►
From YouTube: Istio User Experience Working Group April 16, 2019
Description
Istio User Experience Working Group session April 16, 2019
A
So
I've
opened
up
the
meeting
notes.
Please
add
your
name
to
the
attendees.
What
we
have
for
it
is
we
we
need
to
make
an
epic
for
what
we're
going
to
do
for
1.2.
We
have
75
open
issues
which
is
large
for
a
working
group
and
we
need
to
decide
what
going
into
one
to
I
wanted
to
show
a
demo
of
a
misty
Oh
control
to
Direction
rule
for
common
sto
tasks.
That
I
have
been
experimenting
with.
A
We
have
a
user
survey
that
we
want
to
send
out
and
I
wanted
to
get
feedback
on
the
kind
of
questions
around
that
I
did
this
with
ROM
John
was
asking
about
the
operator
I.
Don't
have
any
information
on
that?
Maybe
someone
else
does
the
rule
manager
Democrats
been
postponed,
I
want
to
ask:
why
is
TOC
Rd,
don't
have
a
status
field,
but
there
may
not
be
time
for
that.
A
Very
few
of
them
are
bugs
they're,
mostly
things
that
or
if
the
air
box
things
that
I
don't
know
how
to
fix
or
how
to
close
like
disc
refiner
thing,
I
think
has
since
been
fixed,
I'm,
not
sure
if
the
reporter
goes
not
to
checks
it
in.
So
we
need
to
sort
of
wrangle
these
down
into
what
we
want
to
do
for
one.
Some
of
them
have
been
set
one
by
me
or
others,
but
we
want
to
make
an
epic
listing
all
of
the
ones
that
we
want
to
do.
A
Flying
is
a
you
know:
there
will
be
a
get
above
epic
I'm
going
to
have
trouble
typing
and
talking
at
the
same
time.
If
someone
wants
to
write
down
these
lead,
you
are
right
he's
not
as
soon
as
we
go
through
them
and
then
create
the
epic
sure
perfect
it.
We
can
go
back
and
fix
up
the
epic,
but
we
want
a
list
of
sort
of
what
we
want
to
do.
There's
75
items
here
since
there
it's
going
to
be
tough
to
discuss
tomorrow,
but
maybe
there's
some
kind
of
first
divisions.
We
can
do
blowers.
A
A
It's
not
telling
them
the
fun
of
that,
so
that
it's
unclear
how
if
we
want
to
keep
all
these
under
users
experience
or
break
these
down
in
some
kind
of,
like
maybe
more
than
one
epic
like
epic
for
helm,
stuff,
epic
for
our
control,
plane,
telemetry
components,
configuration
or
if
we
want
to
double
tag.
Some
of
these
for
configuration
or
that
sort
of
thing
does
anybody
have
any
thoughts
on
that?
A
C
A
I
tagged
everything
because
I
thought
that
was
the
rule
that
was
in
a
few
days.
Someone
had
a
tree,
I
did
is
being
part
of
their
area,
and
then
we
had
to
get
someone
on
it
so
sudden
they
produced
a
problem
when
I
realized
I
had
no
idea
how
to
fix
this
crap
on
a
thing,
for
example-
and
you
know,
I
reached
out
and
I'm
like
I
need
some
help
and
someone
said
it:
they
tried
a
new
version
of
your
phone
and
iframe
works
for
them.
A
A
D
So
I
have
just
question
it
like
such
kind
of
basket
is
part
of
the
user
group
coconut,
because
what
I'm
asking
here
about
is
a
for
example
like
stuff
like
help
shot
installation,
we
got
a
new
SEO
answer
which
is
working
progress,
and
so
perhaps
the
sandbox.
We
should
go
like
to
API
stuff
like
that,
for
example,
I
have
pushed
them
in
something
today.
Oh,
we
just
put
a
link
in
the
chart
to
the
help
shot
and
the
guy
asked
me
to
push
it
to
the
nuance
product.
D
E
A
A
D
A
D
F
As
an
area
of
user
experience
like
the
garbha
distinct
seems
to
be
a
plug
right,
more
than
a
kind
of
a
bigger
experience
to
share
a
big,
bigger
design
issue.
Right.
Do
you
want
to
be
tagging
those
things
as
user
experience,
rather
than
maybe
giving
it
to
the
telemetry
group,
which
I
think
Griffin
and
from
at
issue?
I
would
follow
into
that
bucket
right?
F
C
A
path
here
outside
of
anything
work
is,
is
if
you're
annoying
a
user.
It's
a
user
experience
problem
again.
You
have
to
sort
of
scope
this
to
what
are
fundamental
features
and
things
that
be
running.
We
need
to
do
to
just
make
the
overall
experience
better
and
then,
when
you
started,
look
at
all
these,
they
should
be
assigned
to
whoever
the
team
is
responsible
for
a
completely
no
fundamental.
A
Issue
this
key:
there
is
no
catcher
for
health
install
bugs
and
there
is
no
catch.
There
was
no
catcher
for
is
do
control
bugs
so
I
started
just
tagging
both
of
those
as
user
experience
figuring.
They
were
not
anything
else,
so
we
can
revisit
that.
Let's
try
to
basically
work
back
here
and
maybe
decide
what
has
to
go,
and
one
two
so
I'm
really
big
on
this
templates
for
coming
this
geo
test,
I'm
going
to
show
it
down
with
that.
So
I'm
not
going
to
triage
that
right
now,
a
classic.
C
A
C
C
F
Still
cuddle
used
to
be
on
in
the
tag
the
area
configure
I,
so
we
actually
owned
it.
So
I
think
it
goes
well.
You
can
use
a
day
user
experience
to
track
stick
on
the
related
stuff.
It
probably
makes
a
bit
more
sense
right
so,
but
that
pretty
much
will
be
the
only
component,
especially
in
that
sense
that
truly
belongs
to
the
user
experience.
Here.
F
A
A
A
A
H
A
Okay,
well,
Erica
I've
got
a
PR
that
just
went
through
through
there
and
so
I've
got
a
bit
of
a
working
relationship
with
Steve.
If
you
want
to
send
me,
a
link
to
the
PR
I
can
give
him
a
kind
of
a
friendly
side
thing
on
it.
Okay,
thank
you
much
so
we'll
skip
through
final
one.
The
outlier
check
is
not
that
sore.
That's
it.
He
writes
pretty
Berg
health
check,
rewrite
I,
don't
understand
this
one,
but
it's
been
double
tagged
as
security
and
user
experience,
I'm,
not
sure
why
the
smaller
is
tio
control.
A
Do
we
want
to
target
that
four
one?
Two,
it's
not
particularly
important.
It's
just
a
nice
thing
to
have
what
would
its?
What
is
it
smaller
at
what
there's
two
things
one
is
we
just
found
that
it
would
be
20%
smaller
if
we
compiled
it
with
a
different
link
or
option,
so
that'd
be
nice
to
do
and
then
we
found
we
could
make
it
much
smaller.
If
we
run
this
to
learning
climate
running
the
tool
doesn't
help
much
because
when
we
download
the
thing
we
downloaded
zip
to
anywhere
in
the
appointment.
A
C
F
A
A
F
A
So
when
we
say
that
we've
adopted
is
to
use
CTL,
we
don't
mean
that
we've
all
adopted,
like
the
pilot
debug
endpoint,
that
underlies
a
lot
of
the
SPOC
CTL
proxy
commands
right.
That's
fill
firmly
in
the
networking
area,
I
kind
of
feel
like
the
low,
maybe
maybe
is
used
mostly
as
for
the
user
experience
and
Fritz
do
control,
because
it's
the
you
want
pilot
fun.
A
C
A
Four
one,
seven
two
one
two
I
looked
at
the
pointy
plans,
know
them.
Someone
has
said
that
some
networking
means
precisely
so
it
is
going
to
be
considered
and
multi-protocol
still
have
to
considered
UX
I
think
because
we
are
targeting
UX
and
stability
for
one
two
and
SEO
control
improvements
are
one
too,
so
this
epoch
is
supposed
to
be.
A
Basically
these
things.
The
thing
is
related
to
too
late,
too
simplified
mesh
Federation.
This
do
control
improvements,
because
one
two
is
supposed
to
be
a
smaller
thing.
You
had
a
lot
of
new
features,
mostly
testing,
so
maybe
viewing
this
sidecar
configuration
is
out
of
scope.
Maybe
how
are
done
or
done,
however,
has
already
done
it.
A
A
A
C
A
A
Update
the
validation
tool
to
work
on
the
new
style,
CRTs
I
think
is
probably
the
key
website
item
so
and
I
think
we
do
want
this
validation
tool
for
one
two.
So
Lee,
can
you
add
one
two,
seven,
seven
three,
who
are
epic
perfect
gateways
sample
invalid,
makes
ingress,
go
stale
open
this
one
I,
don't
know
it's
a
bug
so
I'm
going
to
skip
it
for
the.
A
Maybe
it's
got
the
confrontation
skin
I'm.
It's
do
control
version.
This
one
I
want
some
wisdom
from
the
community
on
so
with
one
one.
We
get
two
new
options
on
SD
official
version,
remote
and
short,
but
no
one
knows
they're.
There
I
only
discovered
what
I
was
doing,
testing
of
all
the
options,
and
it's
really
nice
I
encourage
everyone
to
use
it.
It
looks
so
much
nicer
to
do
this.
A
A
B
A
A
It's
my
feeling
that
one
thing
we
should
focus
on
for
one
from
SEO
control,
point
of
view
is
commands
to
show
you,
the
multi
cluster,
set
up
that
you've
done
with
commands
like
multi
cluster
inflow
multiplexer.
This
networks
I
find
it
frustrating
that
you
choose
these
things
when
you're
distilling
with
helm,
and
you
have
no
way
of
seeing
what
you
have
chosen
later
on
any.
C
A
Of
this
originally
for
multi
cluster,
because
I
had
worked
on
some
other
multi
cluster
stuff
that
never
made
it
in
last
year.
So
the
idea
is
first,
you
want
to
get
some
information
about
the
configuration
you
also
may
want
to
change
some
things.
It's
pretty
easy
to
add
networks
and
delete
networks,
given
the
a
bit
without
reinstalling
from
helm.
But
do
we
want
to
put
that
in
sto
control
or
did
want
the
users
to
redeploy
to
helm?
A
There
is
the
generate
service
entry
which
had
worked
on
before
and
that
pair
had
died,
but
could
be
redone
again
to
my
new
generation
piece
and
in
other
commands
that
people
have
asked
me
for
expose
the
remotes,
exposing
services
or
demote
users
list
remote
services,
consumer
remote
service.
All
these
commands
related
to
to
setting
up
gateways
and
exposures
I,
don't
know
how
much
this
we
want
to
expose
I
feel
like
we.
We
want
more
than
we
have
now
yeah.
C
You
know
what
options
and
by
which
features
are
like
turned
on
I,
have
a
gateway
or
not
things
like
that.
So
I
think
we
need
one
one
issue
here:
it
really
capture
the
problem
of
sort
of
making
the
user
experience
of
install,
and
you
know
understanding
what's
installed
for
a
multiplex
or
not
so
it
would
be
better.
So.
C
D
C
Generally,
I
think
the
problem
is
when
you
do
helm,
install
up
according
to
the
way
the
documentation
tells
you
to
do
it.
You
never
actually
even
saved
the
help
file.
The
installation
config
you.
If
you
you
know
you
get
a
help,
file
you're
automatically
applied
based,
we
pipe
it
into
apply.
You
try
to
apply
and
you're
done,
so
you
don't
even
see
that,
but
but
more
generally,
I
think
that
that's
too
little
it
you
know
it's
low
level
stuff
in
there
like
what
these
settings
are.
C
They
say
like
eyes,
because
you
know
in
the
home
Tempest
is
all
these
different
settings
that
you
can
turned
on
or
off
and
depending
on
those
a
bunch
of
things
happen
or
not.
I
think
we
need
to
understand
better
how
to
convey
to
the
user
features
here,
you're
getting
and
turning
on
and
get
you
the
anomaly's
motive.
D
C
A
That
the
user
who's
using
us,
you
may
not
be
the
administrator
who
installed
it,
so
they
might
not
know
what
helm
options
were
chosen
and
it
might
be
hard
to
talk
about
and
did
I
install
it
with
remote
networks
did
I
install
it
with
MPLS
all
that
stuff
needs
to
be
exposed
in
a
way
that
makes
it
easy
to
make
another
one.
That's
the
same.
Okay,
I
think.
A
A
Right
so
I'll
try
to
have
these
things
better
characterized
for
the
next
meeting.
Maybe
when
we
do
this
again,
I
see
we've
already
spent
half
an
hour
on
it.
We
have
given
Lee
very
few
items
to
add
to
the
thing,
but
maybe
leak.
Maybe
you
make
the
epic
with
those
items
that
we've
done
and
then
we're
going
to
be.
B
A
I
D
A
C
C
A
So
one
thing
that
people
keep
asking
me
for
is:
why
is
it
so
difficult
to
do
common
tasks
in
SEO,
exposing
something
we're
like
oh
go
to
the
sample
file
and
take
book,
and
so
I
start
editing,
it
or
I
say
that's
said
some
weights
and
they
go
ok.
Well,
I
spent
half
an
hour
on
that
and
my
tabs
were
screwed
up
and
I
didn't
figure
out
what
I
was
doing
so
I
wanted
to
create
a
easy
way
to
let
us
supply.
A
Tasks
or
ability
with
sto
control
or
some
tool
to
create
these
things
and
I've
been
playing
around
mostly
with
both
edge
service,
so
I
have
a
sort
of
set
of
generators
here.
This
is
all
very
ad
hoc
and
simple
right
now,
so
ignore
sort
of
how
I've
implemented,
but
I
have
this
explosive.
This
generator,
they
can
sort
of,
say,
generate
exposed
service
service
equals
product
page
and
when
you
do
that
it
generates
the
mo
for
for
doing
that.
That's
a
nice
gateway,
I
called
product
page
gateway.
A
There
is
a
virtual
service,
the
products,
the
correct
port,
but
it's
sort
of
figured
out.
If
you
set
gateway
equaled,
some
existing
gateway
that
exists
it
will.
It
will
refer
to
that
properly.
But
if
it
sort
of
doesn't
exist,
it
will
complain
so
if
I
say
sit
gateway
and
misspell
it,
it
will
come
back
and
say
well
that
is
not
one!
I
A
So
what
the
implementation
of
this
is
is
somewhat
strange.
It
is
in
a
few
pieces
because
I
wanted
to
see
what
I
could
do
so.
The
first
thing
I
did
was
I
wrote,
helm
for
this
of
the
output
that
you
saw
when
I
did
this
command,
you
know
literally
gives
a
source.
It
has
all
that
stuff
that
you
see
heritage
and
all
that
stuff
that
comes
out
of
home,
because
I
implemented
this
as
a
helm
piece.
A
C
A
So
that's
the
next
piece,
so
when
I,
the
first
I
first
tried
to
tackle
you
said
the
other
one,
because
it
has
empty
LSD,
it's
the
versions,
one
which
is
in
progress
right
now.
So
when
I
first
did
this
with
helm,
it
was
fun
and
easy
to
do
this,
but
it
didn't
do
any
checking
of
any
kind.
It
didn't.
You
know,
validate
his
parameters
at
all,
so
I
wanted
to
do
a
quick
and
dirty
validation
thing.
So
I.
A
Made
a
like
a
validation
and
I
put
those
in
config
maps,
because
this
geo
control
is
good
at
looking
at
them.
So
let
me
show
you
the
config
map
that
I
have
for
exposed
service,
so
actually
it's
it's
somewhat
poured
right
now.
First,
there's
a
template:
this
template
is
just
a
helm,
command
line,
and
you
know
if
we
don't
like
helm.
I
can
replace
this
with
a
full
template.
The
way
that
the
injector
does
we'll
put
in
the
incomplete
template
in
the
config
map
itself.
A
There's
parameters:
it's
not
structured
right
now,
because
that's
aren't,
but
it
sort
of
says
the
first
parameter
is
a
v1
service.
The
second
parameter
is
an
INT.
The
third
parameter
is
a
networking
v1
alpha
three
gateway
and
that's
all
checked,
and
there
are
also
defaults
that
are
expressed
in
JSON
path,
so
that,
if
you
don't
specify
the
service,
it
will
go
out
looking
at
the
results
from
validating
the
types
and
let
you
pick
up
defaults
that
way,
so
this
turned
out,
at
least
for
the
ones.
A
G
F
If
you
can
describe
what
these,
what
the
co
IDs
look
like
with
these
simplified
templates,
but
does
that
mean
we
should
actually
have
this
higher
level
CRD
with
a
controller
that
dynamically
produces
the
lower
level
c
or
DS,
but
so
effectively
what
this
template
is
doing,
except
it's
done
in
a
static
way?
Yes,.
A
Well,
Reiter
already
the
reason
I
don't
think
it
should
be
that
way,
it's
a
whole
nother
layer
of
things,
for
people
to
learn
and
to
know
which,
which
things
are
high
level
in
which
things
are
low
level,
and
it
also
puts
us
in
a
straightjacket
of
well.
The
high
level
stuff
is
going
to
be
a
particular
way.
What's
this,
the
reason
I
did
this?
I
C
I'm
confused,
which
users
are
we
talking
about?
There's
a
user
who
wants
to
use
this
Chi
wizard
thing
to
get
started:
I,
don't
think
that
user
sees
any
of
this,
it's
the
it's
or
do
you
expect
those
users
can
go
on
at
their
own
custom
kind
of
stuff
into
this?
Using
this
I
thought
this
is
sort
of
lovers.
Adding
new
wizard
features
like
I,
said:
I.
A
Don't
expect
anyone
is
going
to
look
at
this,
except
for
users
who
want
to
add
things
I
expect
most
people
are
just
going
to
say,
generate
or
list
generators.
It's
going
to
say
we
have
these
things,
they'll
be
documented
and
we'll
say:
oh
you
want
to
deploy
your
product
page,
do
generate
expose
service
for
service
equals
product,
page
and
you're
done.
A
A
So
it
was
supposed
to
be
not
a
sick,
servicing
API.
It
was
supposed
to
be
a
better
way
than
sample
so
right
now
the
API
is
I'm,
just
going
to
look
a
little
local
sample
directory
and
something
close
and
then
start
editing
it,
and
this
was
supposed
to
be
a
faster
version
of
that,
but
maybe
others
don't
see
it.
That
way.
A
A
So
I've
asked
people
what
are
the
common
tasks
and
I
got
a
sort
of
list
of
them
which
I
started
working
on
I
talked
I,
I've
talked
about
add
to
measure
think
at
another
one
of
these
meetings,
but
I
guess
the
quartet
destination
rule
is
a
particularly
tricky
one,
because
when
you
deploy
new
commitment
or
we
delete
a
deployment,
you
sort
of
change
your
subsets.
These
promote
things
that
people
were
startling
with.
It
seemed
like
an
easy
way
to
get
them
there.
A
A
Maybe
the
people
that
reached
out
to
me
for
help
aren't
as
skilled
as
the
people
who
are
reaching
out
to
others
for
help,
because
but
these
are
the
problems,
I
often
see
and
I
thought.
Maybe
if
I
made
it
easy
for
people
to
create
any
sorts
of
things,
people
who,
like
a
developers
would
contribute
them
have
been
easy
routes
to
getting
them
into
is
do
control
the
digital
king
or
custom
code
to
be
written.
A
E
A
So
maybe
we
want
to
just
do
a
command
line
to
make
it
easier
to
do
key.
Ollie
I,
don't
know
so.
I'll
take
a
look
at
those
Tet
items
after
the
meeting.
So
the
next
item
we
have
up
is
the
user
survey,
so
these
are
survey
ramen
I
worked
on
and
it's
what
we
want
to
send
out
to
certainly
ideas
customers,
but
maybe
everyone
using
is
to
know
via
discussed
in
all
these
places,
to
sort
of
get
feedback
on
who
is
using
sto
when
what
is
what
is
not
there.
A
E
E
A
A
How
do
you
feel
about
the
terminology?
Do
you
this
do
control
what
area
of
SEO
needs
the
most
improvements
bunch
of
choices
here?
Would
you
recommend
this
do
to
others?
What
we're
hardware
platform
is
where,
where
is
running
SEO
in
terms
of
technology,
it
also
sort
of
in
terms
of
topology
cloud
type
and
scale.
I
E
F
I
E
I
H
A
H
Was
asking
about
it
he's
on
the
call
there
was
an
sto
operator,
I
thought
it
was
a
kick
off
or
I.
Think
some
folks
are
talking
about,
adding
it
to
I.
Don't
know,
if
is
one
two
or
after
I,
don't
know
current
about
it,
which
is
what
I'm
asking
what
you
know.
What
is
how
are
we
going
to
be
using
operators
if
we
are
at
all
okay.
F
So
the
plan
is,
we've
been
working
to
define
a
new
CRD
that
describes
an
Estill
deployment
and
the
operators
purpose
in
life
is
to
read
the
co
D
and
then
actor
lines
it
so
deploy
the
right
components
with
the
right
option
to
support
that
CRD.
The
plan
is
to
make
that
happen
in
the
for
the
1.2
in
the
1.2
timeframe.
It
is
supposed
to
be
the
only
major
feature
that
1.2
adds
is
this
operator.
F
F
F
A
So
the
only
other
item
I
have
on
the
agenda.
We're
finished
with
operator.
People
have
been
asking
and
I've
been
wondering
why
SEO
series
don't
have
a
status
field,
so
this
this
it
has
to
do
a
bit
with
validation.
So
we
have
5000
validation
and
does
a
great
job
for
single
items.
We
have
tools
like
EPP
and
mesh
that
will
sort
of
improve
your
complete
thing.
A
F
But
there
are
a
couple
of
there
are
multitudes
of
issues.
The
first
one
is
for
the
first
line
of
defense,
where
you
know
you
look
at
a
bunch
of
CI
knees
together
and
see
whether
they're
going
to
work
or
not
like
that
kind
of,
like
a
composite
validation.
I
think
makes
a
lot
of
sense
to
implement
this
in
Delhi
and
just
expose
that
information
that
you
haven't
done
it
yet.
F
The
second
set
of
problems
is
the
pilot,
can
actually
have
a
bit
more
of
the
runtime
information
that
goes
into
that
decision,
so
that
information
is
to
be
percolated
up
to
galley
so
that
it
can
be
reflected
on
the
status
fields
right.
So
there's
some
casualties
in
the
MTP
protocol
to
enable
this,
but
you
know
we
haven't
done
the
probe
implementation
of
it.
We
are
still
like
you
know,
making
galley
as
a
stable
component
as
like
a
40
for
the
forward
path,
and
it
turns
into
third
set
of
problems.
F
D
F
F
A
Group
all
right,
I'm,
very
interested
in
what
sorts
of
things
are
going
to
be
there
and
what
sorts
of
things
will
be
able.
So
what's
going
to
be
in
the
status
field
and
then
who's
gonna
be
allowed
to
report.
So
it
would
be
great
if
either
someone
who's
in
the
config
group
could
summarize
this
and
tell
us
here,
because
I
think
that's
going
to
be
how
the
users
are
going
to
start
seeing
problems
that
are
the
single
item.
Validation
can
check
so
just
same
way:
the
pods
I'm.
A
F
It's
Norway,
it's
going
to
be
in
one
two
want
to
say
about
code,
mall
and
there's
a
very
limited
set
of
things
that
we
can
do
for
that.
We
also
need
to
do
an
internal
defect
in
your
galley
so
that
some
of
these
features
can
be
implemented
and
I'm
trying
to
get
it
done.
The
advantage
and
prey
the
most
likely
time
frame
is
1/3.
F
This
is
up
being
a
pretty
high
priority
thing
to
tackle.
We
have
partners
like
a
native
that
want
to
build
stuff
on
top
of
us
and
because
there's
no
way
to
discover
status
of
the
system.
It's
they
end
up
just
putting
time,
delays
and
hoping
for
the
best
to
try
and
orchestrate
thing,
and
that's
just
not
a
robust
strategy.
A
F
So
I
leave
a
so.
The
estimation
of
folks
were
actually
working
on
some
validation
related
stuff,
and
he
actually
you
know
we
could
kind
of
get
in
contact
to
kind
of
get
their
stuff
into
the
mainly
story
phone
right
and
if
you
can't
get
it
in
a
reasonable
way
right
where
you
know
a
set
of
proto's
like
a
graph
of
objects,
then
we
can
analyze
them.
F
You
know
to
having
it
in
that
way,
then
I
think
we
can
use
that
as
a
library
board
in
East
your
cuddle,
as
well
as
in
galley,
so
that
in
Easter
cuddle
you
can
actually
say
here's
a
bunch
of
yellow
files.
Please
go
and
see
if
they're
valid
or
not,
and
then,
if
you
indicate
itself,
we
can
actually
look
at
a
snapshot
of
configuration
and
see
a
you
know.
Given
these
set
of
objects
is,
are
they
valid
together
or
not,
and
one
path
can
be
like
you
know,
exposing
that
as
the
status
fields
right.
F
F
I
A
However,
so
quickly
many
times
the
proxies
come
stale
with
things
that
is,
if
that
can't
detect,
there
needs
to
be
some
way
that
when
we
can
AK
from
a
site
are
we
can
sort
of
log
it
in
a
way
that
we
can
go
back
and
say
well,
look
at
Co
vet
done
a
pilot
or
look
Valley.
You
didn't
detect
this
problem
that
envoy
detected.
We
need
a
way
to
make
sure
that
that
gets
caught
and
the
only
way
we
can
catch
it
now
is
who's
next.
F
So
there
is
a
okay.
This
has
been
a
long-running
discussion
among
various
people.
It
is
a
hard
problem
perfectly
this
information
up,
so
that's
no,
it
didn't
eat
you
venture
boy,
I,
believe,
there's
a
time
delay
like
you
know,
one
of
the
things
that
we
want
to
it
able
to
answer
the
question:
hey
I
push.
This
configuration
is
active
or
not
right,
so
it
starts
from
there
and
from
there
you
can
actually
say
look.
They
say
if
you
are
getting
an
act
from
envoi.
F
A
F
Right,
so
it's
more
like
it,
you
know
you
should
make
it
easy
to
figure
out.
Okay,
this
problem
is
happening,
so
somebody
goes
and
opens
it,
but
for
us
right,
because
an
AK
from
n
voice
should
be
rare,
so
it
like
Danny
were
talking
with
for
me
I.
Think,
like
his
point
was
this
should
be
an
exceptional
error.
That's
coming
from
annual
lack
is
kind
of
reserved
for
that.
So
then
you
know
if
he
fertilize
up
is
the
nice
to
bug,
and
then
we
can
fix.