►
From YouTube: Meshery CI Meeting (March 25th, 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
C
C
Oh,
by
the
way
I
was
gonna
to
gibral,
I
was
gonna
try
to
provide
an
answer
in
chat
when
you'd
asked
this
question
earlier
and
part
of
my
response
was
going
to
be,
like
my
goodness,
I
think.
Initially
we
were
on
alpine
and
then
at
least
for
the
mesri
server
image
and
then
multi-stage
builds
came
out
and
we
started
to
get
more
clever
and
then
I'm
surprised
or
okay.
C
D
Maybe
because
we
moved
to
ubuntu,
because
we
we've
been
using
a
couple
of
binaries
which
are
built
for
ubuntu,
the
the
load
testing
stuffs
like
the
wrk
or
knight
of
bills
that
are
currently
there.
I
guess
they
are
for
those
binaries
are
built
for
ubuntu
and
that's
the
reason.
Probably
we
moved
back
to
or
we
moved
towards
to
towards
ubuntu
image.
C
C
And
and
gibral
go
ahead.
C
Good,
just
you
know,
I'm
I'm
always
happy
to
repeat
stuff
like
al.
These
guys
are
sick
of
hearing
me
talk
so
yeah.
What
I
was
saying
is
you're
totally
right,
like
hey,
we
shouldn't
be
wasting
space
unnecessarily,
that
even
moving
like
abhishek
has
led
the
way
toward
us
moving
some
of
the
meshi
adapter
images
over
to
distro-less
right
distroless
as
the
base
image.
C
And
while
that's
true
for
some
of
the
adapters-
and
it
makes
things
you
know
about
as
lightweight
as
we,
you
know,
super
lightweight
which
becomes
important
when
someone
goes
to
use
meshri
and
it
downloads,
you
know
like
10
images
and
it's
like
oh
geez,
they're,
just
kind
of
sitting
there
waiting
for
the
images
to
download
one
of
the
negative
aspects
of
using
distro
lists
or
using
such
a
small
image
is
what's
really
like.
C
C
It
puts
a
little
more
pressure
on
the
application
itself
to
have
its
own
instrumentation,
to
have
its
own
mechanisms,
for
you
know
debugging
itself
and
or
debugging
or
for
us
just
to,
and
this
might
be
something
that
you
want
to
do
for
someone
to
say:
hey,
you
know
as
the
image
gets
smaller
and
you
have
less
and
less
utilities
like
like
you,
don't
have
telnet
or
you
don't
have
ping
or
ns
look
up
or
dig
or
some
of
these
common
utilities
that
you
might
need
to.
C
You
know
to
use
as
you're
debugging
a
connectivity
issue
or
or
whatever.
Well
then,
hey
that's
fine,
they're,
not
on
the
image
directly,
but
what
the
process
is
that
you
attach
a
utility
container
or
you
sidecar,
a
you,
know
a
debug
container
and
here's
the
and
here's
the
process.
For
that,
like
that's
something
that
would
be
helpful,
that's
something
that
would
make
us
collectively
more
comfortable
with
slimming
down
the
image.
C
B
B
C
Yeah
yeah
totally
and
some
some
of
it's
about
choice,
some
of
it's
like
hey.
If
you're
yeah,
yeah
yeah,
it's
like
hey,
you
go
turn
on
debug
mode,
then
you
can
see
all
the
you
know
for
the
application
for
the
for
measuring
like
go
turn
on
debug
mode
or
yeah
anyway.
There's
a
bunch
there's
a
bunch
of
right
answers
and
one
of
those
correct
answers.
C
Even
if
the
image
is
on
alpine,
it
might
still
be
that
you
don't
have
certain
utilities
so
like
and
not
everyone
who's
using
the
project
or
the
tool
necessarily
is
familiar
with
other
mechanisms
to
you
know
like
attach
a
container
or
do
other
ways
of
figuring
some
of
this
stuff
out.
So
we
don't
necessarily
need
to
teach
kubernetes
or
teach
containers
to
everyone,
but
but
we
don't
have
much
by
way
of
troubleshooting
guides
today,
and
so
so
I
mean
that's.
Just
a
good
thing
to
have
in
general
is
like
hey.
C
What
are
you
actually?
We
should
have
brought
this
up
before,
because
there's
I
mean
yeah.
Is
we
don't
yeah?
It's
a
troubleshooting
guide.
It's
like
you
know.
Mastery
runs
in
a
certain
way.
Can
you
run
mescheri
on
a
service
mesh
meshery
manages
service
meshes,
but
can
you
deploy
measuring
on
a
service
mesh?
Yes,
you
can
do
we
have
that
documented?
C
No,
we
have
a
deployment
file,
a
kubernetes
deployment
file
in
the
install
folder
in
the
mesherie
repo,
but
that's
it
like.
Nobody
knows
that
it's
there,
what
you
know
like.
C
Maybe
we
should
do
that
too.
Anyway,
back
to
your
original
point,
about
the
size
and
and
is
are
the
artifacts
that
are
being
built
here.
Are
those?
C
Are
we
taking
the
image
away
as
the
artifact
or
are
we
taking
a
binary
away
as
the
artifact.
C
Is
this
even
valid
anymore,
abhishek.
D
D
No,
no,
not
the
night
off,
so
I
thought
you
were
pointing
out
knight
of
stock
is
stuff
is
probably
valid
because
I
I've
never
changed
it
like
it's
the
same
thing
that
what
kush
had
put
in
okay.
I
did
not
change
it
after
that.
C
So
on
this,
actually
this
is
a-
and
this
is
an
opportune
time
to
to
tell
gibral
and
piyush
specifically
you
two
guys
about
this
other
project.
The
project
is
called
get
nighthawk,
it's
fairly
young.
We
have
meetings
on
the
project
in
the
cncf.
C
Actually
it's
at
this
it's
at
about
this
time.
It
is
at
this
time,
like
next
yeah
next
week
next
week,
we'll
have
a
meeting
on
get
nighthawk
and
and
so
I'll
send
you
a
link
to
see
if
some
of
the
work
that's
going
on.
There
is
intriguing
because
it
while
it's
a
different
project,
we're
doing
the
project
for
measuring,
so
there's
going
to
be
overlap
and
it
might
be
of
interest
to
you
so.
C
D
A
C
Yeah,
that
would
that
mean
that
there
might
be
two
offering
two
offerings
like
one.
That
is
one
that's
for
people
to
run
and
kind
of
a
development
mode
and
one
that's
offered
as
more
of
a
production
mode
like
if
we
did
two.
C
Yep,
that's
a
thought
in
part
like
jibril,
as
you
think
on
that
a
little
bit
so
by
the
way
gibraltar.
This
is
this
is
the
second
meeting
that
you've
been
on
right.
C
Second
layer:
five
meeting:
yes,
okay,
yeah,
nice,
good,
good,
good,
you're,
you're
hooked,
already
good
okay,
since
this
is
your
second
meeting
you
get
to
you,
get
to
walk
away
with
an
action
item
or
if
you,
if
you
want
it,
if
you
want-
and
that
is
like
one
of
the
things
to
think
about
like
in
general-
is,
if
you
think
so
so
one
thing
I
mentioned
was
there's
kind
of
a
lack
of
a
troubleshooting
guide.
C
There's
kind
of
the
start
of
one,
but
broadly
it's
we're
lacking
in
con
in
that
context,
is
the
same
kind
of
a
context
that
an
effort
like
an
effort
to
produce
two
different
images.
It
would
be
sort
of
in
that
context,
like
this
other.
You
know
one
for
prod
one
for
dev
kind
of
a
thing
that,
along
to
advance
that
thought
a
little
bit
right
now
there
is,
and-
and
rudolfo
can
tell
you
all
about
this-
there's
a
build
and
release
strategy.
C
A
document
that
outlines
what
the
ci
processes
are
for
measuring
all
of
its
components:
the
fact
that
there
are
multiple
release
channels
so-
and
this
may
be
something
to
play
into
so
to
describe
that
briefly-
there's
two
release
channels
that
one
of
which
is
called
edge
and
the
other
one
is
called
stable
and
they
are
what
they
sound
like
ones
out
there
you
know,
and
so
you
might
imagine
that
if
we,
if
there
are
two
docker
builds
of
measuring
of
the
components,
you
know
one,
that's
slimmed
down
one!
C
That's
based
on
ubuntu
that
maybe
the
stable
release
channel
only
ever
uses
this.
You
know
alpine
or
distrolus,
or
you
know
something
slim,
the
other
release
channel
the
edge
one.
Maybe
it
you
know,
uses
the.
I
don't
know
just
just
kind
of
thinking
out
loud
as
as
you
think
about,
as
you
digest
all
of
this,
as
you
kind
of
step
into
it.
C
C
C
It
is
also
a
c
plus
application,
and
no
doubt
that,
are
we
compiling
those?
Are
we
just
grabbing
binaries,
but
the
compilation
of
those,
though,
does
require
you
know
a
more
robust
os
with
a
bunch
of
you
know
a
tool
chain
to
to
generate
those
so
part
of
what
the
project
get
nighthawk
is
about
it's
about
building,
nighthawk
and
making
it
available
in
various
formats.
Various
architectures.
B
Tonight,
did
you
get
that.
C
Yeah,
I
did
I'm
hoping
that
abhishek
talks.
D
Yeah,
so,
first
of
all
on
the
about
the
image.
Yes,
we
are
compiling
wrk
to
during
build
time
and
we
are
not
compiling
nighthawk
because
we're
just
fetching
the
image
from
the
container
and
then
transferring
it
to
meshware's
container.
D
And
that's
about
like
what
was
the
second
question
again.
D
Yes,
partly
I
mean
there
is
a
section
for
invoking
or
doing
performance
testing
on
the
dashboard
and
in
there.
Basically,
it
depends
on
the
type
of
test
you
invoke.
We
do
support
wrk
to
fortio
and
idoc
right
currently,
and
each
of
these
binaries
are
invoked
under
the
hood
to
perform
those
tests.
C
So
the
another
way
of,
let
me
let
me
characterize
that
a
little
bit
differently
just
and
that
is
to
say
that
if
someone
wants
to
compile
and
build,
mesherie
mescheri
does
not
require
wrk2
or
nighthawk
to
be
present.
So
you
can
build
meshri
without
it
when
you
run
meshri
as
a
user
you're
using
meshri.
B
B
C
So
gibral,
how
about
that?
How
about
that
troubleshooting
guide
that.
B
C
A
A
A
C
Yep
he
I
had
spoken
with
him
yesterday
or
the
day
before,
and
he
he
was.
He
was
working
on
it.
That
said,
he
doesn't
get
it
or
he's
not
and
so.
Well,
I
helped
him
along
on
like
but
of
the
questions
he's
asking
like
so
and
also
asuko
has
been
with
the
project
for
a
year
and
something
like
and
he
his
toes
don't
get
stepped
on
or
like
it's
not
he's,
not
gonna,
he
won't
be
upset.
He
will
offer
a
comment
and
try
to
help.
C
You
know
reinforce
work
that
you've
done,
but
but
yeah
it's
slightly
confusing
for.
A
Him
so,
okay
yeah,
so
I
think
I
I
can
talk
with
him
and
and
get
a
solution
together,
because
the
last
time
that
I
talked
with
him,
I
think
we,
if
we
got
a
point
so
yeah,
I
think
we
can
work
together
and
I
will
be
yeah
posting.
The
the
progress
on
github
or
or
in
the
meetings
also.
C
A
A
C
So,
okay,
yeah,
sorry,
oh
yeah,
well,
the
issue
here
is
that
I
think
they
end
up
charging
us
for
those
for
that
minute
or
those
two
minutes
or
whatever
the
30
seconds
or
whatever,
and
by
charging
I
mean
like
eating
up
to
three
minutes.
C
At
this
point
I
don't
it
was
a
really
big
issue
when
we
were
trying
to
work
so
fast
on
layer,
5,
ng,
on
the
new
website
and
any
anymore.
You
know
I
don't
think
we're
running
over
our
minutes,
but
that
was
mostly
why
we
had
wanted
to
not
do.
It
was
just
like
to
not
invoke
netlify
at
all
to
you
know,
because
I'm.
A
Assuming
yeah,
some
some
workflows
were
not
triggered
by
default,
but
someone
that
they
they
pre
they
deployed
preview.
The
only
setting
that
we
can
create
to
to
avoid
that
is
that
checking
this
this
this
rule
so
well
or
or
in
the.
C
Well,
maybe
I
miss-
maybe
I'm
not
seeing
this,
but
in
the
github
actions
themselves.
Is
it
a
github
action
that
is
invoking
this
deployment?
That's
web
hooking.
A
Now
this
this
workflow
is
triggered
by
linking
the
or
linking
the
layer
5
repository
it's
by
default.
A
A
C
Yeah
yeah,
and
so
like
in
concept.
We
would
we
would
leave
this.
I
don't
know
what
you
call
it,
but
this
project
this
site,
but
the
the
deployment
configuration
of
netlify
for
that
site
would
be
not
to
do
anything
or
not
to
be
not
to
be
watching.
C
You
know
the
repo
rather
waiting
to
be
web
hooked,
for
you
know,
like
just
the
start,
point
changes
and
so
that
that's
one
way
of
you
know
another
way
of.
C
C
C
Before
it
was,
it
was,
you
know
they
were
charging
money
yeah.
But
now,
if
we're
not
in
danger
of
hitting
running
over
the
minutes,
then
maybe
we're
doing
fine.
C
A
Oh
yeah,
this
one
yeah
31.
C
Minutes
and
last
month
275,
because
we
had
already
finished
with
see
how
january
like
december
and
january,
were
like
insane
and
then
we
basically
stopped
it
turned
it
all
off
in
february,
because
we
kept
getting
charged
right
turn
it
back
on
wow
anyway,
anyway,
like
yeah,
there's
absolutely
a
way
to
not
have
like
to
not
have
netlify
invoked,
but
to
have
that
controlled
from
github
actions
to
determine
whether
or
not
files
have
changed
under
a
certain
path
and
then
from
there
take
action
under
that
path
and
that'll
actually
be
the
same
approach
that
you'll
end
up
taking
this
weekend
with
like,
irrespective
of
what
asuka
says,
I
mean
he's,
I'm
telling
you
that
yeah
you'll
determine
whether
or
not
there's
been
a
change
to
the
slash,
install
folder
and
then,
if
so,
okay,
it's
time
to.
C
C
You'll
have
to
look
to
see
what
image
tags
we're
using
if
it's
stable
latest
then
thank
you
alonzo.
If
it's
stable
latest,
then
maybe
we
don't
have
to
cut
a
release
every
time,
but
the
reality
is
it
doesn't
really
matter
like
we
could
rev
that
chart
all
day
long.
Who
cares?
I
don't
care,
we
just
need
to
okay
gibra
we
just
need
to,
but
we
need
to
be
getting
the
latest
stuff
out
there
to
people
and
making
sure
that
updates
are
getting
out
and
we're
not.
A
C
C
We
end
up
outputting
a
couple
of
statements
at
the
end
telling
the
user
what
to
do
next,
and
I
don't
know
that
they're
accurate,
so
one
of
the
actually
abhishek
can
you
take
this
note
to
maybe
open
an
issue
about
it?
Is
it
at
as
a
user,
and
maybe
it's
just
documenting
this
process
as
a
user?
If
I
only
want
to
have
like
the
messaging
adapter
for
linker
d,
then
how
do
I
do
that
using
the
helm
install?
C
C
Okay
and
so
nice
so
next
topic,
I
guess.
A
A
That
is
very
we're
very
glad
to
to
help
him
in
the
community,
because
he
he
has
helped
a
lot
in
the
in
the
test
part.
I
think
the
the
next
focus
of
the
ci
group
is
to
have
golang
test
or
golang
unit
test
in
order
to
to
test
everything
that
we
have,
because
we
don't
have
any
unit
testing
in
the
backend
at
all.
So
I
think
that
could
be
one
of
our
main
focus
here.
A
E
C
Yeah
as
part
of
that,
like
one
thing,
we
could
do
next
time
we
meet
or
before
we
meet
is
well,
is
try
to
outline
the
test
like
the
test
plan,
or
or
maybe
maybe
it's
just
the
high
level
test
strategy,
because
in
part,
what
you're
saying
right
now
is
part
of
the
test
strategy
for
guaranteeing
that
releases
are
solid
and
stable
is
that
we
should
test
some
stuff
with
cyprus.
We
should,
you
know,
end
to
end.
We
should
test
some
unit
tests
and
go.
C
We
should
slim
down
the
image
sizes
we
should
probably
just
like
and
so
that
the
strategy
doc
should
sort
of
you
know
speak
to
a
lot
of
that
high
level
and
then
the
other
like
test
plan
spreadsheet,
maybe
is
just,
is
more
of
like
an
itemized.
It's
not
like
it
lists
out
every
unit
tests,
or
it's
not
like
like
more
like
what
that
would
do.
Is
it
would
just
based
on
what
the
strategy
says.
C
It
would
do
something
like
oh
here's,
a
row
for
every
adapter
and
maybe
there's
like
five
high
level,
there's
like
one
high-level
end-to-end
test
or
five
high-level
end-to-end
tests,
and
it's
just
like
one
of
the
the
things
that
we
capture
about
the
adapter
is:
what's
the
current
passing
percentage
of
unit
tests,
so
it
doesn't
list
out
all
the
unit
tests.
It's
just
like.
Yes,
their
unit
tests,
here's
their
status
but
yeah,
just
even
laying
here's
the
function
and
then
basically
like.
C
I
think
I
think
a
lot
of
it
would
be
capturing
what
cyprus
is
doing
like
the
end
to
end
stuff,
because
that
covers
so
much
ground
it's
like
do.
We
have
a
sense
of
what
is
covered
under
end-to-end
tests
and
what
you
know,
what
isn't
part
of
what
we
go
into
the
strategy
dock
is
the
fact
that
we're
using
bats
for,
like
shell
scripting
of
testing
out
some
console,
the
console
adapter,
specifically
good,
we're
only
using
it
for
that
adapter
not
for
the
rest,
bad.
C
C
Yeah
wait,
wait
are
all
the
I
guess.
I
thought
it
was
only
the
console,
adapter,
that's
being
tested
with
bats.
A
A
Lee
do
you
want
to
explain
that
yeah?
Actually,
I
know
vijay
was
just
about
to.
F
Explain
no,
I
wasn't
I.
I
was
just
going
to
say
that.
F
You
know
a
few
weeks
ago,
I
don't
remember
when
I
I
did
write
out
a
strategy.
You
know
on
testing,
I
mean
it
may
not
be
what
you're
looking
for,
but
you
know
I
put
out
a
paper
head.
F
I
I
think
I
covered
a
little
bit
of
bats
and
you
know
what
was
the
other
cyprus
and
you
know
what
each
of
them
does
and
on
which
platform
like,
for
instance,
here
bats
is
good
for
something
else.
So
maybe
you
know
you
have
to
have
and
also
I
laid
out
a
few
things
about
what
are
the
types
of
tests
you
can
possibly
do
I
mean
you
have
like
installability
testing.
You
have
like
performance
testing.
You
have
unit
test.
F
Regression
testing
so
now,
anyway,
those
are
probably
not
things
they're.
Looking
for,
but
whatever
it
is.
You
know
I
just
I
put
out
a
few
thoughts
on
on
a
piece
of
paper.
You
know.
F
Adapters
enable
the
okay
adapters
enable
a
machine
to
connect
to
service
machines,
so
there's.
F
C
C
That
vijay
is
probably
hunting
down
that
that
now.
C
So
vj
on
your
testing
strategy,
that's
fantastic!
Even
if,
for
whatever
reason,
the
project
didn't
end
up
doing
a
particular
type
of
tests
or
tests
that
it's
still
a
good
thing
to
list
down
like
under
out
of
scope,
these
are
out
of
scope
and
like
here's,
why
we
aren't
doing
that
now
or
decided
to
never
do
it,
but
most
of
them
probably
are
in
scope
and
yeah.
Just
a
singular
document
to
point
people
to
repetitively
to
say
you
know,
here's
what
those
frameworks
are.
Here's
the
thought.
C
Here's
you
know,
here's
where
you
go
get
introduced
and
and
to
be
able
to
the
stuff
that
you
put
into,
I
think
the
doc
that
was
just
being
shared
a
moment
ago.
There's
a
there's,
a
document
called
the
master
test,
pla
master
test
strategy
or
something
like
that.
It
definitely
says
master
and
I
think
it's
the
only
one
go
take
liberties
with
it.
I
mean
like
that's.
C
I
was
just
trying
to
create
it
as
a
placeholder
for
just
what
you've
just
done
so
yeah,
please,
you
know
draft
you
know
put
those
in
and-
and
we
can
review
on.
This
call
for
sure
releases
are
like
there's.
The
yeah
I
mean
releases
are
absolutely
what
we're
talking
about
here.
There's
the
build
and
release
strategy.
There's
the
the
the
master
test
plan
strategy.
A
A
C
C
If
I
would
have
it
would
have
been
a
spreadsheet,
because
that's
because
because
I
consider
it
to
be
a
very
itemized
like
specific
thing
and
so
yeah,
first
starting
with
the
strategy
and
then
moving
into
the
specific
plan
and
then
vj
had
their
that
testing
strategy
right
there
vj,
if
you
bj,
had
he
put
it
into
the
build
and
release
folder.
So
we
can
consolidate
what
he's
got
into
the
master
test
strategy.
So
he's
got
one
at
the
bottom
there.
It's
called
testing
strategy
yeah
that
one
there
you
go.
C
And
then
gibril
to
talk
more
to
like
the
architecture.
There's
a
community
drive
to
build
it.
You
should
there's
the
link
to
it
in
chat,
lots
of
docs
to
browse
within
there
the
mystery
architecture
slide.
Deck
is
within
there
as.
C
A
What
else
do
we
wanna
know?
I
think
those
were
all.
A
C
So
I've
got
two.
The
first
one
is
that
aditya,
it's
nice
to
have
you
on
the
call.
C
And
we'll,
if
you
join
tomorrow's
community
call,
we
will
get
to
know
you
even
better
on
that
call.
C
Good
good
good,
good
we've
got
a
good
kid
and
then
ju
jubril,
you
you
did.
You
have
a.
C
C
E
C
I
know
oh
okay,
good
yeah,
so
we'll
talk
more
about
that
either
either
in
tomorrow's
community
call
or
in
next
week's
mastery
server
commercial
development
call
but
either
way.
C
C
The
second
topic
that
I
had
was
more
about
you:
it's
just
that
keep
your
questions
coming
and
if
you
have
other
ones,
let's
talk
about
them
now
else
we
might
take
this.
D
Yep
wait:
hey
phones
are
so
bad.
Okay,
I
just
lost
the
last
part
of
the
voice.
Jubril
was
asking
something
actually
yeah.
C
Okay,
can
someone
repeat
that
yeah?
You
know
you
know
what
actually
let
me
hear
I'll
feel
that
I'll
feel
that
question
great
great
question.
In
general,
your
question
was
and
and
abhishek
interrupt
me
if
you,
if
I'm
stealing
the
thunder
so
but
the
question
was
really
about
like
service
discovery
and
what
all
is
this,
what
all
does
mesh
rediscover?
C
I
part
of
your
voice
cut
out
a
little
bit
jubril.
So
I
think
you
were
saying:
does
mesh
rediscover
exist
service
meshes
like
existing
ones?
Does
it
discover
purposes
that
are
on
those
surface
meshes?
Does
it
maybe
interface
with
like
a
console
as
a
general,
not
console
the
service
mesh
but
console
the
service
discovery?
C
You
know
server
system
like
like
does
mesh
rediscover
vms
and
you
know
stuff
with
that
context.
Abhishek,
do
you
wanna
you
wanna,
take
it
talk
about.
D
Yes,
so
currently
mishri
relies
mostly
on
the
kubernetes
discovery
system,
so
anything
at
all
that
is
detected
or
that
has
been
inside.
The
network
of
kubernetes
is
projected
out
inside,
measuring,
for
example,
from
the
native
resources
to
the
custom
resources
to
any
federated
virtual
machines
that
are
inside
the
service
network
of
kubernetes.
C
So
yeah
so
there's
a
particular
component
inside
so
I'll
make
it
look
sound
more
sexy
then,
because
there's
a
particular
component
of
meshri,
that's
called
mesh
sync
mesh.
Sync
is
responsible
for
synchronizing
service
discovery,
state
information,
any
anything
that
goes
on
not
in
well
anything
that
goes
on
inside
of
kubernetes
that's
of
interest
to
mastery
and
that
would
be
under
management
by
measuring
or
that
measure
you
just
generally
need
to
be
aware
of.
C
Mesh
sync
will
get
that
over
to
a
central
to
to
a
meshre
server
so
that
it
can
reason
over
a
lot
of
that.
So
the
adapters
that
mastery
has
allowed
mastery
to
speak
to
to
go
deep
on
any
one
of
the
given
service
meshes
that
it
manages
and
mesh
sync
that
runs
under
under
control
of
mesh
reoperator.
C
It
interfaces
like
abhishek
was
saying
with
cube
api
and
with
with
all
of
the
things
that
it's
tracking
pulls
that
data
back,
so
that
it
can
reason
over
it
and
and
do
more
intelligent
things
with
by
understanding
the
state
of
the
mesh
or
the
state
of
kubernetes,
the
state
of
the
mesh,
the
state
of
the
workloads.
C
There's
a
visualization
project,
that's
called
mesh
map
that
will
visually
depict
the
topology
of
the
various
service
meshes
and
the
services
that
are
running
on
top
of
them
provide
metrics,
and
you
know
things
like
that.
It'll
also,
eventually,
let
you
design
the
mesh
so
you'll
be
able
to
visually
design
the
work.
The
workloads
that
you
want
on
the
surface
mesh
you'll
be
able
to
take
actions
on
the
surface
meshes
or
maybe
you
know,
do
things
like
performance
tests
against
services
that
are
running
on
the
mesh
or
off
the
mesh
it'll.
C
Let
you
onboard
and
off-board
services
off
the
mesh
field,
basically
that
that
visual
topology
will
do
anything
that
you
can
do
in
measuring
through
the
form-based
ui.
You
should
be
able
to
do
there
as
well,
so
mesh
sync
is
new.
As
of
this
release
that
just
happened
like
yesterday.
I
don't
think
we've
actually
made
an
announcement
about
it,
but
it's
the
the
dot
five
release
and.
C
Yeah,
which
means
that
meshri
will
perform
some
amount
of
discovery.
So
certainly,
if
you
use
meshri
to
provision
a
service
mesh,
mastery
is
aware
of
that
fact.
It
it's
the
one
that
provisioned
it,
but
mesherie
also
supports
brownfield
in
brown
field
environments,
meaning
if
you
already
have
a
surface
mesh
deployed
you
can
deploy
measuring
it
will
discover
the
fact
that
the
mesh
is
there,
pull
back
that
info
through
mesh
sync
it'll.
Also.
C
So
this
type
of
an
architecture
tries
to
help
elucidate
like
what
you
know,
which
components
mesh,
reconsiders
itself,
authoritative
for
what
is
wrong
and
which
aspects
it.
C
If
measuring
owns
an
object,
then
it
owns
its
complete
full
life
cycle,
even
at
that,
even
for
some
of
the
the
objects
that
measuring
might
own
like
a
service
mesh
it'll
provision,
it
it'll
de-provision,
it
it'll
update
the
configuration
still
there's
an
understanding
and
a
recognition
that
operators
of
the
or
users
of
the
service
mesh
may
not
always
use
meshri
to
make
changes
to
it.
So
measuring
needs
to
be
continually
in
sync
with
the
service
meshes.
C
So
it's
an
interesting
question
when
you
place
it
like
that,
jubril
is
that
we
don't
talk
about
measuring,
doing
service
discovery.
Although
yeah,
that's
exactly
what
it's
doing.
I
mean
it's
doing
that
plus
a
little.
You
know,
plus
some
things
to
the
side
of
that
we
had
just
or
for
my
part,
I
hadn't
used
the
term
service
mastery
doing
service
discovery
there,
but
yeah
it
has
to.
B
C
Yeah
one
of
the
things
that
can
be
improved.
There
is
well
okay,
well,
fine,
so
mesh
rediscovers
that
there's
a
service
mesh
that
you've
deployed,
but
that
could
be
one
of
ten
different
meshes.
So
how
does
measuring
know
which
one
that
is
like?
What's
the
unique
identifier?
What's
the
fingerprint
that
it
uses
to
discover
what
service
mesh
it
is?
C
I
think
that
that
mechanism
could
be
improved
if
you
stick
around
in
the
community
long
enough
I'll
tell
you
what
that
mechanism
is,
which
is
my
way
of
both
saying
I've
been
talking
too
much
and
there's
a
cliffhanger
for
you.