►
From YouTube: Layer5 Community Meeting (Feb 26th, 2021)
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
A
A
B
A
Yeah
there's
there's
a
community
member
who's
got
his
his
profile
under
pr
right
now
on
the
layer,
five
io
site,
his
name's
nick
jackson
and
he's
out
of
the
uk
as
well
he's
in
london.
I've
been
listening
to
him,
piss
and
moan.
We
talk
every
week.
He
just
talks
about
the
weather.
He
talks.
He
lives
in
the
he's
like
victorian,
older
style,
victorian
homes
and
he's
his
office.
A
Space
is
set
up
like
right
next
to
this
window,
that
to
hear
him
speak
of
it,
it's
the
most
cattywampus
like
wonky,
fitting
like
obtuse,
not
not
rectangular
but
obtuse-shaped
window,
because
the
house
has
like
shifted.
You
know
over
time,
and
so
he
just
has
this
constant
draft
guys
guys
the
biggest
whiner
I'll
tell
you.
A
Yeah
another
friend
of
mine,
also
in
the
uk,
he's
cto
at
docker
now
and
justin
cormac
also
continuously
taking
pictures
of
the
weather
outside
he
just
like
every
other
tweet
from
him
is
a
an
outside
photo
of
so
good.
So
now
that
I'm
done
complaining
we're
five
minutes
after
was
it
friday,
the
26th,
so
last
one
of
february
and
we've
got
a
number
of
things
to
chat
through
so
welcome.
Everybody
is,
is
anyone
here
for
the
first
time?
A
Has
anyone
not
had
a
chance
to
introduce
themselves
on
the.
A
A
I
don't
know
I
don't
know
whether
to
be
really
just
impressively.
There's
like
20,
20,
old
timers,
on
the
call
or,
if
that's
a
bad
sign,
we're
we're
we're,
not
wait.
A
congregation
will
die
if
we
don't
get
some
new
blood.
If
you
don't,
we
have
to
start
feeding
on
ourselves.
I
think
so.
A
Okay
is
any
root
on
you
know.
I
was
just
talking
to
him.
No
okay,
there's
this
letter
letter
of
wreck,
something
something
happening
over
here.
So,
okay,
I
was
going
to
tease
him
about
it
all
right,
good
wow,
what
a
long
way
to
get
started
there
are.
Aren't
any
newcomers
on
this
call
at
the
moment,
but
there's
been
a
number
who've
joined
in
the
community.
Recently
we
had
given
an
a
workshop
at
istio
con
on
monday,
and
we
might
hear
a
bit
from
adina
about
that
today.
A
A
It
had
about
four
fourth
hour
little
just
shy,
four
thousand
attendees.
Basically,
each
time
I
got
on
there
were
about
that
many
which
is
which
I
guess
you
know
it's
a
covid
number.
It
takes
that's
a
very
high
number
for
first
for
an
inaugural
event,
although
with
virtual
events
and
them
being
more
or
less,
you
know
a
lot
of
times
free,
it's
easier
to
get
some
some
additional
butts
butts
in
the
seats,
but
does
speak
to
the
interest
level
around
istio
around
surface
match
good,
so
welcome
so
there.
A
The
reason
I
mention
that
is
because
there's
a
few
newcomers
who
joined
based
on
having
taken
the
workshop.
So
if
all
of
you
are
hopefully
very
familiar
and
if
you're
not,
I
really
encourage
you
to
go.
Take
this
workshop.
There's
two
of
them.
They
are
self-study
workshops,
they
teach
people.
So
here's
the
one
that
we
did
on
monday
and
they
teach
people
about
well,
especially
well.
They
teach
people
about
service
meshes
this
one's
specifically
istio
and
they
do
it
by
using
measuring.
A
So
we
use
measuring
as
a
teaching
tool
and
that's
part
of
the
genesis
story
for
measuring
is
as
a
teaching
tool.
So
so
one
your
works
were
being
shown
off
and
used,
which
is
great,
actually
husseina.
Who's
on
the
call
is
engaged
with
one
of
those
users
and
helping
figure
out
some
trouble
that
they
were
having
with
gke
and
mesheri.
A
I
had
a
second
point
in
there.
Oh,
but
it
was,
I
guess,
two
go
go.
Take
the
workshop.
It's
you'll
find
links
in
here.
There's
it's
self
study.
It's
there's
a
couple
of
there's
three
repositories
in
the
layer:
five
io
repository
where
it
where,
if
you
spend
some
bart,
actually
says
you're
top
of
mind
as
you
go
to
familiarize
more
with
the
domain
space
and
part
of
the
technologies
used
so
really
and
get
more
familiar
with
meshri
and
and
and
you
begin
to
affect
how
it
works.
A
A
Good
good
good
good,
all
right
well,
does
anyone
else
have
announcements
before
we
get
into
topics?
Does
anyone
else
have
topics
to
add
or
things
to
talk
about
stuff?
That's
happened
to
them
things
they've
seen
in
the
news,
new
tech
that
they've
stumbled.
D
C
E
C
E
Tech,
so
that
was.
E
A
Well
with
that,
let
me
let
me
hand
it
off
to
you
for
the
next
topic.
D
All
right
sure
so
yeah,
I
guess.
D
What
was
shared
during
the
easter
corn
workshop,
I'm
not
too
sure
so
I'll,
just
it's
just
a
quick
recap
and
I'll
just
also
similarly
go
over
some
changes
that
I
feel
would
make
the
process
more
seamless,
so
I've
gone
ahead
and
taken
the
liberty
of.
D
The
steer
service
mesh
already
and
my
internet's
a
bit
choppy
so
hopefully
the
application
as
well
deploys
for
the
purpose
of
the
demo
and
okay.
So
while
this
is
going
on,
essentially
what
I'm
doing
is
yeah,
I'm
just
deploying
the
image
of
application,
and
one
of
the
changes
that
I
have
introduced
first,
is
that
in
the
documentation.
D
There's
really
no
mention
of
how
to
sort
of
you
know.
I
mean
okay,
so
when
you
create
the
virtual
cells
right,
so
it
is
it's
it
directs
to
amazon.com,
but
not
it's
not
like
everyone
doesn't
have
this
ip
necessarily
in
the
hosts
file
already.
So
I
thought
one
good
addition
would
be.
C
Okay,
I
think
it's
give
me
a
second.
G
D
All
right,
I
think
this
is
this-
is
the
cake
that
I
was
working
on.
This
is
like
believe
me,
artwork
all
right,
so
I
just
talked
about
a
bit
more
of
what
I
had
in
my
mind.
D
G
D
D
And
web
image
versions
were.
A
A
One
of
the
one
of
the
projects
that
we
have
it
was
is
so
in
the
spirit
of
learning.
Again,
the
pro
is
a
project
called
image
hub
and
image
hub
is
a
sample
application.
A
A
Some
details
about
the
user
token
that
you
might
have
some
details
about
how
many
times
you've
made
a
request
recently,
maybe
whether
or
not
to
rate
limit
you,
maybe
you're,
being
abusive
of
the
site.
So
you
know
back
off
and
you
know
there's
a
lot
of
different
decisions
that
a
service
mesh
could
make
about
what
to
do
with
that
request.
A
Pass
it
along
expediently,
slow
it
down,
stop
it
change
it,
send
it
somewhere
else.
You
know
a
number
of
things,
and
so
this
sample
app
was
created
a
while
it
was
created
a
while
ago.
It
is
intended
to
demonstrate
the
fact
that
service
meshes
could
can
decide
what
to
do
with
that
traffic
by
by
incorporating
application
level
logic
or
or
business
logic,
and
so
the
sample
app
that
was
created
used
docker
hub
as
an
example.
A
So
we
created
this
sample
app
and
we
were
intending
to
give
we
are
intending
to
and
we
did
give
a
presentation
at
dockercon
so
that
we
made
a
sample
app
called
instead
of
calling
it
docker
hub.
We
called
it
image
hub,
just
sort
of
a
generic
rip
off
of
that
of
something
that
people
would
understand
and
what
we
were
showing
with
imagehub
is
to
intelligently
rate
limit
users
that
who
are
doing
a
making
pull
requests.
A
So
when
you,
when
you're
running
docker
and
you
wanna-
and
you
want
to
pull
down
a
container
image
by
default,
you'll
pull
down
that
container
image
from
docker
hub
docker
hub
has
gotten
fed
up
with
or
or
it
costs
docker
a
lot
to
persist
the
world
all
of
the
world's
images,
everybody's
images
that
they
build
and
push
and
pull
and
people.
Sometimes
people
set
up
continuous
integration
systems
to
pull
down
those
images.
A
lot
you
know
like
programmatically
and
docker.
Has
a
company
they've
got
to
pay
for
that
they
use.
A
You
know
aws,
they
use
s3,
object,
storage
to
store
those
images
and
we're
talking
about
the
world.
So,
like
it's
a
lot,
you
know
it
costs
a
lot
for
the
network
traffic.
It
costs
a
lot
for
the
storage
and
they've
begun
to
rate
limit.
How
frequently
p
you
know,
users
that
are
not
paying
for
a
subscription
they've
begun
to
rate
limit,
how
frequently
those
images
can
be
pulled.
A
A
Do
they
have
the
permission
to
pull
an
image
and
then
more
than
that,
if
you're
using
advanced
network
filters
these-
and
these
filters
are
written-
these
protocol
filters
these
packet
filters
they're
written
using
webassembly
as
a
well
they're
written
in
this
case,
they're
written
using
rust,
but
they
compile
to
webassembly
as
a
small
virtual
machine
that
small
virtual
machine
can
run
inside
of
a
service
mesh
proxy
by
convoy.
A
A
Do
they
have
a
subs
an
active
subscription
and
is
that
the
free
tier
or
is
that
the
enterprise
tier
and
you
know,
depending
upon
which
one
that
is
okay?
Let
them
continue
to
make
a
high
rate
of
pull
requests
or
don't
block
them
rate
limit
them,
and
so
that's
the
back
story
to
what
imagehub
is
as
a
sample
app
and
helping
teach
people
about
the
fact
that
that
you
can
expect
that
type
of
intelligence
from
a
service
mesh,
and
so
if
pranav
ends
up
joining
us
back,
yep
and
so
pranav.
A
A
Cool
we
will
we'll
circle
back
in
case.
Pranav's
connection
starts
to
work
later,
we'll
come
back
to,
but
the
not
to
I
won't
steal,
pronounced
thunder
the
gist
of
what
he's
working
on
is
to
take
that
sample
application
and
make
it
more
generic,
make
it
more
generally
usable,
because
when
we
wrote
the
sample
app,
we
did
really
fast
and
we,
you
know
hard-coded
in
some
values
around
the
enterprise
plan
or
the
free
plan,
or
look
up
the
user
and
expect
this
type
of
cookie.
A
But
maybe
maybe
we
should
be
expecting
a
cookie
or
a
jot
at
jwt,
or
you
know
so
he's
working
to
generalize
that
as
a
general
capability,
with
the
intention
of
enhancing
measuring
with
that
so
mesherie
as
a
service.
Mesh
management
tool
is
capable
of
doing
what's
pictured
here
so
so
measuring
over
here
on
the
left.
It
is
capable
of
on
the
fly,
inserting
a
filter
at
the
filter
of
your
choice
because
which
filter
which
webassembly
app
is
appropriate,
is
dependent
upon.
A
You
know
your
your
application
and
what
your
application
is
doing
and
so,
and
maybe
it's
not
just
one
filter.
Maybe
it's
a
couple
maybe
that
what
the
filter
that
you
need
to
run.
Maybe
it
changes
over
time,
maybe
the
filter
that
you
need
to
run
changes
based
on
the
user,
that's
making
requests,
and
so
anyway,
mesh3
is
going
to
both
help
dynamically
load
and
unload
filters
as
well
as
it
would
be.
A
The
point
of
configuration
for
if
you're
running
this
and
you
want
to
configure-
maybe
bart
has
signed
up
for
the
maybe
bart's
an
employee
and
he
he
does
he's
unrestricted.
Like
he's,
not
only
the
free
plan,
the
team
plan
or
the
enterprise
plan
he's
on
the
employee
plan,
so
just
all
of
his
traffic
gets
bypassed.
Well.
Measuring
can
be
that
point
of
configuration
for
on
the
behavior
of
these
filters,
so
pranav
is
working
to
make
that
a
more
generic
capability
questions
comments.
Ideas
on
this.
D
A
H
D
D
A
D
Yeah,
that's
because
I
stopped
minicube,
so
I
I
don't
think
the
demo
is
going
to
be
possible
today,
but
okay,
so
I'll
just
quickly
talk
about
the
issues
that
you
know
that
I
sort
of
found
it's
just
generally,
as
of
now
I'm
still
getting
into
the
wasm
filter,
part
of
it,
and
I'm
still
talking
around
with
that,
but
in
general,
for
future
workshops
and
so
on.
There
are
a
couple
of
issues
that
I
found.
So
first
was
the
documentation
on
the
image
hub.
D
Part
of
the
job
sampler:
this
is,
how
do
you
say
like
you
need
to
you
know
like
it
uses
imagehub.measure.io,
and
for
that
you
need
to
take
the
ingress
like
ingress
gateway
and
then
do
all
sorts
of
stuff
to
get
the
ip.
So
what
I've
done
is
that
I've
taken
one
of
the.
D
F
It
it
was
using
an
older
version.
D
D
The
versioning
for
the
images
is
all
of
them
are
just
v1,
so
I
have
to
again
make
some
changes
in
the
ammo
for
to
make
sure
it
pulls
the
latest
ones
with
the
latest
fixes.
So
versioning
has
to
be
handled,
and
the
third
thing
is
that
when
you
click
deploy
onward.
D
Basically,
it
runs
the
there's
a
file
located
situated
in
the
measuring
dash
repository
called
imagefilter.yaml.
So.
D
To
sit
and
manually
change
it
and
apply
the
custom
example
once
more
in
order
to
get
the
actual
thing
working.
So
I
was
thinking
that
you
know
instead
of
hardcoding
the
1.7
of
the
the
hdr
version,
you
could
make
that
dynamic
as
well.
So
you
will
alleviate
that
issue
as
well.
So
these
are
just
some
thoughts
on
you
know.
D
You
know
bettering
the
chances
of
being
deployed
successfully.
So
this
is
all
I
have
currently
I'm
working
on
adding
postgres
support
to
so,
like
you
mentioned
the
the
paths
and
all
currently
are
hardcoded,
so
you
have
an
enterprise
project
and
personal.
So,
instead
of
that
it
comes
from
the
postgres
database
and
I'm
still
working
on
how
to
make
the
volume
filter
sort
of
do
that
and
where
to
run
the
postgres
like
what
you
say
contain
our
service
or
whatever.
That's
all.
C
A
C
A
There
it
is,
I
guess,
to
be
launched
like
later
today,
I'm
going
to
go
right
up
four
different
internships,
probably
between
linux,
between
google
and
the
cncf,
and
this
will
probably
be
one
of
them.
A
A
You
ever
felt,
like
you,
had
to
sneeze
for
like
a
minute,
there's
four
of
them,
including
console
and
maybe
some
others
so
we'll
you
know
anything
that
you
do
against
something
like
against
the
envoy.
You
get
bang
for
your
buck,
which
is
good
so
good.
I'm
excited
of
the
two
requests
that
I
have
give
a
crack
to
give
a
try
on
the
read
me
if
you
would
second
to
that,
if
you're
able
to
record
in
your
demo
locally
and
send
I'd
for
sure,
I'd
be
interested
to
watch
it
so.
D
D
I
don't
know
if
you
saw
actually
when
I
was
sharing
my
screen
and
if
it
indeed
did
send.
I
have
like
a
text
document
just
jotting
down
all
the
changes,
all
the
tiny
things
I
need
to
add
and
eventually.
F
It's
going
to
be
a
mistake
where,
whether
or
not
it
will
work
on
council.
D
On
istio
for
sure
you
know,
any
beginner
can
pick
it
up
and
just
implement
it.
That's
my
end
goal
for
the
review
as
well
for
the
recording
immediately
like
tonight.
I
might
not
be
able
to
do
it,
but
within
the
next
two
days
for
sure
I'm
just
you
know
demo
it.
It's
not
gonna,
be
too
long.
A
And
so
yeah,
okay,
so
ukarsh
is
up.
Next
he's
been
working
on
an
area
for
and
enhancing
the
performance
management
capabilities
of
mesherie.
He
wants
to
get
some
feedback
from
everybody.
I
Yeah,
actually,
this
is
still
I
mean
the
ui
is
still
being
worked
on.
So
this
is.
This
is
a
familiar
page
where
we
basically
run
the
performance
states
in
mission,
but
one
thing
that
is
new
in
here
is
this
button,
which
is
disabled
for
now.
So
what
this
button
like?
I
What
I'm
adding
here
in
here
is
the
capability
to
create
a
profile,
a
performance
profile
from
so
that
we
can
categorize
this
and
basically
save
all
of
your
point
test
configurations
at
one
place
this
this,
but
the
model
that
is
being
presented
in
here.
It
doesn't
have
a
lot
of
things
right
now.
Only
things
that
it
like
that
is
capturing
right
now
is
performance.
I
It
is,
it
is
a
profile
name,
but
the
goal
is
to
actually
not
the
goal.
It
will
be
really
soon
added
in
here,
but
it
is
the
ability
to
specify
skills
in
here,
so
that
you
can
specify
your
that.
Basically,
this
is
your
performance
file.
It
is
this.
I
This
is
the
name
of
the
performance
reply
and
at
what
time
you
want
to
run
the
test
in
these
profiles,
along
with
all
of
the
configurations
that
you
have
specified
in
here
like
right
now,
this
I
mean
the
model
is
really
simple,
but
definitely
would
have
a
lot
of
more
features
in
here.
I
Yeah
so
just
like
earlier,
you
can
speak
about
all
the
things
that
we
used
to
it's
just
that
now,
although
on
the
front,
nothing
has
changed,
but
on
background,
or
now
from
now
all
of
the
tests
would
belong
to
a
performance
profile.
If
you
specify
it
good,
I
mean
you
can
reuse
it.
If
you
don't,
then
also
you
will
actually
capture
a
performance
file.
The
name
would
be
randomly
assigned,
but
yeah
we
will
be
capturing
so
that
we
can
just
click
on
like
something
you
said.
I
Okay,
we
ran
a
test
on
this
date
and
we
didn't
have
to
say
the
profile,
but
now
I
do
want
to
say
it.
We
can
go
ahead
and
click
on
it.
So
that's
the
reason
that
every
test
would
now
be
a
part
for
performance
to
fight
so
now
again
submitted.
I
Profiles
yeah,
so
here
you
are
like
I've
created
profiles
earlier
too
so,
and
I
run
you
test
so
now
you
can
like,
when
you
clicked
on
a
profile
this
test
profile.
It
has
the
right
now.
The
ui
is
actually
capturing
only
one
and
one,
but
you
would
soon
be
able
to
capture
a
lot
more,
but
these
fields
are
again
rigid
skills,
not
like,
not
necessarily
that
that
we
are
capturing
in
the
ui
there.
So
the
ui
of
now
is
that
you
have
a
list
of
profiles
that
you've
created.
I
You
can
click
on
the
profile
name,
and
you
would
visit
this
page
and
here
are
all
of
the
tests
that
would
run
under
that
profile.
So
this
test
was
I'm
naming
something
random.
I
didn't
choose
any
service
machine,
and
this
is
the
old
formula
you
want
that
we
had
for
the
test.
It's
just
that
now.
The
test
results
are
categorized
categorized
in
since
that
they
can.
All
of
them
are
kind
of
under
under
a
performance
profiling.
This
undefined
and
that's
not.
I
What
is
what
you
would
see
it
would
be
better
randomly.
This
is
the
bug
that
I'm
working
on,
but
yeah.
So
here
also
so
all
like
this
is
one
of
the
one
of
the
profiles
that
you
didn't
probably
have
created,
but
you
ran
at
this
and
it
just
captured
all
of
the
details
under
our
random
randomly
assigned
profile
name.
I
Oh
yeah,
actually
we're
still
working
profits,
but
this
is
almost
how
it
would
be
looking
like,
or
just
a
few
things
would
be
iron
man,
that
is
actions
thing,
is
not
working.
What
you
would
get
in
here
is
the
ability
to
change
your
profile.
That
is,
you
may
have
mentioned
and,
like
maybe
you,
you
mistyped
the
endpoint,
and
you
might
want
to
change
it
so
like
basically
updating
your
profiles
as
well
as
triggering
a
run
but
stuff
like
that.
D
So
I
just
had
one
question:
so
is
each
profile
going
to
be
like
you
know,
associated
with
one
run
or
is
it
like?
I'm
grouping
multiple.
I
In
the
profile,
so
when
you
are
creating
a
profile
you
are
specifying
all
of
the
at
the
end
point
your
service
mesh,
your
content,
request,
advanced
options.
All
these
things.
They
were
used
to
be
a
part
of
this,
and
now
they
are
part
of
profile
and
all
of
the
tests
that
will
be
triggered
from
that
profile.
They
will
inherit
those
configurations,
so
one
profile
is
going
to
have.
One
is
going
to
have
a
number
of
tests.
It
is
done
so
so
that
you
can
like
specify.
I
You
can
come
in
here
and
create
a
profile
like
fortnightly
tests
on
this
thing.
You
can
name
it
something
like
that
and
you
can
specify
a
schedule
that
okay
run
it
on
every
14
days
or
something
and
now
every
14
days
you,
the
test,
will
automatically
be
triggered
and
you
can
track
it
in
here.
That
is,
you
can
come
in
here.
You
can
click
on
the
fortnightly
test.
Then
you
can
see
all
of
the
tests
that
will
run
on
on
every
14
days
and
you
can
check
all
the
results
so
yeah.
I
The
hierarchy
is
that
these
performance
devices
are
super
simple
and
then
there
are
a
number
of
tests
filling
it.
G
A
Beautiful
okay,
well,
I'm
gonna,
withhold
20
minutes
of
feedback
and
instead
save
that
20
minutes
for
our
next
topics.
A
A
B
Yeah
sure
I
haven't
got
a
lot
to
show,
however,
we're
building
a
mesh
reconfiguration
wizard.
The
purpose
of
it
is
to
help
the
user
set
up
and
connect
measuring.
It's
essentially
like
a
multi-step
form,
like
you
see
in
desktop
applications,
so
kind
of
using
a
material
material,
ui
multi-step
form
as
a
base
and
then
kind
of
adding
adding
stuff
to
it.
So
so
far,
we've
had
the
first
design
and
I've
been
working
on
the
prototype
if
you've
not
seen
the
design.
B
Here,
it's
just
on
the
screen.
Now
I
think
it's
absolutely
amazing.
Let's
look
forward
to
building
it.
A
Oh
nice,
oh
good
bart,
I
am
doing
my
best
not
to
steal
any
of
your
thunder,
so
I'm
just
bringing
this
up
to
help.
No,
no
that's
fine
provide
context.
So
how
many?
How
many
steps
are
we
thinking
about
now.
B
We're
looking
at
four,
they
are
all
all
on
the
screen.
We've
also
been
working
on
like
a
like
a
welcome
screen
just
to
explain
to
the
user
what
what
this
is.
A
Nice,
nice,
okay,
austin
other
contacts
to
add
here
other.
J
Hello,
everyone,
but
also
created
a
live
prototype
from
what
we've
been
able
to
bring
out
and
what
you
see
on
the
screen
is
currently
they
are
second
iteration
of
the
design.
J
We
are
working
on
finalizing
that
and
then
then
we
have
another
iteration
on
making
it
better
so
but
worked
on
a
demo
which
I
can
share.
Let
me
quickly
achieve
this.
J
J
At
the
sense
whereby
you
cannot,
you
cannot
pass
if
you
do
not
complete
the
the
the
first
step
more
more
more
like
a
lot
of
the
rings
and
the
team
like
that
shall
not
pass
if
that
was
not
connected.
The
coordinates
which
we
were
trying
to
to
finalize
what
is
like
the
live
demo
of
the
conversion
wizard
from
the
first,
our
first
prototype.
So
by
the
time
we
finalize
this,
the
second
iteration
of
the
design,
also
to
creating
another
demo
and
prototyping.
J
A
Nice,
nice
good.
I
I'm
gonna
comment
for
just
a
couple
of
minutes
and
we'll
get
to
the
next
topic.
That
is
well,
I
guess
so
so
for
additional
context.
You
know.
A
Hopefully,
hopefully,
in
the
best
of
ways
no
one
ever
uses
the
configuration
wizard
which,
on
the
forefronts,
doesn't
sound
great.
But
hopefully
it's
the
case
that
when
people
are
using
measuri
it
just
it's
intuitive
enough
that
that
they
don't
need
the
configuration
wizard
and
it
turns
out
I'm
not
running
measure
yet
at
the
moment.
But
I
don't
know
that
that's
or
for
sure
that's
not
always
the
case.
Not
everyone
necessarily
knows
where
to
start.
A
A
A
F
A
I
like
measuring,
which
is
great
so
I
said
I
would
try
to
be
quick
and
I'm
not
being
concise.
The
point
here
is
people
drop
into
measuring.
It
tries
to
make
intelligent
guesses
about
how
they're
going
to
run
their
infrastructure
and
how
to
connect
to
their
infrastructure,
so
it
tries
to
auto
connect
to
kubernetes
measure,
isn't
entirely
dependent
on
kubernetes,
but
it
does
really
want
to
use
kubernetes.
When
we
lean
into
we
do
more
or
less
expect
that
people
will,
there
will
be
a
kubernetes.
A
You
can
run
messaging
without
only
some
features
work
without,
though
right
now,
mesh
reprovisions,
all
of
its
service
meshes
on
kubernetes,
so
connecting
there
and
kind
of
starting.
There
is
very
important
when
someone
loads
up
mesh
tree
and
it's
not
connected
not
connected
to
kubernetes
or
it
can't
connect
to
its
adapters.
It
becomes
obvious
here.
A
It
also
becomes
obvious
that
measuring
isn't
connected
to
prometh,
prometheus
or
grafana,
but
there's
nothing
that
steps
the
user
through
how
to
really
make
sure
that
their
environment,
that
their
measuring
deployment
is
configured
and
kind
of
ready
to
go
with
a
base
level
of
of
configuration
and
so
bart,
austin
and
perhaps
nikhil
are
on
a
journey
to
create
this
configuration
wizard.
The
initial
thought
is
that
maybe
there
would
be
an
additional
button
up
here.
That
would
be
our
icon.
A
You
might
want
to
run
this
and
when
you
do
maybe
there's
between
one
one
one
to
four
or
five
steps:
those
steps
being
getting
connected
to
kubernetes,
getting
your
mastery
operator
deployed
on
kubernetes
connecting
either
maybe
deploying
a
service
mesh
as
an
optional
third
step,
and
so
maybe
that
would
look
something
like
this.
Maybe.
A
It
would
you
know
this
configuration
wizard
needs
to
understand
that
you
may
already
have
a
service
mesh
deployed,
and
so,
if
so,
mesh
sync
should
discover
it
when
mesh
sync
would
discover
it
after
you
connect
the
mastery
operator.
So
all
of
these
things
that
the
configuration
wizard
are
doing
is
taking
advantage
of
measuring
capabilities
today,
but
putting
them
together
in
a
sequenced
step-by-step
flow,
so
that
the
user
can
be
hopefully
a
lot
more
successful,
with
measuring
and
with
with
their
deployments.
A
So
so,
while
this
is
a
single
button,
single
switcher
to
provision
an
instance
of
console,
there
are
a
lot
of
other
configuration
ways
you
might
want
to
configure
console.
Well,
that's
not
what
the
configuration
wizard
is
for
that
this
is
for
this
is
the
initial.
Maybe
we
should
put
the
word
initial
in
front
of
it
or
something,
but
it's
the
it's.
A
A
A
Lastly,
I
think
bart
and
austin
have
given
names
to
these
first
couple
of
steps.
So
for
this
first
step
it's
thou
shalt
not
pass.
If
you
don't
get
past
the
first
one,
the
second
one
is
don't
pass.
Go,
I
think,
was
the
second
step,
so
we'll
see
what
they
come
up
with
for
the
third
and
fourth.
A
But
if
you're,
a
lord
of
the
rings
fan
or
a
monopoly
player,
those
terms
will
be
familiar
to
you,
maybe
they'll,
put
in
an
easter
egg
in
there
somewhere,
we
can
see
what
those
steps
are
called.
So
next
up
is
michael
there's
a
talk
to
us
about
the
messaging
messaging
system,
notification
center.
K
So
the
messaging
system
and
notification
center
design-
as
we
all
know,
a
lot
of
the
operations
and
actions
that
we
perform
are
as
asynchronous.
K
They
take
some
time
to
perform
and
the
feedback
of
all
the
various
bits
and
pieces
of
measuring
as
a
whole
might
arrive
sometime
later
than
when
a
happy
user,
for
instance,
sends
an
operation
request
through
the
ui
to
the
mystery
server,
for
instance,
to
install
the
istio
adapter,
as
I've
tried
to
illustrate
in
this
figure
here
and
any
of
those
components
that
are
involved
in
executing
an
operation
or
indeed
any
other
thing
that
happens
around
measuring
and
in
measuring
by
measuring,
might
create
a
different
kind
of
messages.
K
Like
error
messages,
log
messages
they
might
create
notifications
for
other
components
to
do
something,
and
so
these
com,
these
messages,
have
to
be
handled
in
a
in
a
good
way
and
also
in
a
way
that
is
future
proof
and
that
can
scale
so.
K
So
this
is
what
this
design
document
and
is
about.
So
in
this
example,
for
instance,
the
istio
adapter
would
create
an
error
message,
and
the
idea
is
that
this
is
then
sent
to
a
notification
system
and,
for
instance,
picked
up
by
a
notification
center
that
displays
all
these
different
messages
and
and
sorts
them
by
type,
for
instance,
by
originator
and
so
on
and
so
forth.
K
Now
one
thing
that
we
talked
about-
and
that
is
of
course
important-
is
that
we,
when
we
begin
doing
this,
is
to
devise
a
messaging
format
and
that
is
easily
to
extend
and
open
to
extension
and
easy
to
understand.
And
so
we
came
up
with
a.
K
Suggestion
that
looks
a
little
bit
like
this.
All
the
different
messages
have
a
different
class,
so
it's
or
a
message,
type
or
message
class
like
this
one
would
be
an
error.
You
would
have
a
message
id.
You
would
like
to
know
who
sends
the
message
which
system
or
component
was
sending
a
message,
and
then
it
might
also
be
useful
to
have
a
more
human,
friendly
description
of
the
sender.
K
If
that
is
desirable,
if
you
send
a
message
back
to
the
messaging
system,
you
would
like
it
to
show
up
in
the
correct
place
like
in
the
user
id,
so
we
might
like
to
have
what
we
called
here.
I
called
your
requester
id
whether
this
is
a
session
id
or
something
like
that,
so
that
it
ends
up
in
the
correct
end,
user
interface
for
the
correct
end
user
and
displays
it
there.
K
Map
with
a
field
specific
for
this
error
class-
and
I
think
the
sort
of
like
the
motivational
use
case
that
is
very
concrete-
now
is
actually
how
do
we
handle
errors
that
originate
in
a
different
part
of
the
system?
And
how
do
we?
What
kind
of
a
system
do
we
devise
for
number
them
the
error
codes,
and
how
do
we
make
this
system
robust
so
that
we
don't
accidentally
have
same
error
codes
for
two
different
errors
in
the
same
system
or
maybe
across
systems?
K
And
so
one
thing-
and
we
discussed
quite
a
lot
about
that-
and
so
this
here
is
a
suggestion-
how
we
could
sort
of
like
sort
and
structure,
error
messages
and
error
codes.
Sorry
that
we
structure
them
by
a
component,
which
is
the
first
part
component,
for
instance,
would
be
the
adapter
or
operator
or
something
like
that,
and
then
the
concrete
implementation
of
an
adapter,
for
instance,
istio,
and
then
the
error
code
within
that
component.
K
So
the
aim
is
that
the
developer
actually
only
has
to
know
about
and
is
quite
free
to
assign
error
codes
in
this
component,
and
so
it
doesn't
accidentally
overlap
with
error
codes.
In
other
components,
because
it
is
differentiated
by
using
a
component
type
and
the
common
component
name,
if
you
like-
and
this
also
helps
to
avoid
that
you
have
to
have
a
central
registry
of
error
code-
ranges
as
we
have
worked
with
until
now-
something
we
also
discussed
and
is
how
can
you
sort
of
like
make
sure
that
this
is
actually
working?
K
How
can
you
create
create
documentation
about
these
error
codes,
because
when
a
user
or
a
developer
gets
this
code
back,
then
somewhere
should
be
documentation.
That
is
describing
this
error
the
severity
and
the
cause
and
remediation,
and
that
this
documentation,
which
could
be
a
large
table,
for
instance,
in
a
google
document?
K
How
do
we
create
this?
How
do
we
edit
this
without
having
to
create
a
new,
read
a
new
release
of
an
adapter?
If
you
want
to
change
remediation,
or
course,
descriptions
and
so
on?
K
So
this
was
just
in
the
beginning
of
sort
of
like
what
we
discussed,
how
we
could
handle
these
errors
and
and
have
some
and
have
some
automatism
so
that,
like
these
tables,
are
updated
automatically
and
that
we
have
scripts,
that
you
can
run
locally
or
in
a
cicd
pipeline
that
checks
that
all
error
codes
are
documented,
that
extracts
the
documentation
and
puts
them
in
the
documents
or
the
tables
or
the
html
pages,
so
that
it's
not
a
lot
of
this
can
be
automated.
K
A
Components:
hussein,
as
you
reflect
on
your
interactions
with,
I
guess
it
was
raymond
wong,
of
who's
having
challenges
with
gke
and
mesherie.
L
Yes,
this
would
definitely
help,
because
I
was
a
little
bit
lost
with
the
initial
set
of
locks
where
I
was
not
able
to
find
out
the
lock
specific
to
a
configuration
related
one.
So.
A
Nice
yeah
good
yeah.
I
totally
think
so
as
well,
especially
when
they're
unique
and
people
are
getting.
You
know
how
all
of
you
have.
Google
searched
some
specific
error
code
and
yeah
yeah.
L
Yeah
because
I
was
not
I
I
had
I
at
one
point-
I
had
a
doubt
whether
I
am
I
have
to
pull
in
another
report
to
look
for
these
codes,
so
I
was
not
sure
that
this
was
coming
from
the
go
client,
so
yeah.
K
K
K
Yes,
you
would
have
so
all
like
within
a
repository
say
if
that
corresponds
to
a
specific
component
implementation,
then
the
maintainers
of
that
repository
will
be
completely
free
or
maybe
you
know
almost
completely
free
to
decide
on
their
error
codes.
So
you
could
actually
have
that
all
you
know,
service
meshes
could
have
air
codes
from
a
thousand
upwards
or
from
one
upwards
whatever,
and
you
will
need
to
coordinate
all
these
ranges
and
you
would
you
would
oh
you
would
so
like
here.
You
know,
you
see
you're
limited
to
100
error
messages.
K
A
A
We
need
to
add
my
over
engineering
to
this
because,
because
you
guys
are
we'll
do
that
onto
your
own,
which
is
excited,
which
is
why
I'm
excited
you
guys
are
focused
on
it.
F
A
You
know
that
if
it's
and
that
that
the
abbreviation
like
istio
like
error
code,
istio
2000
it
will
that
2000
is
namespace
based
on
the
fact
it
has
istio
in
front
of
it.
A
So
and
it's
just
what
you
were
just
saying,
but
if
something
to
think
about
it,
maybe
that
there
is
a
little
bit
of
error
code
ranges
but
not
based
on
component,
rather
that
each
component
would
agree
that
the
first
0
to
one
thousand
are
with
would
be
with
respect
to
a
particular
genre
of
issue
or
or
that
maybe
from
zero
to
ten
thousand.
A
Are
you
don't
know
what
type
of
error
it
is,
but
it's
an
error
if
it's
from
2000
to
3000
it's
an
info
like
maybe
it's
based
on
severity
or
something
like
maybe
there's
an
easy
way
that
that
doesn't
overcomplicate
the
engineering
of
it,
but
does
signal
to
the
user
that
oh,
this
is
an
http
200..
So
I
can't
remember
201
202,
like
200,
whatever
all
I
know
is
if
it's
in
the
200s,
I'm
good.
If
it's
in
the
500,
something
it's
probably
it's
some
anyway
think
about
that.
If
you
would
yeah.
K
A
A
This
isn't
you
could
take
this
concept
and
apply
it
to
any
well-done
piece
of
software,
because
in
part,
what
husseino
was
going
to
say
was
yeah.
If
I
took
istio
2005
as
a
search
term,
I
think
if
you
go
search
that
today
you
it
may
not
come
back,
it
might
come
back
with
a
github
issue
or
something
for
issue,
but
eventually,
if
it's,
if
it's
mescheri,
istio
2005
like
there's
one
page
and
one
page,
only
that
that's
going
to
bring
you
to
and
that
page
should
have
a
context
of
what
the
issue
is.
A
It'll
have
the
probable
cause
and
suggested
remediations
for
what
to
do
about.
What's
going
on?
Oh,
my
goodness,
what
a
what
a
fantastic
user
experience
that
is
to
be
able
to
reference
that
and
the
notifications
in,
like
first
of
all,
you
don't
have
to
go
to
google
to
find
that
you
can,
but
the
notification
center
that
michael,
was
highlighting
will
manage
the
life
cycle
of
those
notifications.
A
But
alonso,
I
do
consider
that
yeah
at
some
point,
this
messaging
framework
and
notification
center
probably
is
fairly
integral
to
it,
would
be
one
of
those
characteristics
of
a
component
of
measuring
that
someone
would
identify
as
like.
Oh
that's,
that's
part
of
the
that's
part
of
what
measuring
is
because
that
component
has
this
type
of
capability
like
hey,
it's
going
to
have
this
in
this
slash
health,
endpoint
or
the
slash
component
info
endpoint
it'll
have
common
characteristics
about
like,
and
one
of
those
common
characteristics.
A
A
A
H
A
Do
okay,
eta.
A
So
maybe
beneath
I'll
catch
up
with
you
in
in
slack
I'll,
go
to
the
mastery
slack
we'll.
This
is
a
great
thank
you
for
raising
that
up,
we'll
I'll
give
everyone
else
more
context
in
there
as
well
and
we'll
have
to
figure
out.
If
that's,
if
that
should
block
the
release,
yeah
sure,
nice,
okay,
all
right,
it's
a
wrap!
Thank
you
all
see!
You
guys
see
you
guys
post
release.
I
guess,
on
the
other
side,.