►
From YouTube: Meshery Development Meeting (Sept 15th, 2021)
Description
Meshery Development Meeting - September 15th, 2021
Agenda:
- [Nitish] Review of progress toward UI architecture overhaul with introduction of Redux
How can people start contributing ? (Volunteers: Aditya Raj, Moa, Meghana )
- [Ashish] Meshkit updates for k8s component registration
- [Ashish] Seed content capability in local provider (#4173)
- [Rudraksh/Ashish] Seed content for applications
- [Gaurav] Demo(draft) of multiple import options in filter/pattern/application UI
- [Neha] Prometheus, static and dynamic boards
- [Rudraksh] Meshery with EKS (#4190)
- [Aditya/Lee] Milestone v0.6.0
- [Piyush] New changes in mesheryctl perf
- [Piyush] SMP Test with mesheryctlHave there been advancements in the registration spec, Ashish?
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/
B
A
D
Hello,
everyone
yeah,
sorry,
I
I
wasn't
even
able
to
hear
you
first
yeah.
So
oh
yeah,
I
just
joined
the
the
community
yesterday
and
I
am
not
I
I
don't
know
what
is
happening
right
now.
So
I'm
just
trying
to
understand
the
things,
and
I
am
just
in
my
break
from
work
and
I'm
trying
to
to
hear
and
understand
from
you
what
is
going
on
and
if
there
is
no
problem,
let's
see
do
you
want
me
to
to
present
myself
yeah
if
you
could
like,
we
would
like
to.
A
D
E
Yeah
we
yeah
there's
a
couple
of
ui
frameworks
that
get
used
around
the
projects.
Laravel
isn't
one
of
them.
Yet
there
is
some
vue.js.
It
so
happens
that
it's
in
well,
it's
in
this
project
called
image
hub
and
it's
that
project
is
kind
of
hard
to
explain.
Actually
it
was
just
a
anyway
there's
a
this
nominal
use
of
vue.js,
but
there's
a
lot
of
react
floating
around
the
site,
that's
being
shared
right
now,
it's
done
in
gatsby
or
in
react.
E
D
E
B
F
A
F
F
A
A
A
C
H
So
yeah
so,
first
of
all
like
to
give
people
who
doesn't
know
what
is
happening
here
in
context,
so
they
have
actually
been
like.
There
was
an
open
proposal
to
implement
redex
and
some
other
bunch
of
stuff.
To
I
mean
to
include
the
state
management
of
measure
which
is
currently
you
know
it
is
okay,
but
that
process
they're
following
right
now
is
not
scalable.
So
there
was
a
pro
open
proposal
to
change
to
using
redux
and
redux
observable
and
other
frameworks
as
well.
H
So
basically
like
this
is
the
branch
like
that
is
in
which
I
am
we
are
going
to
work
on,
so
it
will
be
kind
of
like,
like,
unfortunately,
we
wouldn't
be
able
to
do
it
in
the
piece
by
piece
so
we'll
have
to
it
will
have
to
be
a
certain
scene.
So
the
branch
that
you're
working
on
is
a
measurement
restructure
and
the
link
for
that
is
also
there
in
the.
H
Take
a
look
at
it,
and
so
actually
I
have
updated
the
readme
as
well,
so
that
the
people
so
that
it
is
currently
in
sync
with
the
structure
and
the
changes
that
I
have
made.
So
basically,
I've
made
some
structural
changes
to
the
code
base
itself,
so
I
have
actually
made
it.
I
mean
I
have
descriptively
written
some
documentation
as
to
what
each
of
those
things
mean
and
how
people
who
are
new
to
this
can
contribute.
H
So
if
people
who
want
to
start
the
contributing
to
this
they,
I
would
advise
every
one
of
them
to
just.
First
of
all
follow
this
thing
and
from
there,
if
you
think
about
what
we
have
to
do
so
since
I
know
that
back-end
integration
is
almost
complete,
so
the
next
we
will
be
thinking
about
like
what
we
should
be
doing
is
to
start
creating
ui
components
and
using
the
data
that
we're
receiving
back
into
display,
and
so
for
that,
like
yeah.
H
So
we
can
actually
start
with
the
dashboard
page
for
now.
So
basically
it
wouldn't
be
any
different
from
what
we
have
right
now,
because
there
is
not
any.
I
mean
there's
no
change
in
the
functionality
or
something,
so
we
would
just
take
the
same
thing
that
the
current
the
current
measuring
dashboard
is
doing,
but
then
implemented.
H
H
Basically,
refactor
the
whole
code
and
implement
it
in
a
more
efficient
way
or
if
you
can
call
that
so
yeah
people
who
are
interested
in
it
can
have
penguin
slack
or
whatever
that
variable
so
and
we
can
get
started
with
this
so
first
like
aiding
the
radiation
would
be
the
first
thing
to
do
and
yeah.
If
you
guys,
still
have
any
questions
on
based
on
that
or
if
you
want
to
ask
something
that
yeah
slack
would
be
the
place,
I
believe,
or
yeah
and
yeah.
E
E
That
is
it
true
that,
within
this
branch,
you've
taken
a
couple
of
components,
not
only
have
you
restructured
the
way
the
ui
is
laid
out
just
in
terms
of
the
file
system,
but
also
you've
taken
a
number
of
components
and
refactored
them
to
to
this
new
style
that
incorporates
redux.
So
if
is
there,
one
component
in
here
that
you
would
say
makes
for
a
good
reference
for
others
to
potentially
take
for
other
individuals
to
take
different
components
through
that
same
refactoring.
H
Yep,
actually,
I
do
have
so
basically
like
the
structure
that
we're
implementing
is,
if
you
want
to
know
more
about
the
structure
that
we
have.
I
have
actually
written
about
that
in
the
readme,
but
if
you
want
to
know
more
about
it,
it's
basically
what
retex
toolkit
addresses
us
to
follow.
Basically,
the
text
team
manages
to
follow,
so
that
is
basically
what
we're
doing
here.
So
if
you
look
at
that,
like
we
have
split
up
the
functionalities,
that
measure
is
providing
into
such
folders
so
that
it
is
easy
for
us
to
manage.
H
So
in
this
case,
if
you
the
machining
component,
is,
I
would
say
it
is
pretty
much
I
mean
like
it's
it's
very
solid
as
of
now
so
we
do
not
have
to.
I
mean
like
yeah.
This
can
be
taken
as
an
example
of
how
we
should
implement
other
components.
Other
components
are
not
totally
as
solid
as
a
measuring
components,
as
this
particular
feature
is
so,
but
still
they
are,
I
mean,
like
the
integration
is
there
are
complete,
but
then
it
is.
H
This
would
be
the
ideal
thing
to
take
as
an
example
for
it.
So
if
you
go
over
here
and
see
the
components
you
have
these
two
components
like
yeah,
this
would
be
the
ideal.
I
mean
this
would
be
an
example
as
to
how
we
should
structure
each
and
every
I
mean.
Basically,
we
have
two.
We
have
two
type
of
components
that
we're
going
to
create
one
one.
H
One
type
is
that
they
actually
do
not
have
one
one
type
of
component,
actually
fetches
the
data
from
the
backend
and
then
passes
into
another
component
and
yeah.
That
structure
is
basically
followed
here,
so
these
two
components
would
be
the
example
for
us
to
you
know.
Basically,
this
is
what
how
how
we
should
create
reactivity
components,
but
then,
if
you
want
to
know
about
creating
the
whole
structure,
then
this
would
be
the
idea
if,
if
I
made.
B
E
E
I
Oh
hi
thanks
lily
yeah.
Actually
I
had
a
question
I
didn't
want
to
interrupt
in
the
flow.
Therefore,
I
just
put
my
raise
my
hand.
Couple
of
days
ago,
I
was
trying
to
install
measuring
on
my
aks
cluster
and
a
couple
of
issues,
so
I
want
to
really
understand
what
is
happening
so,
therefore,
to
understand
what
is
happening
under
the
code.
I
I'm
in
no
way
go
expert,
but
if
I
clone
the
mystery
product,
this
particular
repo
and
try
to
read
co
go
through
the
repo
would
that
be
of
any
help,
but
I
should
make
a
start
to
investigate
the
issues
which
I'm
encountering,
I
think
the
guy
from
here.
They
are
more
across
with
the
issue
which
I'm
having
I'm
not
able
to
install
measure
on
my
cluster.
E
Okay,
good
yeah,
so
that
there
are
two
so
you're
right.
That
is
a
different
topic.
I
didn't
realize
yeah,
I
would
say:
if
you're
looking
to
deploy
measuring,
do
not
clone
the
repo.
That's
not
the
that's,
not
necessary!
You
don't
need
to
build
the
project
you
mastery
ctl
should
be
your
answer,
but
yeah
to
be
concise
that
so
maybe
hold
on
to
that.
That
thought
that
we
could.
I
E
E
E
A
H
H
I
have
introduced
this
callback
function
first
told
me
because
in
future
we
might
need
to
dynamically
modify
the
definitions
and
schemas
that
we
are
producing
from
this
function
and
initially
this
function
only
was
taking
yaml
as
a
input,
but
but
if
I
want
to
generate
components
for
for
kubernetes,
the
kubernetes
api
open
api
actually
gives
us
the
json.
So
we
need
json
so
basically
a
little
bit
of
customization
in
the
message.
So
once
this
pr
gets
messed,
I
can
I
can
write
the
function
for
generating
components.
H
H
H
These
filters
are
pre-built
into
the
mesh
binary
and
these
are
the
patterns,
so
that-
and
I
have
made-
I
had
actually
made
this
pr
yesterday
or
day
before
yesterday,
and
I
have
made
two
changes
in
the
pr
to.
Basically,
I
have
added
this
seed
content
function
as
a
method
in
the
local
provider,
because
the
remote
provider
is
not
concerned
right
now.
With
this
remote
that
thing,
the
seeding
of
content
for
the
remote
provider
will
be
added
in
the
mesli
cloud.
So
this
is.
H
This
is
for
the
seeding
of
patterns
and
filters
now
upon
the
applications
for
the
applications.
Different
adapters
have
different
applications
so
that
we
would
need
to
disc.
First
of
all,
we
will
need
to
basically
bake
those
applications
into
docker
files
of
those
those
given
those
basically
adapters
and
from
here
rudraksha
can
that's
that's
the
ci
thing.
H
So
the
drugs
can
first
go
ahead
and
bake
those
binaries
and
not
combine,
not
binaries,
basically
the
applications
inside
of
adapters,
and
then
I
can
do
the
same
for
the
seating
same
for
the
applications.
I
just
have
to
introduce
one
endpoint
in
the
messy
server
that
will
receive
the
http
request
like
when
the
adapter
spins
up
we
send
in,
as
we
basically
register
the
capabilities
of
the
adapters.
At
the
same
time,
we
can
also
send
the
applications
and
those
applications
can
be
seeded
then
into
the
message.
Server
and
different.
H
J
K
I
haven't
looked
into
the
complete
implementation,
so
I'm
not
sure
if
an
error
here
will
result
in
a
notification,
although
it
can't
it
I
mean
if
the
implementation
wants,
I
think
it
can,
because
there
is
a
channel
available
for
sending
his
events,
which
we
actually
use
for
notification.
At
the
moment,
we
should
be
able
to
buy
this
particular
into
that
channel.
E
E
H
That
the
seated
patterns
can
be
deleted
from
the
database,
yes,
but
they
are
still
at
the
file
on
the
file
system,
so
at
restarts
it
they
will
get
re-registered.
So
there
is
no
way
user
can
permanently
delete
it,
except
for,
of
course,
going
into
the
going
into
the
image
and
deleting
them
explicitly.
E
H
E
The
inaccuracies
in
that
statement
have
me
biting
my
tongue-
good,
yes,
true
and
not
true.
At
the
same
time,
energy
boy,
it's
really
hard
for
me
to
buy
my
thumb.
Let's
talk
about
it
in
this
spec,
it's
good!
It's
good
to
great,
what's
been
shown,
but
does
anyone
else
have
feedback
on
what
was
shown
as
really
exciting?
Actually.
E
I'm
being
facetious?
I
guess
so
for
the
for
the
majority
of
the
folks
that
are
on
the
call
just
to
to
be
aware
that
as
mesherie
in
its
upcoming
release,
the
dot
six
release
will
have
its
its
first
in
some
areas,
but
more
robust
support
in
other
areas
for
web
assembly
filters,
the
management
of
them
for
the
in
the
incorporation
of
real
applications,
real
workloads,
not
just
sample
applications
that
have
been
out
of
the
box
before,
but
workload
for
ingesting
and
managing
workloads.
That.
E
E
As
such,
they
have
some
very
similar
concerns
about
the
life
cycle
of
how
they're
managed
how
that
content
is
managed.
The
reason
that
mesherie
is
being
seeded
with
this
type
of
content
is
to
provide
value
out
of
the
box
like
value
when
you
deploy
measuring
it's
great.
That
measurey
is
something
of
an
extensible
platform,
and
that
means
that
people
can
bring
their
own
content
because
they
can
extend
the
platform
that
way
or
they
can.
E
But
in
order
for
measuring
to
be
initially
useful,
it
would
be
great
if
there
were
things
there.
You
can
imagine
something
like
docker
hub
if
it
was
put
up
and
there
were
no
containers
available
for
people
to
download
run.
It
may
not
have
gotten
much
traction
right
like
so.
You
need
to
need
to
provide
some
content
and
all
of
you
have
been
working
on
various
on
different
content.
Some
of
you
on
patterns,
some
of
you
on
applications.
Some
of
you
on
filters.
E
E
H
Yeah,
basically,
the
function
that
sees
those
contents
returns
the
uuids
of
all
of
the
patterns
and
filters,
the
ones
which
were
registered
so
actually,
I'm
also
doing
a
cleanup.
There's
a
cleanup
function,
because,
if
you're
running
messaging
locally,
which
which
would
be
done
in
the
development
environment,
if
you
keep
restarting
machine
each
and
every
time
the
filters
and
patterns
will
keep
getting
registered
in
the
database
and
it
would
start
to
bloat.
So
there's
this
function,
which
actually
takes
in
the
uuids
returned
by
the
seed
content
function
and
then
deletes
those
every
time.
E
E
E
A
L
E
Yeah,
that's
a
good
yeah,
it's
actually
a
good.
Yes,
I
think
it's
the
answer
and
then
also
it's
a
good
question
for
everyone.
That's
on
the
phone
to
think
about
a
little
bit
and
it's
there's
there's
an
invisible
line
of
responsibility.
That's
drawn
between
prerequisites
that
need
to
be
available
in
an
environment.
J
E
E
E
And
so
good,
maybe
we'll
if
we
can
discuss
this
verbally
for
a
moment,
so
the
proposal
is
there's
sort
of
an
acknowledgement
that
the
prometheus
and
grafana
instances
that
meshri
will
connect
to
and
use
you
know
to
date
or
well.
The
the
built-in
support
for
that
mastery
has
built-in
support
for
provisioning
service
messaging
and
and
allowing
users
to
provision
different
configurations
of
that
mesh.
L
E
And
then
so,
the
for
the
question
becomes
so
there's
an
acknowledgement
from
the
surface
mesh
projects
that
the
prometheus
and
grafana
that
they
include
are
oftentimes
duplicative
of
existing
deployments
and
they're
also
generally
incomplete,
or
maybe
not,
production
ready,
and
so
all
of
that
is
good
context
for
trying
to
for
asking
ourselves
should
we
require
that
users
bring
their
own
prometheus
expecting
that
they
probably
would
have
it
anyway?
Should
we
facilitate
for
deploying
prometheus
and
grafana.
K
L
Caution
yeah,
but
another
question
here
is:
how
do
we
like
support
upgrades
and
all
your
latest
promises
like
like?
If
you
move
to
helm
charts
I
see
like
there
is
a
prometheus
operator
which,
like
very
active
com,
they
recently
put
the
updates
on
the
operator
side
and
we
get
lot
of
new
features
in
that.
So
are
we
supporting
in
the
all
these
and
hdr
down.
E
E
E
Seed
content,
the
so
there
is
meshri,
does
have
a
number
does
interact
with
helm
a
fair
bit.
Actually,
a
number
of
measuring
components
are
deployed
via
helm,
programmatically
and
so
to
provide
support
here
when
you
think
of
it
in
context
of
of
it
being
a
seed.
Application
is
like
like
nothing
like,
but
the
the
statements
they
have,
that
it's
actually
really
there's
a
lot
of
wisdom
in.
In
that
forward-looking
statement
that
you'd
made
about
well
geez
like
what
goes
up
so
then
does
mesherie
since
it
provisioned
that
app.
E
E
That
has
to
be
written
like
whatever
the
answer
is.
It
has
to
be
written
down
and
clear
if
mescheri
uses
it
as
a
prereq
to
collecting
the
full
set
of
service
mesh
performance
metrics
to
satisfy
that
specification
and-
and
so
mesri
is
dependent
upon
prometheus
node
exporter
being
there,
then
the
answer
it
probably
becomes
yes,
mescheri
is
responsible
for
that
upgrade,
but
maybe
not
the
rest.
And
so
where
do
you
draw
the
line
so.
E
E
E
So
neha
did
to
leave
you
hopefully
unconfused.
It's
like
hey,
I
think,
there's
a
lot
of
positive
sentiment
toward
this
particular
issue.
A
couple
of
things
to
potentially
write
down
and
sort
of
clear
draw.
The
line
on
like
upgrades
or
whether
or
not
the
prometheus
operator
is
supported
as
a
seed
content
for
applications
and
then
you're
right,
like
the
other
step
here.
The
other
related
work
here
being
understanding
what's
being
collected
in
static
boards
and
what's
left
unsatisfied
by
the
spec.
L
J
Okay,
so
a
little
bit
of
background
on
what
is
going
on,
it
is
about
deploying
mystery
with
dks
like
making
sure
that
it
works
with
eks
deployments.
It
all
started
some
time
ago,
when
a
user
from
personal
dot
ai
named
paula
sister,
he
was
trying
to
deploy
mystery
on
his
eks
cluster
and
he
was
facing
some
trouble,
so
he
reached
out
to
us
and
we
helped
him
debug.
J
Our
immediate
assumption
with
that
was
it
was
due
to
shelling
out,
like
we
were
using
some
hacky
scripts
in
mystery,
cpl
to
actually
configure
the
user's
cube
config
to
be
compatible
with
the
mystery
server.
But
that
was
not
the
case.
This.
This
error
is
actually
not
from
misery
ctl,
but
from
the
measuring
server.
J
J
Some
of
them
are
supported
by
the
kh
client
library,
natively,
like
as
your
gcp
openstack
and
other,
but,
however
aws
is
not
supported,
natively
and
it
depends
on
the
aws
cla,
so
its
client
library
needs
aws
cli
in
path
so
as
it
can
connect
with
the
cluster
and
interact
with
it.
Basically,
it's
written
here.
This
is
not
a
problem
in
in
cluster
deployment.
At
least
I
haven't
faced
this
problem
till
now.
Measuring
cluster
works
with
the
eks
and
you're
able
to
deploy
adapters
with
service
messages.
There
was
only
a
bug
about
service
account.
J
I
guess
that
be
that
that
will
be
a
little
different,
but
for
now
on
docker
deployments,
so
basically
the
client
library
or
any
cube
cpl
binary.
You
would
use
that
would
need
to
authenticate
using
this
command.
Aws
eks
get
open,
and
this
is
how
a
cube
config
for
an
eks
cluster
looks
like
so
even
in
this,
you
can
see
that
it
depends
on
the
aws
cli.
So
this
token
gets
cased
like
once
a
request
is
made,
it
gets
cased
and
then
it
stays
there
for
further
request.
J
After
some
time
it
actually
expires.
But
then
there
is
there
is
that
aws
credential
like
the
user's
credential,
which
is
kind
of
an
access
id
to
use
aws
using
your
console
or
your
cli,
so
that
refreshes
it,
though
so
that's
not
a
concern
like
expiring
of
this
token,
and
basically
some
changes
would
be
needed
for
this.
J
An
immediate
solution
would
be
shipping
the
aws
cli
in
necessary
stocker
image.
This
is
a
little
painful
like
the
thing
is
that
we
can
let
the
client
libraries
handle
authentication.
They
can
just
call
the
aws
commands
and
do
whatever
they
want
when
they
actually
want
to
connect
with
the
cluster
and
they
interact
with
or
there
is
another
application
binary.
I'm
not
sure
I
haven't
looked
into
it.
There
are
some
things
with
this.
So
one
thing
the
original
aws,
it's
actually
a
python
binary.
J
J
I've
lost
the
error,
but
it
was
looking
for
shared
library,
so
when
using
aws
cli
in
mystery
container,
we
need
to
take
care
of
it
other
than
this.
There
is
this
aws
iem
authenticator,
and
this
provider
is
also
a
binary.
However,
this
might
provide
an
api
I
haven't
looked
into.
Also,
something
would
need
to
be
done
through
the
console,
so
a
token
would
be
have
to
generate
it
first
and
then
this
binary
would
be
using
it
to
connect
with
this.
So
I'm
really
not
sure
about
this
right
now.
J
So
in
this
pr
I
have
implemented
the
immediate
solution
and
the
docker
file
I'll
be
updating
in
some
time
like
once,
I
figured
the
shared
libraries
thing
other
than
that.
The
other
thing
would
be
copying.
The
users
aws
credentials
like
the
one
that
are
stored
in
here
in
this
file.
If
you
see
so,
they
need
to
be
copied
in
the
cluster.
As
well
so
that
they
can
be
used
while
using
the
aws
binary,
so
basically,
this
vr
is
about
that.
J
And
I
won't
actually
do
a
complete
demo
that
would
expose
my
credentials
but
a
little
bit
explanation
on
this.
It
first
looks
for
that
directory
like
the
dot,
aws,
slash
credentials
directory.
If
it
exists,
it
mounts
the
directory,
not
exactly
mounts.
It
copies
the
directory
into
messages
container,
and
if
you
don't
want
to
read
that,
then
it
expects
the
access
key
and
access
secretly,
basically
the
credentials
to
be
supplied
as
environment
variables.
J
J
E
J
E
Sorry
to
interrupt
the,
but
the
problem
here
is
yeah.
It's
just
I'm
sorry
to
be
interrupting,
but
yeah
I
mean
like
the
problem
is:
if
you
go
to
deploy,
measuring
without
using
mastery
ctl,
then
you'll
be
stuck
without
access
to
you'll
have
to
you'll,
have
to
provide
your
aw
like
it'll,
be
the
you,
the
ux
and
the
it's
not
just
the
ux.
It's.
J
They
won't
need
this
for
in
cluster
deployment,
so
helm
is
out
of
the
way
and
other
than
that
like
if
they
manually
apply
the
dot
compose
file.
Do
you
want
to
say
that
like
if
they
do
docker
compose
apply
in
our
compost.com,
because
this
would
be
only
when
misery
server
is
running
out
of
the
cluster?
So.
J
E
J
E
F
J
J
Another
reason
could
be
like
if
some
wrong
credentials
get
mounted
like
this.
This
command
would
also
take
care
of.
If
the
credentials
have
actually
expired,
it
would
stop
the
user
at
that
point,
like
don't
execute
this,
these
credentials
won't
work.
Please
try
to
generate
new
one
if
they
mount
it
at
the
start
time,
there
is
no
guarantee
that
they'll
work
until
and
unless
we
do
something
in
the
misery,
server
itself.
J
E
Yeah
yeah,
it's
not
an
easy
there's,
a
few
kind
of
like
what
neha
was
going
over.
It's
like
it
feels
pretty
good
like
what
you're
doing
you
know
feels
pretty
good,
like,
certainly
by
the
way
our
goal
and
hussein
has
been
trying
to
trying
to
chip
away
at
this
somewhat,
not
not
quite
yet
successfully,
but
getting
there
is
trying
to
get
off
of
the
dependencies
on
external
cli
external
binaries.
J
J
No,
actually,
there
is
no
way
in
the
sdk
right
now,
like
I
couldn't
find
anything.
Neither
it's
mentioned
over
the
internet,
like
people
have
been
struggling
with
this,
so
also
we'll
need
to
then
look
into
how
the
client
libraries
will
actually
interact
with
the
aws
sdk,
because,
right
now
the
client
library
is
only.
E
A
A
A
There
are
some
in
messenger
for
let's
see
what
are
those
muscle
cpl
for
edge,
so
there
were
two
commands
that
were
for
you
and
perf
list.
People
were
really
confused
on
how
to
use
them,
which
one
is
for
profiles
which
one
is
to
you
the
results.
So
we
have
removed
those
two
commands
and
we've
introduced
two
new
commands,
that
is
for
profile
and
perfuser.
A
If
I'll
just
write
my
cctv
profile,
I
will
get
what
should
I
say,
a
standard
25
profiles
returned
from
machine
server.
These
will
be
the
latest
profiles
having
the
changes
on
server,
so
25
will
be
listed
with
per
profile,
but
I
want
to
search
something
I
want
to
search
for
shape.
Let's,
for
an
example.
A
A
A
A
A
A
Okay,
previously,
we
were
having
issues
with
persisting
our
smp
results,
so
I
have
fixed
it
and
I
would
like
to
demo
that
so
I'll
just
show
you
the
file
that
I'm
going
to
use
and
it
has
the
smp
version
the
profile
id.
I
want
to
associate
this
result
with
the
name
of
the
test
I
want
to.
I
can
name
it
for
file
testing
and
the
rest
of
the
information
about
the
test,
like
the
endpoint,
the
load,
generator
protocols,
connection,
rps
duration,
I'll
save
it
and
I'll
fit.
C
A
J
J
E
Yeah,
that's
a
great
suggestion
from
rudraksh.
We
should
make
sure,
and
we
should
actually
just
make
sure
that
the
json
or
the
yaml
that
is
produced
is
one
in
the
same
like
it
is
smp
compatible
or
like
or
like.
It
is
smp
format,
because
this,
because
the
spec
and
measuring
go
hand
in
hand,
we
should
it's
it's
a
if
there's
a
difference
and
it
doesn't
make
sense
to
output
all
of
that
in
json
or
something.
Then
we
should
definitely
have
a
separate.
A
E
Yeah,
so
the
response
from
the
server
for
the
definition
of
a
profile,
yeah,
okay
and
then
yeah-
it's
so
so
like
in
in
terms
of
like
s,
p,
protos
and
what
they
define
they
it
should
yeah.
We
definitely
want
to
make
yeah.
We
definitely
need
to
overcome
any
gaps
that
are
there.
So
we
definitely
need
to
update,
what's
being
output
to
to
match
that
both
the
profile
definition
and
a
specific
set
of
results.
E
Those
should
align
verbatim
with
the
spec
like
they
should
be.
Otherwise
meshrie
is
not
implementing
the
spec,
so
we
we
both
might
need
to
change
the
spec
and
also
change
measuring
to
the
line
and
yeah,
and
I
suspect,
there's
a
number
of
things
that
the
spec
calls
for
that
we're
not
gathering
or
we're
not.
E
E
Yep
ultimate,
ultimately,
both
the
yeah.
It's
most
prominently,
the
it's
the
it's.
The
test
results
that
people
are
most
focused
on
and
with
respect
to
s
p,
but
s
p
as
a
format
as
a
spec
like
you,
if
you
just
provide
someone
test,
results,
they're
meaningless,
unless
you
say
well,
this
was
the
profile
that
generated
those
results
and
and
also
that's
meaningless
unless
you're
saying
here's
what
the
environment
looked
like
like
so
like
it
needs
to
capture
all
of
that
and
that's
kind
of
what
the
spec
calls
for.
C
E
E
Also,
you
might
have
particulars
about
your
application,
which
has
a
different
like
signature
and
behavior,
then,
and
so
the
spec
needs
to
aspire
to
most
of
that,
there's
some
aspects
of
that
that
the
spec
just
needs
to
point
out
like
link
out
to
somewhere
else
like
if
you
really
want
to
fully
describe
an
application.
All
that
it
has
it's
like
look
then
bring
your
cnab
definition
or
your
own
definition,
or
something
else
anyway.
Point
is
between
the
two
things
that
we're
discussing
right
here,
profile,
config
and
individual
test
result.
E
E
It
is
most
at
its
foremost
most
concerned
with,
like
specific
analyses
of
a
given
test
result
like
that's
kind
of
the.
What
most
people
would
expect
and
think,
but
yeah
it
needs
it's
trying
to
standardize
that.
So
it
needs
a
standard
way
of
saying
what
were
your
qps
and
or
because
someone
else
may
have
written
it
down
as
rps
and
it's
being
written
down
as
qps.
And
so
what
like.
E
Okay,
cool,
so
one
other
thing
I
I
probably
missed
it,
but
I
can
miss
the
linkage
between
the
perf
profile
when
someone
runs
special
ctl
per
profile.
They
search
for
a
profile
by
name.
It's
looked
like
a
pretty
good
user
experience
where
they
can.
You
know,
search
for
things.
The
one
thing
where
they
could
search
for
a
multi-word
name
without
using
quotation
marks,
you're
able
to
do
that
because
the
command
accepts
a
flag.
E
At
the
end,
we've
been
using
the
word
view
as
a
sub
command
to
exp
to
perform
and
expand,
and
so
give
thought
to
the
consistency
there.
The
the
linkage
that
I
missed
was
when
you
jumped
from
perf
profile
to
collecting
an
actual
result.
I
saw
that
you
used
the
good
mastery,
ctl
perf
result
and
then
a
guide,
but
I
don't
know
where
the
user
got
the
guide
for
that
specific
result.
E
A
A
A
For
me
to
implement
or
for
us
to
implement,
I
mean
okay,
so
profile
id.
There
is
a
separate
command
so
profiles,
someone
can
search
and
get
the
profile
id
from
there,
and
then
he
can
just
search
for
his
results.
We
want
to
search
for
a
particular
result
having
the
terms
he
can
search
that
too.
If
he
wants
just
to
list
out
and
see
this
data
about
the
results
you
can
just
do
that
also.
E
All
right
you,
by
the
way
you
put
in
a
lot
of
work
here
it
is
this.
Is
you
put
a
lot
of
thought
into
it
as
well?
This
is
really
I
mean.
That's
really
cool
that
column
based
output,
the
here's
a
request
think
on
this.
If
you
would
meshery,
ctl
perf
result
profile,
name
and
then
whatever
and
then
either
the
result
name
or
the
result
id
the
reason
for
profile
name
is:
there's
a
lack
of
there's
a
it's
a
wonky
user
experience.
It's
like
in
one
area.
E
A
E
Way,
yeah
yeah
totally
the
same
way
that
you
do
with
the
profile
name.
When
they
first
look
up
profile,
you
show
them
both
results.
You
say
pick
one
or
or
you
say
you
say,
hey
you
say:
hey,
there's
multiple
of
these
the
same
name.
Can
you
specify
the
id
the
id
please
by
the
way,
here's
the
three
that
have
the
same
name
and
here's
their
three
profile.
Ids
like
make
it
easy
on
the
user.
E
A
E
A
E
E
I
mean
you're
you're,
doing
something
you're
you're,
the
the
I
think
the
thing
you're
getting
snagged
on
is
the
fact
that
you're
supporting
white
space
and
yeah
and
that's
all
that's
cool.
Actually
I
wouldn't
do
it
like
if
I
were
or
knowing
it's
kind
of
an
interest.
This
is
an
interesting
phenomenon.
E
When
you
give
you're
bringing
a
better
user
experience,
I
think
than
most
people
have
with
a
cli,
and
I
wouldn't
do
it.
I
would
expect
the
cli
to
crash
or
to
consider
that
my
where
I
wrote
per
file,
I
would
assume
that
it
would
think
that
file
was
another
command.
I
was
trying
to
give
it
or
something
and
it's
funny,
because
we
collectively
all
of
us
get
held
back
by
you're,
bringing
a
better
ux
than
people
have
experienced
and
you're
being
held
back
by
well.
E
You
know
by
by
the
fact
that
that
that
isn't,
the
common
experience
for
clies
is
part
of
how
you're
getting
held
back.
The
other
reason
you're
getting
held
back
is
because
there's
a
reason
why
it
isn't
a
common
experience,
because
you
can't
tell
the
difference
between
what's
a
flag
and,
what's
the
just,
you
know
you
have
to
have
to
put
it
in
quotes
or
they
have
to
give
it
with
a
flag.
E
It
can
be
also.
The
flag
could
also
just
be
a
command
like
it
go
go
I
this
is
it's
a
good
reflection.
Go
if
you
would
do
docker
network
inspect,
go
inspect
the
docker
networks.
Like
you
go
list
out
docker
list
out
all
the
docker
networks
go
grab
a
specific
name
and
inspect
it,
and
like
look
at
the
fact
that
they're
intertwining
object,
names
and
commands.
Thank
you.
E
Yeah
and
it
needs
it
needs,
you
need
more
feedback
than
just
spot
feedback.
Actually
it's
gotten
you
a
long
way,
but
actually
you
need
navendu
needs
to
you
guys
need
to
get
together
and
actually
go
through
it
go
through
the
and
the
venue.
This
stuff
won't
fly
like
we're.
Having
inconsistencies
being
presented
here
with
the
other,
like
number
one
rule
like
consistency,
if
we
need
to
break
it
like
do
it
break
it,
but
you
need
to
make
sure
that
everything
else
is
consistent
as
well.