two
are
not
getting
resolved.
C
So
if
you
want,
I
can
draw
it
out
and
then
pass
it
to
you
guys
over
a
slack
with
a
diagram.
A
But
I
mean
I
don't
know
if
you
have
followed
slack,
but
there
was
someone
I
was
expecting.
A
A
Actually
wanted
to
start
working
on
layer,
7
routing
and
we
even
had
a
small
like
design
discussion
there,
so
we
expect
in
the
next
in
the
months
to
come
that
this
is
to
to
happen.
I
mean
I'm
not
expecting
this
to
be
a
this
month,
but
probably
next
by
the
end
of
next
month.
Maybe
we
will
have
something
pleasing,
so
I
mean
I'm
just
just
saying
that
this
is
this
is
coming.
There
is
interest
from
the
community
so
probably,
or
we
will
start
pushing
into
this-
also.
C
Sure
sure,
okay,
do
you
have
a
link
to
the
the
google
doc
which
you
are
updating?
Have
you
pasted
it
to
the
chat.
C
C
B
B
Just
just
something
you
mentioned
earlier
on
about
some
something
going
from
version:
two
to
version
three
yeah.
What
what
was
that.
A
This
is
the
envoy
api.
So
essentially,
when
you
open
the
api
here,
you
have
api,
v2
and
v3
okay.
So
today,
yes,
our
whole
configuration
is
essentially
using
this.
This
version
of
the
api,
which
is
getting
obsoleted.
A
It
was
supposed
to
be
like
actually
not
deprecated,
but
I
think
that
they
they
will
stop
the
support
for
it
in
the
next
releases
of
the
framework,
although
I
don't
know
it
stays
here,
but
for
sure,
if
you
look
in
the
locks
of
our
current
deployments
in
the
locks
of
envoy
no
matter
if
it's
universal
or
kubernetes,
you
see
that
it
says
a
message:
okay,
this
is
v2.
It's
going
to
be.
You
know,
obsoleted
by
the
end
of
2020,
so
please
migrate
to
v3,
et
cetera,
et
cetera,
so
yeah.
A
We
are
moving
to
this
new
one
and,
as
I
said,
it
has
some
some
advantages.
Some
novelties
new
things
coming
so.
A
So
we
we
want
to
have
at
least
one
one,
one
or
two
minor
releases
before
we
are
confident
enough
that
you
know
we
can
switch
entirely
to
default
to
v3.
So
we
will
keep
both
for
some
time
so
that
you'll
be
able
to
switch
back
and
forth
in
case
you
need
it
yeah.
That's
it.
A
Okay,
so
austin
regarding
the
helm,
charts
we
we
have
hit
again
with
user.
We
have
hit
a
problem
with
the
upgrades
of
the
service
again
so
yeah.
We
are
kind
of
taking
very
serious
efforts
in
this.
We
are
probably
going
to
do
the
upgrade
hook.
B
Yeah
I
know
I
I
owe
you
an
an
answer
on
that
yeah,
do
you
have
you
been
talking
about
it
internally
about
about
a
strategy.
A
Yeah
I
mean
it's
it's
more
or
less
whatever.
We
have
discussed
the
last
time
that
this
topic
was
open
here,
so
essentially
yeah.
A
I
think
that
we
have
found
some
some
very
huge
yeah,
so
this
is
okay.
I
will
post
it
on
the
chat
too.
So
this
is
a
document
that
is
part
of
some
pr
hearing
hill,
but
it's
rather
a
long
document
explaining
the
relations
between
him
and
sincere
this,
and
why
is
this,
and
why
is
that
and
then,
in
the
end
they
talk
about
workarounds.
B
A
Oh
yeah
I
mean
like,
but
probably
you
should
announce
it
on
the
slide.
So
if
someone
wants
to
join.
B
Yeah
yeah
yeah,
I'm
happy
to
set
that
up
if
you'd
like
and
try
and
get
some.
I
don't
know
pr
work
in
progress
out.
B
Okay,
if,
if
I,
if
I
don't
have
time
I'll,
let
you
know
so
you're
not
just
waiting
on
me,
like
usual.
B
A
Good,
we
are
five
minutes
to
the
end
of
the
call.
Is
there
anything
else
that.
C
Do
you
have
any
documentation
on
how
we
can
use
a
global
control
plane
on
both
gkp
and
universal
mode
together?
Do
you
have
a
example,
or
something
like
that
in
your
documentation,.
A
You
you
mean
like
hybrid
well,
okay,
so
we
call
this
hybrid
deployment.
You
mean
like,
like
you,
have
single
global
control
plane
and
then
you
have
zones
in
like
couple
of
zones
in
kubernetes
and
couple
of
zones
in
universal.
Is
this
what
you
mean.
A
Yeah
yeah,
I
don't
know
how
explicit
are
those,
but
in
general
there
shouldn't
be
that
much
of
like
the
moment
that
you
start
deploying
it.
You
just
follow
the
different
parts
here,
like.
A
Okay,
like
when
you
want
to
deploy
global
control
plane,
you
choose
what
you
want
to
do,
then
for
the
remotes.
You
can
deploy
just
one
remote
in
this
one
remote
with
this.
A
Which
is
not
the
kubernetes
english
is
like
the
kuma
ingress.
It
is
essentially
a
separate
data
plane
that
is
interfacing,
essentially
exposing
the
services
running
within
the
zone.
It's
exposing
them
to
the
other
zones
for
for
the
other
zone
to
be
able
to
consume.
Okay.
C
C
Two
ingresses
one
for
kubernetes
and
one
for
the
the
global
one
for
kubernetes
and
one
for
universal
mode,
or
only
one.
A
No,
no
every
zone
and
by
zone
is
like
in
kubernetes.
It
would
be
a
single
kubernetes
cluster,
but
in
universal
it
would
be
just
a
set,
of
course,
button
data
planes
which
are
connecting
to
the
same
comma
cp.
So
that
would
be
as
wrong
right.
So
I.
C
Think
I
understand
the
zone
and
the
index
concept
I
just
wanted
to.
What
I
want
to
know
is
like
on
the
on
the
gkp
side.
Do
we
also
need
to
install
another
ingress
or
not
needed,
not
required?
Yes,
okay,.
A
Yeah,
so
that's
what
I
I'm
I'm
showing.
I
think
that
my
screen
is
still
shared,
so
when
you
deploy
universal.
This
is
this
step
which
actually
describes
how
to
deploy
the
ingress
and
on
kubernetes.
C
A
Okay,
so
we
are
at
the
top
of
the
hour.
Thank
you
all
have
a
successful
around
here
and
I'm
from
bulgaria.
So
in
bulgaria
we
say
have
a
successful
year,
so
I
don't
know
if
this
is
common
to
other
countries,
but
I
guess
that
it
is
so
good
luck.
Everyone.