►
From YouTube: Meshery Development Meeting (Sept 29th 2021)
Description
Meshery Development Meeting - September 29th 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
All
right,
I
think,
let's
get
started
welcome
to
the
measuring
development
meeting
today
is
september
29th
and
we
have
11
people
on
the
call
right
now
we
have
a
couple
of
books
lined
up
today
and
let's
start
with
the
upcoming
measuring
release
so
measuri
we
are
going
to
release
a
new
version
of
mystery
soon,
most
probably
this
by
this
week.
A
A
If
you
look
at
the
roadmap,
you
can
see
some
of
these
new
features
that
will
be
there
in
this
release.
So
making
such
a
a
big
major
release
requires
a
lot
of
testing
from
our
part.
We
need
to
ensure
we
have
ux
consistency
and
we
have
functionality
across
as
expected
across
all
these
measuring
use
cases.
A
So
we
had
gone
over
this
machery
test
plan
in
last
week's
meeting,
so
I
wanted
to
call
out
this
again,
so
there
are
still
a
couple
of
areas
that
needs
people
to
sign
up
and
test
it
out.
So
basically,
what
this
test
plan
describes
is
a
set
of
actions
that
a
user
can
take
to
interact
with
measuring.
A
So
all
of
these
are
supposed
to
work
so
and
what
a
tester
is
expected
to
do
is
test
out
these
scenarios
and
note
down
the
note
down
the
result
and
possibly
open
up
issues
if
there
is
any
functional
discrepancies
or
ux
inconsistencies.
So
if
the
ux
is
not
after
the
mark
or
it's
not
intuitive,
you
can
also
report
that
it's.
B
A
new
great
question
has
have
the
folks
that
have
been
that
are
signed
up.
Have
we
been
seeing
results
coming
back
just
yet.
A
Yeah
we
are
seeing
a
couple
of
results,
but
most
of
the
results
are
empty
now.
So
I
also
want
to
point
out
that
even
if
a
tester
has
signed
up
to
test
a
particular
component,
it
doesn't
mean
you
can't
sign
up
for
that,
but
please
feel
free
to
sign
up
if
another
person
has
already
signed
up
as
well.
A
B
Okay,
let
me
ask
you:
let
me
ask
about
a
couple
of
specific
ones.
If
I
could
on
rude
rocks,
do
you
do
we
have
either
of
the
two
github
actions
scheduled.
C
No
not
yet.
I
was
doing
a
little
work
on
this
today
and
right
now
they
are
docker
based,
so
they
would
install
mystery
on
docker
and
I
don't
think
that
we
need
that,
so
I'm
moving
them
to
kubernetes.
First
and
after
that,
yeah.
B
B
Cool,
okay,
yeah,
those
the
reason
I
bring
up
those
two,
particularly
the
smp.
The
surface
mesh
performance
github
action
is
that
is
that
it'll
cover
quite
a
few,
quite
a
bit
of
code.
So
that's
good!
That's
good
cool!
B
Just
at
red,
I
don't
know
somewhat
randomly
there's
some
tests
that
neha
neha
is
her
last
name
laha.
I
don't
think
she's
on
at
the
moment,
but
she's
been
in
the
community
for
some
time,
she'd
been
wanting
to
work
on
mesherie's
provisioning
of
prometheus.
B
B
It's
also
not
only
advantageous
to
the
to
users
potentially,
but
it's
also
advantageous
to
mesh
read
that
needs
to
collect
any
number
of
statistics
about
the
nodes
or
or
the
service
meshes
the
traffic
and
things
that
are
going
on,
and
so
it
was
a
well-placed
request
anyway.
Point
is
neha
wanted
to
go
work
on
that
request,
and
so
she
was
working
on
test
cases
here
on
line
190
through
19197
in
her
notes.
B
B
Those
those
actions
are
gonna,
be
quite
helpful,
even
if
you
just
take
the
current
docker
one
and
just
like
apply
a
schedule
to
it
and
I
kind
of
make
a
weird
face,
because
I'm
not
sure
the
github
actions
natively
support
the
concept
of
a
cron
like
the
concept
of
scheduling
they
generally
want
to
be
triggered
from
some
other
event
or
a
manual
invocation,
and
we
can
go.
B
There
are
plenty
of
events
going
on
plenty
of
activities
and
events
going
on
in
our
repositories
that
could
be
a
trigger
or
we
can
separately
use
a
different
system
to
trigger,
but
but
yeah,
if
you
would,
let
me
know
if
there's
a
native
solution
to
scheduling
great,
we
should
probably
lean
into
that.
If
not,
the
cncf
has
given
us
a
zapier
account
to
run
things,
and
so
we
might
go
create
a
cron
job
in
there
to
to
kick
off
the.
C
B
B
Just
put
you
on
put
ask
a
question
of
you
as
well
in
the
venue,
any
particular
any
particular
trouble
spots
that
you've
noticed
or
any
maybe
any
areas
that
you're
that
concern
you
more
than
the
next.
That
need
a
little
more
attention.
A
If
you
are
going
to
deploy
to
kubernetes
by
default,
then
one
of
the
areas
that
we
need
to
fix
is
the
discovering
measuring
endpoint.
So
it
is
broken
as
of
now
so
and
we
we
actually
in
measure
recital
we
actually
show
the
user
a
false
end
point.
So
we
need
to
make
it.
We
need
to
have
a
fail,
safe
or
oh.
We
need
to
fix
the.
B
So
yeah-
maybe
I
don't
know,
maybe
to
michael's
point-
I
don't
know
if
he
made
this
point
exactly
but
yeah
maybe
to
to
use
even
more
precise
language
is.
B
What
would
you
say?
It's.
B
It's
it's
it's
halfway,
functional
like
it
it'll
tell
you
it'll,
give
you
the
correct
answer.
Sometimes,
if
you
ask
slowly
enough
like
like
the
challenge
here
is
really
really
more
well,
it
can
be
around
the
logic.
That's
being
used
inside
of
mesh
kit
to
discern
what
the
actual
ultimate
external
cluster-facing
ip
address
or
reference
to
measures.
B
Endpoint
is,
but
there's
also
a
lot
of
it
has
to
do
with
timing
as
well
like,
if
you
imagine,
measuring
ctl
as
an
asynchronous
client
that
you
know
runs
over
a
a
moment
in
time
where
it's
provisioning
measuring,
like
kubernetes,
is
provisioning
these
systems
and
and
the
cert
the
deployments
and
the
services
can
take
a
little
while
to
come
up
the.
If
you
have
a
load,
balancer
integration
as
well,
that
might
take
a
little
while
to
provision
an
ip
and
expose
it,
and
so
there's
a
my
point
is
outside
of
just
asking
ourselves.
B
If
the
logic
for
identifying
an
endpoint
is
correct,
there's
also
the
troublesome
concept
of
of
timing,
and
it's
not
like
mastery
ctl
can
necessarily
sit
there
forever
and
just
waiting
for
the
cluster
ip,
a
node
port
or
a
load
balancer
to
be
exposed,
because
you
know
sometimes
those
things
those
processes
get
stuck.
So
if
it
was
supposed
to
be
a
load
balancer
in
particular,
you
can
get
into
a
pending
state
and
just
or
kubernetes
can
get
into
a
pending
state
and
just
kind
of
stay
there,
and
so
so
anyway.
B
Hopefully,
that
sheds
a
little
more
light
on
the
the
challenge
it.
It
could
be
that
you
know
there
are
ways
to.
There
are
band-aids
in
the
meantime
of
things
like.
B
You
know
having
separate
ad
hoc
routines
that
people
can
do
to
like
maybe
measure
ctl
gets
the
wrong
endpoint
and
people
try
to
connect
and
they
can't
and
then,
if
they
would,
if
we
made
it
more
a
little
more
prominent
the
fact
that
they
can
run
a
system
check
and
if
the
mastery
ctl
system
check
or
something
similar
would
re-verify.
B
B
It
could
also
be
the
case
that,
like
you,
know
like
a
poor
man's
again,
another
band-aid
here
kind
of
another.
It's
like
at
the
end
of
the
message
where
it
says
we
think
we
can
change
the
language
like
conceptually
or
supposedly
or
purportedly
endpoint.
Is
you
know
here,
if
you're
not
finding
that,
then
double
check
it
with
a
cube,
ctl
get
s
service,
namespace,
measuring
and
then
you
know
verify
like
run
this
command
to
re-verify.
B
You
know
like
there
can
be
one
of
those
commands
like.
Sometimes
this
takes
a
while
rerun
this
command
to
verify.
B
B
So
so
maybe
after
he's
done
with
his
english
exam,
he
can
come
back
and
rewrite
what
the
helm
chart
says
at
the
end.
But
but
the
point
is
like
when
we're
done
through
helm,
chart
installation.
We
also
try
to
tell
the
user
here's
how
here's
the
instructions
from
which
you
can
go
discern
your
endpoint,
so
yeah.
B
Not
that
anyone,
it's
very
evident,
like
from
all
from
a
lot
of
the
users
that
mastery
has
no
one
reads
anything
which
is
like
probably
true
of
all
of
us
like
who
is
ever,
who
has
ever
reached
out
to
the
docs
for
anything,
which
is
why
I
hope
that
doesn't
dismay
a
lot
of
people
when
I
say
that.
B
B
Okay,
so
yeah
gosh,
I
guess
with
all
that.
It's
not
yeah.
B
B
B
A
All
right,
thank
you
liam.
Before
we
move
on
to
the
next
topic,
I
think
we
have
a
couple
of
newcomers
on
the
call
today.
So
I
think
aditya,
like
is
this
your
first
time
on
your
call,
discord,
aditya.
D
D
All
right,
okay,
so
I'm
aditya
and
subheri,
I'm
studying
as
a
third
year
undergraduate
computer
engineering
in
india.
I
originally
live
in
nepal
and
that's.
A
I
think
you
are
new
here.
Would
you
also
like
to
introduce
yourself.
A
Thank
you,
vedanta
welcome
to
the
community.
Is
there
anyone
else
who
is
here
for
the
first
time.
A
All
right,
then,
welcome.
Welcome
to
all
the.
A
F
F
So
yeah
there
we
go
so
this
particular
spreadsheet
has
all
the
components
that
we
have
to
create
in
order
for
us
to
refactor
the
whole
code
base,
and
actually
this
is
just
the
I
mean
these
are
for
now
like
these
are
the
global
components
that
are
that
we
are
going
to
be
reusing
all
over
ui
and
this
list
will
keep
on
adding
up
but
yeah.
This
is
a
start.
F
The
plan
is
to
first
complete
these
things
because,
like
all
the
other
components
that
we
are
going
to
create
will
be,
it
will
be
created
based
on
these
components,
so
we'll
be
using
these
components
to
create
as
a
component.
So
the
plan
is
to
create
these
things
first
and
then
move
on
to
the
more
complex
components.
F
So
there
is
an
open
pr
right
now
for
yeah
people
are
actually
working
on
it.
You
can
see,
and
if
you
guys
are
interested,
you
can
actually
take
these
up
and
start
working
on
them
and
create
a
pr
like
all
the
details
that
you
need
in
order
to,
and
basically
the
contribution
right
guide
to.
This
particular
thing
is
in
the
readme
file
in
the
new
mystery
ui
restructuring
branch.
I'll
just
show
it
right
now.
F
So
yeah
this
particular
readme
file
has
instructions
that
one
has
to
follow
for
yeah.
Particularly
this
section
has
all
the
instruction
that
one
has
to
follow
in
order
for
him
to
contribute
to
restructuring
so
yeah.
If
anyone
is
interested,
then
they
can
go
through
this
and
they
can
start
contributing
right
away.
Yeah
we'll
be
reviewing
the
prs
strictly
and
we'll
make
sure
that
the
code
that
we're
going
to
write
this
in
good
quality
and
all
those
stuff
so.
F
G
H
A
G
This,
like
these
four
kind
of
problems,
so
could
you
give
a
little
bit
introduction
to
how,
like
you,
can
go
through
this
and
create
component
for
it
an
example.
F
Yeah
sure
I'll,
okay,
actually
there
are
some.
I
have.
There
are
some
components
that
are
already
created,
which
can
be
an
inspiration,
so
one
of
them
is
okay,
actually,
basically
like.
We
do
not
have
a
lot
of
components
right
now,
and
these
are
the
ones
that
are
pretty
basic,
but
it
should
give
you
an
idea
as
to
how
we
should
start
writing
components.
So
if
you
take.
F
Yeah,
if
you
take
a
look
at
this
chip
component,
it's
pretty
basic.
It
doesn't
do
anything
else
other
than
like
this,
like
usually
like,
when
we
have
a
usually
like
for
all
the
chip
components
that
we're
going
to
use
usually
need
to
change.
It
change
the
cursor
to
the
pointer
and,
you
know,
add
some
padding
and
we
are
actually
exporting.
I
mean
doing
those
changes
over
here
and
exporting
the
component,
but
this
is
very
basic
in
the
sense
that
we
are
not
expanding
and
we
are
not
making
it.
F
We
are
not
doing
many
changes.
We
are
not
doing
changes
on
top
of
this
thing.
But
when
you
look
at
this
paper
component,
we
are
actually
doing
some
changes
in
the
sense
that
this
custom
paper
we
are
actually
just
making
this
making
the
signing
changes
that
will
be
that
will
be
using
all
over
the
ui
in
this
inside
will
be.
We
are
adding
padding
so
that,
like
whenever
you
use
the
custom,
we
use
the
paper.
F
Then
it
is
like
we
have
the
consistent
padding
all
over
the
ui
and
actually,
if
you
look
at
this
particular
paper
with
title
component,
it
it.
We
are
actually
doing
some
things
on
top
of
custom
paper
since
that
okay,
it
would
be
better
if
I
can
show
it
just.
I
F
So
you
can
see
that
there,
this
component,
that
we're
seeing
over
here
all
these
things.
These
are.
These
are
just
paper
components,
but
you
can
actually
see
that
there
is
a
pattern
to
all
of
these
things
in
the
sense
that
we
have
a
heading
and
we
have
the
content
and
actually
have
some
padding
over
here.
So
basically
we
can.
Actually.
This
is
this.
This
one
will
be
a
common
thing
that
we'll
be
using
everywhere,
even
like
in
the
outer
paper.
F
You
can
see
that
we
have
a
heading
and
we
have
a
discontent.
So
if
you
look
at
this
code,
we
have
actually
separated
out
that
logic,
and
it
is
now
like
you
can
actually
like
this
component
can
be
named
something
else
or
whatever
but
paper
with
it.
We
are
naming
it
paper
with
title
and
you're
actually
yeah.
You
can
pass
in
the
title
and
that
would
be
consistent,
consistent
in
the
sense
that
the
all
the
stylings
that
we
are
going
to
play
will
be
same
for
all
the
people.
F
If
you
are
using
the
same
component
so
yeah
this
is
a.
This
would
be
a
good
inspiration.
I
guess
you
can
take
a
look
at
this
thing
and
actually
we
are
doing
some
other
things
so
that
it
is
more
flexible
in
the
sense
that
we
can.
We
are
not
restricted
to
using
only
text
for
the
title,
note
and
yeah.
We
are
actually
making
it
flexible
by
allowing
us
to
have
allowing
us
to
use
a
custom
components
as
well.
So.
F
Yeah
and
actually
these
three
components
are
yeah.
These
are
not
very
basic
because,
like
these
these
this,
I
don't
think
this
would
actually
belong
here,
because
considering
that
this
components,
folder
is
something
that
we
are
going
to
use.
We
globally
use
everywhere,
because
a
navbar
is
something
that
we're
going
to
use
only
once
in
the
whole
ui.
So
this
might
not
seem
very
what
to
say.
It
might
not
seem
very
intuitive
in
the
first
at
the
first
time.
F
Basically,
we
are.
We
are
actually
following
a
pattern,
so
it
would
be
better
if
you
could
easily
follow
the
same
thing
and.
J
B
Good
good
yeah,
please
do
read
through
the
readme
yeah
good
good
next,
actually,
so
so
to
jump
back.
So
I'm
sorry
interrupt
folks,
but
we
need
to
cut
the
cut
and
move
to
to
one
of
the
top
priorities
is
trying
to
get
the
release
out
for
v060
and
neha.
Your
ears
are
burning.
We
were
just
talking
about
you
a
moment
ago
now
this
could
be
toward
the
end
of
the
call
if
you'd
like.
Maybe
we
could
just
do
a
quick
troubleshoot
of
your
environment
to
yes,
okay
cool.
K
Just
let
me
just
let
me
log
into
my
setup
I'll
share
soon
cool
sounds
good.
A
F
Yeah,
so
the
component
registration
is
currently
in
a
working
state
if
you,
but
it's
not
in
a
working
state
in
the
current
release,
because
of
one
reason
that
the
the
q
open
api
relations
schema
binary.
That
is
required
for
generating
components.
F
Kubernetes
components
is
that
that
was,
I
guess,
0.1.0
version
that
has
been
seeded,
but
we
need
the
0.1.2
and
I
guess
rudrak
has
created
a
pr
for
it
and
I
don't
know
if
it's
merged
or
not
so
after
that
gets
the
error
that
we
were
having
in
the
latest
release
that
would
go
away
and
now
coming
to
the
second
thing.
So
I
was
going
to
demo
it
today,
but
unfortunately
I
am
not
able
to.
F
There
are
a
lot
of
errors
that
I
have
already
that
I
have
already
discussed
with
the
nitish
on
slack
that
I
am
facing,
while
you
by
basically
while
using
the
while
trying
to
use
those
components
in
the
mesh
map.
F
My
mesh
map
keeps
breaking
because,
when
the
components
get
registered
and
nithesh
is
working
on
it
trying
to
figure
out,
what's
what
can
be
done
for
that
and
if
he
can
tell
me
the
structure,
if
he's
not
able
to
make
a
lot
of
tweaks
in
the
ui,
I
would
I
have
actually
a
callback
which
can
which
can
restructure
the
definitions
and
schemas
the
way
we
want
to.
So
if
he
can't
do
it
at
the
ui
part
I'll,
do
it
at
the
back
end.
So
after
we
we
get
it
done.
B
No,
I
was
just
gonna.
No,
I
would
yeah
that's
okay,
yeah,
it's
too
bad
when
we
missed
the
deadline.
Yeah,
please!
Please
do
work
through
that
as
quick
as
you
can.
F
No,
actually,
I
just
know
about
the
json
file
and
I'm
actually
working
on
it.
Yeah
I'll
try
to
figure
out.
If
I
can,
whatever
I
can
do
without
uae.
F
L
No,
but
they
are
interesting,
they
are
not
interrelated,
so
the
same
problem
will
not
cause
any
issues
with
the
server,
but
if
the,
if
for
any
reason,
the
structure
of
if
the
structure
of
the
schema
is
not
valid
in
that
case
definitely
even
though
even
the
server
would
refuse
to
apply
that
particular
pattern,
so
it
would
depend
that
what
the
exact
cause
of
the
issue
is.
Is
it?
Why,
or
is
it
something
wrong
with
the
structure
of
the
scheme.
F
Yeah
that
I
believe
I
am
yet
to
test
out,
because
from
what
I
understood
from
the
error
code,
that
error
that
was
showing
up
for
some
reason,
the
it
was
not
able
to
replace
the
like.
We
are
using
a
regex
to
replace
the
names
so
that
it
is
more
formatted
for
for
the
user
to
see
right.
F
So
that's
where
it
is
breaking
the
sense
that
it's
not,
it
is
not
able
to
replace
that
particular
string,
because
of
which
is,
it
is
returning
and
if
it
is
not
returning-
and
it
is
not
returning
a
string
which
which
breaks
like
execution
there,
so
yeah
I'll
have
to
look
into
it.
Why
it
is
happening.
A
All
right,
thanks
ashish,
for
the
update
next
up,
we
have
an
update
on
the
multiple
import
options
in
the
patterns.
Ui.
Would
you
like
to.
J
Yeah
so
yeah.
Currently,
I
won't
be
able
to
demo
it
because
I'm
logged
in
with
windows
but
yeah.
Definitely
I
will
speak
about
the
update
that
I
demo
day
before
yesterday,
on
the
awesome
call
so
previously
what
I
did
for
the
multiple
import
was
a
model
over
a
model,
so
that
was
not
at
all
an
intuitive
approach,
so
what
I
did
was
having
a
drop
down
menu
with
showing
multiple
import
options
like
upload
from
local
and
browser
plus
github,
upload
and
url
upload.
So
for
that,
regarding
the
github
upload,
I
had.
J
For
the
github
upload,
I
had
a
talk
with
no,
so
he
explained
me
all
the
concepts
for
having
the
ui
and
all
the
test
cases.
A
Okay,
yep-
and
I
think,
like
meghna,
was
working
on
fixing
some
bugs
around
that
part,
and
I
think
she
gave
an
update
on
the
last
called
megan
like.
Could
you
just
mention
that,
on
the
midi
minute,
sir.
M
Yeah
so
so
I
haven't
changed
much
when
it
comes
to
the
component,
like
this,
like
the
recent
recent
pr
that
I
opened,
that
that
shouldn't
change
anything
for
gaurav
actually
so
the
way
it
was
before
he
started
it's
just
like
that.
Only
so
it
shouldn't
be
a
problem
for
him,
but
let
me
just
share
two:
pr:
testing
kills.
B
I
wasn't
able
to
join
mom
monday's
meeting
on
wassum
filter,
so
I
don't
mean
to
make
people
like
re-demo
the
same
thing
that
they
might
have
shown
earlier.
I
just
for
my
own
edification
clarification.
I
wasn't
sure
if
so
the
I
wasn't
sure
how
much
of
the
effort
is
aligned
and
potentially-
and
I
don't
think
it's
overlapping,
but
to
just
re-articulate
part
of
the
goal
here
for
those
that
have
seen
the
new
types
of
content
patterns,
filters
and
applications
in
measuring
in
the
ui
people.
B
Have
you
know
users
have
the
ability
to
upload
and
megana
and
gorov
primarily
have
been
iterating
on
the
various
ways
in
which
people
can
upload
and
manage
their
that
type
of
content,
manage
patterns,
filters
and
applications,
and-
and
I
think
that
we've
congealed
on
a
design
that
more
or
less
on
a
design.
I
thought
that
gorov
had
come
forth
with,
which
was
you
know.
B
You
click
this,
this
import
button
and
you
get
a
modal
that
gives
you
kind
of
four
choices
for
the
various
ways
in
which
you
can
import
a
piece
of
content,
and
I
guess
my
question
is
is
just
is
that
is
just
to
confirm
that
that's
still
the
current
goal,
and
if
it
is
it's,
I
mean
that
component
or
that
thing
that
we
should
be
hopefully
making
as
a
component
is
set
to
be
reused
about
four
times
in
in
the
patterns.
B
B
L
I
can
demo
a
bit
and
probably
be
able
to
also
tell
how
others
can
also
sort
of
replicate
what
I
have,
what
the
setup
that
I
have
for
testing.
So
let
me
show
the
screen.
L
So
this
is
my
machine
setup.
It's
looking
quite
exactly
how
you
would
see
it
when,
when
you
would
deploy
a
mystery.
What
I'm
doing
here
is,
I
have
run
mystery
system
start
hyphen,
p.
Q
leds.
This
is
the
command
that
I
run
this
spin-off
mystery
in
my
kind
cluster.
So
these
are
the
components
that
are
running.
These
are
fewer
than
what
you
would
expect,
because
I
have
my
context
customized
for
me
so
yeah.
Basically,
this
is
how
I
set
up
my
mystery.
L
After
setting
this
up,
I
did
edit
and
then
I
edited
my
deployment
to
change
the
image
name.
It's
a
temporary
change.
This
is
just
for
testing
out
the
multi-cluster
change,
because
it's
a
huge
change,
it's
something
it
has
a
very
huge
surface
area
for
potential
bugs
and
there
would
be
definitely
bugs.
So.
L
That
is
why
there
is
a
temporary
deployment
of
machine
on
this
particular
on
this
particular
github.
The
username
that
is
23,
slash,
mystery,
cool
and
stable
hyphen
latest.
This
is
a
version
of
measuring
which
has
the
changes
associated
with
multi-context
and
it's
using
staging
mystery
crowd,
staging
environment
and
a
few
other
things
that
would
be
relevant
to
get
this
thing
working.
I
have
mentioned
all
of
these
things
in
in
slack
mystery
channel.
L
L
This
error
that
we
are
getting
is
because
I'm
running
the
latest
version
of
kubernetes
in
in
my
mini
cluster,
which
is
my
current
context,
and
it
doesn't,
it
has
promoted
its
custom
resource
definition
resource
from
v,
one
v
to
one
to
I
guess,
v
one.
So
that
is
why
this
error
is
showing
up.
This
is
not
something
associated
with
mission.
I'm
running
mini
cube
cluster.
This
is
my
actually
I'm
running
mini
q.
L
Well,
as
thruster
kind
tester
is
mapping
cluster,
and
that
is
this
is
where
mystery
is
actually
deployed.
While
mini
cube
is
my
secondary
cluster
right
now,
I'm
connected
with
the
mini
queue,
so
let's
say
I
want
to
perform
any
operation.
The
reason
I'm
actually
going
to
deploy
I'm
going
to
apply
a
custom
gun
is
because
it
is,
as
I
was
mentioning
that
kubernetes
version
has
changed.
L
So
there
is
a
custom
resource
definition,
changes
so
actually
right
now
seo
deployment
and
other
services
deployments
are
sort
of
sort
of
not
going
exactly
that
they
are
supposed
to.
So
that's
just
for
the
purpose
of
demo.
Let's
try
to
find
something,
a
very
simple
application.
This
is
nothing
but
nginx.
I'm
going
to
apply
this
and
the
manifest
has
been
applied.
L
It's
deleted
from
my
test
namespace
and.
L
Demoing,
muddy
context
is
sort
of
difficult.
It's
a
tricky,
because
I
have
to
switch
between
multiple
contexts
all
the
time
to
show
how
the
state
of
the
cluster
is.
I'm
applying
the
context,
I'm
applying
the
manifest
I'm
connected
with
mini
cube
cluster.
So
what
I
would
expect
after
this
operation
is,
I
will
have
two
replicas
of
engine
x
running
on
my
cluster
on
my
mini
q
cluster,
and
it
is
it
communicated
with
the
mini
key
cluster.
L
Now,
if
I
would
go
ahead
and
change
the
change,
my
active
context
from
mini
cube
to
interester-
and
I
would
go
back
to
let's
say
seo
and
try
to
do
the
exact
same
thing
that
I
did
in
the
in
the
mini
key
cluster
in
my
client
cluster.
I
would
expect
that
I
have
test
that
is
two
nginx
deployments,
two
nginx
refrigerators
running
in
my
pine
cluster.
L
I'm
deleting
the
problem
because
it
was
already
present
in
my
cluster,
so
it
was
not
much
of
a
demo
because
it
it
was
already
there,
but
now
what
I
did
was
deleted,
the
two
replicas
and
it
did
quite
deleted.
So
basically,
the
update
is
about
that.
How
misery
can
now
support
multiple
contexts?
At
the
same
time?
It's
not
so
in
terms
of
this
particular
functionality
has
been
mostly
added
on
the
server
side.
That
is,
our
server
can
support
any
number
of
context.
L
The
ui,
as
you
may
have
noticed
that,
right
now
we
don't
have
any
buttons
here,
let's
say
which
would
say
that:
okay,
now
right
now,
you're
performing
operations,
a
mini
cube
or
you're
performing
operations
and
fine
cluster,
but
this
is
something
that
would
be
added
very
soon,
the
current
pr
or
the
current
vr
that
I've
created.
That
is
mostly
about
adding
adding
multi-bonding
support
on
the
server
side,
and
this
is
what
the
demo
was
mostly
about
the
reason.
L
The
another
reason
I
did
the
interesting
one
is
because,
as
navidoo
initially
mentioned,
that
now
we
are
moving
more
towards
humanities
as
a
default
platform.
So
this
is
mostly
the
environment
that
you
would
expect
mystery
to
be
running
in
in
the
future,
and
this
is
how
probably
he
would
be
able
to
interact
with
you
can
actually
bring
your
clusters.
Multiple
questions
from
the
ui
and
other
changes
would
be
present
in
the
ui
in
the
future.
So
yeah,
that's
pretty
much
it
from
my
side.
B
I'll
put
a
challenge
out
there,
I'll
say
I'll
say
this:
wood
karsh
is
good,
but
he's
not
that
good,
or
rather
there's
got
a
there's
a
bug
in
here
somewhere
I
can
you,
I
can
smell
it
from
here.
Like
there's
a
hey,
we've
got
a
couple
of.
We
haven't
gotten
to
chat
through
some
of
the
item.
You
know
what
we
have
like
15.
mandy.
What
else
do
we
have
on
the
agenda?
B
L
Actually
so
now
I
it's
it's
not
very,
as
you
may
have
noticed,
it's
not
as
interactive
as
the
rest
of
the
mysteries
ui
at
the
moment,
because
it's
not
using
subscriptions
not
so
now.
I
change
the
context
to
minicube,
so
you
can
see
it
says.
Meshover
is
not
deployed.
The
deployment
is
not
succeeding
because
of
the
error
that
that
popped
up,
but
hey
customer
sales
definition
and
yeah.
L
So
it
is
reporting
the
accurate
status
of
mesh
sync
industry
operator
as
per
your
context,
but
as
I
mentioned
with
some
caveats
that
it's
it's,
it's
not
getting
required
at
the
moment
in
the
in
latest
1.2.2.
A
B
Oh,
I
see
okay
yeah
just
a
couple
of
quick
like
clarifying
questions,
things
that
I'm
that
while
I
enjoy
razzing
utkarsh
and
anyone
else
who
lets
me
razz
them
that
I
these
are
things
that
I'm
that
I
know
these
already
considered,
and
so
the
I
was
just
had
just
circled
darren,
lau's
name
on
a
piece
of
paper
so
darren,
as
some
of
you
all
know,
darren
and
some
of
you
don't
he's
a
recent
community
member,
recently
joined
and
is
just
he
lives
in
new
york.
B
He
works
at
cisco,
he's
relatively
younger
individual
bright
guy.
A
nice
guy
seemed
to
make
an
impact
quickly
and
he
was
working
on
the
message
operator
and
the
helm
charts
from
sri
operator
and
consequently
it
was
touching
mesh
sync
as
well,
which
is
primarily
in
terms
of
its
deployment,
and
so
this
is
a
some
multi-cluster
support.
B
B
B
I
chuckled
because
at
some
point
we
should
have
what's
a
server,
a
leader.
What
are
the?
What
are
the
pc
terms,
leader.
B
A
leader
worker
type
setup
such
that
it's
an
configuration
so
like
technically
maybe
there
would
be
multiple
operators,
but
they
would
be
operating
as
the
same
logical,
operating
logical,
like
brain,
logical
unit
anyway,
anyway,
I
think
it's
follower
leader
follower.
Oh
thank
you,
oh
beautiful.
B
H
B
Sorry,
I
lost
my
train
of
thought,
which
is
there's
a
couple
of
items
as
a
small
item
with
respect
to
mesh
sync's
version.
Mesh
sync
is
actually
on,
like
version
1.22
or
something
like
that,
something
higher
is
not
being
accurately
reflected
here,
which
means,
I
don't
think
it's
being
actively
built
into
the
build
of
the
thing.
B
There's
an
open
issue,
there's
also
yeah.
I
mean
to
the
extent
that
there's
multiple
chips
for
individual
kubernetes
clusters-
that's
fantastic,
they're.
You
know
at
some
point
you
would
expect
it
might
be
well,
there
might
be
multiple
chips
for
operators
and
the
reality
is
or
likely.
My
my
reality
is
that
that
we
probably
ultimately
wouldn't
have
multiple
chips
for
a
couple
of
reasons.
B
One
is
because
it
some
of
the
users
that
are
going
to
put
mastery
to
the
test
they're
going
to
deploy
measuring
in
in
an
80
cluster
environment
or
or
a
400
cluster
environment.
Right,
that's
not
uncommon.
B
Over
time,
the
trend
has
been
smaller
and
smaller
clusters,
smaller
blast
radius
matter
of
fact,
measuring
is
extremely
well
oriented
to
be
your
local
dev,
not
cluster
manager,
but
service,
mesh
manager
for
dev,
facilitating,
dev
and
prod,
and
things
so
point
is
from
a
ux
perspective,
not
just
ux,
but
just
a
usability
perspective
that
you
would.
We
wouldn't
be
able
to
line
up
so
many
chips
in
a
row.
It's
not
just
about
fitting
on
the
screen.
That's
one
issue,
but
the
other
issue.
B
It
becomes
much
more
apparent
when
you
go
to
consider
this,
that,
like
okay,
so
if
we're
going
to
represent,
let's
just
say
four
connections
to
clusters-
that
we
would
also
then
represent.
Four
individual
operators
with
you
know
mesh
sync
and
that's,
and
it
becomes
users
won't
want
one
control.
One
component
on
this
screen
needs
to
control
the
others.
B
So
when
you
switch
to-
and
this
isn't
like
this-
it
really
isn't
feedback
aimed
at
per
se
like
he
did
a
massively
sweeping
change
both
on
the
back
end
and
then
starting
on
the
front
end,
and
so
this
is
just
the
first
time
we're
sort
of
talking
aloud.
I
think
about
how
some
of
how
this
feature
would
have
sweeping
ramifications
across
the
user
interface
as
well.
B
Is
that
they're,
they're
likely
or
again
from
my
perspective,
there
becomes
the
need
to
just
like
when
you're,
using
measuring
ctl
anytime
you're
running
a
command
you're
implicitly
executing
that
command
in
context
of
a
measuring
deployment.
You're
always
running
those
commands
in
a
particular
context.
B
You
don't
have
to
specify
the
context,
because
it's
implicit
like
it's
your
current
context
and
that
functions
very
similar
to
how
cube
ctl
functions
and
how
a
lot
of
other
software
functions
is
like
you're
you're
doing
some
things
and
you're
doing
it
in
under
some
implicit
context,
and
so
right
now
the
content,
the
the
the
literal
kubernetes
cluster,
that's
being
used,
is
called
in
cluster,
and
so,
as
a
user
were
to
navigate
to
other
screens,
they
would
execute
certain
things.
They
would
be
doing
it
in
con
in
context
of
that
content.
B
So,
like
a
quick
fix
to
if
someone
connects
to
10
clusters,
like
a
quick
fix
to
here,
is
to
have
like
a
drop
down
that
they
can
select
from
which
cluster
they
want
to
interact
with
now,
eventually
like
that
drop
down
at
their
400
clusters,
like
you're,
looking
at
a
drop
down,
that's
like
you
scroll
forever.
It's
not
the
best
control
either
fine.
B
You
can
have
a
drop
down
with
a
search
at
the
top
and
all
that,
but
anyway,
point
is
once
that
happens
when
someone
selects
the
current
context
to
interact
with
that,
can
update
the
other
components
on
the
screen
with
information
about
the
operator.
That's
you
know
respective
to
that
cluster,
that
same
kind
of
a
multi-check
box
like
drop
down
if
it
has
check
boxes
as
well,
can
work
well
for
other
screens
in
which
maybe
they're
connected
to
10
clusters.
B
B
B
So
anyway,
this
new
feature
becomes
a
pervasive
consideration
across
an
implicit
consideration,
one
that
people
didn't
have
to
consider
and
the
ui
didn't
have
to
consider
in
the
past,
because
it
was
just
a
single
connection,
but
now
it
becomes
something
that
needs
to
be
surfaced,
so
would
kosher
novendo.
If
you
navigate
to
a
life
cycle
to
any
one
of
the
service
missions,
we
would
probably
expect
again
like
if
we
have
this
pervasive
and
ubiquitous.
B
I
guess
those
are
about
the
same
thing,
but
if
we
have
this
component
context
selector,
we
have
this
confirmation.
That's
ever
present.
That
confirms
this
is
the
context
in
which
you're
operating
currently
and
if
you
click
a
single
click
button,
it's
going
to
happen
in
this
context
is
like
those
that's
the
kind
of
thing
that,
like
there
are
some
uxers
on
the
phone
that
are
on
the
call
of
it.
L
Could
maybe
help
you
so
actually
a
few
things
that
I
meant
forgot
to
mention
because
I
was
demanding
so
a
few
features
have
been
also
added,
not
just
adding
support
for
medical
context.
The
features
are
like.
Let's
say
I
have
this
mystery
instance,
and
I
I
let
this
be
connected
to
the
public
internet.
L
If
someone
gets
access
to
this
and
if
they
can
log
in
so
some
authentication
happens
after
that,
you
would
be
able
to
access
the
context
associated
with
this
particular
mystery
instance,
so
you
would
be
able
to
communicate
with
the
cluster
if
you
are,
if
you
are
part
of
the
same
mystery
instance
the
provided,
so
this
won't
work,
that
is,
this
is
connected
to
public
internet
and
you
are
logged
in
with
none
provider
and
so
on.
So
that
won't
work
because
of
authentication
reason,
but
definitely
context
sharing
is
the
thing.
L
Second,
is
that
second
feature
that
has
been
added
is:
if
two
users
are
connected
or
any
number
of
users
are
connected
to
the
same
measure
instance.
All
of
them
can
have
different
context
and
mystery
would
take
care
of
that.
So
let's
say
I
am
using
many
key
quantities,
but
you
want
to
perform
some
action
on
in
cluster
context
or
the
kind
cluster
you
can't
do
that
the
machine
will
take
care
of
isolating
all
of
the
operations.
L
Another
thing
that
has
been
added
is
that
let's
say
you
because
measuring
will
be
supporting
scheduling,
performance
states
and
which
is
a
very
long
running
task.
That
is,
let's
say
you
do
it
on
every
fortnight
or
something
so
what
you
can
do
is
you
can
schedule
that
thing
using
cluster
context
and
that's
fine
once
it's
scheduled
it's
it
will
it's.
Its
execution
is
going
to
be
completely
isolated.
If,
let's
say
in
future,
you
change
your
context
to
mini
cube
or
context
c
context.
L
It
doesn't
matter
because
that
particular
execution
has
been
it's
sort
of
isolated
from
the
rest
of
the
handlers
that
will
be
triggered
later.
This
is
something
that
that
was
sort
of
possible
on
you,
but
it's
now
far
more
granular
than
it
earlier
was
so
every
operation
is
very
isolated
after
after
this
pr
gets
smudged.
So
these
are
some
more
features
of
this
particular
pr.
B
Cool,
so
we're
about
a
time
there
was
a
couple
there's
some
other
items
on
the
agenda
that
I
wanted
to,
one
of
which
I
wanted
to
hit
very
quickly
with
a
couple
of
the
maintainers
like
michael
and
the
rocks
and
there's
a
collection
of
people
here,
so
luke
joggery
he's
a
friend
of
the
community,
a
friend
of
the
project
luke
had
and
worked
on
an
alternative
method
of
installing
meshery.
B
I
think
I
didn't
really
read.
I
got
to
be
honest.
I
didn't
read
what
he
had
said.
I,
my
guess
is
that
he
worked
on
this
because
he
because
mesherie
is
a
bring
your
own
kubernetes
cluster
tool,
and
I
think
what
he's
presenting
here
is
a
set
of
a
multi-pass
or
a
set
of
scripts.
That
brings
a
kubernete
brings
a
vm,
brings
the
kubernetes
cluster
and
brings
measuring
for
you.
So
the
assuming
that
that's
accurate.
The
question
for
you
all
is:
do
you
think
that
that's
a
good
idea?
B
B
B
Well,
well,
we
don't,
because
the
project
right
now
doesn't
really
have
this.
This
isn't
the
first
time
that
someone
has
brought
this
kind
of
a
thing.
The
last
time
it
was
he
was.
It
was
a
nice
individual
from
russia,
his
name
I
forget,
but
he
had
done
it
using
ansible
and
some
other
things
and
because
he
he
too
wanted
like
a
vm
environment
to
run
meshri
and
run
surface
mesh,
and
at
the
time
we
extended
the
invitation
to
him
to
put
those
scripts
into
the
install
folder
in
the
mesherie.
B
Repo
mesherie,
as
a
project
has,
is
very,
is
extensible
and
has
the
concept
of
third-party
plugins,
which
is
good,
but
but
we
don't,
as
a
project,
have
the
concept
of
gosh.
What
is
it.
B
Boy
words
family
today,
as
like.
B
Related,
but
more
or
less
not
directly
supported
by
the
project,
an
ecosystem
thing
like
what
do
you
call
it
a
cap
or
like
a.
B
B
Yes,
it's
sort
of
a
quick
assessment
like
why
gosh
can
we
sustain
that
and
it's
luke
intending
to
hang
around
to
help
people
as
they
bump
into
inevitable
bugs
and
that
kind
of
thing
and
and
so
from
my
perspective,
just
to
lead
the
question,
I
guess
is
to
say
like
yeah
yeah,
it
is
to
embrace
it
and
say
yes,
this
is
great
to
to
at
minimum
link
to
it
and
and
like
tell
people,
this
is
an
alternative.
You
know
approach
but
maybe
caveat
it
in
such
a
way
that
it's
like.
M
B
It
is
yet
to
graduate
it's
like
a
it's
an
idea
and,
and
we
embrace
it,
but
it's
yet
to
be
fully
incorporated
into
the
test
plan
and,
like
part,
you
know
first
class
in
the
project
or
to
find
just
like,
let's
go
ahead
and
make
it
first
class
and
just
say
hey.
This
is
another
approach
and
listed
as
under
the
getting
started
and
then
come
what
we
may
in
terms
of
bugs.
H
I
think,
from
that
point
of
view,
it
really
makes
sense
to
have
sort
of
like
an
easy
packaged,
all-in-one
learning,
solution
or
approach
so
that
you
don't
have
to
worry
about.
You
know
installing
some
sort
of
kubernetes
distribution,
oakley
and
and
struggle
with
the
end
points
and
things
that
we
struggle
a
little
bit
with.
So
from
that
point
of
view,
I
think
it's
it
would
be
good
and.
H
H
Like
have
a
you
know,
often,
if
I
remember
correctly
it's
you
know,
I
have
like
building
around
it
or
additional
tooling,
and
then
you
have
sort
of
like
unsupported
or
unofficial
or
section
where
tools
like
this
could
be
listed
and
and
that
sort
of
carries
a
certain
weight
because
we
say
yeah.
This
is
you
know
this
is
good
and
we
embrace
it,
but
we
don't
have
to
the
bandwidth
still
sort
of
support
it
directly.
B
Yeah
and
what
yeah
that's
a
occurred
to
me,
michael,
but
but
like
part
of
the
challenge
that
we
have
with
exercising
all
the
test
cases
is,
was
like
well
in
the
github
workflows
that
we
execute.
We
need
to
spin
up
a
via
spin
up
a
kubernetes
cluster,
deploy,
measuring,
deploy
and
run
through
exercise
its
code.
H
B
I
know
that
part
of
luke's
thinking
we've
had
a
com
he'd
been
really
interested
in
the
learning
paths
that
the
community
has
been
working
on
that
have
yet
to
be
officially
released.
But
I
dropped
a
link
to
it
in
the
chat.
J
B
I
think
he
was
he's
all
his
thought.
Process
was
sort
of
in
this
same
learning
vein.
So
yeah,
okay,
anybody
else,
any
other
comments,
michael
or
does
anybody
else
have.
B
A
Yep
we
have
one
more
topic
to
discuss
that
needs
discussion,
so
additional
european
polar
cast,
labor
leveler
needs
some
discussion.
I
guess
like
if
rudraksha
could
talk.
Oh
yeah.
D
Yeah-
and
he
said
we
need
to
talk
about
it
so
yeah
the
I
had
previously
made
a
full
request
on
enabling
this
pull
request
for
the
release
crafter
action
right,
but
the
problem
was
the
the
github
token
we
were
using
so
that
secret
personal
access
token
was
not
enabling
that
the
labels
to
appear
correctly
so
that
was
preventing
it
and
the
documentation
suggested
that
we
use
a
different
token.
But
when
I
suggested
that
he
said
that
it
may
you
know
affect
the
release,
release
draft
part
of
the
tower
right.
D
It
may
affect
the
release
right
because
he
shared
me
a
talk
which
said
that
that
certain
permissions
were
not
adequately
given
by
the
github
actions.
Token,
so
you
could
not
change
it.
Thus,
I
created
a
new
workflow
for
the
same.
B
So
I'm
just
reading
through
rudraksha's
notes,
but
a
pat
can
be
responding
to
his
comments
like
that
makes
sense.
A
pat
can
be,
and
we
have
in
the
past
pasted
a
pat
into
a
secret.
Then
secret
is
available
to
workflows
and
then
therefore,
so
is
the
path.
G
B
I
just
wasn't:
I
just
wasn't
sure
like.
If
we
do,
I
mean
I
don't
know
like
the
go.
The
goals
here
are
that
we
want
to
have
auto,
auto
categorize
release,
notes
that
take
care
of
cleaning
themselves
up,
which
we
generally
do,
and
that's
done,
based
on
labels
that
that
contributions
are
categorized
based
on
labels.
B
Okay,
if
people
don't
label
prs
which
most
of
us
don't,
then
then
the
release
notes,
don't
they're.
Just
everything
is
just
in
miscellaneous
in
the
really
sense.
B
B
It
documented,
like
in
a
bunch
of
a
few
different
places
about
the
fact
that
you
ultimately
need
a
personal
access
token
to
overcome
the
security,
the
the
way
that
github
has
the
workflows
working,
and
so
it's
not
uncommon
for
projects
to
use
projects
or
just
any
software
project
to
have
a
robot
account
or
a
service
account
of
which
meschery
does
have
a
robot
account.
If
you've
ever
you
you'll,
I
won't.
B
I
won't
spoil
the
surprise
you
can
go
find
out
which
one
it
is,
what
the
name
of
it
is,
but
that
robot
account
is
a
is
a
common
resource
for
the
project,
and
so
it
is
empowered
with
admin
privileges
to
be
able
to
execute
on
things
just
like
this,
and
so
a
pat
from
that
account
is
documented
and
accessible
to
maintainers,
and
so
that
pack
can
be
used.
You
know
from
the
service
account
rude
rocks.
B
B
Anyway,
yeah.
B
Anyway,
I'm
really
sensitive
to
the
fact
that,
like
we
have
a
horrible
history
or
a
record
track
record
of
going
way
over
time,
so
like
we're
over
time
on
this,
like,
hopefully
that
helps
you
guys,
let's
hit
it
again,
if
that
doesn't
get
us
past,
where
we
need
to
go
neha,
can
we
hit
your
thing?
Real,
quick?
Is
that
because
you're
getting
stuck
on
this.
K
B
B
Need
to
drop,
we
definitely
had
a
great
call
today.
You
know
please
hang
out,
but
if
you
need
to
go
it
was
nice
to
see
you
all.
K
Yeah,
so
here
here
what
is
happening
here
is
like
I
have
deployed
the
prometheus
and
paraphernal
in
the
mastery.
This
is
your
system
main
space.
If
you
see
there
are
foreigners-
and
I
have
couple
of
apps
running
as
well
like
the
product
ratings
app
button
for
app,
I
mean,
and
it
has
all
those
spots
and
and
also
if
you
see
I
have
like
necessary
running
in
the
machine
in
space
measuring
server
with
all
of
the
adapters
like
easter
and
all
so.
K
What
I'm
doing
here
is
like
if
I
access
this
mesh
tree
on
my
docker
platform,
I
could
see
that
the
connection
is
successfully
connected
to
mini
cube,
as
I
have
a
mini
tube
cluster,
but
the
problem
here
is,
if
I
go
inside
life
cycle
is
still
here.
It
should
show
me
all
those
life
cycle
I
mean
earlier.
It
was
showing
me
like,
but
today
I
mean
from
yesterday.
Sometimes
you
should
sometimes
it
does.
Not.
K
K
And
the,
and
what
and
I'll
show
you
like
if
you
see
my
grafana
dashboard,
the
data
sources
promises-
and
I
have
like
a
couple
of
the
static
easter
boards
here
and
I
could
see
the
data
in
control
plane
dashboards.
If
you
see
because
I
mean
I
could
see
because
of
all
the
istio
components,
I
have
running
control,
plane,
components,
that's
why
I
see
the
data
here,
but
if
I
come
here
to
workload
dashboards,
I
I
mean
I'm
missing
few
of
the
data,
so
I
wanted
to
understand
on
this
part.
K
Like
I
mean
I
have
a
booking
for
applications,
so
am
I
missing
anything
here
to
populate
the
data?
That's
what
because
what
I
understood
from
my
debugging
is.
The
prometheus
is
missing.
Few
of
the
targets
here
like
if
you
see
like
I
have
42
cars,
but
only
four
active
service
endpoints,
also,
if
only
one
is
active,
so
I
think
this
is
this
might
be
the
issue
here.
That's
why
it's
not
exposing
the
data
here.
B
It
seems
like
it
was
a
great
walkthrough,
I
mean
it
seems
like
you
when
you
go
to
the
control
plane,
dashboard,
it's
also
empty.
I
think
you,
oh
okay,
so
this
is
positive.
Good.
K
K
That's
it's
showing
me
data
and
if
you
see
here
I
have
like
in
prometheus,
I
could
see
all
these
two
endpoints.
B
Cool,
can
you
go?
I
didn't
I
missed
it.
Can
you
go
back
to
so?
This
portion
is
working.
That's
great!
The!
If
you
go
back
to
your
cube,
ctl
output.
Did
you
end
up?
Do
you
have
a
you?
Have
the
sample
application
deployed
and
there's
an
element
that
utkarsh
was
noting
about
the
notion
that
the
so
so
this
is
good.
This
is
entirely
expected.
We
don't
have
any
bugs
yet
there's
no
workload.
Data
metrics,
because
the
sample
app
itself
is
not
it's
off
the
mesh.
It
is
not
running
on
istio
mesh.
B
One
way
that
you
can
tell
from
the
output
here
is
that
there
are
the
applications,
the
service,
the
pods,
rather
details,
reviews
ratings.
They
would
have
like
a
two
of
two.
They
would
have
two
containers
running
a
secondary
sidecar.
B
Totally
easy,
you
know
like
a
totally
easy
thing
to
overlook,
but
that's
why
that
particular
board
isn't
working.
You
can
still
further
your.
I
mean
one.
You
can
go
back
and
you
know
get
on
board
the
sample
app,
but
two.
You
can
also
verify
that
in
meshes
ui-
and
this
is
probably
the
thing
to
do
right
now
like
if
you
go
back
to
meshes
ui.
If
you
go
over
to
settings,
you
go
to
metrics.
C
K
K
B
K
B
B
If,
when
you
do
see
the
control
plane
in
the
drop
down,
you
should
then
get
metrics
and
that
that's
good
and
then
yeah
once
you
onboard
your
apps,
you
should
be
able
to
get
metrics
from
the
workloads
board
as
well,
but
the
issue
that
I
would,
if
I
were
you
and
I
would
go
file
in
a
bug
on-
is
this
right
here,
like
th.
This
is
like
whether
you
were
getting
metrics
or
not.
You
should
be
seeing
the
boards
from
the
first
start.
K
I
would
surely
like
file
the
box
for
this,
and-
and
this
part
here
is,
do
we
need
something
here
like
yeah
there's,
a
friend.
K
B
You
know
yeah,
it's
a
crappy
ux.
This
is
a
super
powerful
thing.
There's
a
ton
of
power
in
these
two,
these
two
things
and
we
just
the
the
there's
a
couple
of
bugs
and
bad
ux
getting
in
the
way.
But
what
this
lets
you
do
is,
if
you
have
a
prior
investment
in
grafana
boards
and
you've
curated
your
own
set
of
metrics
that
you're
looking
at
and
you
like,
but
you
want
to
run
some
performance
tests,
and
so
you
want
to
see
those
same
metrics
in
measuring
at
the
same
time.
B
What
that
prometheus
paste
that
that
text
field
lets
you
do
is
on
the
prometheus
tab,
and
even
this
is
a
little
bit
confusing.
I
think
for
people
because,
like
what
you,
what
you
would
do
is
you
can
go
into
grafana
grab
the
json
out
descriptor
of
a
given
board,
and
then
you
can
have
your
own
custom
board
in
like
very
quickly
incorporated
into
measuring
and
even
eventually
yeah.
So,
okay.
K
This
is,
this
is
fine
like
we
can
anyways
customize
this
board
like
we
can
have
our
own
board
here
paste.
B
That's
exactly
the
thing
yeah
when
you,
I
believe
that,
right
now,
the
operations
for
the
istio
adapter
are
such
that
if
you
provision
the
sample
app.
K
K
Here,
it's
still
actually
I
yeah
there
are
like.
I
could
see
a
few
of
those
ui,
but
this
is.
I
think
this
is
also
a
bug
or
not
sure
right
there.
I
could
see
this
like
blank,
don't
say
anything
adapter
list
here.
B
K
Oh
so
this
is
okay
default
name
space.
I
will
click
hto
system.
B
B
B
K
K
Okay,
fine,
fine
I'll
do
this
like
and
I
will
verify
like.
I
could
see
the
workload
data
as
well
in
my
draft
and
right
and
just
a
last
one
more
query
about
this
performance
test.
So
what
I'll
do
is?
I
will
just
trigger
that
product
page
endpoint
here
and
I
will
run
this
test
and
I
will
verify
like
if
I
see
some
of
the
errors
you
were
saying
some.
There
is
some
network
connectivity.
K
That's
why
you
were
seeing
the
errors
in
this
performance
results:
yep
yeah!
How
do
we
debug
that,
like?
What
was
the
network
connectivity?
Do
you
mean
like
different
some
of
the
like
rules
and
all
firewall
you
mean,
or
I
did
not
got,
that.
B
Yeah
the
end
point
that
you
had
had
in
your
screenshot
was,
I
think,
localhost.
It
was
basically
measury
generating
load
against
itself.
I
think-
and
that's
fine,
but
I
don't
know
that
the
particular
endpoint
that
you
had
listed
was
necessary.
It
the
way
that
the
output
read
was
like
oh
100
failure,
because
the
the
load
generator
inside
of
meshi
server
couldn't
actually
hit
reach
the
endpoint
that
you
were
trying
to
generate
load
against
so
so
I
would.
K
K
K
B
K
Yeah,
so
the
other
thing
I
want
to
tell
you
about
the
work
you
have
sent
me
to
like
investigate
the
difference
between
the
smp
metrics
and
the
node
exporter.
So
on
that
I
want
to
discuss
with
you
like
a
few
points
there.
So
can
we
have
another
call
for
that
like
any
other
time
or
do
you
want
to
discuss
now?
I
can
go
ahead
now
as
well.
B
Yeah,
I
don't
I
wouldn't
yep
on
the
community
college.
Pro
is
maybe
the
best
spot
because
we're
because
we're
about
to
have
a
ux
conversation
with
a
few
folks
who
are.
K
Okay,
sure
sure
so,
just
I
wanted
to
understand
from
there
is
like
I
have
seen
few
of
the
like
smp
metrics
already
code,
some
of
the
coding
mystery
where
it
shows,
I
think,
the
parameters
which
we
use,
but
I
did
not
understood
the
static
boards.
Do
you
mean
these
are
my
static
boots
or
these
are
my
dynamic
modes?
That's
I'm
not
clear
between.
F
F
B
I
just
for
everyone
else
who's
watching
here.
I
can't
help
myself
but
say
this
that
there's
a
reason.
There's
a
there's.
A
couple
of
reasons
why
neha
is
being
is
talked
about
when
she's,
not
here
in
the
in
a
good
way,
and
why
she's
being
successful
right
now
is
because
she's
tenacious
like
she's
curious
about
it.
B
She's
going
after
it
she's
getting
her
she's
asking
her
questions,
she
get
them
answered,
and-
and
it's
been
like
that
for
a
long
time-
and
she
just
keeps
she
just
keeps
being
more
of-
she
just
keeps
growing
in
in
the
impact
of
the
work
that
she's
doing
and
it's
tremendous
and
like
a
lot
of
other
people.
Just
I
don't
get
it
and
like
hit
the
you
know,
give
up,
and
so.