►
From YouTube: Meshery Development Meeting (March 31st, 2021)
Description
No description was provided for this meeting.
If this is YOUR meeting, an easy way to fix this is to add a description to your video, wherever mtngs.io found it (probably YouTube).
A
A
Hello,
yeah,
looking
you
don't
look
tired
at
all.
You
don't
look
like
it's.
What
time
is
it
by
the
way.
B
Almost
almost
10
o'clock,
yeah
yeah,
I
I'm
I'm
an
office
yeah.
It
looks
like.
C
B
A
B
A
Yeah
well
asuka
we're
we're
just
about
to
start
and
there's
a
number
of
people
who
have
seen
you
in
github
and
seeing
you
in
comments
and
other
things,
but
there's
a
few
folks
on
the
phone
that
might
not
have
gotten
a
chance
to
see
you
to
see
you
in
person.
A
Did
you
wanna,
yeah,
say
hi
and
then
introduce
just
say
a
little
bit
about
yourself.
B
Hi
everyone,
I'm
asko
material
of
the
lawyer,
file,
organized
and
notice.
I
joined
the
gnu
hard
project
as
my
senior
and
I'm
and
now
I'm
retiring.
My
english
exam.
It's
a
hard
work
for.
A
Me
that
was
perfect.
That
was
fantastic,
yeah
yeah,
oh
nice,
good!
Well,
yeah!
All
of
you
all
of
you
have
seen
zuko's
active
in
the
community,
but
but
the
time
zone
difference
makes
it
difficult
for
him
to
join
plus
he
has
slave
drivers
for
managers
just
and
then
also
also
he's
been
dramatically
improving
his
his
english
over
the
last
year
or
so,
and
so
so
we're
fortunate
that
he
just
happened
to
be
awake
and
able
to
join.
So
this
is
nice.
A
A
We
do
a
better
job
than
a
lot
of
other
projects
about
documenting
things
that
we're
doing
about
like
actually
there's
a
lot
of
google
docs,
but
but
still
some
of
what
we
discussed
doesn't
get
captured
or
even
though
we
take
meeting
minutes
like
it's,
not
the
full,
it's
not
all
of
what
was
said,
and
so
so
this
is
nice.
So
we
have
a
suco
today.
A
Let
me
let's,
let's
get
going.
Let's
share
these
meeting
minutes,
put
a
link
into
the
chat
for
those
that
might
not
have
the
minutes,
but
please
please
jump
in
take
notes
as
we
go
there's
a
number
of
you
who
have
who
have
topics
listed
today,
which
is
great,
and
so,
if
you're
on
the
call,
you
know
a
couple
of
things,
one
just
a
reminder:
we
record
the
meetings
and
post
them
on
youtube.
A
Also,
you
know
if
you're
on
the
call
record
your
name
down,
get
your
attendance
in
there
and
with
that
we're
off
to
our
first
topic
so
piyush.
I
think
I
think
this
is
you.
A
E
A
E
It's
coming
up,
yeah,
okay,
so
like
we
can
see
that
this
is
the
part
of
the
code
which
is
actually
failing,
like
failed
to
add
authentication
details
to
it
to
the
request.
Okay.
E
A
E
A
I'll,
let
I'll
let
someone
else
speak
to
that.
F
I
believe
mystery
system
start
would
be
a
prereq
to
start
system
of
performance
command
right.
E
F
E
A
Yeah,
basically,
it
yeah
the
the
essence
of
it
is.
This
is
pretty
similar
to
a
lot
of
other
software
that
hey
when
you
wanna,
when
you
want
to
communicate
with
its
api.
In
this
case,
it's
rest
api
that
you
need
to
have
an
authenticated
session
and
meshfreectl
is
while
it's
a
first
class
client.
It's
a
client
all
the
same,
just
like
measure
ui.
So
before
anyone
is
allowed
to
use
measure.
U
ui,
they
first
have
to
authenticate.
They
have
to
establish
a
session,
and
so
the
same
thing
for
a
mastery.
A
This
injects
a
small
manual
process
currently
where
yep,
if
you
want
to,
if
you
want
to
use
measuring
ctl
for
most
of
its
functions,
you've
got
to
present
a
token
and
in
order
to
get
that
token,
as
a
user
today
to
get
the
to
get
the
token
you
you
sign
in
establish
a
session
download
the
token
from
the
ui
and
then
from
there.
A
Token,
in
the
future,
in
as
a
matter
of
fact,
we
might
talk
about
the
roadmap
here
in
a
little
bit,
but
there's
hope
for
supporting
oauth
as
a
more
convenient
way
of
still
securely.
You
know
allowing
measuring
ctl
as
a
client
to
securely
communicate
with
the
rest
api
and
that
oauth
flow
would
be
a
little
nicer.
A
The
the
flow
would
be
more
like
you
would
go
to
invoke
a
mastery
ctl
command,
and
you
you,
you
know
it
would
recognize
that
you
don't
have
a
token
or
a
valid
token
at
the
time,
so
it
would
automatically
open
up
a
small
session
of
measuring
ui
have
you
off
and
then,
when
you're
done,
authing
download
the
token
to
your
local
file
system.
A
It
would
just
sort
of
grease
the
wheels
and
you
still
have
to
establish
a
session
and
get
so
that's
one
thing
in
the
future
that
might
help
another
thing
in
the
future
that
might
help
is,
is
yeah
is
some
way
of
making
it
either
longer
lived
tokens.
A
So
you
only
have
to
do
the
process
once
or
twice
in
part.
The
hope
is
that
people
are
using
measuring
in
their
pipelines
in
their
ci
pipelines,
and
so
one
of
the
easiest
ways
for
them
to
do
that
is
to
script
around
measuring
ctl
rather
than
the
rest
api,
but
either
way
that
they
do
it.
They'll
need
to
have
a
valid
token.
So
we'll
need
to
facilitate
the
process
of
like
of
programmatic
authentication.
You
know
like
making
that
a
little
bit
easier.
A
Yeah
so
anyway,
the
short
of
it
is
yeah.
It's
a
good
idea
that
you
had.
I
think,
we're
gonna
find
that
we're
gonna
have
to
just
you
know:
let
the
user
go
grab
the
token
and
bring
it
back,
and
so
understanding
that
you're
saying
that
this
air,
this
particular
command,
is
looking
for
a
token
in
a
particular
location
on
the
file
system.
Although
yeah,
doesn't
it
still
have
it?
A
It
has
a
token
flag,
though
right
like
you,
can
you
can
say
minus
t
and
then
tell
it
where
your
token
is,
which
doesn't
have
to
be.
It
can
be
in
any
location
right.
E
A
E
A
I
guess
maybe
if
if
this
is
okay
and
nintendo,
I'm
kind
of
looking
to
you,
but
this
is
okay,
that
we
might
as
we
walk
through
the
rest
of
the
topics.
If
you
could
establish
a
session
with
measure,
ui
download
your
token
and
then
try
to
run
this
command.
But
with
with
that
fresh
token,
you
know
as
a
flag
and
then
that's
and
then
maybe
we'll
circle
back
toward
the
end
of
the
call
and
see
where
we're
at.
A
Sure
and
there's
probably
room
for
improvement
by
the
way
I
mean,
there's,
there's
a
reason
why
navendu
is
asking
you
to
take
a
look
at
it,
because
I
think
I
think
he
he
collectively
we
sort
of
anticipate
that
there
are
some
rough
edges
and
so
yeah.
Please,
please
try
to
break
it.
I
think
you
know
please
try
to
identify
where
it
can
be
improved.
A
E
A
Any
other
feedback
for
push.
E
G
G
Yeah
one
feedback
might
be
that,
like
we
download
the
odd.json
file
in
the
downloads
folder
so
maybe
set
that
that
path
as
the
default
path
here,
instead
of
the
dot
meshery
folder
yeah.
What.
E
But
I
think
like
we
can't
set
it
up
like
that.
Well,
like
we
have
to
access
the
api
now
to
download
that
token.
G
Yeah
like
like
leaving
the
like
getting
downloading
the
token
part
right
now
and
like
the
user,
have
to
manually
download
the
json
file
from
the
machine
ui
and
after
that
they
can
just
use
the
measuring
ctl
proof
and
slash
the
dash
d
and
provide
the
path
to
the
token
file.
G
E
A
Yeah
yeah
right.
That
is
what
we
tell
them
to
do
today.
Actually
what
I
mean,
what
what
the
venue
was
just
saying
is
like
to
to
your
point:
push
about
trying
to
make
it
convenient
for
the
users.
A
Is
that,
instead
of
having
the
instead
of
assuming
that
that
to
look
for
the
token
under
this
file
name
in
this
directory,
maybe
just
continue
to
like
if
the
user
runs
the
command
with
no
token
flag,
we'll
just
make
the
assumption
that
they
have
that
they
do
have
a
valid
token
named
auth.json
and
that
it
is
under
the
path
home,
slash
downloads,
just
as
a
guess,
like
maybe
50,
of
the
time.
A
That
would
be
the
case
or
and
if
it
works
great
and
but
if
it
doesn't,
then
we
just
give
them
the
same
error
that
we
give
them
today,
which
is
you
make
sure
that
you
have
a
valid
token
go
down,
go
download.
It
he's
just
saying,
update
the
file
for
the
the
default
directory
from
auth.json
to.
A
Like,
oh,
okay,
you
know
what
let
me
I'll
we'll
type
it
up
and
if
andy
will
type
it
out
to
you.
A
Good
thanks:
okay,
any
other
feedback
for
abuse.
A
All
right,
karsh,
you're
up
next
performance
profiles.
D
Is
it
there?
It
is
okay
yeah,
so
this
is
a
pretty
short
update.
Recently
I
mean
yesterday
I
added
a
new
section
in
mystery
docs.
It
is
related
to
performance
profiles
where
I
am
describing.
What
performance
profiles
actually
is
so
a
brief
statement
about
it
that
it
is
basically
designed
for
scheduling
performance
tests
as
well
as
saving
test
configurations
and
stuff
like
that
in
here.
D
I
have
also
mentioned
that
the
scheduling
that
performance
files
is
going
to
provide
me
as
fast
powerful
as
strong,
but
without
the
hassle
of
remembering
the
syntax.
So
so
I
mean
this
is
so
it.
D
The
reason
I
mean
so
so,
actually,
when
I
say
cron
has
popular
spawn,
it
does
inherit
a
few
disadvantages
of
drawn.
So
something
like
that
so
cron
has.
So
when
we
talk
about
cron
us,
can
you
even
stop
using
corn?
We
are
not
using
prawn
internally,
but
we
would
be
we'd
be
internally
using
drawn
expressions.
So
it's
inherent
inability.
Is
that
the
minimum
unit
of
expressing
something
is
minutes
and
not
seconds?
D
So
if
you
want
just
getting
something
for
do
it
every
15
seconds,
you
cannot
do
that
because
tron
expressions
does
not
support
that.
So
that's
actually,
the
only
limitation
kind
of
we
will
be
facing
because
of
the
limitations
of
cron
is
like
you
don't
get
logs
and
you
cannot
pass
environmental
vehicles.
We
don't.
We
are
not
going
to
face
such
kind
of
problems
in
here,
but
that
is
one
kind
of
disadvantage
that
we'll
be
inheriting.
If
we
go
ahead
with
cron,
that
is,
first
thing
is
called
exhibition.
D
The
right-
and
if
it's
not
please
do
so,
is
that
whatever
the
exhibition
is
going
to
make
like
is
a
powerful
ass
corn,
because
I
mean,
as
far
as
I
could
think,
cron
was
the
crown
expressions
not
crunch
off
front
abs.
Quan
expressions
were
far
like
the
most
descriptive
thing
as
well
as
so.
Basically,
they
capture
a
lot
of
information
in
a
very
kind
of
minimal
expressions,
so
that
is
why
we
are
using
forward
expressions.
D
Something
else
can
also
be
used.
What
something
is.
First,
that's
the
primary
thing.
The
second
thing
is
that
is
this:
does
this
make
sense
that
has
powerful
is
strong,
provided
that
it
also
means
that
we
are
inheriting
one
of
the
disabilities
one
of
the
disadvantage
of
crimes?
A
A
Utkarsh
to
me
it
it
it's
the
the
way
that
it
sort
of
rolls
off
your
tongue
kind
of
sounds
kind
of
slick
like
you're
sort
of
calling
to
the
inner
nerd
a
lot
of
people
like
yeah,
a
question
for
you,
as
you
described,
that
that
scheduling
capability
or
what
what
would
become
the
scheduling
capability
is.
Is
it
the
case
that
you'd
be
able
to
have
a
performance
profile
scheduled
to
you
know
performance
tests
to
be
run
a
performance
profile
to
be
run?
A
A
D
So
it's
like,
if
any
scheduling
can
be
described
in
one
expression,
not
that
you'll
be
writing
one.
So,
like
you
said
that
do
everything
so
basically
one
test
every
wednesday,
so
yeah
basically
run
it
everywhere.
Next,
they
ask
us
every
friday,
after
the
visit
today
that
can
that
can
be
expressed
in
one
expression,
so
that
would
be
one
schedule
so
the
unit
of
one
schedule,
the
the
like
the
unit
of
so
one
schedule
in
the
back
end
in
the
way
that
we
are
representing
it,
it
is
basically
a
con
expression.
D
That
is
how
we
store
it,
not
exactly
how
it's
going
to
work
so
so
the
skin
is
actually
just
described.
That
is,
do
it
on
wednesday,
friday,
saturday
or
maybe
yeah.
So
basically,
every
month
on
these
few
days,
or
maybe
every
fourth
night
on
these
two
days
that
can
be
expressed
in
one
corn
expression
and
then
definitely
that's
going
to
come
out
thanksgiving.
D
So
I'm
not
being
extremely
clear
here,
because
you
can't
afford
a
lot
of
things
in
chronic
missions.
A
A
different
way
of
asking
is:
can
you
take,
can
you
create
two
independent
schedules
and
assign
those
to
a
performance
profile,
irrespective
of
whether
or
not
they
could
have
been
expressed
in
a
single?
D
Right
now,
performance
profiles
can
be
so
performance.
Profiles
only
have
one
schedule
if,
if
it's
not
an
imitation
of
the
uis
and
limitation
of
how
it's
been
implemented
in
the
back
end,
so
if
others
are
saying,
if
somehow
it
can
be
expressed
in
one
expression,
it's
going
to
be
once
given
and
that's
what
we.
What
can
be
done.
Performance
device
can
be.
D
Not
as
of
now,
but
we
can
definitely
extend
on
that
part,
that
is
supporting
any
number
of
schedules
with
performance
profiles
being
able
to
display
clearly
in
gui,
is
going
to
be
a
bit
of
challenge.
I
guess
I
mean
letting
users
know
that
now
you
are
creating
a
second
schedule,
or
so
it's
giving
me
something.
A
Yeah
right
right,
okay,
good
yep-
this
is
good
food
for
thought.
Going
forth
as,
like
the
as
people
get
an
intuitive
interface
like
this
to
schedule
their
performance
tests
that
that
there
may
be.
You
know
the
people
might
want
to
schedule
the
test
to
run
one
time
they
might
want
to
schedule
it
to
run
recurring
on
a
daily
or
a
weekly
or
a
monthly
or
something
you
know,
fortnightly
or
whatever
basis,
but
then
they
may
also
have
you
know
they
may
also
want
the
test
invoked
or
the
performance
profile.
A
A
A
A
different
user
might
create
the
same
schedule
for
the
same
performance
profile
potentially
or
the
same
user
that
owns
that
particular
profile.
They
might.
They
might
want
to
express
something
that
isn't
they
might
want
to
have
a
schedule
that
isn't
necessarily
expressible
in
a
single
schedule,
and
so
so
food
for
thought
about.
You
know
how
it
is
that
your
the
underlying
system
is
being
designed.
A
So
some
of
you
may
not
have
seen
some
utkarsh's
work
here
on
the
performance
profiles
and
the
way
that
you'll
be
able
to
look
back
at
the
history
of
what
those
tests
have
been,
and
you
understand
when
those
when
they
were
run.
A
D
Actually
that
functionality
was
present,
I
had
disabled
because
things
were
getting
duplicated
a
lot.
So
we
already
have
like
a
preview
where
you
capture
the
cards
and
see
the
jails,
and
then
we
have
a
table
view.
So
I
disabled
it
here
because
it
was
like
the
third
view
that
was
coming
in.
That
is,
you
can
come
to
the
calendar
and
just
see
it
right
here.
So
I
mean
it's
up
to
the
community
like.
If
everyone
has
it
is
not
yeah,
it
would
be
a
nice
idea
to
have
a
third
grade.
D
You
can
just
calendar
click
on
to
the
date
and
see
the
results.
Definitely
I
just
added
back.
A
Okay
and
then
good
yeah-
I
guess
that's
yeah,
so
just
and
then
you'll
also
be
able.
Is
the
intention
for
this
interface
for
someone
to
be
able
to
maybe
double
click,
to
create
a
schedule.
Yeah.
D
A
D
Yeah,
I
do
I'm
not
sure
if
it's
I
mean
actually,
it
may
require
basically
a
bit
of
more
brainstorming,
so
if
we
can
shift
it
to
last
and
see
if
we
can
squeeze
it.
A
All
right,
very
good,
nice
thanks
for
that
abhishek
and
asuka.
A
A
G
Yeah,
I
guess
so.
Basically,
the
main
person
who
was
working
on
this
is
was
michael.
So
since
he's
not
on
the
call
I'll
brief
what
the
exact
change
was
or
what
his
intention,
what
his
pr
was
so
should
I
share
my
screen
or.
A
G
All
right,
so
his
changes
were
around
the
protobuf
that
we're
using
in
measuring
for
that
that
is
used
between
machine
measuring
adapters.
So
he's
changed.
The
he's
introduced
a
new
function,
which
is
which
is
called
component
info,
which
takes
in
the
request
component
for
request,
which
is
currently
blank
and
which
gives
out
information
about
the
component,
the
component
here
being
the
adapter
which
are
the
type
of
the
adapter,
which
is
if
it
is
hd,
adapter
or
link
or
whatever
it
is.
G
Oh,
no,
sorry,
the
type
would
be
the
name.
The
type
of
the
component,
which
is
adapter
here
and
the
name
would
be
the
name
of
the
adapter,
is
to
your
service,
mesh
or
command
or
whatever
it
is,
and
then
the
version
here
and
and
then
the
git
share.
So
basically,
this
is
the
new
protocol
free
work
that
he
has
worked
on
basically
yeah,
like
that.
That's
pretty
much
it.
So,
if
he's
also
worked
on
the
adapter
library
part,
I
guess
so.
G
G
Yeah,
so
that's
that's
pretty
much
the
pr
in
here
so
yeah.
We
need
inputs.
G
No,
as
in
we
definitely
have
to
go
forth
with
this.
But
when
is
the
question
because
we'll
need
to
do
the
it's,
not
this
single
pr,
but
it's
a
combination
of
this
and
adapter
library
and
the
all
the
adapters
all
going
together,
because
it's
the
product
of
change.
G
You'll
need
to
update
the
adapter
library
version
in
each
of
the
adapters.
That's
another
thing.
A
To
do
yeah
do
we
have
so
asuka
is
here
who
can
help
with
carrying
that
across?
Yes,.
A
Some
of
you
are
familiar
with
the
fact
that
there's
a
a
spec
out
there
for
a
messaging
framework
and
notification
center
is
that
this
is
related
in
some
respects
to
what
we're
trying
to
achieve
with
that
larger
system,
which
is
you
know,
a
common,
a
common
transit
and
format
for
exchanging
messages.
Like
a
message
like
what?
What
component
are
you
and
what's
your
version?
A
A
This
seems
like
a
logical,
straightforward
step
that,
if
the
other
system
engulfed
it
that
later,
we
could
switch
over
to
that
other
system
or
that
that
other
system
would
enhance
and
augment
these
messages.
G
A
B
I'm
not
familiar
with
the
message
start,
but
but
I
think
I
believe,
the
the
another
grpc
service
just
for
transfer
the
information
about
the
component
information,
which
is
it's,
maybe
a
good
way.
I'm
not
sure.
B
A
Makes
sense
so,
to
recap
this,
the
the
current
state
of
what
we
have
is
that
each
of
the
most
of
the
individual
measuring
components
well
they've
been
migrating
to
use
a
common
library
mesh
kit,
mesh
kit
helps
each
of
the
components
have
some
uniformity
and
part
of
that
uniformity
is
like
or
would
be
reporting
on.
What
who
the?
A
What
the
component
is
like,
what
it's
it
should
be
able
to
self-report
like
what
its
name
is,
what
its
version
is
in
the
case
of
adapters,
what
operations
it
supports
and,
and
so
they,
the
adapters
all
do
that
or
almost
all
of
them
do
that
today,
any
of
them
that
are
using
mesh
kit
that
report
what
their
name
is,
what
operations
they
support.
A
A
It
kind
of
yep,
and
so
that
there's
like,
if
that's
a
full
thought
like
that,
that's
what
this
whole
thing
is
right
here,
that's
the
in
totality
that's
what
this
is
in
on
a
related,
but
to
object's
point
like
separate
set
of
discussions
around
a
common
format
for
exchanging
messages
for
like
tracking
the
state
of
operations
and
activities
across
the
components.
A
A
Well,
in
my
mind,
they're
kind
of
related
or
could
be,
but
but
it's
a
separate
system
today
and
we're
just
looking
at
expanding
that
system.
So
do
we
lose
a
suco
yeah,
I
know
so,
let's
go.
Does
that
help?
Does
that
make
sense
to
differentiate
between
the
two.
B
Yes,
this
this
makes
sense.
A
G
Okay
but
yeah,
basically,
we
will
have
to
do
it
concurrently,
but
yeah.
I
can
like
basically
create
a
roadmap
for
it.
G
We
are
constrained
to
do
it
concretely
because
of
the
product
of
changes
like
because
the
contract
changes
and
if
we
release,
if
we
release
merchandi
or
the
adapter
first,
it's
gonna
break
the
link
between
them.
So,
okay.
F
G
To
give
a
background
on
this
issue
from
a
user's
perspective,
a
user
can
have
multiple
clusters
running
and
he
can
have
multiple
service
meshes
running
across
these
clusters,
or
he
can
have
service
measures
spanning
across
clusters,
and
he
can
ideally,
he
can
use
measuring
to
manage
all
these
multiple
service
meshes,
but
measurey
is
deployed
onto
a
single
kubernetes
cluster.
G
So
there
comes
a
point
where,
where
measure
is
deployed
into
one
kubernetes
cluster,
but
it
has
to
manage
multiple
service
meshes
running
in
other
clusters.
G
G
Should
be
applied
or
should
be,
should
be
applied
onto
the
cluster
into
which
measure
is
deployed
and
other
service
mesh
management
commands
in
measuring
cdl
should
also
be
aware
of
the
other
kubernetes
clusters
that
are
running
and
which
it
has
which
are
service
meshes.
That
mashery
is
managing
so
so
yeah
currently
measuring
assumes
that
the
service
mesh
or
the
measure
or
measure
is
to
be
deployed
to
the
default
kubernetes
cluster
specified
in
the
cube
config
file.
G
So,
instead
of
that,
we
are
providing
the
users
with
an
option
to
specify
the
path
to
their
kubernetes
config
file,
and
they
can
also
specify
the
context
inside
that
config
file.
So,
even
if
the
user
is
managing
multiple
clusters,
he
can
even
if
their
default
cluster
is
something
else.
He
can
just
force
measure
ctl
and
measure
it
to
be
deployed
inside
one
cluster
and
all
of
the
measuring
ctl
commands
will
be
applied
onto
that
cluster
and
we
actually
have
two
two
designs
for
this.
G
One
is
to
have
the
the
path
to
configure
context
inside
the
inside
each
context,
or
we
could
have
it
as
a
separate
entity
which
is
outside
of
each
context,
and
we
can
just
provide
the
name
to
each
platform
configuration.
G
So
that
is
something
we
we
need
to
figure
out
and
see,
which
is
best
and
coming
to
the
other
point
where
measuring
needs
to
be
able
to
manage
meshes
that
are
deployed
across
multiple
clusters
and
which
may
be
outside
the
cluster,
in
which
measures
measure
ctl
measure
is
running
so
for
that,
instead
of
programming
measuring
ctl.
To
do
it,
measure
ectl
can
use
the
api
endpoints
exposed
by
measuring
server,
which
will
use
the
machinery
operator
to.
G
To
make
api
calls
to
service
meshes
in
ins
that
applied
in
other
clusters,
so
these
are
the
two
user
stories
for
this
issue
and
yeah.
That's
it
any
input
I'll
just
share
the
design.
A
F
F
F
So
how
do
we
know
about
how
to
connect
to
the
other
clusters
and
how
do
we
manage
the
keys
or
art
tokens?
Are
anything
of
that
sort
of
the
other
cluster?
This
is
something
which
I
am
not
very
sure
from
this
design.
Spec,
maybe
I'll
sync
up
with
navendu
offline.
A
Yeah,
it's
it's
confusing,
or
rather
there
isn't
a
clear
answer
at
the
moment
there
is.
There
is
thinking
like
you're
thinking
on
it
I'll
toss
in
a
couple
of
thoughts
to
to
hopefully
help,
and
that
is
oh
by
the
way
the
the
fireworks
make
for
a
nice.
Some.
A
This
is
a
little
bit
of
a
mind,
bender
a
little
bit
of
our
brain
teaser
and
the
vendor's
trying
to
help
us
navigate
through
it
by
starting
off
with
a
definition
of
terms
and
holy
cow.
What
a
what
a
fantastic
thing
to
do
to
before,
even
starting
the
discussion,
let's
agree
to
like
not
confuse
ourselves
and
and
call
you
know
and
then
use
some
terms,
I'm
still
trying
to
swallow
the
the
second
term,
but
I
don't
have
a
better
suggestion.
So
I'm
going
to
have
to
swallow
it.
A
Great
so
measuring
ctl
system
when
it,
when
any
of
those
commands
start
stop
restart
logs,
etc
when
when
they,
when
they're,
starting
or
stopping
or
deleting
or
or
updating
a
mesh
redeployment,
they're
concerned
with
the
platform
that
mesh
reserver
and
its
components
are
running
on
and
there's
only
there's
really
like
in
in
our
world
right
now,
there's
sort
of
only
one
of
those.
It's
like
it's
either
a
set,
a
docker
host
or
a
kubernetes
cluster.
A
A
Great,
so
all
the
system
commands
they
only
care
about
one
cube,
config
and
in
most
respects
like
they
really
only
care
about
one
kubernetes
context.
One
kubernetes
context
inside
that
cube
config
cool.
If
the
world
was
that
simple,
we
may
not
need
to
have
this
discussion
that
the
problem
becomes
when
people
start
to
use
other
measuring
commands
like
the
perf
command,
that
piyusha
is
looking
at
or
the
mesh
command
like
like.
A
If
someone
wants
to
use
mesh
free
ctl
to
provision
a
new
service
mesh,
okay,
well,
they
might
have
24
kubernetes
clusters
and
that's
not
a
large
number.
They
might
have
86
and
that's
still
not
that
large
of
a
number
it's
decent
sized,
but
there
are
organizations
out
there
with
hundreds
of
kubernetes
clusters
heck
each
and
you
know
so
anyway.
Point
is,
if
they're
seriously,
using
measuring
to
manage
their
service
mesh
deployments,
their
service
mesh
and
they're,
using
their
service
mesh
deployments
to
do
some
meaningful
things.
It's
going
to
be
a
multi-cluster
thing.
A
Multiple
clusters
are
going
to
come
in
here:
okay,
fine,
they're,
also
going
to
have
different
environments.
You
know
production
and
staging
and
development
and
other
things
that
they're
gonna.
You
know
they
might
they
it's
quite
possible
that
they're
gonna
have
multiple
measuring
deployments,
so
multiple
kubernetes
clusters,
multiple
measuring
deployments,
all
right,
so
things
just
got
complex
the
complex
because
of
not
be
because
of
the
the
other
life
cycle,
management
commands
for
a
service
mesh
or
or
for
an
application
or
for
performance
tests.
A
Because
when
someone
says
I'd
like
to
deploy,
link
or
d,
they
may
not
want
to
deploy,
link
or
d
on
the
same
cluster
that
they've
deployed
meshri
as
a
matter
of
fact
that
cluster
that
they
deployed
measuring
on
that
might
be
like
a
management
tools.
Cluster,
like
a
utilities,
cluster,
a
cluster
that
has
monitoring
and
management
software
like
meshri
on
it.
It
might
be
like
you
know,.
A
Know
it
might
be
separate
from
their
workload
clusters
that
are
running
other
applications.
Organizations
have
a
bunch
of
different
ways
of
divvying
up
where
they're
running
what
workloads
and
name
spacing,
and
all
this
all
this
stuff?
Okay.
Well,
so
if
I
want
to
run
mestre
ctl
mesh
start
linker
d,
but
I
don't
want
it
to
linker
d
to
run
on
the
single
cube
config
that
I
know
about
like
I
need
some
moniker
some
way
of
telling
measury
hey.
A
Actually,
as
you
go
to
execute
this
command,
I
don't
intend
for
that
to
happen
on
the
same
cluster
that
you're
on
I'd
like
to
give
you
this
piece
of
context.
This
string
this.
Basically,
I
want
to
give
you
this
cube
context,
this
a
string
that
matches
a
particular
cube
context
that
you
should
have
loaded
up.
You
should
be
aware
of,
like
some
a
user
may
have,
and
this
these
are
things
that
have
yet
to
come.
Husseina,
but
these
are.
This
is
kind
of
the
start
of
the
environments
discussion
in
measuring
server.
A
F
Each
yeah
each
context
would
have
a
platform
and
name
and
then
from
there
we
can
go
to
the
correct
platform
type
and
pick
the
cube
config,
and
maybe
we
could
use
the
wiper
calls
to
load
that
up
and
read
the
cert
related
information
or
anything
of
that
sort
to
connect
to
the
actual
workload
clusters
right.
A
That's
about
yeah,
that's
about
right,
except
for
the
one
item
that
might
be
a
little
bit
off.
Is
that
if
I'm
a
user
on
my
laptop,
I
might
have
some
cube
configs
in
some
contexts
locally,
but
meshre
servers
deployed
like
all
the
way
over
there
in
u.s
east
one
in
eks,
and
that
mesh
re
server
may
or
may
not
have
the
same
credentials.
A
The
same
cube,
configs
same
context,
so
as
a
user
like
the
flow,
the
user
flow
for
this
might
be
something
like
okay,
I
would
like
to
provision
linker
d
onto
my
development
system.
My
development
kubert
eks
cluster,
and
so
I
need
just.
Let
me
just
double
check.
Let
me
just
do
a
mesh
receipt,
meshery,
ctl
platform,
ls
platform
list
or
meshri
ctl
deployment
ls,
like
give
me
the
names
of
from
mesh
retrieved
from
mesh
reservoir.
A
All
the
names
of
the
environments
measuring
ctl
environments
list
all
the
names
of
the
environments
that
I
know
that
this
meshry
server
this
deployment
has
access
to,
because
actually
I
have
a
different
mesh
free
server,
that's
deployed
in
gke,
and
it
knows
about
different
environments.
So
I
actually
need
to
switch
between.
A
A
A
Yeah
something
like
that.
That's
a
possible
scenario
right,
like
what
I
was
trying
to
point
out,
is
that,
like
we
like,
I
don't
know
that
pointing
to
cube
configs
here
and
like
I
may
have
led
nvendu
or
australia
is
some
in
the
conversation
we
were
having
before
about
keeping
track
of
references,
file,
path,
references
to
different
cube,
configs
and
to
their
context-
and
maybe
that's
helpful,
and
maybe
it's
also
you
know
I
I
don't
know
you
you
might
I
don't
I.
I
think
I
think
I
don't
know.
A
I
think
I
don't
have
the
right
data
points.
It
might
be
the
case
that,
like
while
I
said
that
mesh
res,
like
the
user
and
my
laptop,
might
have
all
of
my
cube
configs
that
I
mean
it
probably
has
all
the
cube
configs
that
I'm
have
access
to
and
I've
probably
loaded
those
some
of
those
into
mastery
server.
A
It's
possible
that
there's
a
discrepancy,
it's
possible
that
I
forgot
to
load
them
or
that
I
wasn't
ever
given
them,
but
I
was
given
access
to
mesh
free
server
to
use
it
to
do
some
stuff,
even
though
I
was
never
directly
given
the
cube.
Config
to
do
the
okay,
so
some
of
this
is
a
little
bit
of
an
unknown.
A
Some
of
it
we
don't
have
to
over
complicate.
I
might
have
just
made
something
super
complicated
that
our
challenge
becomes
when
you
simplify
it,
it's
just
hey
when
you
run
a
when
you
want
to
tell
mesri
to
do
measuring
ctl
when
you
want
to
use
it
to
do
something
that
other
than
manage
the
lifecycle
of
the
mesh3
server
deployment.
A
Oh
okay,
so
in
mesherie
you
will
have
created
environments.
That's
the
li
dev
east
that
I
call.
I
call
that
lead
of
east
and
so
I'll.
Just
simply
like
it's
very
simple,
I
don't
need
to
think
around
with
cube
configs
locally
and
stuff
I'll
just
say.
I
want
a
new
link
or
d
set
up
in
li
dev
east,
like
based
on
the
fact
that
I
set
that
up
in
measuring
servers
and
I'm
just.
A
A
A
The
path
that
you've
been
on,
I
think
yeah,
is
probably
the
same
that
you
know
it
makes
sense
for
trying
to
make
sure
that
mesherie's
systems
start
any
of
the
system
commands
that
mesh
ctl
is
cognizant
of
what
type
of
kubernetes
cluster
is
there.
I
think
this
doesn't
change
that
path.
I
think
that
work
you
know
continues,
as
is
as
we
as
it
from
mesri
ctl's
perspective,
as
you
start
to
look
at
other
commands.
C
A
Nice,
okay,
I
didn't
mean
to
aim
all
that
at
you
or
but
necessarily
I
mean
unfairly
but
but
yeah.
It's
any
other
thoughts
on
this
that
other
people
have
about.
A
Good
yeah,
it's
a
lot
to
digest.
Thank
you
for
putting
on
paper
navindu
like
you've,
got
a
call
for
call
for
comment.
I
guess
a
request
for
comment
all
right.
We
just
had
time
anything
that
I
know
we
have
like
five
more
topics
that
didn't
get
covered.
A
A
So
push
you
saying
you
tested
the
command
pointing
to
local
auth.json
and
okay,
good,
so
yeah
push
great
said.
Thank
you.
You'll
put
your
notes
into
slack.
It
sounds
like
a
little
bit
later,
so
cool.