►
From YouTube: Layer5 Community Meeting (Aug 13th, 2021)
Description
Layer5 Community Meeting - Aug 13th, 2021
Join the community at https://layer5.io/community
Find Layer5 on:
GitHub: https://github.com/layer5io
Twitter: https://twitter.com/layer5
LinkedIn: https://www.linkedin.com/company/layer5
Docker Hub: https://hub.docker.com/u/layer5/
A
Welcome
everyone
to
the
life
community
meeting
today
is
13th
of
august,
and
we
have
a
couple
of
new
folks
join
the
community
this
week,
and
so
here
we
have
a
tradition
where
newcomers
gets
to
introduce
to
the
community
and
gets
to
know
what
all
projects
are
happening
in
laravel
and
the
community
also
can
get
introduced
to
newcomers.
B
Hello,
hello,
am
I
audible?
Yes?
Yes,
yes,
you
are
hello,
everyone
I
am
bharan
jabanath.
I
am
from
india.
I
am
pursuing
a
beta
game,
electronics
and
and
from
the
last
2-3
months
I
am
just
exploring
the
web
development
field
today.
My
first
pr
merge.
So
I
am
happy
about
that
yeah.
That's
me.
A
Awesome,
thank
you
varun
and.
C
Is
that
is
that
your
first
open
source
pr,
or
is
that
your
first.
B
C
B
A
That's
great
to
hear
and
congratulations
on
your
first
open
source
beer.
Did
you
get
a
chance
to
introduce
on
this
call.
E
Yeah
sure,
hello,
everyone
I
am
from
hakka
bangladesh
and
I
have
recently
graduated
like
completed
my
bachelor's
as
a
csu
student,
and
I
have
been
into
open
source
a
little
for
like
quite
a
year
now
and
last
winter.
I
was
an
outreach
participant
for
the
project
of
mozilla
and
I'm
looking
forward
to
contribute
to
this
community
as
well.
E
A
E
A
Next
up
we
have
weerish,
I
think,
with
you
just
joined
the
community.
Would
you
like
to
introduce
yourself.
F
Yeah,
like
I'm
very,
like
studying
a
third
year
of
csc
from
bangalore,
like
I'm
more
of
a
front-end
guy,
like
recently
been
learning
the
devops
and
stuff
yeah.
That's
it.
I
wouldn't.
A
A
If
you
are
new
here
like,
if
you
have
mic
problems,
then
you
can
just
type
it
in
chat
all
right.
Let's
get
started
with
the
agendas
for
today.
So,
as
we
have
mentioned
in
previous
meetings
measuring
and
service
mesh
performance,
which
are
layer,
5
community
projects
are
now
cncf
adopted
by
cncf.
A
So
we
have
a
community
group
in
community.cnc
dot
io.
So
here
we
will
share
all
the
events
that
is
happening
around
the
community
and
you
can
watch
the
recordings
and
all
those
other
stuff.
So
if
you
haven't
joined
the
community
yet
you
can
join
right
now
the
links
in
the
chat-
and
we
also
have-
we
also
have
an
upcoming
webinar
with
cncf
on
measuri.
So
this
will
be
on
august
19th
and
this
will
be
hosted
on
youtube.
A
C
I'll
add
a
little
bit
of
flavor
to
that.
If
I
could
and
that's
to
say
if
it
isn't
evident
by
some
of
the
neighbors
some
of
your
neighbors,
some
of
the
other
folks
that
are
on
the
call
people
that
are
on
the
call
that
students
are
most
welcome
here
as
well
like
as
a
matter
of
fact,
there's
well,
I
think
if
you
tallied
up
the
majority,
there's
a
there's
a
lot
of
students
here
and
whether
you're,
formerly
learning
in
a
university
or
your
graduated
university
and
you're
still
learning.
C
There's
a
ton
of
learning
that
goes
on
here.
We're
dealing
with
new
tech
service
messages
are
not
ubiquitously
adopted,
yet
there's
tons
of
things
to
learn:
they're
very
powerful,
while
you're
trying
to
learn
and
wrangle
those
things,
there's
docker
and
kubernetes,
and
prometheus
and
graphona
and
jaeger,
and
there's
a
and
cloud
events
and
blah
blah
blah.
There's
like
ton
of
technology
being
used
within
the
projects
here,
it's
in
part
because
the
core
project
that
we
focus
on
or
the
largest
project
that
we
you
know
tend
to
focus
on
is
measuring
mesherie.
C
Well,
it
has
a
saddle.
It
has
a
number
of
satellite
projects
if
you
will
that
those
satellite
projects
get
talked
about
on
the
newcomers,
call
on
thursdays
to
introduce
everyone
to
them
because
they
use
different
languages
and
different
folks
are
interested
in
different
aspects
of
the
projects.
C
So,
if
you're
not
familiar
with
them
all
like
go
to
the
newcomers,
call
you'll
get
introduced,
it's
good
there's
one,
that's
coming
up
that
will
be
in
c
plus,
and
a
number
of
you
have
are
interested
there.
So
that's
good,
I
interjected
to
say
that
everybody
here
is
is
learning
including
myself,
and
so
what
we're
gonna
try
to
we're.
Gonna
have
a
couple
of
sessions,
as
as
the
cncf
has
recently
formed.
C
Well,
the
cncf
hasn't
formed
this.
This
is
more
just
there's
a
band
of
cncf
ambassadors
or
just
nerds,
who
are
recognized
as
being
nerdy
toward
cncf
projects
that
created
a
twitter
handle.
That
then
also
created.
I
forget,
I
don't
know
you
call
it.
I
guess
a
stream
on
twitch,
that's
part
of
the
cncf
stream,
so
it's
really
just
like
a
loosely
managed
channel
there
and
twitter
handle,
but
it
speaks
to
students
broadly
and
many
of
you
are
students.
C
That's
the
goal
of
that
session.
We
thought
we
might
have,
and
so
by
the
way,
if
you
have
suggestions
or
things
that
like
would
be
a
better
tip
or
a
better
trick,
like
please
drop
it
in.
If
you
you
enjoy
sharing
with
others,
it
might
be
that
it's
your
time
to
share
a
tip.
C
We
thought
we
would
have
probably
a
couple
of
other
sessions,
assuming
that
this
goes
well.
Those
other
sessions
might
be
introducing
people
to
measuring
as
a
project
and
going
through
measuring
a
bit
more
deeply
and
not
not
just
sort
of
generically
about
how
to
get
involved
in
a
cncf
project.
That's
what
this
session
is
about.
C
The
next
one
might
be
about
measuring
the
one
after
that,
maybe
might
be
about
introducing
people
to
service
meshes
in
general
as
a
concept
we
give
and
have
given
lots
of
workshops
on
how
to
get
started
with
a
service
smash.
So
maybe
we
would
do
that.
I
I
kind
of
an
open-ended
there's
a
few
proposed
sessions,
and
maybe
how
far
we
go
with
some
of
this
might
depend
upon
how
involved
some
of
you
get
with
it.
A
Thank
you
lee,
so
I
have
shared
the
links
to
all
the
sessions
and
all
the
details
in
the
chat,
so
you
can
take
a
look
at
it
there,
so
we
have
a
couple
of
help
needed
areas
in
measuring,
so
one
of
this
is
to
work
on
the
measuring
workflows
in
in
measuring.
So
basically
the
issue
is
that
our
workflows
run
unnecessarily.
A
A
C
Sure
yeah,
I'm
not
sure
if
we
have
any
like
super
special
call
outs
on
well
on,
like
which
areas
are
more
needed
than
others,
but
there's
there's
something
to
be
said
about
documentation
and
about
things
like
unit
tests
or
things
like
fixing
a
bug.
C
I
think
I
think
those
get
a
negative
stigma
that
is
actually
undeserved.
C
You
can
be
a
super
bright
software,
back-end
software
engineer,
writing
the
craziest
functionality
and
all
condensed
into
two
lines
of
code
that
no
one
will
ever
understand
because
it
was
so
dense
and
and
that's
actually
the
the
point.
That's
actually
the
problem
is,
if
you
don't
go,
if
your,
if
your
fingertips
are
too
sensitive
to
go
type
out
some
documentation
as
to
what
that
does
well
ain't,
nobody
gonna
use
it
again.
Then
it
doesn't
matter
at
all.
It
doesn't
matter,
people
don't
use
it.
It
doesn't
matter.
C
So,
anyway,
the
point
is:
there's
a
lot
that
you
all
have
done
on
this
project
and
and
the
other
ones
as
well,
and
this
there's
a
ton
of
docs
that
are
missing
in
here.
It's
what
a
great
way
to
learn
the
tech
is
by
I'm
documenting
some
stuff,
and
so,
if
you're
thinking
hey,
I
might
turn
my
hand
to
that
and
go
you
know
document
some
things
great
there's
issues
for
that,
but
but
also
there's
just
go
use
the
project
and
if
you
find
a
bug,
go
file,
a
bug.
C
C
Or
there's
just
totally
a
page,
that's
entirely
missing,
there's
a
few
pages
that
are
entirely
there's
some
things
that
all
of
you
can
well
make
a
name
for
yourself
by
like
there's
a
there's
a
whole
another
thing
that
we
do
in
the
community
about
having
folks
that
want
to
write
blog
posts,
there's
a
place
to
to
publish
those,
a
blog
post
on
tips
and
tricks
on
how
to
use
mastery
cto
I
mean
the
blog
post
can
be
very
short,
be
very
long.
C
There
could
be
a
video
if
you
don't
want
to
write
a
post.
It
can
be
so
those
things
are,
those
are
all
very
much
needed.
It's
I'll
talk
to
some
very
you
know
well-seasoned
engineers
who
will
come
over
and
look
at
the
project
and
go
okay.
Well,
so
where
do
I
start
like?
What
am
I
supposed
to
do?
I'm
running
it
now,
but
then
like
what's
next
and
it's
kind
of
it's
kind
of
surprising?
How
often
that
sentiment
is
conveyed,
and
it's
important
and
so
docs
are
important.
I
guess
what
I'm
saying.
C
C
C
There
was,
I
think,
while
we're
on.
I
think
this
is
kind
of
a
general
call
for
a
couple
of
areas
that
are
in
need
of
assistance
and
so
mesherie's
ui,
if
you're
into
figma,
I'm
sorry
if
you're
into
ux
or
individual
design
either
one
there's
always
need
for
assistance,
and
it's
not
just
mesher's
ui.
There
are
six
different
sites
that
the
community
has.
That
always
need
some
help.
C
But
yeah
there's
one
other
item
that
I
would
call
out
that
isn't
also
isn't
always
obvious
to
people.
I
was
just
talking
to
neha
about
a
meeting
that
goes
on
once
every
two
weeks.
It's
the
continuous
measuring,
continuous
integration
meeting.
We
sometimes
talk
about
devops-centric
things
there
as
well.
The
the
biggest
open
need
in
terms
of
artifacts
that
in
terms
of
ci
for
measuring,
is
well
the
fact
that
the
measuring
container
image
is
like
1.7
gigabytes.
C
It's
healthy
is
the
most
polite
way
of
saying
it.
It's
pretty
thick
a
little
overweight
obese,
it's
morbidly
obese.
It's
it's
a
put.
It's
a
turn
off
to
people
who
want
to
use
the
project.
They
have
to
wait
for
the
thing
to
download
and
while
waiting
for
the
main,
the
main
adapter,
I'm
sorry
the
main
container
image
to
download
they're,
also
looking
at
the
other
mastery
server,
I'm
sorry
mesh,
readapter
containers
download
and
they
start
to
get
impatient
and
there's
so
slimming
down
that
container
image
is
open
item.
That's
in
need
of
help.
A
Thanks
lee,
so
suhani
also
has
another
call
for
action.
Suhani.
Would
you
like
to
talk
about
this
smash
labs.
H
Sure,
okay,
so
this
is
about
interactive
service,
mesh
labs
and
it
has
guided
scenarios
where
we
can
learn
how
we
can
use
mashri.
H
And
now
we
are
looking
for
volunteers
who
can
help
us
in
improving
the
existing
labs
and
also
help
us
in
creating
the
new
labs
and-
and
we
have
few
volunteers
who
have
already
signed
up
and
their
names
are
sorry
if
I
misspell
your
name,
it's
okay,
it's
a
surya,
shankar
das,
vijay,
aditya,
chatterjee,.
H
So,
which
of
the
scenarios
you
want
to
pick
like
you
want
to
improve
the
existing
labs
or
create
new
labs,
which
is
mentioned
here.
H
Okay,
great
so
I
didn't
get
your
name.
Sorry.
A
J
A
This,
so
that's
like,
please
jump
in
and
pick
it
up.
A
Thank
you
suhani
for
that.
Thank
you
for
leading
this
up
anita.
Would
you
like
to
talk
about
the
community
handbook
next.
J
Yeah
sure,
hello,
everyone,
let
me
try
to
share
my
space.
E
J
J
Longer
to
load
but
yeah,
we
my
point
exactly
why
I
brought
this
up.
Is
the
handbook
right
now
the
one
on
the
website
it's
confusing
even
to
persons
in
the
community
who
would
want
to
read
it,
and
that
is
not
what
we
want
to
achieve.
We
want
a
handbook
that
if
anyone
comes
up
and
goes
through
it,
they
can
be
able
to
understand
the
community
without
needing
an
extra
guide
and
then
so
far
these
documents
outlines
everything
that
we
need.
We
have
so
many
people
contributing
like.
J
There's
so
many
persons
right
now
goodness
daniel
ruth
myself,
krishna
right.
I
think
I
got
me
right.
Yes,
so
many
other
persons
are
working
on
this
hundred
right
now,
but
we
need
more
hands
because
if
we
needed
to
come
out
faster,
we
need
one
hand
to
help
with
the
documentation,
especially
because
that's
the
major
aspect
with
compared
to
the
previous
one.
It
was
so
disorganized
and
put
together
everything
in
this
table
of
contents,
but
it
still
needs
like
to
be
moved
from
the
previous
dot
to
this
one.
J
So
everything
is
organized
according
to
this
table
of
content,
so
we
don't
have
to
have
the
challenge
of
doing
that
manually
on
the
website.
We
just
move
everything
from
this
doc
directly
to
the
website
and
yeah,
I'm
just
calling
for
more
contributors
to
help
us.
We
have
someone
who
has
taken
on
the
about
page.
Another
person
is
working
on
the
community
community
heading
another
is
working
on
the
project.
J
Another
person
is
taking
another
repository
overview.
They
learn
layer,
five,
so
all
those
are
assigned,
but
yeah.
We
could
still
need
more
hands
working
on
this
and
also
the
ui.
We
still
need
a
lot
of
ideas
on
the
best
way
to
put
together
the
ui,
because
there's
no
fixed
design
for
it.
So
we
we
just
it's
not
a
very
complex
design.
J
Actually,
we
just
need
more
hands
persons
who
are
very
who
have
a
good
idea
of
design,
maybe
on
the
front
and
and
also
persons
who
have
an
idea
of
documentation,
and
you
don't
have
to
be
an
expert
at
it.
Actually,
so
just
the
idea
is
enough:
how
how
we
would
best
read
a
document
to
help.
You
is
the
way
we
want
this
to
look
like.
So,
if
you
think
the
current
state
right
now
is
good
enough
for
you
as
a
contributor
already,
then
you
should
picture
it.
J
How
someone
who's
coming
to
the
community
for
the
first
time
would
feel
reading
document,
and
for
me
I
don't
think
it's
good
enough,
so
we
could
use
more
hands
on
the
suggestions
and
best
ways
to
improve
this
yeah.
So
that's
just
about
the
handbook
right
now.
We
just
need
more
volunteers
to
help
us
out
here.
A
Thank
you
there.
There
is
a
couple
of
feedback
in
the
comments.
One
was
to
indicate
the
languages
in
the
repository
overview
page
to
indicate
the
core
language
or
the
core
text
stack
of
each
of
the
projects.
J
Yeah,
we're
walking
was
that
currently
I
was
actually
looking
at
the
repository
and
the
project
overview
before
now,
and
I
I
actually
wanted
to
include
that
consensus
on
the
website.
It
doesn't
have
the
languages
indicated,
which
is
actually
core
experience
for
somebody
who
would
want
to
like
hop
on
any
of
our
projects
that
will
be
included
as
well.
We
still
need
more
suggestions.
You
can
drop
your
suggestions
in
the
slack
channel
to
any
of
the
mesh
meets.
Actually
you
can
reach
out
to
me
or
any
of
the
measurements
on
them
would
implement
it.
A
Sounds
good
sounds
good,
so
reach
out
to
anita.
If
you
have
any
ideas
and
suggestions
before
we
move
on,
would
anyone
do
anyone
else?
Has
any
comments
right
now.
A
H
Yeah
so
yeah,
so
actually
there
is
the
issue
like
which
opened
for
promises
and
rafa
now.
So
I
would
like
to
like
let
samir
start
first,
because
he
has
opened
that
issue
so
he'll
give
the
context
what
we
want
to
do,
and
then
I
mean
actually
we
wanted
to
understand
from
the
code
as
well
like
because
we
wanted
to
move
from
the
east
to
add
on
so
somewhere.
Would
you
like
to
go
ahead.
F
Hi,
yes,
so
what
exactly
happens
now?
The
prometheus
and
the
grafana,
which
has
been
installed
from
the
ui
of
missouri
or
and
from
the
mercedes
ctl
cli
that
installs
the
prometheus
and
grafana
from
the
like?
You
know
from
the
ist
add-ons
so
wherein
it
is
a
dependency
on
istio
package.
F
But
if
somebody
wants
to
use
any
other
service
mesh,
then
they
may
not
be
able
to
install
that
because
it
has
to
be
available
from
the
istio
addon
folder,
which
is
been
pointed
in
the
like
various
codes.
So
that
is
where
we
like
you
know.
I
I
just
had
a
thought
which
I
converted
regarding
if
we
can
just
install
it
from
the
community
and
chat
which
is
there,
so
that
is
why
this
particular
issue
is
there.
C
Oh
yeah,
oh
yeah,
what
ticket
so
people
when,
when,
when
users
run
meshery,
we
want
to
try
to
reduce
the
time
to
value
to
as
little
bit
as
possible,
which
is
to
say
as
soon
as
they
run
the
installation
command.
A
meshery
will
actually
download
itself
it'll
start
itself,
it'll
open
up
your
browser.
It'll
like
get
the
user
pretty
far.
C
So
how
how
to
make
sure
that
prometheus
is
there
and
prometheus
is
collecting
interesting
information.
Well,
sometimes,
a
lot
of
times
measuring
will
be
connected
to
a
cluster
that
has
an
existing
deployment
of
prometheus
good
mesh.
Sync,
which
is
a
component
of
measuring
it'll,
discover
that
there's
an
existing
grafana
existing
prometheus
retrieve
information
about
how
to
connect
to
those
and
pull
that
into
measuring
it'll,
make
it'll
and
measure
your
attempts
to.
If
there's
just
one.
C
Actually,
I
don't
know
if
measuring
takes
it
this
far
a
minute,
and
we
should,
if
there's
just
one
prometheus
and
one
grafana
meshrey,
should
try
to
auto
connect
and
actually
to
try
to
pull
over
some
interesting
stats
like
immediately
so
that
way,
the
time
to
value
is
reduced.
People
are
hopefully
at
some
point
delighted
by
the
software
delighted
that
mesherie
is
you
know,
being
informative,
is
enlightening
them
okay.
Well,
if
prometheus
or
grafana,
these
two
metrics,
the
metrics
collector
and
this
visualizer
of
metrics.
C
If
those
aren't
available
well,
each
service
mesh
can
come
with
their
own
sort
of
slimmed
down
version
of
prometheus
or
slim
down
version
of
grafana
and
good.
You
know
mesh
recon
lets
people
deploy
those
slimmed
down
versions,
that's
helpful!
You
can
get
some
metrics
from
them,
it's
good,
but
those
slimmed
down
versions,
one
they
may
not
be
there
right
away
so
so
it
may
take
users
additional
time
to
get
to
that
value,
to
deploy
those
things
and
connect
to
them.
Two
after
they
deploy
them.
C
The
slimmed
down
versions
that
come
with
those
service
meshes
aren't
collecting.
C
C
When
someone
runs
a
performance
test
in
measuring
it's
imperative
that
it's
not
just
characteristics
of
latency
and
throughput
of
the
of
the
requests
that
are
generated
and
analyzing.
Those
requests
like
it's
good
to
analyze.
How
long
did
it
take
for
each
of
these
requests
to
come
back?
How
many
requests
could
measuring
the
load
generator
push
through
at
one
time?
It
can
characterize
the
performance
of
a
mesh
in
that
way,
that's
good,
but
if
you,
if
measuring,
doesn't
also
capture
the
node
level
resources,
the
number
of
cpu
cores
number
cpu
sockets
the
memory.
C
If
it
doesn't
capture
and
how
many
nodes
there
are
well,
then,
unfortunately,
that
makes
the
performance
test
not
comp.
You
can't
compare
it
to
someone
else's
performance
test
or
you
can't
compare
it
across
environments.
If
you
have
two
different
environments,
because
you
don't
have
you
didn't
capture
all
of
the
detail,
that
was
needed
to
really
characterize
and
provide
a
frame
of
reference
for
the
performance
of
the
mesh
at
that
time,
running
on
ten
nodes
or
one
node
or
an
overloaded
node
or
what
was
what
was
the
deal
so?
C
Hence
I
guess
what
I'm
getting
to
is
for
a
couple
of
different
reasons:
almost
guaranteeing
that
when
you
deploy
meshri,
you
kind
of
by
default
also
get
prometheus
with
a
node
exporter.
C
Now,
there's
a
little
bit
of
gotchas
in
there.
It's
like
well,
okay,
if,
if
mastery
deploys
to
an
existing
kubernetes
cluster,
maybe
the
default
behavior
is
to
deploy
an
instance
of
prometheus
with
a
node
exporter.
So
it
can
start
collecting
and
mastery
can
automatically
connect
to
it
and
then
grab
the
exported
metrics
and
then
give
people
that,
like
whiz
bang,
give
them
that
delightful
experience
and
capture
the
necessary
data.
As
someone
runs
a
performance
test,
that's
one
scenario:
well,
the
other
scenario
is
no
no
like
this
deployment
has
been
using
prometheus
for
decades.
C
You
know
for
sin,
you
know
for
a
while,
like
they've
got
prometheus
deployed,
they've
got
it
and
maybe
aren't
running
under
the
prometheus
operator,
and
they
got
running
as
a
damage
that
they've
got
whatever
configuration.
They've
got
okay
so
find
then,
then,
in
order
of
sequence,
maybe
measuring
mesh
sync
should
run
first,
identify
whether
or
not
it
there
is
an
existing
prometheus
which
it
does
already
and
it
and
it
collects
information
about
how
to
connect
to
it.
That's
what
I
had
said
a
moment
ago.
C
C
Let's
go
ahead
and
deploy
a
prometheus
with
that
node
exporter,
because
it's
it's
just
really
important.
Now
we
can
give
people
controls
over
oh,
like
they
don't
like
that.
They
didn't
want
meshri
to
do
that.
They
have
a
different
setup
and
they
want
to
do
whatever
fine.
We
can
expose
controls
over.
You
know
over
those
settings
for
when
someone
installs
measuring,
but
by
default
mesherie's
opinionated
deployment.
C
I
would
propose
to
you
all
that
it
should
go
ahead
and
include
prometheus
node
exporter
and
and
deploying
that
using
helm,
makes
tons
of
sense
and
and
deploying
with
a
particular
configuration
that
is
of
interest
to
to
measuring.
So
so
there's
a
couple
of
parts
to
that
issue.
I
think,
like
okay,
what
what
statistics
are?
If
we
do
the
node
exporter
great,
is
it
doesn't
need
any
additional
subsequent
configuration,
probably
not,
but
before
a
node
exporter
is
deployed.
C
First
verify
that
mesh
sync
has
is
done
with
its
discovery
and
doesn't
have
an
existing
prometheus,
okay,
great
or
maybe
if
it
does
have
an
existing
prometheus,
but
there's
no
node
exporter.
Okay,
then,
should
we
provision
a
node
exporter,
and
so
samir
I've
just
walked
through
a
lot
of
things
that
I'm
sure
you
already.
You
know
that
was
already
part
of
your
thought
process.
There.
F
F
H
Yeah,
so
so
liam,
and
can
we
note
down
the
like
steps
here
how
to
start
implementing
this?
Because
it
has
few
like
sub
steps?
I
mean
the
task.
Actually,
we
can
start
because
I
will
need
to
check
like
the
code
where
we
I
need
to.
I
mean
implement
this
part
in
which
repository
and
all
those
things,
so
that
would
be
great
here
to
get
started.
H
Yeah
yeah
I
mean
first,
we
will
try
to
like
complete
the
installation,
part
using
helen
chart
right
and
then
we
can
go
for
exposing
metrics.
Also.
H
Yeah,
but
I
wanted
to
understand
from
this
like
measuring
repo,
like
how
do
we
start
with
this
implementation?.
C
Sounds
great
there's
I'm
gonna
put
a
link
to
I'll
put
a
link
that
will
get
us
going
on
writing
a
functional
specification
in
which
we
can
capture
some
of
the
logic
that
I
just
verbally
walked
through
and
then
it
can.
C
We
can
also
capture
like
where
the
best
section
of
the
code
is
to
place
this
logic
and
yeah
and
in
the
meantime,
like
yeah,
we
you,
you
know
doing
a
manual
deployment
of
the
prometheus
and
of
the
node
exporter
kind
of
familiarizing
familiarizing,
with
what
metrics
come
back
and
things
like
and
then
we
can
identify
which
ones
are.
You
know
critical,
which
ones
should
we
just
by
default,
show
in
the
dashboard,
because
those
would
be
you
know
generally
interesting,
like
yeah.
All
that
info
can
be
captured
in
the
design
spec.
C
I,
as
it
turns
out,
it
won't
be
a
small
thing
like
in
in
the
best
of
ways,
that's
great
like
because
we'll
need
to
expose
control
to
users
to
like
de-provision
that
prometheus,
they
don't
want
it
we'll
need
to
yeah.
There's
it's
good.
It'll
include
kind
of
the
full
gamut
of
concerns.
H
H
Okay,
so
I'll
I'll
wait
for
the
updates
here.
Yeah
I
mean
we
can
then
get
started.
I
mean
I
will
also
check
on
the
prometheus
side
that
helm
chart
and
all
those
things.
C
Sounds
great
and
we've
got
yep
that
sounds
good
I'll
help
I'll
get
into
that
design.
Spec.
If
you
guys
want
to
create
a
new
one,
you
feel
free.
If,
if
you
don't
I'll
I'll
create
one,
and
there
is
prior
art
in
the
community,
drive
right
now
for
well
we'll.
We
will
probably
end
up
using
some
terms
and
we'll
probably
end
up
talking
about
static
boards
and
dynamic
boards,
and
so,
if
dhruv
is
listening,
he
might
point
us
to
a
brief
document
on
what
I'm
talking
about
so.
A
Yeah,
thank
you.
Thank
you,
samir
and
thank
you
lee.
The
link
to
create
a
design
spec
is
in
the
chat
like
you.
Can
the
directions
to
create
a
new
respect
is
in
there
all
right.
Moving
on
to
the
next
topic,
would
you
like
to
talk
about
the
recent
updates
in
the
adapters,
as
well
as
measuring
ctl
mesh.
D
I
This,
the
mystery
ctl
that
I'll
be
using
for
this
demo
is
not
yet
released.
So
this
is
my
own
compiled
version
now
to
deploy
a
service
mesh,
you
can
use
the
deploy
command.
Basically,
some
of
the
arguments
that
it
takes
is
the
name
of
the
adapter
and
the
token
also
name
of
the
service
list
specified.
As
I
mean,
the
name
should
be
compliant
with
smp
protocols,
because
that's
what
we
aim
to
standardize
amongst
everywhere.
I
I
It's
basically
used
for
performing
a
smi
conformance
test,
possibly
some
other
specifications
in
future,
but
right
now
on
the
sma
conformance
test
and
both
of
these
commands
use
a
watch
flag.
So
this
is
a
thing
that
we
should
know
this
watch
flag
right
now
is
just
to
make
sure
that
we
wait
for
a
response
from
the
event
streamer
api,
because
we
still
have
to
build
a
good
nesting
system
in
here.
So
right
now
this
is
a
little
bit
hacky
but
reliant.
I
I
I
just
lost
connection
to
my
ssh,
okay,
so
yeah.
I
had
an
error
while
running
a
smi
conformance
test,
and
that
happened
because
I
didn't
have
a
load
balancer
in
my
system,
so
it
returned,
then
oh
one
exit
code,
which
means
that
the
previous
command
didn't
execute
successfully
and
this
would
be
used
in
the
kit
of
action
to
fail
it
like.
I
I
Osms
list
and
I'll
just
verify
these
so
yeah.
This
fits
the
latest
osm.
That
is
0.9.1,
and
this
is
what
the
next
topic
is
about.
That
is,
improvement
in
the
osm
adapter
to
fetch
the
latest
release,
rather
than
the
hard
coding
of
the
versioning
that
was
there
earlier
so
and
just
take
out
the
tr
real,
quick.
I
And
now
we
are
installing
using
helm
charts,
which
is
the
latest
instruments
repo
and
for
side
car
in
which
can
we
know
some
cli,
but
rather
doing
it
by
annotating
the
community
space
and
labeling
it
appropriately
other
than
that
there
was
some
cleanup
and
we've
happened
in
this,
and
this
now
ensures
that
osm
adapter
should
work
fine
with
latest
releases
and
yeah.
Basically,
that's
it.
There
is.
I
There
is
still
one
thing
that
is
to
be
done
in
one
of
the
miss
sub
commands
and
that
is
bringing
in
these
changes
to
mystery,
ct
and
miss
removed,
which
will
remove
the
service
mesh.
So
if
we
have
it
right
now,.
I
It
would
look
something
like
this.
We
have
this
right
now,
but
it
isn't
like
same
as
the
deploy
and
the
validate
commands.
It
should
wait
for
the
service
mesh
to
stop
and
then
exit
and
some
of
the
changes
have
to
be
replicated.
So
if
someone
wants
to
work
on
this,
there's
a
call
for
help
and
open
up
an
issue
in
a
bit
and
you
can
get
started
with
this.
A
A
K
So
basically,
it's
a
very
small
update.
I've
recently
created
a
pr
so
for
the
context.
If
anyone
has
tried
to
use
mini
cube
with
with
mystery,
you
might
have
stumbled
upon
a
issue
that,
if
you
try
to
do
so
without
a
flattening
or
modifying
your
cube
contact,
that's
that
is
basically
flattening
or
when
you
find
your
cube
control,
I'm
actually
sort
of
throws
an
error,
because
it
runs
in
a
container
and
does
not
have
access
to
the
many
q
files
to
flatten
that
thing.
K
The
configuration
so
now
with
this
pr.
You
have
the
support
out
of
the
box.
That
is
you
can't
just
basically,
you
can
use
mini
cube
and
it
will
automatically
flatten
it
internally.
That
is
so.
In
order
to
do
so,
I
had
to
start
mounting
mini
cube,
mini
cube
directory,
so
mini
cube
directory
now
gets
mounted
on
mystery
container,
not
all
of
the
adapters.
K
It
gets
mounted
onto
a
machine
container
and
mystery
internally
flattens
the
cube
container
and
then
passes
on
that
configuration
to
the
adapter,
so
that
adapters
can
actually
use
that
one
in
the
flattened
cube
conflict.
So
in
this
way
you
don't
have
to
basically
flatten
or
minify.
Your
duplicate
manually
measure
will
take
care
of
that.
So
partially,
one
of
the
update
is
that
now
you
can
actually
start
in
a
mini
use,
mini
cube
out
of
the
box
without
doing
any
further
steps.
K
The
second
update
is
actually
now
mystery
will
be
able
to
support
multiple
context.
At
the
same
time,
by
same
time,
I
basically
mean
that
we
now
have
apis
using
which
you
can
actually
switch
context.
Just
like
you
would
do
in
cube
cut,
so
you
can
send
a
put
request,
not
that
you
would
be
using
the
api,
but
now
there
are
apis
which,
using
this
particular
api,
that
is
sending
a
put
request
to
this
certain
endpoint.
You
would
be
able
to
switch
context
and
the
adapters
will
also
follow.
K
Basically,
if
you
send
this
particular
request
and
say
that
hey
I
want
to
now
use
my
ccp
cluster
to
see
the
configuration
will
be
sent
to
mushroom
server
and
message
server
propagate
the
same
information
to
adapters,
just
like
it
used
to
and
another
basically
endpoint
that
has
been
added
is
to
get
all
of
the
context.
We
had
similar
end
point
for
getting
context,
but
it
was
from
extracting.
It
was
about
more
about
taking
your
cue,
config
and
extracting
context
out
of
it
and
sending
back
now.
It
is
more
about
that.
K
Hey
mystery
knows
that
these
contexts
exist
and
we
are
letting
you
know
that
hey.
We
are
aware
of
these
particular
context.
So
basically,
this
this
is
the
short
update
out
of
box,
mini
cube,
support
and
ability
to
switch
context
right
from
mystery.
K
Yeah
yeah,
so
I
think
right
now
we
have,
I
don't,
have
mystery
running
movies
right
now.
We
have
system.
K
We
actually,
even
today,
support
basically
display
all
of
the
contacts
that
we
are
aware
of.
So
one
of
the
updates
that
we
can
do-
and
that
is
the
reason
I
basically
created
this
particular
endpoint-
was
that
now
we
can
start
showing
more
than
one
chips
and
you
should
be
able
to
ping
all
of
them,
because
now
we
can
switch
context
on
the
fly
as
it
is
loading
up.
K
So
one
of
the
updates
that
I
would
suggest
on
the
ui
is
to
be
is
to
show
more
than
one
tips,
at
least
here,
and
second
update
would
be
to
show
context
names
here
so
right
now
I
have
actually
only
one
context,
but
you
what
you
should
be
able
to
do,
because
we
have
support
for
changing
the
context
from
this
particular
endpoint
and
listing
all
of
the
context
from
this
particular
endpoint.
You
should
be
able
to
like
the
ui
should
be
able
to
list
all
the
contexts
in
here
you
switch.
K
You
should
be
able
to
switch
it
here.
At
the
same
time,
you
can
probably
not
sure
about
this
one,
but
you
should
probably
be
able
to
have
a
chip
here
which
says,
for
example,
if
I
mini
keep
running
and
kind
of
running,
you
should
be
able
to
see
both
of
them
right
there
and
all
of
that.
Basically,
all
of
that
should
be
achievable
using
these,
but
two
endpoints
that
we
have.
K
Okay,
it
does
not
have
any
cube
right
now,
but
what
would
happen
is
that
if
the
so
the
pr
is
about
that,
if
the
pr
sees
that,
if
the
code
actually
sees
that
you
have
some
file
references
in
in
the
in
your
cube
config,
then
it
will
try
to
resolve
them.
I'm
not
sure
if
I'd
be
able
to
show
it
but
yeah
it
is
yeah.
So
what
the
it
doesn't
actually
look
specifically
for
mini
cube
the
yeah.
The
code
is
a
bit
more
generic
than
that.
K
What
it
will
do
is
it
will
look
for
it
will
find
the
potential
candidates
for
flattening.
So
it
will
see
that
is
there
something
that
looks
like
a
filebot.
If
it
does,
then
it
will
go
ahead
and
try
to
confirm
that
does
the
file
actually
exist
because
there
can
be
it's?
Basically,
it
can
be
a
false,
false
positive,
also
because
you
can
have
a
stream
paths
there,
so
it
just
goes
and
try
to
check
if
there
is
a
actual
valid
path
from
the
file
system.
K
And
if
it
does,
then
it
will
go
ahead
and
try
to
resolve
that
thing
and
by
resolving
it
basically
means
that
it
will
read
that
particular
file
and
after
reading
it
it
will
modify
your
cube
config
not
on
disk.
It
will
modify
it
in
memory
so
that
the
disk
conflict
can
can
be.
It
can
stay
intact
and
messy
doesn't
actually
overwrite
it
or
change
it
in
any
sense,
and
just
replaces
it
with.
K
Basically,
it
adds
hyphen
data
because
that's
that's
a
convention
that
coupon,
because
if
it's
client
type
insert,
then
it's
actually
by
parts
difference
in
incline
pipe
insert
hyphen
data
and
it's
actually
the
embedded
data.
It
will
encode
into
b64
and
then
we'll
just
place
that
in
the
cube
config
and
all
of
this
isn't
just
in
memory
so
that
the
file
system
doesn't
get
depicted.
I
K
A
A
Please,
please
fill
out
the
community
member
form
and
you
will
get
access
to
all
the
community
resources
and
we
also
have
a
newcomers
meeting
every
thursday.
A
So
you
can
join
that
as
well
and
we
have
meetings
almost
I
mean
literally
every
day,
so
you
can
join
this
as
well
to
get
oriented
with
the
layer,
5
projection,
yeah,
and
so
thank
you
guys.
F
K
Actually,
we
won't
change
cube,
config
2.0
grid
will
change,
but
it
will
be
completely
in
memory.
The
file
system,
quantum
won't
change,
so
it
will
stay
in
dark.
That's
not
untouched.
C
Because
I'm
admit
to
not
hearing
everything
you
said,
and
so
did
you
say
that
this
implementation
would
allow
people
to
at
least
the
back
end
that
you're
delivering
it
will
allow
people
to
upload
more
than
one
cubeconfig
file
that
might
have.
K
No,
I
know
this
is
more
about
that
you
have.
We
have
the
same,
keep
conflict
now,
it's
just
that
we
never
used
to
persist
all
of
the
context.
Now
we
actually
keep
it
in
memory
and
we
can
return
it
back.
You
can
actually
switch
context.
That
is
again
something
that
was
not
supported
earlier.
It
was
like
that
you
can
use
q
code
to
change
the
context
and
then
delete
the
chip
reload,
the
page
so
that
we
have
a
refreshed
cluster
connection.
Now
you
can
do
it
just
using
mystery.
K
A
I'll
write
any
other
comments
here.
K
Yeah
in
case
of
actually,
if
it's
spot,
then,
if
it's
spot,
then
it's
using
inclusive
configuration.
In
that
case,
we
are
not
actually
dealing
with
a
tube
config
and
if
it's
a
container,
then
if
it
happens,
actually
what
happens
is
that
if
you
find
your
cube
config,
then
that's
fine,
and
if
it's,
if
we
do
not
find
it,
we
go
ahead
and
try
to
read
it
from
the
disk,
and
at
that
point
we
actually
flatten
it
and
store
it
in
coupons
like
in
memory.
K
So
at
no
point,
I
believe
a
user
would
actually
be
basically
prompted
that
hey.
We
cannot
flatten
this
thing
or
something
like
that.
It
all
happens.
The
same
flow
yeah.
I
C
Yep,
but
think
about
this,
if
you
would
so
a
natural
progression
here
might
be
that.
Well,
you
know
if
a
remote
provider
so
chooses
to
persist
those
cube
contexts
that
won
over
time
as
as
mastery
goes
to
inevitably
support
users.
C
Loading
multiple,
like
loading
up,
multiple
cube
configs
that
have
any
number
of
contexts
in
them
that
a
natural
evolution
is
that
mescheri
would
a
allow
people
to
load
up
multiple
config
files
and
just
whether
in
memory
for
the
local
provider,
if
you
track
those
and
let
people
you
know
manage
through
those
and
and
then
for
a
remote
provider,
that
does
offer
persistence
and
does
have
more
security.
Around
user-based
access
to
you
know
certain
users
uploaded
certain
contexts
and
they
have
certain
credentials
that
were
provided.
C
You
know
at
that
time,
but
yeah
a
natural
progression
would
be
for
the
remote
providers
to
potentially
persist
and
then
make
it
convenient
for
users
to
not
have
to
like
in
the
case
of
in
samir.
You
know
one
of
the
examples
that
sameer
has
is
an
aks
deployment
and
samir
your
do
you
experience
any
hiccups
around
the
the
notion
that
you
need
to
provide
mesh
reserver,
a
cube
config.
Is
that
something
you
have
to
manually
up?
Are
you
doing
an
in-cluster
deployment,
so
you
get
to
kind
of
bypass
the.
F
C
Anyway,
I
guess
my
point
was
going
to
be.
Is
that,
like
one
of
the
intel
engineers,
that's
working
with
messier
right
now
is
their
use
case.
Their
deployment
is
a
little
bit
different,
so
their
deployment
is
on
a
remote
vm
there's
a
mini
cube
instance,
and
then
on
that
remote
vm
for
remote
from
their
laptop.
So
their
laptop
is
not
running
the
vm
there's
a
remote
vm
they
deploy
mini
cube.
They
also
deploy
meshery
in
this
case.
They
like
to
do
it
not
deploying
meshri
in
cluster,
but
on
docker
on
that
same
vm.
F
Yes,
I
I
did
it
at
the
beginning.
So
when
I
was
kind
of
evaluating,
then
that
needs
few
more
security
groups
and
few
more
configuration
to
like
you
know,
connect
it
with
your
on
cloud
kubernetes
cluster,
which
was
pretty
tedious.
So
I
thought
why
not
to
move
it
inside
the
same
cluster
so
yeah,
because
if
you
create
the
vm
and
you
have
to
create
the
networking
between
the
your
kubernetes
cluster
and
vm
to
yeah
connect.
C
Okay,
exactly
yeah
he's
running
into
those
same
kind
of
hiccups
about
networking,
also
in
this
environment.
This
scenario,
when
we
talk
about
cubeconfig
and
loading
up
cubeconfig,
is
that
he,
as
a
user,
may
have
certain
cube
configs
on
his
laptop,
which
he
might
want
to
provide
to
upload
to
mesheri,
which
is-
and
I
guess
what
I'm
getting
to-
is-
that
hey
at
some
point
we're
going
to
want
to
let
people
upload
multiple
cube,
configs.
C
C
A
Yep,
I
think
barik
is
on
call
barik.
Would
you
like
to
talk
about
it.
L
L
L
Okay,
sure,
all
right,
so
what
I've
been
doing
is
I've
been
developing
some
of
the
the
things
that
we
haven't
been
covered
on
the
service
mesh,
so
in
a
in
a
service
mesh
services,
there
are
two
sections.
Basically,
so
there
are
so
there
is
basically
a
control
plane
and
also
a
data
plate.
So
in
measuring
right
now
we
are
only
showing
the
control
pane
yep.
L
So
what
we
are
we
were
trying
to
do
here
is
basically
we're
trying
to
add
some
of
the
features
for
for
for
seeing
how
the
data
plane
for
each
of
the
control
planes.
Basically
so
I've
written
a
pr
about
it
and
I
think
it's
sort
of
done
actually.
So
what
what
I?
What
I
am
doing
is
basically
I,
when
you
hover
on
one
of
the
whatever
one
of
the
control
plane
you
cannot,
you
can
actually
see
what
is
the?
L
What
is
the
data
planes
that
are
consist
inside
of
that
control,
plane
yep,
so
we
can
see
it
right
now,
so
at
grafana
control,
plane,
there's
no
step
when
this
is
running
and
then,
if
we
see
on
the
ingress
gateway
so
yeah,
so
we
are
seeing
that
istio
proxy
and
then
status
ready.
L
What
is
the
image
and
what?
What
is
the
ports
inside
of
that
data
plane?
So
for
the
actual
implementation
of
how
I
did
it,
it's
basically
from
all
the
containers
inside
the
inside
the
services.
Basically,
so
I
I
filter
out
the
name
if
the
name
of
the
containers
includes
the
name
includes
proxy
I'll,
I'll
filter
it
and
then
I'll
get
the
information
for
that
particular
container,
meaning
that
that
container
is
a
proxy
for
that
control.
Planes
yep!
L
So
that's
what
I've
been
doing
yeah
so
currently,
I
can
just
show
this
kind
of
fields.
I
think
I
think
making
it
into
a
tooltip
is
a
great
way
to
to
to
know
the
relationship
of
data
plane
and
control
play
yep.
So
I
think
yeah.
So
that's
what
I
think
all
right!
Can
you
feedback.
C
C
Got
you
cool,
okay,
yeah
I'll,
definitely
dig
in
with
you.
This
is
this.
Is
nice
it's
a
good.
It's
a
nice
idea
that
you
know
one
of
the
things
that
you're
conveying
is
that,
like
hey,
there's,
there's
a
workload,
that's
running,
maybe
it's
an
application.
Maybe
it's
a
control
plane
component.
Maybe
it's
a
visualize!
Maybe
it's
a
grafana
prometheus,
it's
like
and
each
one
of
those
like
what
you're
essentially
designating
here
with
the
tooltip.
Is
this
workload
whatever.
C
C
C
The
format
of
like
the
sunday
call
would
be
to
grab
a
mesh
ready
to
play
like
basically
try
to
take
the
the
test
plan
that
I
the
spreadsheet
that
we
shared
like
run
through
some
deployments,
some
vanilla
deployments
of
like
of
measuring
and
look
to
see.
If
there's
any
hiccups
going
on
identify
those
you
can
you
can
do
it
haphazardly
or
you
can
do
it
following
it.
According
to
the
test
plan,.
C
So
it's
good,
so
it's
a
good
way
to
get
familiar
with,
measuring
as
well.