►
From YouTube: Ambient Mesh WG meeting 2023 07 12
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
All
right,
everyone
welcome
to
the
July
12th
Wednesday
occurrence
of
the
mbn
work
group
meeting.
A
So
first
up
on
the
agenda
is
follow
up
from
a
couple
weeks
ago,
we
were
reviewing
the
drive
mbmesh
to
Beta
Docker,
the
one
that
then
started,
and
so,
first
of
all,
once
we
go
through
the
stock
and
the
purpose
of
this
exercise
that
myself,
then
Keys,
Nathan
and
John,
undertook
yesterday
is
to
prioritize
the
items
outlined
in
the
stock,
so
that
we
have
a
clear
understanding
of
what
must
be
in
the
MDM
beta
MVP.
A
What
are
the
good
two
halves
and
the
items
are
not
worth
considering
for
beta,
so
we
are
about
halfway
through
this
document
and
once
we're
completed
through
it
will
require
approver
approving
from
the
TLC
members.
A
So
one
thing
we
realized
that
the
stock
is
structured
into
sub-feature
areas
and
one
thing:
we
realized
that
in
the
in
one
hour
that
we
had
yesterday,
it's
not
possible
to
go
through
and
go
into
details
feature
areas.
A
So,
instead
of
what
we
did
is
we
prioritized
the
existing
items
on
there,
but
also
nominated
Community
leaders
to
go
and
own
Drive
each
one
of
these
areas
in
order
to
get
to
the
one,
the
list
of
actual
items
that
are
needed
here
and
most
importantly
created
and
make
them
into
issues
and
have
descriptions
for
each
of
the
issues.
A
So
one
thing
I
want
to
use
the
first
time
it
was
the
first
part
of
this
meeting
is
to
go
through
the
dock
here
and
make
sure
that
we
have
owners
for
each
one.
They
said
the
own.
The
community
owners
here
are
willing
to
take
up
this
responsibility
so
stabilizing
the
ambient
apis
keys.
They
already
on
that.
So
thank
you
very
much.
The
second
part
about
extending
ambient
Beyond
single
cluster
kubernetes
workloads.
A
So
looking
through
the
Section,
we
decide
that
this
is
none
of
these
are
must-haves
for
nbm
beta,
and
so
we
don't
have
a
community
lead
for
this.
Yet,
although
each
one
of
these
sub
items
already
have
owners
next
to
them,
so
any
questions
on
this
so
far.
A
All
right
so
moving
on
compatibility
was
at
least
one
major
cni,
so
we
the
conclusions.
We
need
to
support
at
least
one
of
the
cnis.
When
you
fully
support
it,
we
just
don't
know
which
one
it
is
and
for
Community
leader
with
nominal
Iris,
for
this
looks
like
iris
is
not
here.
A
Oh
first
of
all,
the
question
is:
is
there
anyone
else
since
Iris
isn't
here
today?
Is
there
anyone
else
more
qualified
than
Iris
to
take
on
this
work?.
A
All
right,
if
no
comment,
then
let's
move
on
trust
and
security
hardening.
This
is
a
big
area,
especially
with
the
Piezo
cases
of
handling
the
edge
cases.
So
the
conclusion
we
came
to
is
we'll
we'll
need
someone,
so
the
the
names
that
we
put
on
here
for
now
Jackie
she's,
pretty
new.
We
do
need
more
experienced
community
members
to
handle
and
Lead
this
area.
So
right
now,
Keith
you
are
on
this.
B
Yeah
so
I'm
just
kind
of
going
to
be
the
cat
Hunter
trying
to
find
mentorship
Jackie's.
You
know
pretty
capable,
but
just
with
just
kind
of
new
to
ambient.
So
if
she
can
get
someone
to
to
kind
of
give
her,
the
Letterland
she'll
be
able
to
tackle
I.
Think
a
couple
of
these.
C
A
So
actually
John
can
you
speak
to
other
ability?
I
think
you
had
a
lot
of
comment
about
this
and
on
in
terms
of
who
should
lead
this.
D
D
Really
solved
like
the
overall
strategy
of
like
what
observability
we
want
to
admit.
So
it's
hard
to
talk
about
some
of
the
details.
Like
you
know,
we
implemented
some
things,
but
we
never
actually
really
sat
down
and
thought.
Do
we
still
want
reporter
source
and
destination
for
waypoints?
It's
about
a
third
reporter,
you
know:
do
we
want
all
the
same
metrics
still?
How
do
we
deal
with
the
fact
that
now
we
have
kind
of
two
layers,
TCP
and
HTTP?
Do
those
report
metrics?
Are
they
overlapping
all
these
sorts
of
problems?
A
F
Yeah
and
and
I
think
how
how
we
position
with
versus
open
Telemetry
is
a
big
question.
I
mean:
do
we
continue
to
support
all
the
20
different
apis
vendors
on
my
mechanisms,
who
are
doing
our
line
behind
open,
Telemetry
and
and
follow
us
there
naming
conventions
our.
A
So
I
think
there
are
conclusions.
Ideally
we
need
someone
on
the
TLC
to
seeing
deeply
about
the
strategy
or
and
drive
this
one
right
now,
I'll
just
put
John
on
it.
Do
we
have
any
other
members
of
TLC
who
can
take
on
this
work?.
C
I
can
probably
help
with
requirements
and
refining
we're
not
going
to
be
a
ton
of
use
in
terms
of
cat
hurting,
PRS.
C
F
No
is
there
an
option
to
decouple
Telemetry
and
ship
ambient
with
telemetic,
as
it
is
with
the
plan
to
just
move
to
open
Telemetry
later
I
mean
not
because
Telemetry,
you
know
in
gateways
it's
what
it
is.
We
essentially
share
implementation
with
the
istio
Gators
and
we
want
to
be
compatible
and
for
Z
tunnel.
It's
not.
That
is
that
critical
for
for
initial
release.
B
I
think
the
problem
with
that
is
that
it's,
it
will
be
very
difficult
to
ship
Ambience,
the
topology
as
data
and
have
the
metrics
not
be
as
data
as
the
rest
of
ambient
and
so
beta
telemetry
means.
We
support
that
for
going
forward
and
we
lose
the
ability
to
make
any
kind
of
breaking
API
changes
if
we
need
to
so
I
think,
that's
the
emphasis
for
for
aligning
on
the
overall
strategy
for
ambient
Telemetry.
At
this
point
before
we
get
to
ambient
beta.
F
Yeah,
because
my
point
was
that
again
with
the
community
standardizing
around
the
open,
Telemetry
one
way
or
another,
the
future
is
pretty
clear
that
it's
going
to
be
based
on
open
Telemetry
at
some
point.
So
the
move
to
open
the
limit
I,
don't
think
it
can
be
avoided.
What
I'm
suggesting
is
to
ship
with
a
telemetic
we
have
today,
let's
call
it
one
zero
and
it
can
be
V1
can
be
stable.
F
B
I,
don't
know
it,
it
feels
and
it
changes
a
lot
about
seo's
metrics,
exactly
these
metrics
SEO
Telemetry
or
potentially
changes
a
lot
just
because
you're
in
a
different
topology,
I
think
at
the
very
least
I
I,
just
I,
don't
feel
comfortable
shipping
Telemetry
as
it
is
without
thinking
about
some
of
the
edge
cases
in
in
a
beta,
a
beta
project
that
feels
that
feels
like
not
enough
due
diligence.
In
my
opinion,.
C
B
If
we
do
keep
the
same,
if
we
do
keep
the
same
apis,
then
we
are
acknowledging
that
moving
forward
after
data
those
apis
and
the
apis
out
in
symmetry
like
Capital
Sync
API
I
mean
the
actual
metric
scheme
or
metric
scheme
a
lot
whatever
we
are
essentially
saying
this
is
what
that
steam
is
going
to
be
moving
forward
and
it
doesn't
feel
like
we
want
to
do
that
unless
we
sit
down
and
think
pretty
pretty
hard
about
that
decision,
it's
actually
like
if
we
just
move
forward
kind
of
the
customer
saying
you
just
use
things,
as
is
that's
a
decision
point
right,
we're
making
the
decision
to
inherit
whatever
it's
we're
dealing
with
now
in
the
istio
sidecar
mode
and
just
kind
of
refer
to
it
with
an
ambient
and
feels
like
we
don't
want
to
do
that.
F
No,
no
I
completely
okay,
I
I,
I
I
completely
agree.
We
shouldn't
give
the
Impressions
that
we
are
going
to
support
that
model
forever.
My
my
point
is
that
it's
a
big
decision,
Point
here,
is:
if
we
adopt
the
standards
or
if
we
invent
our
own
ambient,
that
would
be
the
main
decision
point.
If
we
decide
to
go
the
standard,
you
know
the
standard
has
a
field
as
API.
We
don't
have
to
do
anything.
F
E
E
G
F
E
D
H
So,
there's
a
reason
why
the
whole
semantic
conventions
are
not
apis
because
they
are
not
schematized
in
general
they're,
not
really
part
of
the
AI.
So
maybe
what
it
means
is
that
we
need
to
adopt
the
same
approach
and
make
them
not
be
API,
but
something
like
configuration
so
there's
we
will
not
dictate
the
kind
of
activities
we
use.
We
just
say
you
can
use
these
values
and
you
can
name
them.
However,
you
want
and,
as
the
conventions
evolve,
we'll
either
adapt
to
the
to
the
new
changes
in
the
information.
F
H
F
H
F
Doesn't
mean
necessarily
just
you
know
a
Proto
I
mean
it
can
be
an
API.
You
know
the
scheme
of
the
URL
in
kubernetes.
That's
an
API.
F
What
I
mean
what
I
mean
is
they
have
some
keys
that
are
predefined,
that
are
in
Broad
used
by
elasticsearch
and
they
are
part
of
the
I
mean
that's
really
the
point,
the
key
names
and
what
should
be
put
inside
as
a
value.
That's
really
the
what
they
are
trying
to
define
a
standardized
across
vendors.
H
C
C
Given
if
the
hotel
stuff
is
in
flux
custom,
it
might
be
a
little
premature,
so
we
need
to
go.
Do
an
assessment
of
it
right.
How
stable
is
this
stuff
actually,
and
maybe
we
need
to
go
talk
to
the
hotel
folks
about
this.
C
G
So
I
guess
the
question
is:
is
beta
the
right
point
to
tell
users
you
can
upgrade
your
existing
service
mesh
to
ambient
or
or
is
that
something
that
we
wait
to
communicate
to
stable
at
beta?
It's
good
for
production
use
in
Greenfield
cases
and
it's
stable.
You
can
upgrade
your
existing
service
mesh
I.
Don't
have
a
strong
opinion
there,
but
we
do
need
to
come
to
a
consensus.
C
Or
are
we
carve
Telemetry
out
right?
We
say
you
know
ambient
is
beta
for
runtime
and
traffic.
The
telemetries
alpha,
okay,.
G
B
I
think
we
just
got
I
think
the
scope
up
on
this
a
bit
and
we
are
now
kind
of
rehash
from
a
fundamental
question
about
what
does
beta
versus
stable,
mean
I
think
as
far
as
observability
bullet
point,
the
the
questions
here
originally
I
think
are
about
how
much
does
to
what
degree
is
ambient
going
to
break
Telemetry
for
sidecars
I?
Don't
know
that
we
have
a
solid
answer
on
that.
Like
John
mentioned
some
of
the
reporters
stuff,
the
degree
of
symmetric
and
half
of
the
Waypoint
versus
utonal
Telemetry.
B
D
John
yeah:
this
is
how.
D
Really
saying
maybe
telemetry's
Alpha
but
Ambience
beta
I
I've
been
thinking
about
this
a
bit
and
I
a
bit
worried
about
approaches
like
that.
Like
I,
think
the
saying
that
is
fine,
but
what
I
would
want
is
kind
of
he
opted
into
the
alpha
parts.
So
if
we
just
say
in
some
Doc
somewhere,
yeah
Ambience
beta,
but
these
eight
features
are
alpha
or
experimental,
but
they're
still
there
and
used
it's
kind
of
implicit
I'm
worried
that
it
kind
of
doesn't
matter
that
we
said
it
was
Alpha
right.
Users
will
start
depending
on
it.
D
D
D
D
C
D
C
Document
the
limitations
of
what
right
release
them
with
the
beta
state
and
then,
if
we
want
to
standardize
on
Hotel
the
user
has
to
either
we
put
a
right.
We
separate
those
things
right
and
if
the
user
wants
to
adopt
otel
right,
we
say
or
adopt
the
hotels
convention
right
then
it's
an
explicit
opt-in
to
do
that
right.
That
would
be
the
safest
approach.
G
F
I
I
agree
that
that's
what
I
think
as
well
does
that
you
know
we
need
to
support
some
migration
anyway.
It
doesn't
matter
if
it's
now
or
later
so
we
need
to
support
the
schema
at
least
a
subset
of
it,
to
be
sufficient
for
migrations
and,
at
the
same
time,
against
the
future.
I
think
is
clear:
I,
don't
think
we
we
can
survive
as
having
our
own
schema
where
everyone
else
is
using
the
difference.
So
there.
F
F
The
same
thing,
semantic
conversion
is
only
single
because
for
interoperability,
that's
what
matters!
If
you
have
a
you
know,
Waypoint
Gateway
from
issue
and
one
from
someone
else,
engineering
link,
Rd
or
someone
else.
We
want
the
user
to
be
able
to
see
kind
of
or
some
vendor,
so
we
want
them
to
have
the
same
views
to
be
able
to
craft
include
all
the
information
to
not
be
kind
of
distorted
view
with
each
of
them
having
different
representations.
Basically,
so.
F
H
F
F
C
I
agree
with
cost,
and
the
conventions
are
probably
the
more
important
thing
right.
Having
a
an
option
to
support
push
versus
poll
doesn't
affect
most
users
or
it
affects
the
admin
right,
but
pretty
much
just
then
the
schema
affects
everybody
Downstream
of
that
information
right,
whether
it's
open,
whether
it's
Automation
and
CI,
CD
Pipelines.
H
H
You
know:
telemetic
production
systems
to
emit
standardized,
open,
telemetry
right
to
do
an
adaptation
right
right,
I
mean
that's
what
people
do
existing
gke
systems
I
think
it's
kind
of
the
whole
point
of
the
collectors
to
that
administrator,
so
not
care
about
conventions
that
are
used
in
applications.
Instead,
they
label
them.
F
C
So
we
need
to
let
Francis
get
to
his
list.
Obviously
there's
a
deep
conversation
to
be
had
here,
but
we
should
probably
put
a
pin
in
it
just
so
we
get
through
the.
A
Top
levels,
so
the
list
kind
of
ends
here,
because
we
only
got
so
far
the
observability
well,
the
second,
the
last
integration.
We
think
that
Stephen
Landell
is
a
great
candidate
to
lead
this,
but
we
still
need
to
confirm
with
Stephen
who
is
not
on
the
meeting
today.
A
So
I
think
one
thing
that
came
out
of
this
discussion
here
is
hotel
or
observability
in
general
is
a
pretty
it's
contentious
point.
So
Louis.
Can
you,
as
a
representative
of
the
TLC,
can
you
help
to
gather
requirements
here
and
set
some
strategic
directions?
Working
with
John
yeah
I
can
work
on
them
all
right.
Thank
you.
A
So
the
rest
of
this
list
we
didn't
get
through
yesterday,
but
the
intention
is
once
then
is
back
from
traveling
she'll
work
with
Nathan
Keith
and
John
offline
to
go
through
the
rest
of
this
list
and
then
we'll
review
this
in
one
week's
time.
B
Yeah,
this
might
be
quick
I'm
going
to
fix
myself,
though
so
we'll
see,
I
just
wanted
to
check
and
see.
If
anybody
has
a
can
you
give
me
information
on
the
current
status
of
ambient
labels
and
annotations
I
saw
the
doc
apis
I
linked
to
it
there
in
the
meeting
notes
where
we
say
no,
we're
not
going
to
support
these
annotations,
but
we
do
want
to
support
these
annotations.
Do
we
have
a
sense
of
how
much
of
that
work
is
actually
completed?
How
much
is
it
in
progress?
B
A
So
Keys
is
the
asked
to
see
which
one
of
these
sections
are
sufficiently
Rich
consensus,
so
we
can
start
so
you
can
start
adding
issues.
Is
that
I
asked.
B
I'm
sure
Okay
so,
first
of
all,
there's
another
item
here
to
documents
on
the
tips,
but
the
first
one
would
be
set
cardio
that
one
pretty
obviously
feels
like
it
shouldn't
belong
in
ambient.
B
Ambient
redirected,
pods,
sorry
non-stack
car
pods
shouldn't
be
valid
proxy.
That
istio.config
it
says
no
I,
don't
know.
B
Something
thinking
about
things
like,
for
example,
maybe
a
Gateway,
topology
or
other
proxy
settings.
How
to
how
else
would
you
deploy
those
to
the
Waypoint
without
proxy
config
today,.
B
I
mean
there's
a
bunch
of
stuff
in
proxy
config
The
annotation,
not
the
crd.
B
But
I
thought
that's
an
open
question,
so
maybe
not
no!
For
that.
Yep
connect
timeouts
keep
alive
tracing
access,
logging
cross
domains.
We
probably
want
some
money
to
edit
that,
but
locality
load
balancing,
so
it
was
like
there
might
be
some
things
that
you
want
to
configure
on
the
Waypoint
with
that
annotation
John,
then,
if
you're
still
around
is
am
I
missing.
Something
obvious
as
far
as
why
you
put
why
that
was
set
to
no
initially.
D
I,
don't
know
how
to
write
any
of
this
part.
So
I
don't
know,
I
think
that
we
would
probably
want
proxy
config.
B
B
Excluding
excluding
ports,
19
ranges
interfaces,
there's,
probably
some
subset
that
we
are
getting
to
exclude,
but
I
can
see
this
being
useful
for
Waypoint
Pods.
At
least
you
know
anybody
have
any
opinions
on
any
of
these
three
so
far,.
B
I,
don't
disagree
because
it's
rare
rare
ish
that
you
would
want.
Oh,
maybe
I,
I
I'm,
trying
to
think
of
a
scenario
in
which
you
would
want
different
settings
for
particular
Waypoint
instances
and.
B
B
B
So
yes
for
some
but
I,
don't
what
I
wouldn't
want
is
to
have
to
have
multiple
Gateway
classes
for
for
waypoints.
Just
for
some
of
these
configs,
so
I
think
that's
probably
The
annotation
stay
useful.
C
So
like
some
of
the
examples
you're
reading
all
from
like
traffic-
that's
like
her.
Some
of
those
seem
like
transitional
things
like
keep
alive
is
right.
Was
there
to
prevent
psychers
from
keeping
too
many
connections
open
to
the
down
the
upstreams,
and
that
was
made
sense
in
a
situation
where
the
number
of
client
proxies
outnumbered
the
number
of
upstreams,
and
so
it
was
going
to
be
a
resource
consumption
problem
for
the
client,
but
with
waypoints.
That's
completely
different
right.
The
topology
Dynamics
are
totally
different.
So
why
bother
right?
C
You
could
see,
keep
lives
being
important,
maybe
because
you
had
too
many.
The
upstreams
had
to
keep
too
many
connections
open
and
they
were
getting
overwhelmed,
but
the
number
of
waypoints
is
typically
equal
to
or
less
than
the
number
of
upstreams.
For
that
connection,
so
again
not
likely
a
problem,
so
it
seems
like
they
could
be
retired.
C
C
G
That
makes
sense
to
me.
These
annotations
have
been
problematic
when
we
were
inside
car
mode
alone.
The
complexity
of
applying
them
in
an
ambient
World
means
that
we
should
probably
be
carrying
forward
as
few
as
possible.
D
D
You
know
we'll
we'll
expand
and
violate
that
I'm
sure
you
know
we
started
off
saying.
Even
maybe
we
only
support
Gateway
API,
for
example,
but
in
general
I,
think
it's
a
good
principle
because
we
can
always
add,
but
it's
basically
impossible
to
remove.
D
This
is
probably
our
one
real
chance
to
do
anything
like
this
project
for
quite
a
lapse.
G
E
D
B
B
D
Yeah
I
mean
some
of
those
things
like
they're,
not
even
really
like.
We
could
configure
them
at
the
global
level,
but
we
can't
configure
them
out
per
pod
basis
anyways.
So
it's
kind
of
right,
tricky
I
mean
unless
we
say
like
if
you
enable
the
global
ambient
flag.
You
also
change
all
this
other
Global
Behavior.
C
E
H
If
I
remember
many
of
these
auditions,
we
had
it
as
a
Warcraft
for
issues,
and
but
we,
what
are
the
actual
issues
we
have
with
waypoints
right
now,
do
we
know,
can
we
can
we
can
we
size
them
correctly?
Can
we
deploy
them
correctly?
Can
we
do
lifecycle
correctly
for
them
I?
Think,
like
the
sort
of
a
lot
of
this
annotations
might
be
needed.
If
we
cannot
do
any
of
those
things
right,
but
since
we
don't
know
yet-
or
maybe
we
do
maybe
I'm
just
not
aware
of
it,
it
might
be
premature
to
remove
them.
H
C
B
What
does
this
allowing
us
to
do
and
does
it
still
make
sense
as
a
workload
annotation
or
you
know,
Waypoint
pod
level,
annotation
in
an
ambient
world,
and
then,
if
we
still
feel
that
we
need
that
functionality,
then
we
have
to
go
build.
Something
to
you
know
some
API,
some
way
of
configuring
that
in
ambient
world,
so
just
to
take
proc
to
keep
alive.
For
example,
let's
say
we
had
a
component
use
case
for
people
live.
Maybe
that
is
a
a
Gateway
class
or
change
the
controller.
Or
what
have
you
just?
E
E
If
you
need
this,
raise
a
bug
right
and
then
we'll
add
it
back
as
Ryan
was
there
Lewis
was
saying
so
like
I
think.
Can
we
just
do
that?
Do
we
have
to
actually
decide
that
we
need
to
add
it
and
bring
it
in,
or
can
we
wait
for
people
to
to
yell
when
it's
not
there,
because
we're
not
shipping
a
stable
product?
Yet
so
we
can
have
things
that
are
not
implemented
right,
it's
okay!
To
do
that.
C
E
So
yeah
like,
if
it's
just
a
catalog,
it
can
be
like
we
absolutely
will
never
need
this
in
ambient.
We
might
need
this.
If
somebody
wants
it,
they
can
raise
a
bug,
but
here
it's
not
just
FYI.
It's
these
things
are
not
supported
directly
in
ambient
at
this
time
when
we
ship,
whatever
we're
going
to
ship
right
now,.
B
But
yep
must
be
telling
me
at
next
Adam
thanks
everybody.
A
Okay,
so
next
item
is
Mitch.
G
Hey
yeah:
this
is
just
a
public
service
announcement.
We
are
going
to
be
hosting
an
ambient
end
user
feedback
Summit
that
we've
been
talking
about
some
time.
We've
chosen,
Tuesday
August
15th
from
9
to
11
that'll,
give
us
two
hours.
We
are
looking
for
three
to
five
end
users
who
have
used
ambient
in
a
significant
way,
such
as
a
staging
environment.
G
Obviously
we're
not
asking
our
users
to
put
this
into
production
yet
so
we're
not
looking
for
production
experience,
but
we're
looking
for
something
Beyond
book
info,
like
I,
took
my
Dev
environment
that
contains
a
full
copy
of
my
service.
Mesh
and
I
deployed
ambient
on
it,
so
I'm
reaching
out
to
users
to
sort
of
curate
a
list.
G
If
you
have
users
in
mind,
if
you
would
send
them
to
me
or
or
give
me
some
referral,
information,
I'll
work
together
with
you
and
then
at
this
point
like
we're
looking
for
for
users
and
I,
don't
expect
to
have
too
many,
but
running
them
through
me
keeps
us
from
accidentally
inviting
say
9
or
12
users,
which
would
sort
of
overwhelm
the
summit.
If
we
do
have
too
many
users
interested,
we
can
always
run
a
second
Summit
at
a
later
time
to
collect
more
and
different
feedback.
G
Also
I'll
be
working
on
a
list
of
questions
for
those
users
to
sort
of
prompt
their
responses
fairly
open-ended,
because
we
want
to
collect
a
lot
of
feedback
and
spend
most
of
our
time
with
our
users
talking
about
their
experience
rather
than
us
talking
to
them
again
look
forward
to
any
help.
The
community
wants
to
provide
on
that.
A
So
Mitch
the
the
ask,
is
for
the
community
to
get
in
contact
with
you
and
provide
you
with
at
least
two
users.
G
Yeah
I
mean
I
I
have
some
that
I'm
pursuing
on
my
own,
but
if
there
are
some
users
whose
feedback
you
think
would
be
particularly
valuable,
please
do
refer
them
to
me.
Reach
out
to
me.
Let
me
know
and
we'll
coordinate.
A
B
There's
a
link
if
you
scroll
up
a
little
bit
more
Francis
the
resource
annotations.
Oh
this
guy
yep.
C
B
B
So
quick
question:
the
revision
stop,
maybe
I'm
I
dropped
at
it
and
it
worked
for
a
little.
While
did
we
come
to
a
conclusion
on
Waypoint
revisioning?
It's
just
going
to
be
tied
to
the
control
plane.
D
It's
the
same
as
gateways
which
is
so
you
label
the
Gateway
object,
and
it
will
be
controlled
by
that
revision.
It's
control,
plane.
D
That's
correct,
yeah.
B
B
C
The
other
thing
I
think
we
will,
let's
just
talk
about
real
world
usage,
maybe
instead
of
the
alpha
beta
labeling,
because
we
haven't
progressed,
haven't
spent
as
much
time
progressing
these,
but
it
certainly
gives
us
a
reasonable.
B
C
C
E
G
C
C
C
B
Yep
there's
a
CA
certificates
Pim
that
will
give
you
extra
root
certificate
for
workload
to
workload
and
I.
Think
for
zero
downtime
route
start
rotation.
You
need
that
or
the
accompanying
it's
also
in
the
mesh
convicting
map.
So
you
could
use
mesh
config
to
take
care
of
that
as
well.
Instead
of
an
annotation
right.
C
C
D
Yeah,
it's
pretty
minimal,
I!
Think
it's
mostly
really
just
bootstrapping
the
connection
to
East,
Jody
and
I
mean
the
reason
it's
so
low
is
really
that
it
has
one
important
field
which
is
an
arbitrary
environment
variables,
and
then
we
shove
a
bunch
of
stuff
in
there.
So
it's
only
so
small
because
of
that
escape
hatch,
I
think
but
yeah.
A
lot
of
the
stuff
was
kind
of
envoy,
specific
or
Telemetry
related.
A
Either
runs
on
the
other
baby.
Can
someone
else
take
over
the
presentation.