►
From YouTube: Meshery Development Meeting (Aug 11th 2021)
Description
Meshery Development Meeting - Aug 11th 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
everybody
to
the
measuring
development
meeting
today
is
11th
of
august
and
we
have
a
schedule
today.
A
lot
of
demos
so
sit
back
and
bring
in
those
critics
inside
you.
So
so
the
first
couple
of
items
is
related
to
measuring
ui.
So
nedish,
would
you
like
to
get.
A
B
So,
actually,
regarding
the
first
item
like
sometimes,
we
are
actually
having
issues
with
the
department
of
missing,
so
me
and
dhruv
were
actually
discussing
about
this-
that
whether
it
would
actually
be
it
would
actually
benefit
us.
If
we
have
a
separate,
you
know
functionality
in
the
ui
to
deploy
that
manually.
B
That
is
sort
of
something
that
we
thought
we
could
discuss
it
in
the
community.
I
mean
discuss
it
in
this
meeting,
so
yeah
like
we
are
like
I'm
actually
waiting
for
here
to
get
your
opinion
on
this
on
whether
we
should
proceed
than
this
or
you
know,
whatever.
C
Yeah
so
like
to
give
you
a
a
small,
a
brief
detail
on
what
is
happening.
Basically,
when
you
connect
to
meshing,
there
are
three
processes
happening
underneath
first,
is
you
deploy
mystery
operator,
which
indeed
also
starts
a
lab
server?
Then?
The
second
thing
which
happens
is
you:
is
the
meshi
server
tries
to
connect
to
that
net
so
as
a
client
and
the
third
thing
which
happens
is
after
it
is
successfully
able
to
connect
to
that
particular
nat
server?
It
will
deploy
missing
now
in
any
of
this
three
different
steps.
D
C
Meshi
operator
is
deployed,
then
we
consider
that
operator
is
deployed
and
everything
is
open
to
give
them
specific.
C
Use
cases
for
them
to
connect
themselves,
so
that
was
what
we
had
in
mind.
We
could
separate
them
into
three
stages
and
we
can
allow
them
to
allow
allow
them
to
redo
each
stage
if
they
like.
C
E
This
page
is
painful
when
you
don't
get
feedback,
so
this
is
a
proposal
right.
C
C
E
Yep
there
is
additional
control
needed
over
mesh
tank.
Mesh
sync
is
the
linchpin
to
the
rest
of
mesher's
functionality,
working
and
so
between
between
cubeconfig
and
having
it
prepared
and
accessible
to
measuring
server
and
be
between
that
and
mesh
sync.
These
are
the
the
two
foremost
areas
where
people
using
measuring
stumble
into
pitfalls,
and
so
there
there
is
some
thought
that
is
needed
and
some
and
it
needs
to
be
written
down
and
also
diagrammed.
E
It
could
just
be
a
simple
wireframe
for
a
couple
of
utilities
that
dhruv,
I
think
the
dish
are
proposing
that
for
people
to
use
and
it's
the
fact
it's
there's
a
couple
of
scenarios,
there's
a
scenario
in
which
for
every
chip
in
the
for
every
connection
that
meshery
has
to
a
component,
whether
it's
its
own
component
or
an
external
externally
managed
component.
E
When
any.
Every
time
you
get
a
connection
established,
you
get
measuring,
will
show
you
a
chip
in
the
ui
and
if
anyone
has
it
up,
we
could.
We
could
take
a
look
just
as
we
discussed
this,
because
we
might
be
able
to
arrive
at
like
at
least
one
or
two
specific
action
items
here.
Part
of
the
reason,
I
think
that
this
might
need
to
be
written
down.
Is
there
been
some
enhancements
to
empowering
users
with
additional
tests
and
like
health
checks
that
they
can
run
against
mesh
sync
specifically,
which
is
great.
E
E
E
Now
we've
provided
the
to
date,
we've
provided
a
fairly
common
user
experience
for
users
when
they
go
to
click
on
a
chip.
There's
a
quick
ad
hoc
connectivity
test
that
is
invoked.
That's
consistent
across
all
of
the
chips
except
for
the
operator
and
mesh
sync.
E
They
do
not
have
a
quick
connectivity
test,
they
have
a
more
robust
test
happening
and
it
creates
a
dis
and
which
is
good,
but
also
negative,
to
the
extent
that
it
creates
a
discrepancy
in
user
experience
and
if
we've
learned
anything
in
our
meetings,
it's
that
a
lack
of
consistency
is
poor
user
experience.
So
we
want
to
train
the
users
to
expect
a
particular
thing
when
they
click
or
right
click
or
whatever
they're
doing
and
and
to
date
it's
been
there's
a
quick.
Can
you
click
on
the
kubernetes
chip?
Please
pink!
E
Sound
that
goes
with
it.
That
would
be
great
just
a
little
bit,
but
anyway
that's
nice.
It's
really
it
feels
I
mean
to
me.
I
don't
know
how
everyone
else
feels
it
feels
great
like
okay,
yeah,
it's
there
is
it
there.
It's
there
good
good,
good
like,
and
it's
fast,
it's
responsive.
It's!
You
know
it's
just
the
lightest
of
tests.
Just
basically,
when
you
click
that
button
there
are
like
three
things
that
are
affirmed.
E
One
is
that
mesh
reserver,
wherever
it
is
deployed,
has
network
connectivity
to
kubernetes
good.
If
you
get
a
positive
response
rate.
Second
thing,
that's
confirmed
is
that
measuring
server
has
or
that
cube.
Api
is
in
fact
responding
and
is
up
and
available
and
accessible
to
measure.
It
is
responding
to
measuring.
So
it's
it's
awake.
Good
third
thing
is
that
mescheri
server
has
the
right
credentials
to
be
able
to
authenticate
to
that
api,
make
a
quick
test
and
come
back.
So
that's
like
three
things
in
like
a
second
and
a
half
two
seconds.
E
There
needs
to
be
a
written
proposal
about
what
checks
we
have
in
place.
What
ad
hoc
connectivity
tests
we
have
for
the
operator
what
one
we
have
for
mesh
sync,
what
actually
is
performed
in
those
two
ad-hoc
connectivity
tests
and
then
a
list
of
other
operations
that
are
possible,
like
de-provisioning
the
operator,
wiping
mesh
sync
database,
invoking
a
resync
of
just
mesh
sync:
should
we
directly
control
allow
the
user
to
directly
control
the
lifecycle
of
of
sync
like
to
delete
mesh
sync
and
re-install?
E
E
Okay,
but
there
are
other
checks
as
well
like
well
great
mesh
sync
is
there,
but
is
data
actively
flowing?
Are
information
updates
actively
being
pumped
in
from
mesh
sync
through,
like
there's
a
number
of
checkpoints
in
there
and
so
yeah?
That's.
I
think
that
that
is
a
great
proposal
and
it's
so
great.
E
It's
that
there
needs
to
be
a
warning
flag
like
that's
fairly
prominent.
It's
like
you
can
still
use
meshri.
As
a
matter
of
fact,
let
me
just
take
a
quick
look
at
the
attendees.
E
You
know,
there's
a
there's,
an
engineer
from
intel
who's
using
measuring
actively
and
they're,
not
using
meshri
as
it's
connected
to
kubernetes
like
they.
Don't
that's,
not
functionality
that
they're
concerned
with
that
they
care
about.
So
it's
totally
valid
to
not
have
to
use
measuring
without
mesh
sync
connected
it.
Just
it
is
a
degraded
state
for
much
of
the
functionality
not
for
the
functionality
that
they're
concerned
with
so
this.
E
I
think
I
think
we
need
to
get
a
couple
of
things
down
on
paper
to
make
sure
that
we're,
but
you
guys
are
barking
up
the
right
tree
from
my
perspective,
so
I
just
keep
parking.
E
Good
feedback
too
much
feedback,
but
not
what
you're
expecting.
E
No,
no,
no,
that's
not
a,
but
that's
like
hey,
it's
entirely
appropriate
to
just
surface
it
briefly.
Is
it
worth
writing
down
and
spending
that
time
yeah
totally?
It
really
is.
You
guys
are
striking
on
a
very
important.
It
would
be
one
thing
that
would
be
really
helpful.
Is
I
don't
know
if
everyone
on
the
call
is
aware,
but
or
even
but
but
mesh
sync,
you
know
the
information
that
messry
discovers
about
kubernetes
and
all
the
things
going
on
in
kubernetes.
E
Mesh
sync
facilitates
that
discovery
and
retrieves
that
data
pushes
it
into
mastery
server.
So
the
message
server
can
do
intelligent
things
based
on
that,
sometimes
that
synchronization
gets
out
of
sync
or
mesh,
think
it's
out
of
sync
and
you
might
need
there
might
be
circumstances
in
which
you
need
to
wipe
that
database
clean
and
kind
of
resync
it
like
initiate
a
resync.
E
B
Okay,
so
I'll
move
on
to
the
next
stop
with
them,
which
is
you
know
like
it's
regarding
the
same
thing
that
we
discussed
right
now,
since
we
are
actually
not
sure,
like
the
thing
that
we
are
showing
in
the
status
over
here,
it's
not
actually
connected.
I
mean
it's
not
actually
checking
whether
machining
is
working,
and
I
mean
the
data
is
flowing
through
machines
or
not.
B
It's
just
checking
the
operator
whether
the
operator
has
deployed
successfully
and
at
the
status
of
the
operator
is
fine,
so
I
was
thinking
we
should
actually
rethink
about
this
thing
that
you
know
only
when
the
measuring
only
when
the
missing
is
totally
healthy.
We
should
actually
show
it,
so
it
does
a
connected
thing.
Otherwise
it
does
not
make
any
sense,
because
if
it
is
not
up,
I
mean
if
it
is
not
doing
its
job
properly,
then
I
don't
think
it's
good
to
assume
that
it's
connected
and
it's
healthy
right.
B
So
that's
actually
a
part
of
the
same
thing
that
we
were
discussing
the
first
place.
I
guess
so
we'll
think
about
it.
In
the
I
mean,
those
two
things
are
will
get
combined
into
one
thing
and
yeah
we'll
try
to
see
what
how
to
go
about
implementing
that.
E
E
I
don't
think
that
there's
any
necessarily
any
well
from
my
perspective,
my
vote
would
be
that
there's
no
additional
discussion
needed
about
changing
about
what
nation
just
said
is
like
the
the
gating
criteria
for
whether
or
not
someone
should
be
allowed
to
move
to
the
third
step
in
the
connectivity
wizard
in
the
connection
wizard
the
fish.
I
think
that
that's
a
great
proposal
that,
while
measuring
operator,
might
be
healthy,
that's
great,
but
what
really
matters
is,
is
mesh
sync
connected
and
that's
that's
probably
what
your
proposal
sounds
good.
B
Yeah,
so
moving
on
to
the
next
one.
Actually,
this
connection
wizard
is
actually
it
has
been.
You
know,
released
in
the
last
missionary
version.
I
mean
it's
it's
traditional
last
week.
It
says
if,
if
people
doesn't
know
about
this,
it's
already
there.
If
you
go
to
your
like,
if
you
have
the
age
version
running
and
then,
if
you
go
to
the
system
connection,
you
would
actually
see
this
thing
and
you'll
be
able
to
experiment
with
it.
So
basically
we
actually
need
a
lot
of
people
to.
B
You
know
open
this
and
try
this
thing
out
and
see
how
we
can
improve
this
thing
and
how
we
can
go
about.
You
know
there
are
a
lot
of
scope
in
this
area.
That's
what
I
think,
because
when
you
think
about
it
like
this
could
potentially
be
a
point
where
you
can
where
it
can,
like.
You
know,
diagnose
the
whole,
diagnose
all
the
problems
that
that
a
person
is
facing,
while
you
know
deploying
measuring
components
and
stuff.
B
So,
if
like,
if
you,
if
people
are
interested-
and
I
urge
you
guys
to
you-
know,
take
I
install
it
and
see
and
go
through
each
and
every
step
and
whatever
feedback
you
have
just
you
know
that
would
be
very
helpful
in
you
know,
improvising
this
whole
component,
this
new
step
that
we're
taking
towards
so
yeah
right
now
we
see
that
this
ui
is
not
up
to
the
mark,
and
the
reason
for
that
is
because
there
is
only
there's
not
a
lot
of
people
working
on
this
thing,
and
if
we
can,
you
know,
get
together
and
discuss
this
thing
and
you
know
come
up
with
some
new
ideas
and
there
are
definitely
a
lot
of
ideas
that
we
can
come
up
with
to
display
it
more
elegantly
than
what
it
is
right
now.
B
B
So,
according
to
the
functionality
device,
I
don't
think
like
it.
It
is
in
the
basic
stage
right
now
like
it's
in
the
first
step,
I
would
say
we
can
go
in
the
future
like
we
might
have
some
other
functionalities
as
well.
We
might
change
something
here
and
there,
but
then
the
we
will
have
to
come
up
with
the
fine
tuned
ui
that
we
are
going
to
stick
with
some,
at
least
for
some
time
right.
B
So
it
would
be
great
if
you
guys,
can
you
know,
run
this
thing
and
see,
give
us
give
the
feedback
on
how
it
is
working,
whether
it
is
working
on
some
environment,
and
it
is
not
working
on
some
other
environment,
so
we
can
actually
debug
it
and
get
it
working
yeah.
That
is
what
I
wanted
to
share.
B
A
Thank
you
this
would
you
like
to
continue
from
here.
G
Yeah
yeah
sure
so
give
me
one
so
about
the
progress
on
subscription
for
performance
subscription.
So
actually
I
made
some
roadblocks
during
the
few
days
ago,
so
it
is,
it
is
mostly
about
so.
The
question
is
all
about
like
how
do
you?
How
do
you
publish
a
message
from
the
handlers
so
that
the
subscription
know
that
a
new
performance
result
is?
G
Is
there
already?
Basically
so
this
one
I'm
looking
for
and
and
initially
what
I
thought
is
I
I
could
just
create
a
new
kind
of
resolver
for
mutation,
but
then
I
thought
that
there
is
too
much
yeah.
There
is
too
much
and
I
can
just
I
can
just
put
the
publish
message
onto
the
handlers
itself,
but
then
the
issue
here
is
that
the
there
is
a
cyclic
loop
import
that
is
happening
if
I
just
import
the
library
directly
onto
the
onto
the
handlers
package.
G
So
I
I
talked
about
this
with
group
before
and
and
yeah.
I
think
we
yeah.
So
basically,
we
are
still
in
the
discussion.
C
So
a
brief
discussion
on
what
barik
said
over
here
was:
how
do
a
subscription
basically
know
that
hey
data
is
updated,
go
give
the
ui
a
new
data
and
to
do
that,
he
was
talking
about
how
if
something
happens
in
one
particular
handler,
how
will
a
resolver
know
whether
it
is
happening
or
not?
C
So,
basically,
for
people
who
don't
know
in
general,
we
use
channels
to
communicate
whether
something
is
happening
or
not
happening
so
that
the
subscription
know
that
hey
something
new
has
happened
and
the
ui
needs
to
change,
so
it's
in
simulator.
So
that
was
what
ferry
was
talking
about.
C
I
said
this
because
I,
if
someone
is
interested,
they
can
look
into
this
new
design,
spec,
which
we
have
from
moving
to
channels
to
a
more
broadcast
way
where
you
can
have
multiple
listeners.
Basically,
if
multiple
subscriptions
have
the
same
trigger
that,
let's
say
if
you,
if
a
performance
test
is
successful
exit,
then
there
are
five
things
in
your
current
ui
that
needs
to
be
updated.
Let's
say
one
is
the
count
of
total
number
of
tests
and
second
would
be
the
test
profile
itself.
C
E
A
quick,
quick
clarification
so
borrow
the
shift
to
subscription
for
performance
test
results
that
is
not
dependent
upon
a
change
from
independent
channels
to
broadcasts
correct.
E
It
is
independent
good.
Is
that
done
then.
E
Okay,
all
right
is
the
is
the.
C
E
Cool
okay
but
but
yeah,
that's
a
good
good
update
from
both
of
you
guys.
E
Oh
bark
there's,
while
you're
in
there
there's
one
other
thing
I
don't
know:
if
does
anybody
have
measuring
ui
up
at
the
moment
like
you
could
pull
up
a
performance
test
so
for
all
of
you,
who've
run
a
performance
test.
If
you've
had
there's
an
open
issue
on
this,
it's
probably
a
real.
I
think
we've
opened
it
a
few
times.
It's
like
a
really
old
request
and
it's
fairly
straightforward.
E
E
What
endpoint
is
it
that
you
would
like
to
generate
load
against
and
you
can
type
it's
a
free
form
field,
so
you
can
type
in
whatever
url
you
want
and
that's
great
we
want.
We
want
for
people
to
type
in
anything
they
want.
Essentially,
not
everything
would
be
supported,
but
but
you
know-
and
there
is
form,
validation
and
things,
but
but
there's
something
significantly
missing
here.
That's
right
at
our
fingertips.
There's
a
lot
of
data
inside
of
the
mesh
sync
database,
the
mesh
reserver
database,
that's
populated
by
mesh
thing.
E
The
ideal
here
is
that
in
the
service
mesh
drop
down
so
conceptually
if
someone
fills
in
a
profile
name
which
they
don't
have
to
it'll
auto
generate
if
someone
is
running
a
service
mesh
that
is
discovered
and
denoted,
this
drop
down
is
rearranged
and
based
on
what
service
mesh
that
they
will
have
chosen.
There's
a
certain
subset
of
discovered
services
that
can
be
populated.
E
Does
that
use
case
make
sense
to
you,
yeah
yeah?
It
can
be
done
nice.
One
thing
I
can't
tell
you
that's.
I
can't
tell
you
how
many
times
I've
described
that
over
the
last
for
I've,
but
there's
one
thing
that
is
yet
to
be
done,
and
that
is
that,
like
yesterday,
we
were
demoing
measuring
to
a
collection
of
service
mesh
users
and
at
the
on
the
system
that
we
were
using.
E
There
were
three
different
types
of
service
meshes
provision,
and
you
could
tell
because
you
know
for
a
couple
of
reasons
you
could
tell.
But
one
one
reason
is
because
when
you
click
the
drop
down
for
service
mesh,
it
would
show
you
the
three
that
were
provisioned
and
it'll
draw
a
line
underneath
saying
these
are
provisioned,
and
these
haven't
been
and
that's
great.
But
when
you
first
loaded
the
form
it
would
still
be
an
empty
drop
down.
It'll
still
show
you
empty.
E
B
A
B
B
H
B
Yeah,
the
only
thing
that
is
weird
is,
if
I
close
this,
this
is
what
is
happening.
Everything
else
is
fine.
B
So
the
navigator
here,
okay,
wait
refresh.
I
Is
my
is
my
screen
not
visible?
It's
it's
feasible,
but
it's
is
it
active
screen
like
you
have
been
on
mesh
like
pull
request.
C
No,
so
that
we
are
able
to
see
the
screen.
Maybe
your
internet
is
slow.
B
I
So
is
it
only
me
there
who
cannot
see
through
his
screen.
I
E
Okay,
maybe
since
abi
is
since
it's
important
for
abby
to
see
this
now,
do
you
mind
just
stopping
sharing
and
restarting
real
quick
yeah
yeah.
I
Yeah,
so
is
there
any
problem
with
the
navigator
menu
since
there
was
a
slack
message
from
yeah?
This
issue
is,
like
I
haven't
seen
it
earlier,
but
yeah
it's
an
issue
like
the
overlapping
of
chips,
right
yeah,.
I
I
So
there
is
no
problem
with
navigator
right
now
right!
Oh
I
that's
what
I
got
confused
with
you
said
then
navigate
there's,
no
problem
with
the
navigator.
The
problem.
I
Actually
you
some,
I
I
don't
know,
I
actually
don't
remember
like
you
was
the.
I
Was
made
by
me
on
navigator,
so
I
was
like
what
has
been
this.
This
is
really
weird
to
see
so
yeah
this.
This
is
something
that
I'm
seeing
for
now
and
yeah.
We
can
make
this
like.
I
haven't
made
any
changes
to
this
operator
thing,
but
still
we
can
do
collectively.
B
So
other
than
like,
if
you
would,
can
you
create
issues
regarding
this
thing.
D
F
Yeah,
so
one
of
the
issues
that
we
noticed
here
was
that,
because
now
the
calendar
is
a
different
card,
absolutely
as
soon
as
you
are
crafts.
Basically,
you
add
grafana
boards,
this
right,
the
card
on
the
right
top
side
right
now,
which
is
which
it
says,
is
no
service,
mesh,
metrics
confirmation,
yeah,
yeah
and
the
current
just
comes
at
the
bottom.
B
So
yeah,
so
it's
not
like
this
card
is
not
of
a
fixed
height,
so
I'm
pretty
sure
if
it
extends
the
calendar
would
just
go
down.
F
I
It
would
yeah
yeah
and,
and
one
more
thing
like
just
a
nitpick,
that
if,
if
he
has
card
where
there
are
run
test
and
manage
profiles
like
it's
very
less
information
that
is
imposed
on
two
separate
cards,
so
this
can
be
like
made
in
one
card
with
a
you
know:
divider
vertical
divider.
That
would
look
much
sleek
because
yeah
there
is
some
card
over
card
and
that's
yeah.
I
C
B
E
E
You
end
up
with
a
ton
of
wasted
landscape.
You
know,
underneath
the
left
hand
column.
So
one
thing
that
could
be
done
is
that
the
calendar
is
rearranged
to
like
a
week
or
a
day,
and
it's
shown
just
it
is
shown
consolidated
in
the
left-hand
column.
You
know,
underneath
the
results
in
the
profiles
that
way
we
can
guarantee
that
people
can
actually
still
see
what's
going
on
and
also
see,
excuse
me
see
the
fact
that
there's
a
calendar
okay,
what
if
the
user
had
selected
month
as
like?
That's
what
they
want
to
see?
E
E
You
know
stuff
like
that,
like
that,
some
of
that
is
like
cheating,
it's
the
it's.
The
cheat
way
to
get
to
the
answer,
which
is,
I
don't
know,
just
give
the
user
the
tool
and
let
them
you
know,
try
to
do
your
best,
but
then
give
them
the
tool
and
so
to
abby's
point
by
the
way.
Hopefully,
everyone
on
the
call,
especially
those
are
those
that
are
relatively
fresh
or
those
that
are
already
making
an
impact
like
megana,
is,
but
still
you
know,
relatively
new
to
the
community.
E
E
It's
okay.
We
want
to
to
be
better,
so
the
card
on
card
type
thing
that
avi
was
just
calling
out
is
it's,
like.
You
know,
nails
on
the
chalkboard.
It's
like
yeah
yeah.
There
is
a
div.
There
is
a
distinction
between
test
results
and
performance
profiles,
and
so,
but
we
don't
need
to
overly
distinctify
by
you
know
stacking
cards
on
card.
The
other
thing
too
is
like
well
do
we?
It
would
be
great
if
someone
did
an
assessment
to
look
across
the
other
pages.
E
You
know
to
use,
I
think,
to
use
material,
ui
nomenclature
or
vernacular
that
there's
there's
paper
and
there
are
cards,
and
maybe
the
distinction
is
that
a
card
is
just
paper
except
rounded,
on
the
corner.
With
a
shadow
point
is
I
don't
know
that
we're
being
consistent
with
our
use
of
paper
and
cards
and,
what's
being
elevated
and
what's
not-
and
you
know
not,
everything
needs
to
be
on
a
card
now,
if
we're
going
to
let
people
drag
and
drop,
rearrange
some
stuff.
E
Oh
okay,
okay,
but
yeah.
So
just
in.
E
To
do
a
review
across
them
and
look
yeah,
I
think
in
general
there
are
some
common
features
that
we
would
like
to
be
able
to
empower
people
with.
That
is
the
ability
to
ultimately
start
to
have
some
preferences
on
what
it
is
that
they
would
like
to
see
some
people.
The
the
nice
thing
about
a
calendar
is
one
meshery
underneath
the
covers
it
has
the
ability
to
do
to
allow
people
to
define
schedules.
E
There
are
apis
that
define
schedules
or
that
allow
us
to
define
schedules
already,
there's
nowhere
in
the
ui
that
we're
allowing
people
to
have
a
plus
sign
and
like
double
click
on
the
calendar
and
create
a
schedule,
a
recurring
performance
test,
what
an
awesome
capability.
That
is,
that
you
don't
won't
necessarily
find
in
some
other
tools.
E
E
So
if
you
have
some
extra
love
and
you
have
any
amount
of
ux
talent
yeah
anyway,
so
I
wanted
to
call
that
out
and
say,
like
more
than
what
abby
had
said
about
the
cards
being
stacked
and
a
suggest.
A
proposal
about
you
know,
a
simple
slender
divider
is
that
it's
granted
people
are
on
the
page
in
context
of
performance.
That's
the
menu
item
that
they
clicked
on.
That's
the
title
of
the
page.
E
Well,
I
guess
actually,
if
you
click
on
the
performance,
so
right
now
you're
on
profiles
right,
but
we're
not
showing
that
in
the
menu.
No,
it's
pissed.
You
know
that's
too
bad.
Ideally.
D
E
Another
thing
to
think
about
is
like
see
how
the
we
have
the
grid
view
in
the
upper
right
hand,
corner
and
the
table
view.
It's
not
obvious
to
a
lot
of
folks
that
that's
actually
a
thing
and
is
available
to
them
and
the
fact
that
they
actually
get
a
lot
more
data
when
they
go
to
a
table
view
like
that
or
there's.
I
And
yeah,
and
one
more
thing
like
if
you
are
seeing
this
card,
especially
so
a
lot
of
space
has
been
wasted
right.
So
I
think
we
can
use
that
space
to
make
it
more
compact,
yet
more
clear
to
our
eyes,
because
a
lot
of
space
has
been
wasted
like
after
the
four
results,
the
total
space
that
has
been
wasted.
We
can
just
view
this.
I
Our
test
result
can
be
uplifted
a
bit
above
and
and
this
information
can
be
on
new
line
like
last
run,
but
still
that
would
be
much
shorter
information
and
we
could
actually
show
two
three
card
three
cards
or
more
than
two
cuts.
At
the
same
time,
with
extra
information
on
the
dashboard
itself,.
A
What's
next
yeah,
I
did
like
the
next
item.
Like
upsell
has
reported
a
bug.
I
mean
trying
to
reproduce
a
bug.
So
if
you
could
test
that
out
quickly
and
see
if
it
is
a
bug,
it's
the.
A
So
the
bug
basically,
is
that
the
notifications
are
not
falling
into
correct
filters,
so
yeah,
if
you
could
test
it
out
asynchronously.
As
we
move
on
to
the
next
topic
and
and
reach
out
to
ourselves,
then
it
would
be
great.
B
Yeah,
I
just
wanted
to
give
this
update
that
right
now
we
have
pre-committee
installed
with
our
ui.
So
whenever
you,
you
know,
come
and
try
to
commit,
I
mean
first,
you
have
to
run
npm
install
so
that
the
hook
can
be
configured
with
your
dot
directory.
So
after
that,
if
you
guys
like,
if
you,
if
you
want
to
commit,
then
we
would
actually
it
would
actually
render
lint
command
npm
run
link
command
and
it
would
check
the
formatting
for
it
and
yeah.
B
I
just
wanted
to
give
this
update
that
this
is
actually
a
thing
right
now
and
you
guys
before
yeah.
Before
doing
any
development,
you
guys
have
to
run
npm
run.
K
A
A
I
Actually,
last
night
I
was
on
call
with
kash
and
we
just
discussed
how
to
you
know
populate
all
the
things
that
I
was
not
able
to
populate
earlier
and
he
managed
to
like
tell
me
what
was
the
problem
and
actually
it
was
the
host
problem
that
was
not
establishing
the
connection
between
our
machine
server
and
the
adapter
itself.
So
that
thing
has
been
done
and
I'm
I'm
now
integrating
those
things
in
the
ui
itself,
so
that
work
is
yet
not
completed
so
yeah
once
that
will
be
done.
A
Right
that
sounds
good.
Okay,
you
opened
up
this
pr
and
it's
still
in
draft
okay
cool.
Would
you
like
to
talk
about
the
navigator
menu
right
now,
like.
I
Yeah,
so
actually,
I
was
not
sure
about
what's
the
requirement
for
the
navigation
menu
for
now
like
is
there
something
like
there
should
be
an
accordion
type
thing
in
the
collapsed
view
also
like.
That
was
something
that
I
just
read
on
slack
so.
E
On
the
the
navigator
menu
being
there
yeah
avi,
let
me
I'll
there's
a
let
me
catch
up
with
you
or
you
know.
Let
me
catch
you
in
slack
or
something
so
we
don't.
A
All
right,
I
believe,
megan,
is
on
the
call
megan.
I
would
you
like
to
show
the
import
by
url.
M
Yes,
so
I
actually
can't
show
it
live
right
now,
I'm
outside
and
I'm
on
my
phone.
So
but
I
have
updated
my
github
issue
there.
I've
shared
the
screenshots
of
how
it
looks
right
now.
So
if
you
could
go
to
the
issue
assigned
to
me
so
you'll
be
able
to
see
it.
M
J
M
D
C
Basically
rejects
to
evaluate
whether
the
url
input
was
indeed
a
url.
M
Oh
no,
I
I
didn't
add
that
there
was
something
about
the
github.com
needed
some
other
set
setting.
I
don't
know
we
discussed
something
on
that
call,
but
again
we
mentioned
that
for
the
first
version
it's
not
required,
so
I
didn't
do
anything
else
and
I
just
showed
the
current
look
so
I'll
add
that
it
won't
be
a
problem.
E
That's
about
it,
I
think
it's
just
a
change,
maybe
of
verbs.
E
And
then
yeah
other
than
that,
like
yeah,
I
think
you're
you're
hitting
it
on
the
head
that
they're
it
github.
If
the
url
is
at
github.com,
that
will
be
a
special
use
case,
but
it
to
your
exact
recollection,
that's
a
step
two
or
like
it's
enough
that
just
to
allow
anyone
to
import
from
an
http,
url
and
so
yeah.
I
guess
after
they're
done
importing.
You
you'd
probably
send
a
a
notification
to
the
ui
just
to
confirm
that
an
if.
E
M
E
What
is
it
does
that
the
this
is
the
filters
api,
so
when
a
new
filter
is
being
imported,
is
the
is
it
the
end
point?
That's
already
has
some
smarts
about
identifying
whether
or
not
it
was
a
url
at
github.com
versus.
F
So
the
request
differs
if
you,
if
you
are,
if
you
want
to
import
from
a
url,
then
the
then
the
object
that
you're
sending
would
would
appear
a
bit
different,
but
but
the
endpoint
doesn't
basically
endpoint
covers
the
thing
that
should
it
should
it
do
special
stuff
for
github.com,
or
should
it
just
do
a
simple
http
request
for
importing
raw
content?
F
So
that's
something
that
the
endpoint
would
recover,
but
you
will
definitely
have
to
specify
that
hey
import
using
the
url
or
just
taking
this
particular
file
name.
So
I
I
actually
point
out
a
document
where
this
this
stuff
is
documented.
M
That
that's
what
I
wanted
to
know
is
there
anything
else.
You
want
me
to
add
any
other
changes.
M
M
So
right
now
it's
a
regex
checking
and
then
I
need
to
change
upload
import
and
then
the
curve
or
the
card
and
and
the
other
document
that
utkarsh
will
be
sharing.
I
think
I
got
the
name.
A
Yeah
but
sounds
good
meghana.
H
H
C
Yeah,
so
about
this
I
wasn't
really
sure
how
we'll
do
it,
but
I
abhishek
just
tweeted
and
just
commented
on
the
slack,
and
I
think
there
is
a
way
to
do
it,
but
is
basically
manipulating
the
smi
tool
in
a
way
that
it
won't
then
send
a
grpc
request.
H
Yeah,
I
guess
the
problem
here
is
that
this
leads
to
the
tool
not
working
with
different
kinds
of
environments
like
in
mini
cube,
and
I
guess
in
kind
as
well.
You
don't
have
load
balancers,
there's
something
notepad
and
either
the
tool
should
figure
it
out
or
whatever
we
are
using
for
deploying
stuff
so
or
we
may
get
rid
of
this.
E
It
will
never
connect
in
these
environments
like
that
load,
balancer
it'll
stay
pending
for
ip
address
perpetually
the
there's
no
use
for
the
smi
for
learn
layer
5
for
that
smi
performance
test
other
than
running
the
test
for
the
mesh
from
which
it
was
provisioned,
adding
and
so
like.
Why
does
it
need
to
additional?
E
Why
does
it
need
to
be
told
after
it's
deployed
to
actually
do
the
thing
that
it
was
deployed
to
do?
But
I
don't
know
if
there's
a
logical
argument
is
that
the
change
in
update
inside
of
each
adapter
should
be
nothing
more
than
the
name
of
the
adapter.
The
adapter
knows
what
it
is.
It's
an
adapter
for
a
specific
mesh.
It
should
just
pass
along
its
mesh
name,
we're
looking
at
like.
E
H
E
The
real
issue-
yeah,
that's
just
the
synth-
I
mean
like
the
real
issue-
is
in
mesh
kit
and
understanding
what
how
well
it's
not
necessarily
mesh.
It's
the
fact
that
we're
hard
you've
got
hard-coded
load
balancer
as
the
network
type
for
that
particular
application
for
smith
performance,
which,
in
some
environments
will
work
and
others
won't
and
and
the
having
a
more
intelligent
system
is
using
something
like
well
functionality
stemming
from
mesh
kit,
but
flowing
into
the
adapter
to
more
intelligently,
understand
like
how
that
particular
service
should
be
exposed.
E
H
A
Since
we
are
out
of
time
I'd
like
to
thank
everyone
for
joining,
but
we
I
think
we
will
go
for
a
couple
of
minutes
more
to
discuss
some
of
these
other
items
on
the
minutes.
So
if
you
are
other.
A
K
Yeah,
actually,
this
issue
is
with
the
bash
installer,
so
on
some
operating
systems,
unzip
command
is
not
present,
so
we
are
basically
packaging,
machine,
ctl
and
me
and
license
together
into
zip
package,
and
this
script
would
fail
on
systems
which
does
not
have
unzipped
command
installed
by
default.
So
we
are
planning
to
change
the
flow
to
package,
only
measuring
ctl,
and
then
there
won't
be
a
need
to
unzip
the
package
and
you
can
just
move
the
command
and
start
executing
it.
So
I
made
some
changes
in
go
related
files.
K
I
hope
you
can
open
that,
so
I
I
needed
some
help
with
how
to
test
this,
so
I
have
reached
out
to
rudraksha
and
has
also
given
some
comments
on
this.
So
if
this
particular
change
works,
I'll
also
make
the
bash
installer
script.
Changes
as
well
to
remove
any
references
to
unzip
command
are
the
mystery.js,
so
yeah.
E
E
E
If
there
are
some
sample
filters
or
sample
apps
or
sample
patterns,
that
we
will
that
we
do
and
will
ship
with
bundle
into
mesherie's
packaging
today,
that
bundling
is
built
into
the
image
the
docker
container
image
from
mesh
reservoir
or
into
the
images
of
each
measuring
adapter,
and
so,
even
as
we
go
to
expand
that
and
bundle
in
additional
things
in
the
future,
it
doesn't
seem
to
be
the
use
the
need
for
a
zipped
package
here.
So
I'm
just
kind
of
reinforcing
that
I
don't.
E
K
Yes,
actually,
while
going
through
the
go
releaser
code,
also,
there
are
so
many
things
os
specific
and
whether
to
skip
this
for
a
particular
platform
or
not
just
because
some
format
doesn't
exist
on
a
platform.
K
H
A
Okay,
so
so
ajithi
has
a
proposal
to
use
cash
to
manifest
in
measuring
ctl
for
deploying,
so
this
could
be
a
first
step
to
possibly
make
be
able
to
deploy
it
in
a
air-capped
environment.
So
she
has
made
a
proposal
for
this
and
I
think
she'll
be
working
on
it.
So
some
comments
on
this
feel
free
to
talk
to
other
tea,
about
it
all
right
piusa.
D
A
A
You
can
see
more
like
five
issues,
so
you
can
work
and
if
you
are
like
interested
in
working
in
any
of
these
issues,
you
can
either
contact
me
for
some
help
or
can
go
through
this
talk,
which
kind
of
includes
all
of
basic
info
on
how
to
start
with
writing
unit
test
for
message:
ctl,
okay
and.
A
If
you
want
to
check
on
what
are
the
commands
we
have
already
covered
or
not,
you
can
go
to
this
sheet.
That
is
necessitable
command
tracker.
We
have
a
column
here,
maybe
test
written.
We
have
mark
yes,
which
we
have
already
covered.
The
gray
part
need
to
be
covered,
so
if
you
are
interested
in
any
of
these
things,
just
right,
what
should
I
say
a
comment
in
any
of
the
any
of
the
mentioned
threads?
We
will
send
it
to
you
and
you
can
get
started
with
it.
A
Okay,
let's
move
on
to
the
rest
doc
status
we
actually
have
covered,
mostly
all
of
it.
The
only
the
only
let
me
go
through
the
file
we
recently
covered
applications
and
the
rest
we
have
to
cover
is
only
filter,
endpoints
and
then
this
whole
part
of
open
api
specification
documentation
will
be
finished.
A
E
E
C
So
for
graphql
api,
like
not
really
to
be
honest,
I
will
fix.
A
E
E
Yeah,
thank
you.
Yeah.
They
have
been
doing
work.
I've
been
advancing
on
the
interactive
labs
and
they've
been
learning
stuff
so
honey
it
might.
E
It
might
be
good
to
have
that
as
a
topic
on
the
community
call
on
friday,
for
you
to
kind
of
like
I,
you
did
a
great
job
last
time
all
telling
people
what
they're
about
inviting
people
to
participate
and
devonch
is
there.
So
that's
good,
but
I
don't
know
that
you've
gotten
a
lot
of
feedback
on
what's
working,
what's
not
working,
so
just
getting
some
more
air
time
on
the
community
called
and
highlighting
the
fact
that
those
resources
are
there.
E
K
Okay,
I
will
let
you
know
yeah.
A
Sounds
good,
alright,
everybody
thank
you
for
staying
this
long.
A
I
guess
we
will
see
you
next
time.
Bye.
C
So
like
before
going
if,
if
an
smi
tool
like
utkar,
helped
me
point
this
out,
but
if
the
adapters
are
outside
the
cluster,
how
will
you
get
the
result
from
the
smart
if
you
don't
have
a
load,
balancer
yeah,
if
you
don't
have
a
grpc
connection
which
is
through
the
load
balancer?
L
C
Yeah,
so
it
is
currently
bi-direction.
The
adapter
is
a
client,
but
flying
sends
a
request
for
the
with
the
specifications
of
each
of
the
tests
and
it
gets
the
result
back
from
the
gear.
E
And
it
is
good,
that's
good
justification
for
not
shifting.
You
know
for
continuing
to
need
the
you
know
ex
service
exposed
externally
and
if
it
continues
to
need
to
like
like
while
there
are
ways
to
circumvent
that
or
to
work
on
that
like
it's,
probably
not
it's
not
worth
the
effort
to
change
that
and
the
suggestion
to
auto.
I
think
that
you
know.
E
E
I'm
sorry
we
were
addressing
the
symptom
of
not
having
it
and
it
auto
initiates
the
test.
The
root
cause
is
like
because
the
service
isn't
being
exposed,
and
so
working
on
that,
probably
you
know,
can
be
helpful
to
other
things
that
we're
doing,
maybe
occasionally,
because
eventually,
it's
not
too
far
in
the
future.
As
a
matter
of
fact,
I'd
like
to
put
this
up
for
grabs
now,
but
we
won't
it
is
dhruv-
can
tell
everyone
about
dynamic
boards
and
static
boards.
E
These
two
constructs
inside
of
nestery
that
will
measure,
will
query
a
node
exporter,
a
prometheus
node
exporter
for
metrics
about
the
nodes,
but
unless,
if
it's
not
there,
then,
if
there
isn't
a
prometheus
node
exporter
there,
then
ashrae
can't
get
that
data
and
mastery
can't
get
that
data.
Well,
some
of
the
the
performance
test
reports
are
really
lacking
a
lot
of
context.
E
Okay,
so
then
maybe
mastery
should
auto
deploy
a
prometheus
if
one
isn't
present
and
connect
to
it
and
then
automatically
start
showing
people
a
bunch
of
nice
graphs
in
the
chart.
Okay.
So
if
mastery
is
going
to
provision
a
prometheus
outside
of
the
prometheus
synchrofaunas
that
come
with
your
service
mesh,
then
it
too
would
have
an
open
need
for
exposing
a
service
external
to
the
cluster.
So
that
logic,
you
know
essentially
is
located
in
mesh
kit-
is.
E
Server
to
deploy
its
own
instance
of
prometheus
and
its
own.
E
Because
the
instances
that
come
with
the
service
meshes
aren't
inclusive
of
all
the
info
needed
and
if
ministry
is
going
to
do
that,
it
needs
to
expose
prometheus
external
to
the
cluster,
and
so
then
that
same
logic
that
we
would
use
to
figure
out.
Should
it
be
a
load,
balancer,
node,
port
or
cluster
ip
same
logic
can
be
used
to
make
sure
that
smi
performance
is
provisioned.
C
So
about
the
load
balance,
if
I
remember
correctly,
we
changed
it
from
node
4
to
load
balancer,
because
mini
cube
wasn't
working
properly
with
node
banks.
I
don't
remember
so
I
remember
having
the
talk
with
abhishek
of
doing
such,
but
that
was
the
case.
I
don't
know
the
problem
at
that
time,
but
yeah.
E
D
E
Different
ways
to
handle
that,
like
either
check
the
environment
before
like
in
mesh
kit,
though,
is
like,
is
where
this
or
in
the
adapter
library
either
one.
It
needs
to
be
defined
and
there's
some
logic
that
says
to
check
the
environment.
Our
load
balancer
is
working
or
is
no
report.
The
right
thing
is,
or
whatever
the
approach
is,
but
but
it
has
to
dynamically
check-
and
there
is
some
of
that
already
in
mesh
kit,
where
it
will
verify
connectivity,
there's
some
amount
of
intelligence
there
that
can
be
built
upon.