►
From YouTube: Meshery Development Meeting (May 26th, 2021)
Description
Meshery Development Meeting (May 26th, 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
Good
evening,
good
morning,
good
afternoon,
everybody
this
is
the
measuring
development
meeting
today
is
may
26th
and
we
are
three
over
and
we
I
think
we
can
get
started
so.
First
up
we
have
gaurav
who
will
be
demoing
on
filters
ui,
or
do
you
want
to
share
your
screen.
A
While
gaurav
does
that
to
get
give
a
bit
of
context,
we
recently
added
support
for
filters
that
is
awesome.
Filters
in
measuring,
so
governor
has
been
working
on
building
a
ui
that
will
where
the
user
can
manage
their
filter
files
and
import
and
deploy
filters.
So
yeah
got
all.
A
B
B
B
So
yeah
hello,
everyone
I'll
be
giving
a
walk
through
the
filters
ui.
It
is
a
work
in
progress,
so
you
may
consider
it
as
a
draft,
so
it
is
built
pretty
much
similar
to
the
patents
ui.
B
B
So
currently
it's
not
supported,
because
I
am
specifically
designing
it
for
the
jsonpam
format,
because
in
the
patterns
ui,
we
can
see
that
it
is
only
supported
the
yaml
and
the
yml
format,
so
we
will
be
integrating
the
data
with
a
react.
Json
schema,
so
the
user
can
create
itself
here
and
add
up
the
values
for
the
filters
so
which
will
be
a
part
of
this
data
table.
Also
so
after
the
data
table
is
ready,
we
can
integrate
it
with
the
with
the
web
assembly.
That
is
awesome
filters
using
credit
commands.
B
C
Garage
nice!
Yes,
this
is
good
to
see
this
nice,
that's
an
healthy,
looking
icon
that
you
have
as
well.
That's
great.
I
was
busy
chatting
while
you
were
giving
part
of
the
updates.
Let
me
let
me
recap:.
D
C
Little
bit,
if
I
could
to
make
sure
so
one
one
thing
is
that
for
some
of
us
this
is
for
some
of
us,
and
this
usually
includes
myself-
we're
set
we're
asking
ourselves
what
the
heck
is
this
guy
talking
about.
You
know
like,
and
so
in
this
case,
you're
talking
about
filters.
C
Okay,
network
traffic
filters,
okay,
so
in
service
mesh
land
request,
user
requests
like
http
requests
commonly,
but
they
don't
have
to
be
http
like
these
requests
are
being
sent
to
someone's
wordpress
blog,
that's
being
hosted
on
a
service
mesh
as
an
example
and
service
meshes,
have
the
ability
to
capture
those
requests,
filter
them
or
apply
filters
to
like
intelligent
filters.
C
Deeply
intelligent
filters
at
an
application
level
to
well
affect
the
traffic
like
stop
the
traffic
speed
it
up,
send
it
somewhere
else
change,
maybe
change
the
css
that
comes
back,
so
maybe
we
can
apply
a
this
is
going
to
be
kind
of
a
pun.
I
guess,
but
a
if
you
ever,
you
know
doing
photography
and
you
have
a
filter
on
your
camera
filter
on
your
lens,
like
we
could
apply
the
dark
mode
filter
to
a
website
without
changing
the
without
the
website
itself.
Changing
a
measuring
filter
could
bring
dark.
C
There
could
be
a
dark
mode.
Filter
brings
dark
mode
to
websites
without
changing
code
of
the
website,
that's
kind
of
cool,
and
so
so
that's
just
a
random
example
of
like
what
these
filters
can
do
in
order
for
and
filters
are
things
that
developers
can
write
that
there's
an
example
of
one
that
was
demonstrated
a
year
ago
at
the
last
dockercon
by
the
way
dockercon
is
like
starts
tomorrow,
we're
not
presenting
this
time
around.
We
presented
the
last.
I
don't
know
three
four
dockercons,
I'm
giving
workshops
at
those
as
well,
but.
C
The
link
will
help
people
understand
and
give
context
to
what
you're,
showing
and
then
again
think
more
specifically
to
what
you're,
showing
you're,
showing
the
ability
for
measuring
to
take
different
network
traffic
filters
and
insert
them
into
service
meshes
to
start
with.
Mastery
needs
basic
crud
operations.
Basic
like
create,
read,
update,
delete
operations
for
these
filters
and
you're
intentionally,
creating
this
capability
in
measuring
in
a
very
similar
fashion.
As
to
how
patterns
have
been
created.
C
If
you
don't
mind,
can
you
go
to
the
patterns
ui
that
menu
for
a
moment,
and
so
actually
to
start
with
before
the
data
populated
there
like?
It
was
really
hard
to
tell
the
difference
like
these
are
very
similar,
and
this
is
that's
great,
like
that
means
that
you're
doing
a
quality
job,
it's
quality,
because
these
two
uis
will
be
consistent
and
the
user
will
have
a
consistent
experience
and
that'll
be
great.
C
Now
you're
intentionally,
I
think
not
providing
certain
functionality
for
filters
because
they
work
differently
than
patterns,
and
you
know
but
yeah.
One
thing
that
it
strikes
me
that
I
don't
think
that
we've
seen
just
yet
in
the
demos
that
you've
been
giving
is
anything
in
the
table
and
I
know,
and
so
have
you
been
able
to
import
a
file
or.
B
C
No,
it
doesn't
like
forget
about
that
that
doesn't
matter
at
all,
just
let
any
file
get
uploaded.
It's
not
going
to
be
just
json.
If
is
pranav
on
no
that's
funny,
because
he
was
asking
for
another
meeting,
but
he's
not
here.
Speaking
of.
C
Anita,
do
you
mind
pinging?
Well,
the
delusional
optimist,
or
maybe
abhishek?
Can
you
do
that?
Since
I
know
you,
you
know
who
rudraksha's
and
see
if
he
can
jump
on
but
nope
it
doesn't
matter
like
hey,
even
if
that
were
the
case
still
doesn't
matter
still
you
want
to
progress
and
push
forward
to
show
that
you
can
upload
a
file.
You
can
retrieve
it.
You
can't
show
data
about
it
later.
D
C
It
needed
to
only
be
json
okay,
but
that
won't
be
the
case
in
this
case.
It
needs
to
be
well
anything.
Why?
Because
filters
come
in
a
lot
of
different
packages,
we'll
be
able
to
whittle
it
down,
but
even
at
that,
like
I'm,
not
sure
how
far
it'll
take
us
a
long
time
to
wheel
that
down
right
now,
we'll
just
let
the
user
shoot
themselves
in
the
foot
if
they
want
to
meaning
we
won't.
What
do
you
call
it?
C
If
you
would
yeah,
thank
you,
do
you
have
you
ever
just
in
your
testing
this
out?
Have
you
ever
tried
loading
a
file.
B
B
And
and
it
it
did
upload
or
no,
no,
it
didn't
upload
it
because
I
think
there
must
be
some
path
issues
so
I'll,
be
sorting
that
too.
C
Nice,
okay,
yeah
yeah.
Well,
I'm
I'm
super
encouraged
to
see
this
garage.
This
is
it's
it's
looking
good.
It
looks
like
you're
you're
diligently
following
along
on
the
right.
C
Yeah
I
mean
yeah,
it
would
be
good
to
yeah
yeah,
that's
kind
of
a
whole
nother
set
of
discussions
around
file
formats
and
what,
if
anything,
to
do
with
them
that
it's
actually
in
part,
why
the
singular
largest
difference
between
patterns
functionality
and
filters
functionality
will
be
that
for
patterns.
If
you
don't
mind,
can
you
click
on
that
ellipse
on
the
right
hand,
side
on
any
of
the
rows,
the
dot
dot
dot?
C
B
C
Now,
some
of
what
some
of
what
comprises
a
filter,
it
might
be
like
two
or
three
files
that
are
belong
to
the
same
filter.
One
of
those
might
be
a
json
file
that
provides
configuration,
maybe
but
probably
not,
and
that's
why
there's
a
whole
second
there's
a
second
addition
to
what
you're
doing
like
right
now:
you're
doing
crud
and
you're
just
trying
to
get
stuff
into
a
table
and
provide
simple
stuff
as
soon
as
you're,
successful
and
you're.
C
Next
time,
you're
demoing
you're,
showing
the
ability
to
upload
and
delete
you
know
create
up
upload,
delete,
yeah,
we'll
go
through
and
design
collect.
You
know,
collaboratively
we'll
design.
C
Another
interface
will
squish
the
table
into
its
own
section,
but
we'll
design
another
interface
in
which
people
will
be
able
to
use
your
ui
to
configure
the
filters,
and
so
instead
of
people,
editing
yaml,
they
will
edit
the
configuration
of
the
filter
and
so
it'll
be
pretty
neat.
When
we
get
there
but
okay
yeah
does
anyone
else,
have
feedback
for
or
questions
for.
C
C
Cool
so
hopefully
everyone
saw
the
link
to
the
concepts
that
garage
is
working
on.
I
will
post
it
again
in
the
chat
just
so
that
everybody
can
digest
it
a
little
bit.
It's
really
really
interesting.
It's
actually.
It's
actually
somewhat
slightly,
futuristic
and
extremely
interesting
as
a
matter
of
fact
like
if
it.
If
this
type
of
functionality
grows
in
popularity,
then
people
will
be
writing
applications
differently.
C
C
A
A
Yeah
yeah,
nice,
okay,
so
last
time
I
display,
like
demo
the
same
command,
and
I'm
doing
it
again
just
for
some
context.
This
command
is
to
list
out
either
the
performance
profiles
for
performance
results,
or
we
can
list
out
all
the
results
concerning
a
performance
profile
like
yeah
like
right.
Now,
I'm
going
to
list
all
the
performance
profiles
stored
in
the
data.
A
A
A
Setup,
do
we
have
a
json
output
for
this
yeah?
We
have
both
yaml
and
json
output,
supported
for
both
both
the
commands,
results
and
performance
profiles.
Should
I
display
it
yeah.
Let
me
share
that.
E
A
All
the
results
are
like
in
this
time
we
are
actually
showing
them
of
almost
25
results,
that's
the
limit
and
we
are
printing,
all
of
them
and
similarly,
with
json.
F
So
this
particular
json
or
yaml
file
like
do
do
you
have
the
it
only
shows
us
the
first
only
for
the
first
page
itself
or
do
you
have
the
option
to
query
like
a
page
to
page
two.
A
We
don't
have
the
to
query
the
page
too.
We
are
just
like
printing
the
maximum
limit
that
is
25
right
now.
F
A
Can
you
can
you
say
it
again?
I
couldn't
understand
what
you're
trying
to
say.
F
Like
the
query,
based
on
number
of
pages
and
like
the
number
of
one
pages,
was
developed
by
keeping
your
ui
and
the
browser
ui
so
like
maybe
some
changes
might
help,
I
don't
know
to
query
all
of
them
or
is
it
not
at
all.
A
Like
if,
in
this
format,
thing
now.
F
A
Yeah,
I
think
we
can
add
a
separate
one
more
like
page
thing,
because
I'm
using
one
page
variable
that
might
help
us.
F
A
F
Yeah
that
that
is
all
good,
but
what
my
problem
was.
When
you
did
the
json
thing
right,
I
it
was
not
getting
all
of
the
crystals
or
it.
F
A
F
A
Yeah
drew
like
if
the
user
has
a
lot
of
profiles,
then
that
will
be
because
we
need
to
get
all
those
details
from
the
server.
So
that
is
one
issue
for
a
non-paginated
approach.
A
C
Opinions,
no,
I
just
I
just
wanted
to
just
want
to
be
really
clear
and
pull
my
microphone
really
close.
No,
I'm
saying
that
looking
groove
right
in
the
eyes
just
get
drew.
Can
you
no
all
right?
Okay,
this
is
well
so
push.
Has
this
hard
challenge
of
like
of
designing
a
new
user
experience
new
in
that.
C
Meaning
that
it's
not
that
we
have
been
this
challenge
has
been
behind
the
scenes
for
a
little
while
and
he's
trying
to
make
it
right
he's
trying
to
bring
forth
an
experience
in
which
he
had
he's
gaining
a
lot
of
data
and
like
trying
to
parse
through
a
human
trying
to
parse
through
that
data
on
a
cli
is
not
really
what
I
don't
know.
You
know
it's
more
difficult
from
the
cli.
Let
me
toss
out
a
few
examples
of
prior
art
in
the
space,
so
I'll
call
upon
the
docker,
cli
and.
C
C
I'm
curious
if
you
mute
for
a
second.
C
Second,
okay,
thanks
feel
free
to
unmute
anytime,
that
you
want
to
say
something:
it
was
there's
just
some
white
noise
going
on
in
the
background,
and
so
all
right.
Well
so
so
part
of
the
guiding
principles
of
design
for
mastery
ctl
is
well.
It
focuses
around
user
experience.
What
is
measure
ctl?
It's
a
user
experience,
it's
a
client,
it's
a
client,
just
like
maestro
ui
is
dhruv
has
heard
me
say
this
more
times
than
he
cares
to
and
as
such
everyone's
opinion
on,
this
call
is
valid.
Everyone
is
a
user.
C
C
C
The
fact
you
have
to
take
a
user
on
a
journey
you
have
to
let
them
evolve
like
hey.
Why
are
we
still
using
these
stupid
mice
like
why?
Why
are
we
not
just
using
training
our
eyes
on
the
screen?
Why
is
it
not
looking
at
my
pupils
and
figuring
out
what
I'm
looking
at,
like
you
know
like
eventually,
we'll
have
different
inputs
and
all
these
things,
but
right
now
we
still
have
clies
right
from
the
1950s
or
whatever
all
right,
fine.
C
What
does
the
1950s
have
to
do
with
today?
Well,
a
lot
as
it
turns
out,
unfortunately,
now
some
of
our
very
relevant
clis
in
the
space
that
others
have
used
that
you
all
have
probably
used
if
you're
not
using
a
couple
of
these
cli's,
then
I
recommend
that
you
do
particularly
if
you're
going
to
hang
around
on
this
community
it'll
just
benefit
you
in
general.
Here's
three
top
of
mind:
one
is
cube
ctl
and
by
the
way,
it's
notably
not
cube,
cuddle.
It's
cube,
ctl.
We
don't.
We
don't
hug
our
infrastructure
around
here.
C
We
control
it
so
just
for
the
record,
obviously
measuring
ctl
but
cube
ctl
docker
as
a
cli.
What
is
docker,
why
did
docker
become
so
popular?
Well,
you
could
say
a
couple
different
reasons
concisely,
but
one
of
them
is
just
ux.
They
brought
a
ux
to
containers,
they
made
them
portable
as
well.
So,
notably,
what
is
docker
cli?
C
It's
another
client,
it's
another
ux,
it's
so
looking
at
and
understanding
the
docker
cli
great
place
to
soak
it
in
all
of
you
are
going
to
use,
cube,
cto
and
all
of
your
classmates,
whether
they
like
it
or
not.
They
just
don't
know
it
yet
at
some
point,
they'll
touch
it.
C
C
C
If
you
aren't
there
yet
I
know
a
number
of
you
are,
but
if
you
aren't
there
then
go
try
it
out.
So
why
am
I
going
pontificating
upon
this
other
stuff
instead
of
letting
soham
demonstrate
some
really
interesting?
Things
is
because
the
user
experience
is
super
critical.
It's
in
part,
it's
why
people
will
use
meshri
in
the
first
place.
C
What
do
you
mean
by
that?
Well,
like
people
can
go
deploy
istio
if
they
want
to
go
ahead,
it's
got
a
user
experience,
it's
got
a
cli,
so
why
would
they
use
mesh?
We
see,
you
know
mysteries,
well
we're
going
to
bring
some
stuff
on
top
some
stuff
that
you
can't
get
anywhere
else
and
we
are
doing
that.
C
But
we're
also
going
to
do
at
the
same
time
is:
hopefully
have
people
fall
in
love
with
what
you're
doing
and
the
way
that
they
do,
that
is
by
having
it
be
delightful,
feel
good,
they're
going
to
tell
other
people
about
it.
Oh
this
is
nice.
It
was
so
easy
like
it.
Okay,
was
it
easy
to
pagnate?
Through
pages
walls
of
json
in
my
cli,
no,
it
wasn't
did
I
enjoy
that?
No,
I
didn't
as
much
as
I
enjoyed
getting
my
you
know
my
vaccine
shot.
C
Well,
what
would
I
like
better?
Well?
Maybe
an
interactive
set
of
prompts
to
be
able
to
paginate
through
things
yeah,
maybe,
but
even
then,
I
probably
don't
want
to
be
looking
at
json,
like
I
really
like,
I
I
I
feel
like
my
nerdiest
just
sitting
here,
looking
at,
what's
being
shared
on
the
screen
right,
all
right,
so
long
story
short
the
reason
that
we
don't
so
so
this
all
that
all
still
is
more
of
like
a
yes
to
what
drew
is
saying.
C
The
reason
it's
probably
a
no
is
that
drew,
is
headed
toward
the
right
place
and
he's
highlighting
some
pain
that
he's
just
watching
and
seeing
what's
going
on
here.
So
there's
a
few
different
and
by
the
way
a
lot
of
this
is
already
written
down
on
the
issue,
a
lot
of
it's
written
down
the
acceptance
criteria
of
the
issue-
and
a
lot
of
this
has
already
been
described
specific
to
answer
some
of
these
questions.
So
we
should
be
showing
the
user
when
they
list
out.
There's
there's
two
constructs
that
they're
working
with
here.
C
C
C
Users.
Here
need
to
be
able
to
look
over
their
profile.
The
performance
profiles
that
that
they
want
to
run
these
tests
against.
They
also
need
to
look
at
any
historical
test
results
results
of
tests
that
they
run
in
context
of
those
profiles.
Someone
only
executes
a
test
in
context
of
a
profile.
So
if
you
think
about
the
hierarchy
of
these
constructs,
there's
a
profile
and
and
subsequent
to
that
there
are
tests.
C
Hence
that's
why
push
has
the
cli
design
in
in
this
way?
So,
if
he's
doing
mastery,
ctl
perf
list
right
now,
what's
he
what's
he
talking
about
the
profiles
or
the
test
results?
Well,
this
may
need
some
revisiting
because
to
me
it
looks
like
he's
talking
about
profiles,
because
he
didn't
specify
a
particular
result.
What
it
looks
like
to
me
he's
specifying
is
a
specific
profile
id
which
needs
to
go
away.
We
need
to
be
first
supporting
people
with
display
with
names
that
are
human,
readable
names.
C
C
To
drew's
answer
is
something
like
a
search
so
the
reason
I
called
out
gh
docker
and
cube
ctl.
These
are
good
examples
of
different
clis
that
have
done.
They
have
lots
of
data
as
well.
They
have
lots
and
lots
of
containers,
you're,
probably
managing
lots
and
lots
of
containers
with
docker
lots
and
lots
of
containers
and
nodes
with
cube
ctl.
They
have
the
same
issue
right
now.
How
are
they
dealing
with
it?
How
many
open
issues
do
we
have
in
measuring?
C
Pr,
that's
really
hard
to
say:
that's
a
if
you
wanna
use
gh
to
iterate
through
those
and
look
at
them,
we're
talking
about
pages
and
pages
and
pages
of
pr's
to
look
through.
Well.
How
do
you
navigate
that
with
gh?
C
Well,
you
don't
navigate
through
page.
You
don't
go
you
don't
say
yeah.
I
think
it's
on
page
2807,
that's
probably
the
page
that
I'm
going
to
find
the
interesting
issue
on
no
you're
going
to
search
you're
going
to
use
some
metadata
or
some
interesting
thing
like
well.
I'm
looking
for
a,
I
know
it
probably
like
started.
I
had
something
to
do
with
bash
because
I'm
having
an
issue
with
bash,
I'm
going
to
search
for
bash.
Let
me
do
a
measuring
ctl,
perf
search,
soak
test
and
whatever
matches
soap
test.
C
Great
now
I'll
see
a
set
of
those
results,
and
then
I
want
to
dig
into
that
particular
one.
We're
probably
not
ready
to.
I
mean
we're
ready
to
take
the
cli
to
a
search,
and
it's
not
too
difficult.
There
are
other
things
we
probably
need
to
get
to.
Like
that's
a
good
road
map
item,
we
should
probably
get
to
it
for
now
providing
people
the
like,
not
letting
people
hammer
on
a
remote
provider
for
like
thousands
of
test
results
and
showing
them
a
massive
wall
of
text
that
their
terminal
buffer
couldn't
handle.
C
Anyway,
it's
like
oh
yeah.
We
should
be
capping
it
at
25,
letting
them
work
through
that
they
can't
read
even
one
of
them
anyway,
if
you
just
output,
just
one
of
these
test
results,
you
couldn't
read
it
and
so
yeah
issues
yeah,
not
pr's,
I
mean
yeah,
so
it's
just
a
little
bit
confusing
like
so.
There
have
been
three
thousand
issues
and
prs
in
total.
C
Put
on
merceri,
so
long
story
short
is,
is
difficult.
You've
had
to
explore
a
couple
of
different
ways
of
trying
to
provide
an
interactive
pagnation
like
that
type
of
an
experience
in
the
ui
and
ultimately
like
one
of
the
the
stronger
answers
is
a
search.
Yes,
we
can
do
interactive
pagination
with
like
letting
the
people
use
their
arrow
keys
to
cursor
over
things
and
choose
what
page,
and
you
know
that's
helpful
for
certain
situations.
C
I
was
giving
examples
as
to
why
it's
generally
not
as
helpful
as
something
like
a
search,
but
the
other
reason
why
and
when
adida
brings
this
up
in
a
little
bit
that
we'll
talk
about
interactive
prompts
in
the
ui
is
like
they're,
very
pleasant
for
a
human
and
the
cli
is
meant
for
human.
C
Unfortunately,
must
much
of
us
in
the
industry,
all
of
you
as
well
ignore
the
fact
that
clis
are
made
for
humans
and
we
script
over
them
as
if
they're
an
api,
so
interactive.
C
Cli
experiences
need
to
be
designed
with
that
caveat
in
mind
that
there
needs
to
be
an
alternative
programmatic
approach
so
anyway
drew
got
me
all
riled
up
like
normal,
and
so
like
I'd.
Think
of
anything
that
I,
that,
from
my
my
feedback
and
and
and
droves,
is
equally
as
meaningful,
is,
I
think,
you're
there
or
I
think
like
like
of
the
70-something
comments
you
have
on
that
pr.
It's
probably
time
that
we
merge
that
thing.
I
think
that
you're
there.
C
I
think
that
there
is
something
to
consider
around
the
ability
to
page
through
things
and
it
doesn't
have
to
be
searched
like
I
was
suggesting
there
are
other
solutions
for
that:
the
documentation,
the
how
how
in
your
design,
spec
ph
you've,
accounted
for
the
experiences
that
I
was
mentioning
from
docker
and
gh
and
things
that's
really
good.
We
don't
have
to
copy
exactly
from
there,
but
this
leads
us
back
to
what
I
was
saying
initially
about
a
cli
being
an
experience
and
what
and
what
makes
a
quality
experience.
C
And
I
was
talking
about
consistency,
but
it
consistency.
But
I
was
also
talking
about
familiarity.
So
if
you
do
something
that
people
are
also
already
familiar
with
it's
helpful
because
they
you
can
change
tweak
it,
you
can
improve
it,
but
it's
helpful
because
they
sort
of
have
a
starting
point
and
so.
C
So
the
summary
is,
like
I
think,
you're
there.
We
probably
aren't
the
reason
that
we
would
ever
be
really
interested
in
the
use
case
where
someone
is
outputting
more
than
one
test
result
at
a
time.
E
A
F
Currently,
there
are
really
less
grass
fluid
graphical
endpoints,
which
you
can
see
in
between
over
here,
which
are
currently
being
deployed,
but
we
are
planning
on
creating
many
more,
so
the
people
can
use
them
if
they
want
to
right
and
as
a
part
of
them.
Currently,
we
have
given
priority
to
some
of
them,
which
we
want
to
start
with,
and
we
are
going
to
start
with
them
pretty
soon.
F
Any
of
all
of
you
are
invited
to
look
at
this
graph
and
give
your
opinions
on
how
we
are
documenting
or
naming
our
endpoints
and
whether
they
should
be
there
or
not,
and
actually
this
refactoring
is
a
two-part
thing.
The
first
part
is
refracting,
the
uis
changing
their
nomination
properly
and
stuff
like
that,
and
second
is
documenting
this
particular
endpoints
correctly.
F
So
to
the
graphql
point,
we
have
arrived
somewhere
where
we
are
automatically
generating
the
graphql
docs
using
the
schema
itself
schema
and
the
comments
in
this
keyboard.
All
together
framework
is
all
based
on
ruby
through
which
our
graphql
works,
so
to
give
you
a
gist
of
how
it
works.
This
is
the
current
schema
which
you
have
now.
It
takes
the
comments
which
you
have
for
your
subscriptions
and
use
that
particular
thing
to
make
dots
and
it
uses
a
template
which
I
have
created
earlier.
F
It
is
a
html
file
for
the
people
who
have
worked
with,
I
guess
in
django
if
they
have
certain
templates
in
which
you
can
fill
in
stuff.
If,
if
you
want
correctly
so
basically,
there
are
some
functions
which
I
have
defined
in
ruby,
which
I
can
run
it
from
over
here
and
it
will
automatically
fill
those
kind
of
spaces.
F
So
you
can
think
of
this
as
a
custom
mesh
res
compiler
for
craft
build
logs.
So
it
is
pretty
simple
from
the
user's
point
of
view.
There
is
a
simple
rake
command
which
you
can
run,
which
you
have
to
run
inside
talks.
F
Yeah
and
this
particular
rake
command
will
get
the
input
from
the
schema.graphql
inside
your
backend
server
itself
and
create
a
md
file,
create
a
md
file.
Md
file
is
basically
a
static
file
which
j
which
our
doc
file
uses
to
create
a
particular
page.
In
this
case,
it
was
a
graphql
reference
argument,
so
use
that
particular
template
to
create
this
md
file,
which
is
basically
a
static
page,
showing
all
the
documentation
for
the
graphql
schema,
and
it
is
currently
up
in
live
in
our
docs
itself.
So
if
I
go
to.
C
F
Is
our
automatically
generated
documentation
now,
if
any
of
you
guys
are
even
slightly
familiar
familiar
with
graphql
and
have
documentation
about
it,
it
would
be
greatly
appreciated
if
you
can
go
through
it
and
give
us
a
first
review
or
something
like
that,
because
I
believe
I
would
have
made
certain
mistakes
while
writing
the
comments
in
the
graphql
and
so
once
perfect.
So
you
can
look
through
it
and
give
us
some
first
review.
That
will
be
very
great.
F
We
can
update
the
comments
on
that
particular
thing,
and
even
if
you
don't
like
the
format,
then
also
you
can
ping
me
up,
we
can
see
if
we
can
do
anything
about
it
and
if
you
are
very
good
at
graphql,
you
can
help
us
in
this
particular
gigantic
amount
of
refactoring
and
adding
new
graphql
link
points
in
meshi.
F
If
you
are
interested
in
that,
you
can
ping
me
too.
Currently,
gaurav
has
is
going
to
help
us
as
shown
his
hand,
to
help
caught.
F
A
F
D
A
Thank
you
through,
since
we
are
running
out
of
time,
let's
move
on
to
soham
for
a
demo
on
meshmap
designer,
so
all.
G
Okay,
so
hey
guys,
so
this
is
mesh
map
designer
it's
one
of
the
features
that
is
literally
unique
to
us
and
there
are
two
components
to
mesh
mob
designer.
There
is
a
designer
view
which
you
are
currently
looking
at
and
there
is
a
visualization.
C
So
I'm
so
sorry
for
interrupting.
I
am
I'm
only
doing
that
because
just
an
in
order
of
sequence
of
like
of
this
functionality
or
is
it
okay.
C
H
H
So
let
me
share
a
pattern
right
quickly.
I
hope
pattern
is
visible,
so
just
to
give
a
brief
context
on
what
the
work
is.
I've
been
recently
working
on
adding
a
new
type
of
functionality
to
patents.
That
is
now.
You
should
be
able
to
deploy
applications
using
just
mystery.
So
what
here
this
particular
pattern
file
is
doing.
Is
I'm
trying
to
deploy
up
an
application
which
is
essentially
a
docker
container
and
I'm
trying
to
deploy
it
using
just
mission,
no
other
tools.
H
So
basically
that's
a
that's
a
short
con
context,
other
things
that
would
be
soon
added
to
this
thing,
which
are
not
yet
present,
although
they
are
kind
of
present
in
the
code,
I
will
be
demoing
that,
because
I'm
not
sure
about
it
here,
so
what
you
would
be
able
to
do
is
you'll
be
able
to
specify
the
deployment
strategy.
That
is,
you
can
specify
if
you
want
to
go
ahead
and
do
calorie
or
grouping
that's
two
things
that
I'm
trying
to
cover
for
now.
H
So
I'm
trying
to
apply
a
new
pattern
on
this
new
rollout
demo..
This
is
the
file.
I'm
saying
that
I
want
this
application
to
be
deployed
which
his
name
is
going
to
be.
C
H
And
I'm
also
trying
to
provision
is
the
image
not
that
it's
required,
I'm
just
trying
to
show
that
depends
on
kind
of
works.
So
now
I
have
hit
the
I
have
hit
enter.
This
particular
process
takes
a
bit
of
time
because
it
tries
to
provision
istio
mesh
and
tries
to
provision
svc.
This
could
have
been
executed
confidently,
but
they
will
be
executed
sequentially
because
I've
explicitly
stated
a
dependency
in
here.
H
H
A
service
also
so
yeah.
Basically,
it
has
completed
the
task
that
we
assigned
it
and
that
was
to
declare
an
application
and
also
let
you
access
it
using
low
bands.
The
low
balancer
is
kind
of
the
default
right
now,
although
I
do
plan
to
add
the
settings
in
there.
So
there
is
a
setting
speed.
H
You
should
be
able
to
specify
increase
in
here,
and
that
is
a
and
measuring
would
automatically
detect
that
what
ingress
controller
you
are
using
if
it's,
if
you're
using
the
default,
is
the
gateway
or
you're
using
nginx
ingress.
That's
the
tool
I
have
in
my
head,
I'm
not
sure
if
aws
or
something
ingress
I'd
be
adding
support
for
for
rapper
right
now,
but
this
is
the
first
iteration
and
first
iteration
is
going
to
cover
these
kind
of
features,
and
this
is
still
working
progress.
That's
a.
A
That's
that's
nice,
like
queries,
thought
suggestions
on
this.
G
Of
course,
we
can
also
upload
our
pattern:
files
on
michigan
ui
right.
I
One
thing
is
the
ingredients
you
talked
about,
which
you
could
able
to
do
it,
so
that
is
mostly
like
you
know,
into
let's
say
I'm
using
sto,
so
it
will
create
the
ingredients
for
sdo
using
stew.
H
Yeah
I
mean
was
that
yeah.
If
the
question
was
that
if
you
are
using
istio
ingress
gateway,
which
would
I
think
I
mean
it
does
come
with
a
default
profile?
So
if
you,
if
this,
if
messy,
would
figure
out
that
you
have
a
single
gateway
present
in
your
cluster,
it
will
it
will.
H
It
will
actually
configure
this
if
this
gateway,
so
that
you
can
access
your
application,
leveraging
the
english
gateway
that
you
have
right
now,
it
assigned
the
load
balancer,
because
because
this
particular
thing
hasn't
been
it's
it's
not
completed
but
yeah.
Ideally
that's
what
and
that's
what
is
going
to
happen,
but
it's
going
to
configure
your
english
gateway
for
you,
so
that
you
cannot
access
your
application
using
that
you
would
be
able
to
configure
that
thing,
not
not
to
its
fully
full
extent.
H
You
would
be
able
to
configure
things
like
get
it
make
it
accessible
on
this
particular
host
these
kind
of
things,
exposing
the
entire
configuration
for
interest
controller
would
be
a
bit
difficult
because
all
of
the
english
controller
kind
of
exposed
different
kind
of
functionality.
So
the
goal
would
be
to
expose
as
many
functionality
as
we
can,
while
keeping
same
defaults.
I
Okay,
so
that
is
what
I
was
asking.
Let's
say:
I'm
on
azure,
I'm
using
a
guest.
I
have
apart
from
service
message
as
your
comes
with
out
of
the
box
solution
for,
let's
say,
front
door
or,
let's
say
application,
gateway,
ingress
and
a
lot
of
things.
So
how
soon
my
message
will
understand
which
ingress
I
want
to
expose.
H
Okay,
yeah,
actually,
if
it
detects
more
than
and
actually
even
for
this,
actually
it's
going
to
right
now,
it's
going
to
make
you
make
choices
for
you
for
service
mesh,
so,
for
example,
it
detects
more
than
one
service
mesh.
It's
going
to
actually
use
sd
in
that
case
and
same
is
going
to
be
for
ingress
that
it's
going
to
make
a
choice
for
you.
What
the
default
is.
That's
what
I
have
put
like.
That's
not!
H
I
have
hardwood
yet,
but
the
thing
is
that
just
like
now,
you
all
have
a
choice
to
specify
that
go
ahead
and
use
istio
mesh
you
can.
This
is
a
this
is
a
very
valid
feed.
You
can
ask
mystery
to
specifically
go
ahead
and
configure
sd
or
use
istio
to
do
this
deployment.
You
would
also
have
the
have
the
ability
to
basically
force
mystery
to
use
a
particular
english
controller
that
you
want
to.
I
H
Yeah,
actually,
the
thing
is
that
this
mesh
thing
this
is,
I
was
trying
to
emphasize
that
this
mesh
thing
is
this
would
be
used.
So,
for
example,
this
would
be
used
to
create
virtual
sources
for
you.
But
what,
if
you
don't
have
this
theory,
it
would
use
linkedin.
This
is
not
exactly
related
to
english
controller
configuration
for
in
for
configuring
english
control.
You
would
have
some
other
field
like
it
would
be
here.
It
would
be
in
trades,
but
you
would
have
some
other
feed
like
you
want
to
use.
H
Let's
say
engine
exchange
controller,
so
you
would
have.
You
would
be
able
to
tell
me
that
do
not
make
choice
for
me
or
do
not
do
anything
on
your
own
until
you
go
ahead
and
configure
engine
engineering,
service,
controller.
I
C
Maybe
yeah
you
don't
only
have
six
minutes
left
so
samir
if
I'm
gonna
I'll,
follow
up
and
we'll
just
make
sure
that
that
that
and
I
suspect
that
only
partially
answers
your
question.
So
we
will
just
chat
you
up
more
to
make
sure
it
does
and
and
to
make
to
make
sure
it's
what
you
need
as
well,
because
so.
A
Awesome
like
let's,
let's
move
on
to
the
next
agenda,
I
guess,
since
we
are
running
out
of
time,
so
do
you
want
to
continue
your
demo.
G
Yeah
yeah
yeah
I'll
be
just
sharing
my
screen
again.
Is
it
principal?
Can
someone
confirm
yeah,
it's
physical,
so
yeah?
I
was
talking
about
us
having
two
components.
Currently,
one
of
them
is
the
visualizer
view
where
you
can
actually
visualize
your
deployed
service
meshes
and
the
one
that
I
would
be
explaining
today
is
the
designer
view
so
like
on
the
left
side,
we
have.
G
What
we
call
a
designer
pane
and
in
the
center
is
what
we
call
the
canvas
now,
let's
fall
back
to
what
was
explaining
was
talking
about
patent
files,
and
you
can
also
upload
your
pattern
files
on
the
meshi
ui.
So
suppose
you
have
uploaded
a
pattern
file
over
here.
The
pattern
file
that
you
just
uploaded
over
here.
You
can
just
see
that
the
pattern
files
are
also
fetched
over
here
and
as
soon
as
I
click
on
any
pattern
file.
G
The
designer
view
does
a
lot
of
things
like
it
reads:
the
pattern
file
it
converts
it
into
a
cytoscape
json.
It
reads
all
the
elements
inside
the
pardon
file
like
the
pods,
the
services,
whatever
you
have
written
in
the
part
file,
and
it
actually
visualizes
them
inside
the
canvas
now,
so
you
can
visually
see
what
the
button
file
exactly
looks
like,
but
that's
not
why
we
call
it
the
designer
actually
yeah.
So
as
soon
as
you
load
a
pattern
file,
you
can
do
a
lot
of
stuff
with
it.
G
So
on
the
left.
Side
of
the
first
tab
is
where
there
are
a
lot
of
components
listed
so
now,
for
example,
these
are
service
mesh
components.
For
example,
we
have
service,
we
have
deployment
now.
What
we
can
do
with
them
is
what
I
can
do
with
them
is
I
can
just
click
and
drag
and
drop
these
components
inside
the
canvas
I
can
make
edges
from
one
service
to
the
other
service
I
can
zoom
out.
I
can
zoom
in,
I
can
fit
the
service
I
can
fit
inside
the
view.
G
I
can
take
a
screenshot
of
it.
I
can
maybe,
if
I
want
to,
I
can
group
them.
I
can
move
the
group
around.
I
can
make
edges
from
a
group
to
a
service.
G
I
can
also-
maybe,
if
I'm
not
happy
with
like
I
don't
if
I
don't
want
this
component
to
be
here,
I
can
just
delete
it
or
maybe,
if
I
did
it
as
a
mistake,
I
can
just
undo
it
redo
it
and
do
a
lot
of
stuff
like
that
is
why
we
call
it
the
designer
now
it
does
not
end
there.
G
Now
suppose
you
have
edited
your
cameras,
the
way
you
wanted
suppose
you
have
the
desired
design
that
you
want
now,
if
you
look
look
right
over
here
right
now
what
it
says
that
that
you
have
updated
your
cameras
and
it
gives
you
an
option
to
save
it.
G
So
what
it
does
is,
once
you
click
save
over
here
and
give
a
name
to
this
design,
or
this
pattern
for
this
design
that
you
want
to
save
what
the
designer
view
does
is
the
designer
you
convert
this
design
into
a
pattern
file
and
then
saves
it
over
the
patterns
that,
like
the
ones
that
you
saw
earlier,
but
unfortunately,
since
the
machine
cloud
has
not
been
updated,
I
guess,
like
I
talked
to
it
was
yesterday
and
since
it's
not
been
updated,
I
won't
be
able
to
convert
this
into
a
pattern
file
right
now,
but
yeah
the
desired
functionality
which
has
been
implemented
is
that
I
can
just
save
this
as
a
patent
file.
C
It's
what
what
is
this
may
this
isn't
christmas
time?
What
the
is
this
this
is
like
a
christmas
come
early.
This
is
very,
very
early
functionality
super
compelling
from
my
perspective
and
samir
you're
grocking
it
spot
on
that.
If
it
was
intended
to
do
the
the
exact
same
things
as
keali,
it
would
be
like
well,
it
would
be
really
difficult
to
justify
writing
yet
another
kiali-
or
you
know
like,
but
there's
a
couple
of
things
that
are
significantly
different
here,
some
only
some
of
which
are
apparent
at
the
moment.
C
It'll
become
a
lot
more
apparent
just
how
dramatically
different
this
is
when
you
I'll,
I
won't
spoil
some
of
those
but
we'll
say
two
significantly
different
capabilities.
That
should
be
fairly
obvious
at
this
point
are
this:
is
for
every
mesh,
not
istio
or
not
just
istio.
C
C
G
Yeah
yeah
you
can
like
recently
utkash
made
a
pr
from
which
you
can
basically
fetch
pardon
files
from
any
remote
location.
You
just
need
to
give
the
measuring
ui
the
url
to
like
it's,
it's
not
yet
implemented
in
measure
ui,
but
it's
implemented
in
memory
ctl.
I
suppose,
and
you
can
just
give
the
url
of
the
where
the
file
is
located
and
also
flag.
If
you
want
to
save
the
file
or
not
and
jack,
would
basically
fetch
the
file
and
save
it
as
a
pattern
file.
C
Most
certainly
there's
another
component
of
that
to
to
restate
that
it
is
that
not
only
will
does
meshri
allow
you
to
and
import
from
any
generic
http
endpoint,
but
mastery
also
will
interface
with
github's
api,
so
it'll,
oh,
if
you
owe
auth
to
messerie
through
with
your
github
account
eventually
you'll,
be
able
to
give
messary
permission
to
browse
through
your
repo
find
anywhere
where
you
have
kubernetes
manifest,
or
these
service
mesh
pattern
files
it'll
import,
those
you
can
then
pull
them
up
like
soham
was
just
showing
here
in
the
visual
designer
work
on
them,
iterate
on
them.
C
When
you're
done
making
some
changes,
perhaps
you
don't
want
to
apply
those
right
away?
Perhaps
the
most
appropriate
thing
is
to
submit
a
pr
on
those
so
measuring.
Will
our
mesh
map
will
facilitate
that
as
well?
Ultimately,
when
you
go
to
click,
save
it
will
also
say:
did
you
want
to
generate
a
pr?
You
know
if
this
one
is
based?
I
And
all
these
graphics
are
this
generic
for
any
of
the
service
or
maybe
because
this
is
the
same
kind
of
graphics,
which
cali
uses.
C
Yeah
this
is
we
are,
I
almost
would.
I
would
almost
not
comment
on
what
this
will
be
like.
This
is
extremely
early
like
there
are
thoughts
given
to
a
letting
the
user
upload
their
own
graphics
for
to
represent
whatever
that
service
is
be
yeah,
ideally
not
using.
You
know,
I
don't
actually
smear
your
input.
There
might
be
very
helpful
because
we're
on
the
yeah,
because.
C
C
None
of
that
going
on
for
sure,
there's
no
way
in
hell.
Anyone
could
ever
copyright
a
triangle
anyway,
but
but
even
at
that,
like
no,
moreover,
like,
why
is
a
service,
a
triangle?
What
is
that
to
say
if
it
makes
sense?
No,
probably
not
now
we'll
probably
end
up
with
some
polygons
with
some
of
those
types
of
shapes
just
just
because
you
have
to
show
something,
but
but
it's
food
for
thought.
C
We
might
even
make
that
like
a
user's
choice
where,
if
they
want
to
use
the
a
suite
of
polygons
to
represent
something
or
their
own
custom
icons
or
something
more
like
an
icon
like.
C
C
But
yeah
no,
I
was,
we
were
I'll,
show
you
some
screenshots
of
some
of
the
things
that
my
teams
have
done
before.
Keali
came
along
and
we've
got
a
long-standing
relationship
with
those
guys,
the
folks
behind
kiali
before
they
started
the
project
but
yeah,
but
you're
right.
I'm
samir
people
will
most
readily
compare
this
to
kiali,
because
kali
is
really
the
only
significant
I
mean
he
had
the
only
significant
open
source.
I
I
C
C
One
of
these
is
not
only
because
we
want
to
accomplish
some
significantly
different
things
like
dry
run
is
one
of
those
one
of
those
is
around
like
performance
testing,
which
is
you
want
to
do
dry,
runs
or
do
exam
like
what
happens
when
you
tweak
your
configuration,
and
does
your
performance
go
up
or
down,
and
you
might
want
to
like
measure,
is
built
to
facilitate
those
very
types
of
experiences
or
and
help
you
optimize
toward
that
there's.
Another
aspect
of
which
keali
is
was
fairly
heavily
centered
around
prometheus
and
it
should
be.
C
You
need
to
get
tons
of
metrics
from
prometheus
about
what's
going
on,
but
it
is
inherently
set
up
to
as
a
polling
architecture
which
sometimes
you
got
to
do
that.
Basing
your
entire
architecture
on
that?
C
It's
not
what
I
would
do
and
that's
not
what
we're
doing
this
is
event
driven
by
and
large,
that
and
that's
what
mesh
sync
is
about
like
mesh
sync
is
pumping
info
in
into
so
next
time
we
give
a
demo
of
mesh
map.
Maybe
a
good
thing
to
do
is
like
this
is
the
designer
view.
Maybe
we
should
take
a
look
at
the
visualizer
view
and
watch
as
like
the
visualizer
view
is
more
akin
to
kiali,
where
you
know
it's
watching
the
active
deployment,
and
so
we
should
see.
C
Yeah,
do
you
mean
like,
like
it's
saying,
I
think,
there's
two
ways
in
which
that
question
could
be
interpreted.
One
is
that
mesh
sync
itself
is,
you
know
designed
to
be
focused
on
things
concerning
service
meshes,
but
things
that
you
know
objects
and
kubernetes
that
concern
service
meshes,
which
actually
means
everything
eventually
like
stateful
sets
and
volumes
and
other
things
you
might
want
to
visually
represent
like
at
some
point.
C
If
you
expand
on
the
designer
here,
it's
like
well,
my
application
uses
jobs
and
other
you
know
is
that
what
you
meant,
or
did
you
mean
a
volume
to
persist?
Your
configurations,
you.
C
C
A
Shooting
yeah
yeah
yeah,
I
guess
like
we
are
out
of
time.
Like
always,
I
guess
so.
Thank
you
everybody.
This
was.
This
was
a
good
meeting
and
we,
I
think
everyone
got
a
good
feedback
on
their
works
and
yeah.
I
guess
I'll
meet
some
of
you
tomorrow
on
the
machery
ctl
bug
hunting
session,
which
is
right
after
the
newcomers
meeting
or
we
will
meet
friday
on
the
community
card,
and
if
you
are
a
newcomer,
we
also
have
a
newcomers
call
tomorrow
so
be
sure
to
check
out
them.