►
From YouTube: Layer5 Community Meeting (Dec 03rd, 2021)
Description
Layer5 Community Meeting - December 3rd, 2021
Join the community at https://layer5.io/community
Find Layer5 on:
GitHub: https://github.com/layer5io
Twitter: https://twitter.com/layer5
LinkedIn: https://www.linkedin.com/company/layer5
Docker Hub: https://hub.docker.com/u/layer5/
A
So
welcome
everyone
to
the
larify
community
meeting
today
is
the
3rd
of
december
and
we
have
some
important
announcements
to
kick
off
this
meeting.
But
before
going
to
that,
let
me
welcome
all
the
newcomers
who
joined
us
to
join
the
community
this
week
and.
A
So
our
shield
has
been
here
for
some
time
and
I
don't
think
I
see
any
other
newcomers
on
the
participants
list
today.
So,
let's
get
on
with
the
announcements.
So
I
have
I
open
the
floor
to
need
to
make
the
announcements
solely.
B
Yep,
hey,
hey
very
briefly:
deepa
yawn,
have
you
been
on
the
community
call
before
just
a
double
check.
B
B
Good,
you
have
to
give
me
some
grace,
since
it's
only
your
second
all
right,
good,
good,
good,
good
you're,
starting
to
develop
a
pattern.
That's
patterns
is
something
that
we
talk
about
in
this
community.
B
There's
we're
going
to
try
to
create
a
catalog
of
service
mesh
patterns,
so
we'll
actually
have
something
to
talk
about
at
cubecon,
china
that
about
service
mesh
patterns.
So
I
won't
spoil
this
prize
but
rather
say
that
it's
very
nice
to
have
everyone
on
the
call.
This
is
the
latter
five
community
call
we
meet
once
a
week
and
gosh.
What
do
we
do?
We
do
a
number
of
things
on
this
call
like
the
community
call
is
where
we
talk
about.
Well,
almost
anything
goes.
B
Sometimes
we
invite
other
other
open
source
projects
or
other
just
engineers
or
technologists
to
come
and
speak
to
learn
about
technologies
that
they're
they're
using
this
community
has
a
real
theme
of
learning.
Hence
why
we
invite
people
to
come
talk
sometimes.
Hence
why
you
find
me
going
off
on
a
massive
tangent
to
talk
about
something
that
I've
just
learned
or
something
that
I
think
that
might
be
valuable
for
everyone
else
to
to
learn.
B
So,
as
you
think
about
it,
the
agenda
is
driven
by
all
of
you.
Barak
is
driving,
probably
half
of
today's
agenda,
I'm
just
guessing,
which
is
going
to
be
great
there's.
I
think,
there's
a
number
of
things
to
learn
from
what
he's
going
to
go
through.
B
But
the
point
of
saying
this
is
that
you
all
should
feel
free
to
drop
agenda
items
onto
the
meeting.
Some
of
that
can
be
about
learning.
B
Some
of
that
can
be
about
suggestions
on
projects,
suggestions
for
new
projects,
some
that
can
just
be
questions
about
some
of
the
goals
that
the
projects
have
or
or
or
things
to
do
inside
of
the
community
and
we'll
have
we'll
talk
a
little
bit
about
things
to
do
inside
the
community
and
what
and
that's
actually
with
starting
with
an
announcement
that
novendu
was
trying
to
get
me
to
say
in
the
first
place,
which
is
we
have
some
new
community
managers?
B
There's
a
few
roles
that
people
get
to
play
in
the
in
our
broader
community,
some
of
that's
contributing
by
way
of
code,
someone,
that's
contributing
by
way
of
any
number
of
things,
someone
that's
contributing
by
way
of
like
helping
steward
the
well.
How
do
I
not
use
the
word
community
and
helping
make
sure
everybody
has
a
good
time
here
understands
what's
going
on,
where
the
resources
are
who's?
B
Who
who
to
talk
to
get
an
introduction,
a
quick
overview
of
the
projects
kind
of
as
you're,
particularly
as
anyone's
trying
to
set
up
their
development
environment
and
do
their
first
successful
build
of
the
projects
like
community
managers?
Do
any
number
of
things
and
there's
three
of
them
that
I
wanted
to
highlight
today,
all
of
whom's
names
you
might
recognize
so
deepak
gupta
or
mr
wick.
B
As
he's
fondly
referred
to
john
wick,
for
those
of
you
who
haven't
seen
that
movie
deepak
is
joining
us,
the
guy
has
been
helping
a
lot
of
people
and
we
can
only
encourage
and
support
that.
It's
really
appreciated.
We
we
met
as
a
group
with
bami
who's,
also
taking
on
this
role
formally
as
well,
and
with
devil,
priya
and
and
there's
a
collection
of
existing
community
managers
that
all
met.
B
We
kind
of
went
through
some
recent
training
about
how
the
about
the
ins
and
outs
of
the
culture
of
the
community,
some
of
the
tooling,
that
we
use
to
try
to
make
sure
that
things
flow
smoothly.
It
may
not
look
like
it,
but
there's
a
fair
bit
of
thought
that
goes
into
trying
to
make
sure
that
it's
more
than
just
an
empty
slack
or
an
empty
discussion
forum,
or
that
the
projects
aren't
well
organized
or
people,
don't
know
what
they're
supposed
to
do.
What
they're
allowed
to
do?
B
Can
anyone
here
just
come
and
open
up
a
github
issue
when
they
do
should
they
start
working
on
it?
Could
they
assign
it
to
someone
else
like
what
are
the
what's
kosher
and,
what's
you
know
what
what's
encouraged
and
what's
sort
of
discouraged?
B
What's
what's
really
helpful
and
what's
not,
and
so
we
went
through
a
lot
of
these
things,
a
lot
of
that's
just
self-evident
if
you
hang
around
and
you
just
sort
of
watch
some
of
the
interactions
that
are
going
on,
you
kind
of
you
kind
of
get
that,
but
these
three
individuals
have
been
dedicating
quite
a
bit
of
time.
We
want
to
recognize
their
efforts
and
empower
them
even
more
and
so
to
put
two
of
them
that
are
on
the
call
on
the
spot,
double
priya
and
deepak.
B
I'm
gonna
make
two
suggestions.
Actually
it's
interesting.
Both
devil,
priya
and
deepak
are
doing
some
writing
what
yes
non-code!
Writing
if
you
haven't
seen
it
and
you
might
not
have
because
it's
new,
if
you
went
to
layer,
five
dot,
io,
slash
resources.
B
We
have
a
ton
of
info
about
the
tech
that
gets
used
here
that
I
was
like,
for
example,
on
wednesday,
a
principal
engineer
at
cisco
had
asked
to
meet
privately.
B
Well,
if
you
go
to
layer,
five
dot,
slash
something
something
like
it'll
answer
that
question
for
you
like,
like
there's
a
lot
of
resources
out
here,
come
see
them.
One
of
it
was
like
go.
Try
the
interactive
labs
like
they're,
free,
they're,
curated
they'll,
get
you
up
and
going
on
many
of
the
meshes
go.
Try
the
learning
paths
go.
B
Do
these
things
so
that's
again,
a
testament
to
like
the
focus
on
learning
and
on
sharing,
and
so
deborah
priya
and
deepak
have
been
advancing
this
particular
area
and
there's
a
lot
more
resources
for
us
to
get
out
and
write
about.
B
So
having
said
all
that
we're
gonna,
like
one
of
the
discussions
that
barak
will
have
with
us
today,
will
be
on
a
couple
of
particular
technologies
that
we
might
want
to
write
about
and
discuss
and
and
like
publish
the
design
decisions
are
all
interesting
and
like
having
written
about
it,
reinforces
your
own
learning
and
helps
others
who
come
behind.
You
answer
their
own
questions
about
like
how
does
this
tech
work
some
of
it's
like?
How
does
how
does
one
of
the
our
the
the
projects
that
we
steward
here?
B
B
Thank
you
for
spending
all
the
time
that
you
are
thank
you
for
publishing
for
others
as
well.
Thank
you
for
doing
the
grammar
that
some
of
us
just
can't
do.
A
Awesome,
thank
you.
Thank
you
for
that
lee
and
thank
you.
The
buck,
bami
and
devopria
congratulations
on
being
on
board
as
community
managers
all
right.
So,
let's
get
on
to
the
topics
that
we
have
for
today,
so
cubecon
china
is
on
next
week.
A
So
by
the
time
we
have
the
community
meeting
next
week,
we'll
have
the
tag
network
and
and
the
service
working
group
sessions
at
kubecon,
as
well
as
the
machary
and
smp
project
office
servers
at
kubecon,
so
both
measurey
and
smp
are
projects
from
the
lyrify
community,
as
well
as
their
cncf
projects.
Now
so,
like
the
previous
coupon,
we
have.
We
also
have
a
project
office
hours
session
where
people
can
come
in
and
ask
questions
to
maintainers
and
we
can
demo
the
projects
and
yeah.
A
So
the
the
timing
and
the
rsvp
details
are
available
in
the
meeting
minutes.
You
can
check
check
out
the
links
and
we
also
have
a
tag
network
and
service
mesh
working
group
meeting
as
well.
So
most
of
the
or
a
lot
of
the
projects
that
we
work
in
the
layer,
5
community
are
being
represented
in
the
service
mesh
working
group.
A
So
we
have
that
as
well,
and
we
also
have
mesh
made
anita,
giving
a
talk
at
kubecon
on
solving
the
service
mesh
mesh
adapters
dilemma,
so
anita
will
be
talking
about
measuring
and
how
measury
can
help
service
mesh
adopters
answer
a
lot
of
their
questions.
So
we
have
this
at
kubecon
as
well,
so
sign
up
and
show
some
love
for
anita.
B
A
quick,
quick
question:
anita:
do
you?
Would
you
like
to
push
the
pdf
copy
of
your
slides
to
the
layer?
Five
io
slash
events
to
that
entry,
it
might
be
it'd,
be
good
to
have
out
there.
B
Since
we're
talking
about
presentations,
I'm
just
we
might
have
shown
it
on
this
call,
I
don't
know.
I
really
encourage
any
interview,
all
of
you
that
are
on
the
call,
since
you're
spending
time
here,
like
not
only
check
out
flashmate
anita's
talk
like
it's
a
great
intro
to
meshary
and
to
the
space,
but
also
there's
a
pdf
of
the
presentation
for
the
cncf
service.
Mesh
working
group
go
check
it
out.
You
gotta
check
it
out
like
you'll,
be,
I
think,
you'll
be
pleasantly
surprised.
B
B
Go
go,
go
let
it
soak
in
some
pretty
cool
things
that
you
all
are
doing.
E
B
The
whole
talk
pretty
much
talks
about
what
the
the
projects
here
so
so
I
won't
I'll.
I
won't
speak
to
it,
but
you
can
watch
the
talk.
You
know.
A
All
right,
yeah,
you
can
check
out
the
the
slides
that
we
mentioned
in
the
chat
as
well
all
right.
So
we
have.
We
have
a
new
integration
in
slack
with
the
discus
forum,
so
to
give
a
quick
idea
about
what
the
discuss
forum
is
like.
We
have
a
discuss
forum
in
layer
5,
which
is
basically
a
forum
where
people
can
come
in
and
ask
questions
and
basically
have
a
conversation
going.
So
this
this.
The
conversations
we
have
on
slack
are
not
exactly
permanent.
A
So
what
what
we
use
discuss
is
is
so
that
we
can
have
a
more
permanent
record
of
our
conversation
and
point
people
to
discussions
so
that
yeah,
we
don't
have
to
repeat
ourselves
and
what
we
have
now
is
a
new
command
on
on
slack,
which
you
can
get
by
typing
the
slash
discuss.
A
So
I
think
this
command
has
a
lot
of
features
like.
Basically,
we
can
open
up
new
discussions
directly
from
slack.
So
if
you
are
having
a
conversation,
a
synchronized
conversation
in
slack-
and
you
won't
want
the
discussions
discussion
to-
I
don't
know-
maybe
like
you.
If
you
want
to
move
the
discussion
over
to
the
discuss
forum,
you
can
you
can
just
type
in
slash
discus
and
type
in
the
yeah.
B
You
know
so
it's
interrupt
again
by
the
way
is
it
double
prior
or
deepak
either
the
two
of
you
want
to
demo
this
now
how's
your
chance.
A
Actually,
I
I
didn't
really
look
into
like
the
discus
command,
so
maybe
navendo
early
could
like
take.
B
B
B
If
you
don't
mind,
sharing
your
slack,
there's
there's
a
problem
that
we
have
with
slack,
and
that
is
that
it's
forgetful.
B
And
we
have
some
excellent
discussions
in
slack
and
that
jared
might
be
being
polite
waiting
for
you
to
stop
sharing,
never
do
but
jared.
You
can
just
start
cool.
B
So
we
have
this
problem
where
we
have
some
excellent
discussions
in
slack
and
then
slack
forgets
about
it
and
then
we
repeat
ourselves
and
we
repeat
ourselves
and
we
repeat
it
and
that's
pretty
annoying
like
if
it
isn't
annoying,
listen
to
me
say
that,
like
it's
enough,
so
sometimes
we
have,
you
know
so
it's
like
well
so
actually
one
of
the
maintainers
here,
his
name's
michael
cafeller,
he's
he's
out
on
vacation,
but
but
he's
often
pointed
out
like
he
tries
to
keep
up
and
it's
like
well
there's
just
so
much
going
on.
B
I
can't
keep
up
you
know
and
even
after
we've
had
the
discussion
like
where
are
we,
you
know
persisting
that
documentation
and
a
lot
of
times
the
answer
is
like.
Well,
that's
google
talks,
that's
still
the
case,
but
sometimes
it's
also
just
good
to
catalog
it
on
the
discussion
forum
or,
like
the
discussion,
isn't
done.
B
B
B
I
get,
or
is
it
a
back
side
forward?
Slash,
I
guess
technically,
is
what
it's
but
yeah.
It's
the
other
direction
that
you're
doing
okay
yeah,
and
so
we
have
some
a
couple
of
custom
ones
actually,
and
this
is
one
of
them,
so
you
don't
have
to
like.
I
think
for
all
of
all
of
you
that
are
on
the
call,
the
only
prep
that
you
have
to
do
to
be
able
to
use
this
command
is.
B
Just
have
an
account
on
actually
I
don't
yeah.
I
have
an
account
on
the
disqus
forum.
Let's
see
what
happens,
though,
so
the
command
is
slash,
disqus
and
the
sub
command
is
post
and
what
what
you
do
is
like,
if
you
you
put
in
a
number
like
the
number
of
messages
in
whatever
channel
you're
in
the
number
of
messages
that
you
would
like
to
roll
up
and
create
a
draft
post
on.
So
if
you
say
three
for
you
know
and
hit
enter,
let's
see
what
it
does.
B
It
is
taking
the
last
three
messages
so
starting
three
messages
ago:
a
message
from
sahil
ending
and
whatever's
in
between.
So
that's
the
first
message
there's
one
in
between
and
then
it
ends
on
that
one
and
what
it
should
do
is
did
your
browser
pop
up
and
did
it
start
to
create
a
your
web
browser?
I
don't.
B
B
Okay,
okay
got
it
so
yeah
that
sounds
like
there
is
kind
of
a
requirement
about
the
so
very
good.
Let
me
let
me
do
this.
Let
me
also
share
a
different
experience.
Real
quick.
If
I
could
so
I'll
do
the
same
thing.
I'm
hanging
out
in
general
just
got
I
hit.
I
hit
backslash
discuss
post
three,
because
we
just
had
a
great
discussion
and
anita
said
some
things
that
we
should
catalog.
B
B
You
know
keeps
some
formatting,
it
does
looks
like
daria
has
an
account
on
discus,
so
that's
kind
of
cool
how
it
you
know
it
just
so
happens.
He
has
the
same
username.
I
guess
so.
Nice,
nice,
nice
way
to
catalog
or
continue
the
discussion
or
maybe
pull
others
into
the
discussion.
So
all
right
give
it
a
try,
give
it
a
twirl,
I'm
gonna,
try
it
out.
B
There's
yep,
there's
also
another
command
in
there.
That
will,
I
don't
know
if
everyone
will
be
able
to
do
this
or
not.
I
kind
of
hesitate
to
show
everybody,
because
this
could
be
bad,
but
you
could
there's.
You
can
have
any
time
that
there's
a
message.
B
A
new
thread
created
out
here,
there's
a
notification
that
can
be
sent
into
slack.
We
don't
really
have
we
only.
We
don't
have
many
of
those
configured
right
now,
so
we
didn't
want
it
to
necessarily
be
annoying
to
people,
but
we
might
want
to
figure
out
that,
like
hey,
if
somebody
creates
a
topic
inside
of
measuring,
maybe
we
should
you
know,
let
let
the
community
know
and
hear,
and
that
way
we
help
bridge
the
divide
between
the
two
places,
so
cool
and
just
okay.
B
No,
so
at
one
point
I
think
I
was
messing
around
with
like
hey.
B
What's
the
latest
number
of
performance
tests
that
have
been
run
by
the
you
know,
I'm
out
there
and
it's
5
000
something
something,
and
so
I
figured
it'd
be
nice
to
have
a
custom
command
where
you
can
just
check
the
number
of
performance
service,
mesh
performance
tests
that
have
been
run,
and
so
those
types
of
custom
commands
can
be
created,
cool
anybody
get
any
questions
or
anything
else
to
add
other
than
other
than
I
assume
jared
now
has
an
account
on
discuss
so
yeah,
very
good.
C
All
right,
I'm
gonna,
share
my
spoon.
C
Yeah,
if
you
can
see
my
screen
so
basically,
if
you,
if
you're,
not
aware
of
the
background,
the
topic
itself,
so
here
we
have
a
migration
of
the
golan
channel
to
the
broadcast
phase
separation.
So
I
provide
the
spec
in
the
in
the
in
a
many
minutes
and
also
the
video
shared
it
on
the.
C
This
is
on
chat.
So
if
you
you,
if
you'd
be
kind
to
open
the
subspace
and
and
in
this
call,
basically,
what
I'm
trying
to
do
is
that
so
this
pack
is
I've
created
it
quite
a
few
weeks
ago.
I
think
so
here.
What
I'm
trying
to
do
here
is
in
this
meeting
is
that
I
was
hoping
to
get
a
action
of
action
items
and
then
some
conclusion
that
we
can
basically
start
with
for
this
design
spec.
C
So,
basically,
the
idea
of
this
whole
spec
is
about
migrating
the
gun
channel
to
broadcast
base
solution.
So
why
is
that
the
case?
Because
so,
if
you
so
just
give
use
so
for
the
details,
you
can
actually
read
the
spec
itself,
but
I'll
just
get
on
the
practical
side
of
it.
C
So
if
you
see
on
this
code
itself,
so
I
think
so,
if
you
know
a
bit
of
a
golem,
it
will
be
good
because
I'm
going
to
give
you
an
example
in
golem,
so
here
from
line
eight
until
line
38,
we
have
a
graphql
subtraction
yesterday.
C
Yeah,
I
hope
it's
it's
seen
clearly.
So
in
line
838
we
have
a
separation,
graphql
separation
and
then
what
we
have
basically
how
this?
How
does
the
whole
thing
work
in
the
graphical
program
so
in
in
in
the
resolve
itself?
In
the
function?
We
we
create
a
channel
of
the
boolean.
So
here
I
have
a
I
have
a
result.
It
provides
a
boolean
value
or
for
a
continuous
polling
time.
C
So
here
I
I
create
a
boolean
channel
and
then
at
the
end
of
the
and
then
I
also
provide
the
go
function
called
go
routine,
that
that
has
a
params
of
the
channel
itself.
So
basically,
on
the
ui
side,
we
they
receive
this
channel
and
then
they
listen
to
this
channel
from
a
period
of
time,
so
this
go
routine
will
run
and
then
and
then
we
have
to
on
the
on
the
back
inside.
C
What
we
need
to
do
is
that
we
need
to
input
the
the
messages
onto
this
channel
so
that
the
client
will
get
the
the
message
that
we
want
to
infer
to
them.
Basically,
so
basically,
you
see
that
this
provides
a
lot
of
content
that,
if
we
want
to
have
like,
if
you
want
other
component
to
also
input
a
message
through
this
channel,
so
I'm
talking
about
like
this
channel.
If
you
want
other
components
to
also
input
message
or
struct
to
this
component,
we
need
to
pass
around
this
channel
to
other
component.
C
C
Basically,
so
what
I
propose
is
that
what
we
have
we
want
to
have
something
like
a
event:
bus
architecture
in
our
measuring
server,
in
which
we
we
don't
really
have
yeah,
in
which
we
don't
really
have
right
now
and
so
in
the
go
routine
itself
so
in
in
in
the
implementation
of
this
whole
spec
is
that
we
implement
an
event
bus
that
has
a
particular
particular
subject
or
a
particular
pipeline
that
we
can.
C
We
can
create
a
channel
and
then
listen
to
the
whole
messages
that
is,
that
is
coming
from
those
particular
channel,
and
so
this
this
channel
is
very
is
the
struct
of
this
channel
is,
is,
is
it
can
be
applied
in
with
anything?
Basically,
so
it
can
be
applied
with
boolean,
stroke
or
any
kind
of
data,
and
what
we
want
to
have
is
that
we
want
to
have
a
filter
based
on
on
some
messaging
framework
that
we
have
provided
so
so
this
particular
premium.
One.
C
So
this
particular
design
spec,
is
it's
quite
tightly
coupled
with
the
other
design
aspect
that
michael
one
of
the
maintainers
of
missouri
has
created
quite
some
time
over
as
well.
So
this
messaging
framework
finally
couples
with
this
fact
in
terms
of
like
this
broadcaster
implements
the
messaging
framework
that
michael
has
created
before
yeah,
so
in
this
in
this
pipeline
or
messaging
pipeline.
What
we
want
is
that
we
have
a
filter
based
on.
Where
is
this
where's,
this
data
coming
from?
C
What
is
the
time
that
this
data
is
is
sent
out
and
basically
the
details
of
the
message?
Basically
so,
based
on
this,
you
can
see
you
can.
You
can
actually
think
that
this
particular
pipeline
is
that
when
we
receive
a
message
that
interests
us
in
terms
of
like
this,
we
define
a
filter,
and
then
this
filter
converts
to
the
one
that
we
have
the
one
we
desire.
C
So
we
we
get
this
message
and
and
and
then
we
we
basically
process
the
message,
the
message
data
and
then
we
check
whether
it's
actually
is
actually
true
and
then
this
data
will.
This
data
will
be
sent
out
to
this
channel
that
we
have
sent
over
to
the
measure
ui
so
in
this.
So
basically,
this
is
the
current
implementation
that
I
have.
C
It
has
a
lot
of
a
lot
of
weakness
in
terms
of
like
this
is
a
polling
kind
of
like
system,
so
you
can
see
that
this
is
a
go
routine
and
it
runs
an
indefinite
indefinite
loop
for
that.
Obviously,
it's
indicated
so
we
do
a
polling
for
each
10
second
and
we
basically
check
the
connection
of
the
so
this
resolver
is
checking
the
connection
with
the
broker.
That's
not
broker,
so
every
10
seconds
we
check
whether
the
connection
with
the
broker
is
is
broken
or
not.
Is
it
connected
or
not?
C
So
if,
if
the
connection
is
not
so
if,
if,
if
there
is
any
any
change
in
the
broker
connection,
then
we
send
the
message
to
the
ui
yeah
and
we
change
the
reference
to.
We
change
the
reference
from
the
previous
variable
to
the
current
one,
so
yeah.
So
this
has
a
lot
of
weakness
in
terms
of
like
this
is
not
very.
C
C
Yeah,
so
I
think
that's
the
main
idea
of
this
design
spec
and
the
the
outstanding
concern
that
we
have
is
that
what
is
the
best
way
to
implement
the
broadcaster?
Are
we
using
the
nuts
are?
Are
we
or
can
we
use
another
kind
of
framework
that
we
haven't
implemented
yet
in
measuring
yeah?
So
that's
the
the
main
concern.
C
Yeah,
actually
from
this
particular
spec,
I
I
use
the
implementation
of
the
broadcast
itself.
We
use
what
drove
has
brought
us
soldier
is
one
of
the
excellent
gamers
of
mercury,
so
dhruv
provided
us
with
an
implementation
of
broadcast
using
a
native
online
channel,
which
I
think
is
quite
good
but
in,
but
I
think
that
I'm
I'm
kind
of
leaning
towards
the
next
subject,
subject
based
messaging,
so
if
you're,
so
this
is
where
the
discussion
actually
lies
on
so.
C
I
think
it
can
be
replaced
easily
with
the
nuts
with
a
particular
subject.
So
basically
we
have
a
convention
if
one
subject
is
regarded
as
the
main
channel
or
mind
mind
pipeline.
So,
for
example,
a
subject
of
broad
a
subject
of
broadcast
or
string
broadcast.
C
B
I
don't
know
that
it
does
maybe
yeah.
I
don't
know
that
it
does
yeah,
I
also
like
so
yeah.
B
So
the
reason
I
say
I
don't
know
that
it
does
is
because,
if
we're
using
subject-based
streams,
then
if
a
client
desires
to
receive
to
be
subscribed
to
all
streams
and
then
it
can
subscribe
to
all
streams
right,
I
don't
know
if
there
is
a
built-in
mechanism
into
gnats
for
saying
subscribe,
all
or,
but
but
that
can
be
facilitated
through
any
of
the
clients
to
query
what
the
topics
are
and
then
subscribe
all
right,
and
so,
as
such,
the
client
can
then
choose
to
be
to
receive
everything
basically
subscribe
to
a
broadcast
channel,
even
though
that
there
isn't
necessarily
one
dedicated
for
that
type
of
a
broadcast.
B
D
Yeah,
actually
so,
if
we,
so,
if
we
use
nasa
as
the
even
bus
or
message
is
even
bus
and
we
use
it
for
internal
communication,
for
let's
say
this,
is
we
receive
data
from
mesh
thing
and
we
process
it,
and
then
we
send
to
our
top
again.
Someone
else
is
listening
it
and
then
we
redirect
it.
So
basically,
what
is
happening
here
is
because
naz
is
being
treated
as
the
main
event
bus
for
internal
com
for
internal
communication
as
well.
D
That
also
means
that
we
rely
on
nuts,
which
relies
on
network,
and
it's
sometimes
so.
Basically,
internal
communication
fails
just
because
the
network
was
not
broker
was
not
reachable
is
sort
of.
I
mean
it's,
so
basically
it
will
leave
a
mystery
in
in
an
undesirable
state,
even
though,
even
though
mystery
could
have
performed.
Basically,
it
would
have
worked
just
because
a
mystery
broke,
a
non-smoker
was
basically
not
lance
broker
was
not
reachable.
That's
assuming
that
nuts
is
being
proposed
as
the
event
buzz
that
is
being
used
for
internal
communication.
D
While
the
second
scenario
is
that
nazis
being
proposed
as
not
the
even
bus,
rather
than
go
channels
are
being
still
being
used
internally
for
communication,
the
knots
would
be
used
only
for
street.
So
so
I
mean
there
is
proxy
for
every
protocol
that
exists
like
you
can
use
technical
grpc
on
the
browser
by
processing
it.
So
similarly,
we
can
use
not
not
protocol
as
well.
We
can
proxy
the
data.
We
can
proxy
the
data
that
came
into
nuts
to
the
browser,
and
that
would
be
fine.
So
now
we
are
relying.
D
We
are
relaying
data
that
is
coming
through.
The
issue
with
that
particular
approach
would
be
that
now
we
cannot
do
any
kind
of
processing
on
that,
so
that
is
the
data
came
in
and
we
are
just
piping
it
to
the
ui
without
any
processing.
What
what
that
means
is
that,
right
now
what
happens
is
that
data
comes
in
and
we
do
actually
send
it
to
the
ui
and
we
store
it
in
database
so
that
later
you
can
actually
query
it.
D
But
right
now,
what's
happening
is
that
you
are
missing
the
missing
the
basically
right
now
you
are
missing
the
querying
portion.
You
are
just
drilling
every
day
that
is
coming
to
the
ui,
so
that
actually
that's
only
downside.
That
is
if
we
treat
nuts
as
another
system
for
relaying
data
that
is
coming
quickly
from
meshing
or
something
while
if
it
is
the
even
bus
in
that
case,
I
think
that
there
will
be
far
more
issues,
because
networking
can
I
mean
feed
networking
shouldn't
cause
the
mysteries
internal
communication
failure.
That's.
B
Yeah,
I
think
in
both
in
both
scenarios
like
in
that
first
scenario,
the
location
of
the
broker
needs
to
come
into
question.
The
location
of
the
broker
today
is
potentially
remote
or
potentially
over
a
local
network,
of
which
you
wouldn't
really
expect.
You'd
expect
high
resiliency
of
that
local
network.
If
it's
within
kubernetes,
but
but
to
that
point
we
might
move
the
location
of
the
broker
if
it
becomes,
if
it
totally.
If
all
you
know,
component
to
component
communication
goes
through
through
gnats.
B
So
we
have
to
yeah,
I
mean
in
our
we'd,
have
to
reevaluate
the
deployment
of
nets
or
deployment
of
the
brokers
or
make
sure
that
we
have
a
multi-broker
scenario
in
which
the
local
broker
is
intelligently
being
used
for
local
or
you
know,
for
topics
that
are
intended
to
be
done
locally.
B
B
It
might
be
that
in
the
other
scenario,
if
I
have
to
think
on
this
more,
but
if
we're
using
jet
stream-
and
we
use
a
store
and
forward
mechanism,
if
we're
first
saying
that
the
measuring
measures
server
by
the
way
for
those
that
are
trying
to
keep
pace
like
open
up
the
meshry
architecture,
deck
and
it'll
give
you
some
visuals
as
to
what
we're
referring
to
you'll,
be
able
to
digest
what
we're
saying
a
lot
easier.
B
So,
if
yeah,
we
would
need
to
make
sure
that
we're
storing
and
forwarding,
we
need
to
make
sure
that
we
want
to
have
our
cake
and
eat
it
too.
Some
of
that
might
be
facilitated
through
jet
stream.
B
Having
our
cake
and
eating
it
too
means
we
want
to
have
event.
We
want
to
have
highly
efficient
event,
driven
messages.
B
B
You
know
we
have
to
evaluate
each
of
these
message,
each
of
the
topic
topics
and
the
messages
for
what
would
be
appropriate
for
the
use
cases,
but
there
may
be,
but
but
I
don't
know
that,
but
we're
already
having
to
store
and
forward
today,
as
as
we
as
measuring
server,
receives
and
persists
to
database,
the
there
may
be
multiple
clients
involved
in
a
particular
transaction.
B
B
And
it
needs
to
publish
those
out
a
subscription
to
a
javascript
client
yeah.
We
just
need
to
funnel
those
messages.
We
just
need
to
persist.
Those
updates,
at
the
same
time,
that
we're
fun
I
shouldn't
say
at
the
same
time
we
need
to
sequentially
persist
and
then
leverage
the
intelligence
of
the
graphql
queries
and
the
sql
queries
that
are
being
run
on
the
database
to
like
in
some
in,
and
there
may
be
some
exception.
B
We
may
say
well
for
a
particular
topic:
let's
go
ahead
and
circumvent
the
let's
either
mirror
the
message
or
circumvent
the
persistence
of
that
message
entirely
because
it
was
just
like
it
was
just
a
current
count
of
the
tcp
traffic
between
two
nodes
between
this
service
and
this
service,
and
so,
if
we
skip
a
beat
in
terms
of
like
updating
the
user
interface
to
show
the
current
stat,
like
you
know,
oh
well,
like
the
trade-off,
was
slightly
degraded,
ux
versus
slightly
degraded
performance
of
the
system.
B
Overall,
italy,
what
gets
even
more
fun
is
when
is
when
use.
There
are
any
number
of
user
agents
browsers
that
are
opened
up
to
measure
ui.
You
know
you
know
multi-user
environment,
in
which
there's
you
know
in
which
we
just
need
to
walk
through
that
scenario,
to
determine.
B
So
the
so
of
this
is
just
you
know,
so
so,
for
my
part,
like
of
the
concerns
that
barack
is
raising
it's
like
yeah.
These
are
the
these
need
to
be
addressed.
These
are
legitimate,
there's
a
legitimate
set
of
a
variety
of
scenarios
that
need
to
be
addressed
and
they
have
the
in
and
they
have
I'm
sorry.
B
The
architecture
that's
chosen
has
a
significant
implication
on
how
those
scenarios
are
fulfilled
and
they
ultimately
have
can
draw
really
hard
lines
on
what
meshri
is
capable
of
or
not
capable
of,
and
consequently
that's
why
we
have
that's
why
a
lot
of
this
stuff
is
being
written
down.
B
While
I,
too
would
like
to
drive
to
conclusion
today
right
here
now
and
move
forward,
I
don't
like.
I
do
think
that
we
can
take
some
really
healthy
steps.
B
So
one
of
those
is,
I
mean,
utkarsh,
just
presented
kind
of
two
scenarios:
okay,
and
so
if
we
write
down
those
scenarios
and
the
cab,
you
know
the
caveats
that
he
was
mentioning
there's
a
little
bit
of
there's
a
this
is
this
is
like
a
spike
if,
for
those
that
do
have
done
agile,
like
there's
a
little
bit
of
research
that
needs
to
happen,
but
that
needs
to
happen.
We
don't
want
to
walk
away
from
the
conversation
right
now
and
say.
Well,
we
need
to
study
it.
It's
like
no,
no!
No.
B
We
need
to
write
down
those
scenarios
write
down
the
things
that
we're
trying
to
achieve
where
it's
okay,
to
lose
some
messages
where
it's
not
where
it's
not
okay,
I
mean
the
very
first
thing
that
lukasz
had
pointed
out
was
like
under
our
current
architecture.
B
It
is
entirely
not
okay
to
have
to
traverse
the
network
for
one
internal
component
to
talk
to
another
component,
like
no
yeah,
I
mean
anyway,
so
we
need
to
go
through
the
scenario
a
little
bit
more.
This
is
a
this
is
a
this
discussion
that
barack
has
brought
up.
It's
a
it's.
A
line
item
on
the
v070
roadmap:
that's
how
significant
it
is.
The
decisions
that
are
done
here
are
pervasive
for
all
of
the
components
how
they
communicate
with
each
other.
B
All
right.
The
decisions
have
pervasive
ramifications,
I
mean,
and
I
my
sense
is
that
the
majority
of
us
that
are
looking
at
this
I'm
gonna
put
words
in
everyone's
mouth
is
that
we
should-
or,
from
my
perspective
I
I
think
we
should
go-
prove
why
nats
and
potentially
a
re-architecture
of
how
we're
using
that
is,
does
not
work,
and
that
might
be
the
reality
like
we
might
prove
that
it
doesn't,
but
I
think
it's
a
safer
position
to
start
from
the
right
position.
B
B
We've
got
a
poor
man's
use
of
of
eventing
through
channels
today,.
C
Yeah,
so
I
yeah.
Actually,
I
just
realized
what
the
cars
just
say
and
then
it
it
basically
changes.
My
whole
view
of
using
knots,
because
because
currently
knots
is
an
external
instance
of
measuring
and
we
are
basically
trying
to
use
it
for
talking
inter
components
inside
the
measuring
server
yeah,
so
is
it
safe
to
assume
that
we
need
to
take
a
step
back
and
check
whether
we
need
to
revamp
how
we
use
how
we
start
nas
or
basically
the
whole
connection
of
knots
and
the
measure
server.
B
I
think
that
that's
it
yeah,
it's
fair
that
that's
within
scope.
The
place
where
we
really
need
to
start
is
with
the
scenarios
like
do
we
have
a
scenario
in
which
an
internal
measuring
server
component
needs
to
use
anybody
in
the
event,
bus
or
use
these
channels?
C
We
cannot
close
that
possibility.
C
D
D
So
there
is
so
unless
and
until
we
agree
on
the
fact
that
we
will
use,
keep
on
using
code
channels,
and
we
will
also
use
an
even
bus.
It
would
be
fine,
but
if
you
decide
that
even
bus
is
the
thing
that
is
substitution
based
messaging
is
the
thing
that
is
going.
We
are
going
to
use
for
inter
routine
communication.
In
that
case,
so
I
mean
that
is
basically
that
will
become
an
issue
we
were
actually
trying
to.
D
We
were
trying
to
one
of
the
concerns
of
this
thing
was
actually
the
blocking
nature
of
golan
based
channels,
which
one
of
the
ramifications
of
it
was
that
for
a
while
mystery
had
a
bug
where
we
were
not
able
to
persist
the
data.
If
the
client
was
not
connected
because
the
channel
was
blocking,
we
cannot
send
the
channel
message,
so
persistence
was
also
not
happening.
So
actually,
this
is
where
actually
it
started
from
and
then
we
came
to
broadcast
so
that
we
dropped.
D
So
we
moved
to
a
non-blocking
channel
using
broadcast
base,
and
then
we
actually
realized
that
probably
we
need
subscription
based
things
and
then
we
came
to
even
bus
and
now
which
even
passes
the
probably
the
discussion
so
yeah.
There
is
an
interview,
routine
communication.
B
D
So
I'm
not
sure
about
the
implementation,
like
I
don't
remember
the
implementation
of
performance
test,
but
it
does
happen
in
mesh
thing.
I
will
have
to
actually
pinpoint
where
exactly
it
was
happening,
but
yeah,
and
there
is
a
implementation
where
we
do
realize,
sending
messages
via
channel
and
consuming
it
and
instead
of
sending
it
forward.
So.
D
B
I
suppose,
keeping
the
documents
separate
for
the
moment
makes
I
don't
know
that
it
really
makes
sense.
I
really
want
to
ke.
This
is
like
we're
talking
about
the
messaging
framework
and
there's
an
aspect
of
the
in
the
other
doc,
the
messaging
framework
doc
it
its
focus,
is
more
toward
topics
and
message
format
and
the
data
that's
contained
within
those
those
messages.
B
So
if
we
can
delineate
the
two
design
specs
based
on
mechanics,
which
is
the
what
this
broadcast
spec
is
focused
on
versus,
what's
on
the
wire
in
the
message
format,
you
know
in
the
in
the
other
spec,
the
two
could
can
be
separate,
they're,
both
yep,
both
the
two
could
be
separate.
Otherwise
we
probably
want
to
combine
them,
so
we
don't
so
that
we
account
holistically
for
all
the
use
cases
involved.
B
So
we
should
either
combine
these
two
specs
into
one
because
they're
both
trying
to
talk
about
how
to
message
between
components
or
agree
that
one
of
them
is
speaking
to
primarily
to
the
mechanics
and
the
the
messaging
system
itself,
while
the
other
design
spec
the
messaging
framework,
one
is
focused
on
message:
format,
number
of
topics,
what
they're
called
the
format
of
the
the
messages
that
are
within
there,
the
data
structure
and
then
how
that
interplays
with
er?
B
How
you
can
build
error
codes
on
top
of
that
and
how
you
can
build
a
robust
notification
center.
On
top
of
that,
like
that,
if
we
can
delineate
between
the
wire,
I'm
sorry
yeah.
What's
on
the
wire,
which
is
the
messaging
framework
design
spec,
and
what
the
wire
is,
the
mechanics
of
whether
it's
go
channels
and
mats
and
other
things,
we
can
have
two
two
separate
specs.
If
we
can't
clearly
delineate
that
they
need
to
be
the
same
design,
spec.
B
B
Yeah
sounds
good.
I
think
that
yeah
sounds
good.
I
think
they
do
work
separately.
We
just
I'm
just
saying
it
out
loud.
That's
that
we're
on
the
same
so
everybody's
aware,
like
that's
the
message
format,
what
we're
doing
with
it,
how
we're
how
many
topics
we
have
all
that
and
all
that
other
stuff
versus
this?
B
It's
good
for
those
again
like
for
those
that
were
following
along
the
mesry
architecture.
Deck
really
helps
paint
a
picture
of
where,
where
the
communications
are
and
what
those
various
flows
need
to
be.
B
B
You
want
to
see
that
data
flowing
through
and
that's
one
of
the
scenario,
one
of
the
reasons
why
such
a
system
is
needed-
and
it's
probably
it's
like
the
top,
the
core
driving
scenario
but
there's
another
there's
any
number
of
other
internal
mechanics
like
as
to
the
operations
that
measury
performs,
when
you
say,
run
up,
run
a
perf
test
for
two
hours
and
then
you
don't
want
to
sit
there
watching
the
paint
dry
on
the
wall.
For
two
hours
you
want
to
like
go
start
another
test.
B
Another
two
hour
tests,
another
two
hour
testing,
so
you
tell
them
actually
to
do
like
10
things
at
once.
Well,
it's
going
to
have
it's
going
to
be
a
multi-threaded.
You
know
multi-routine
multi-um,
potentially
multi-channel
that
needs
to
communicate
between
those
routines
understand
who's
done,
and
so
that's
the
other.
That's
the
there
are
other
use
cases
for
these
types
of
internal
mechanics.
B
B
I'll
help
I
mean
I'll
jump
in
and
help
I
mean
the
specs
are
fairly
well
written.
Today
we
do
need
to
catalog
the
scenarios.
B
Okay.
Well,
that's
another
data
store,
there's
more
state
going
on.
What
is
that
key
value
store
and
I
miss-
and
is
it
possible
to
use
postgres
for
that?
I'm
assuming
not
I'm
assuming
it's
more
of
it,
but
those
are
more
questions
that
we
should
be
commenting
and
answering
here.
So
we're
going
to
need
a
nats
nerd.
B
What
there's,
like
all
the
four
people
that
I'm
looking
at
on
the
call
like
all
of
you
guys,
have
glasses,
and
I
don't
I'm
feeling
I'm
feeling
out
brained
today.
It's
not
that
cool
hey
once
in
a
venue,
instead
of
like
where
barack
any
other
commentary
that
we
like
I'll
go
I'll,
go
capture
a
few
of
the
action
items
that
we
just
said
for
sure
to
try
to
help
advance
this,
because.
C
A
So,
thank
you
barik,
like
I.
I
think
we
are
out
of
time
today,
so
maybe
like
we
can
send
out.
I
think
we
will
send
out
the
v0.7
roadmap
more
broadly
to
the
community.
A
Meanwhile
like
before
we
end
the
meeting
for
this
weeks.
I
wanted
to
point
out
that
we'll
be
having
a
a
tutorial
on
basically
a
walkthrough
session
on
contributing
to
larry
projects
on
the
next
week's
newcomers.
Call
so
like
the
deep
dive
session
today,
like
really
emphasizes
the
need
for
more
people
to
be
properly
onboarded
and
like
properly
have
given
some
guidance
on
how
you
how
they
can
start
contributing.
A
So
we
really
want
new
contributors
to
jump
in
and
just
start
working
or
start
contributing
to
layer,
5
projects
so
to
to
try
and
help
some
of
those
efforts
we'll
be
having
regular
tutorials
on
contribution
and
may
like
layer.
A
5
has
a
lot
of
projects,
so
we
will
be
talking
about
some
of
these
projects
and
how
they
can
make
some
contributions
and
they
how
they
can
actually
be
a
practical
walkthrough
on
what
they
can
do
so
yeah
like
if
you
are
new
here
and
if
you
are
interested
to
learn
something
new
and
like
really
jump
in
and
start
working
on,
things
do
join
the
call,
and
there
is
also
a
a
spreadsheet
on
the
meeting
minutes.
So
basically,
this
spreadsheet
tracks
some
of
the
tutorials
that
were
requested
by
newcomers.
A
So
if
you,
if
you
have
some,
if
you
have
any
project
that
you
would
like
to
work
on
or
any
particular
topics
that
you
need,
you
want
to
be
covered
on.
The
call
one
of
the
maintainers
or
the
community
managers
will
will
schedule
that
session
and
we
will
schedule
that
on
the
community
call
so
feel
free
to
drop
in
so
so
that
we
know
what
you
want
and
we
can
try
to
try
and
set
up
a
set
of
those
those
sessions.
A
So
with
that
yep
with
that
I'll
I'll
come
to
a
close
for
this
for
today's
meeting.
Thank
you.
Everyone
again
for
joining
or
joining
this
call
and
congratulations
to
the
new
community
managers.
A
We'll
see
you
guys
next
week,
hopefully
at
kubecon
and
feel
free
to
jump
in
on
the
project
office
hours
and
hang
out
with
us.
Alright
see
you
bye.