►
From YouTube: Meshery Development Meeting March 10th, 2021 HD 1080p
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
Okay,
great
well,
hey
we're
we're
four!
After
and
there's
a
couple
of
people
that
we'll
be
joining,
it
looks
like
they'll,
be
doing
it
late,
so
we'll
have
to
make
them
feel
awkward
when
they
join
late.
That's
just
the
punishment,
that's
just
the
natural
natural
consequence
so
welcome
everybody.
Is
anyone
new
on
the
call
for
the
first
time
today
like
for
the
first
time
on
this
particular
call,
considering
that
we
have
quite
a
few
of
them
we'd
love
to
welcome
you.
If
so,.
A
A
Nice
beautiful
beautiful,
well!
Well,
it
starts
with
you
doing
what
you
just
did,
which
is,
which
is
great,
so
hey
welcome
to
welcome
to
the
community.
Welcome
to
the
the
mesri
development
call,
just
very
briefly
for
you
and
I
know
well,
and
I'm
going
to
pause
for
just
a
minute
because
there's
another
newcomer
who
could
benefit
from
this
just
as
soon
as
their
audio
is
connected.
A
B
A
A
So
that
way
everybody
can
get
to
know
you
and
you
can
get
to
know
them
as
well,
so
to
josh
was
just
saying
hi
and
saying
it's
the
first
time
for
him
on
on
this
particular
call
that
he's
here
kind
of
listening
in
and
getting
a
kind
of
like
you
do,
jump
rope.
You
guys
have
you
guys
ever
do
like
the
dual
jump
rope
you
know
where
one
person's
standing
one
side,
somebody's
standing,
the
other
side
and
you're
sort
of
standing
there
getting
ready
to
jump
in
you're
kind
of
getting
the
flow.
A
A
But
that
means
that
it
includes
kubernetes
and
docker
and
rust
and
web
assembly
and
envoy
and
and
and
and
that
list
keeps
going
for
a
long
time,
and
so
it
takes
a
lot
of
you
to
for
us
to
do
meaningful
things
and,
I'm
very
pleased
to
say,
you're,
all
doing
a
lot
of
meaningful
things.
Actually,
one
of
the
projects
is
well
in
about
two
weeks
should
head
into
the
cncf
and
so
that'll
be
and
that'll
be
a
nice
milestone
for
anyone.
Who's
helped
with
service
mesh
performance.
A
D
Yeah
for
sure,
so
my
name
is
terry
parson
and
I'm
a
cs
undergraduate
in
my
senior
year,
I'm
from
hyderabad
india
and
I've
been
doing
open
source
for
about
a
year
now,
and
most
of
my
contributions
are
on
facebook's
chess,
which
is
a
dressing
framework
for
javascript,
and
my
expertise
lies
in
javascript,
react
and
a
little
bit
of
python
as
well.
But
it's
it's
my
first
time
you
know
joining
this
meeting,
so
I
thought
I'd
see.
What's
going
on
in
just
you
know,
say:
hi
nice.
A
Very
nice,
that's
great!
Okay,
there's
a!
I
don't
know
if
I
mentioned
react
or
which
I
didn't,
but
there's
some
there's
a
couple
of
different
react
projects
that
are
going
on.
Maybe
three
depends
on
how
you
count
them,
so
one
of
them
is
one
of
the
websites
is
done
in
gatsby,
which
which,
which
is
you
know,
react
based,
but
is
for
a
specific
purpose.
It's
another
one.
The
measuring
ui
itself
done
in
react.
A
A
A
You
have
to
get
introduced
around
as
well,
be
good
all
right!
Well,
hey
a
couple
items!
If
so,
if
you
don't
have
your
name
here
on
the
list,
please
slap
it
down
after
you've
been
here
for
a
week
or
so.
If
you
just
put
your
first
name
on
the
list,
then
it's
a
fair
warning
right
now
that
you
may
find
that
one
of
us
makes
up
a
last
name
for
you,
and
so
anyway
put
your
full
name
down.
A
If
you
go
there,
you'll
find
a
calendar
of
the
meetings
that
go
on.
Some
of
them
are
run
in
the
cncf.
Some
of
them
are
run
here
and
it's
kind
of
hard
to
tell
the
difference.
After
a
while,
you
can
subscribe
to
the
calendar
there.
Then
you
can
also
have
a
there's,
a
quick
reference
of
what
meetings
are
happening
when
and
what
the
topics
are
where
the
meeting
minutes
are
how
to
connect.
A
The
other
thing
I'd
mentioned
was
service
mesh
performance
smp
this
particular
this
is
the
project
I
was
mentioning
that
will
be
put
up
for
consideration
inside
of
the
cncf
in
a
couple
of
weeks,
and
it's
sort
of
long
been.
So
I
will
let
everyone
go
read
about
what
that
project
is
and
get
involved
in.
If
you
want
to
it's
a
good
time
to
get
involved
in
a
project
just
before
it
lands
in
the
cncf,
it's
long
been
stated
that
that's
the
goal
for
measuring
as
well.
This
probably
is
headed
toward
cncf.
A
E
E
E
So
this
is
a
user
facing
dock
which
basically
has
all
the
measure-residual
commands
and
its
function
and
it's
using,
but
the
current
implementation
is
not
intuitive
and
it
also
misses
some.
Some
commands
like,
for
example,
like
we
have
a
stat
command
here
and
we
again
have
a
stack
command
with
the
reset
flag
here.
E
So
this
might
be
like
a
little
bit
confusing
for
the
user,
so
we
have
set
up
a
proposal
to
change
this
block,
to
something
similar
to
the
command
metrics,
where
we
will
have
the
command
flag,
function
and
usage
so
like,
for
example,
in
the
system
command.
We
have
a
global
flag
context
so,
and
it
has
all
these
completion
start
stop
and
all
these
sub
commands
and
say
for
this
chart
command.
We
have.
A
E
D
E
Yeah,
actually,
both
like
we
will
have.
First,
we
will
have
a
doc
like
a
page
similar
to
this
one
like
a
table
structure
where
we
like
have
all
the
commands
like
like
we
have
in
this
spreadsheet.
It
will
have
the
command
flag
function,
usage,
yeah,
only
till
usage,
wheeler
command,
flag
function
and
music.
So
we
will
have
a
table
structure
like
this
and
also
we
will
have
an
individual
page
for
each
other.
So
it's
kind
of
good.
A
A
A
Nice,
okay,
let
me
let
me
add,
maybe
a
little
more
flavor
to
what
navendi
was
saying,
to
make
sure
that
this
sort
of
this
all
settles
in
for
everyone.
Everyone
understands,
oh
how
they
can
participate
where
they
can
jump
in.
So
some
of
you
that
have
been
here
for
a
while.
This
might
be
a
little
bit
of
a
repeat
but
I'll
try
to
make
it
short,
and
that
is
that
so
nibindu
has
event
has
been
in
the
community.
A
A
Own
or
have
a
significant
influence
on
and
nevindu
ended
up
spending
a
lot
of
time
toward
mastery,
cto
and
so
he's
on
a
path
very
shortly
to
becoming
a
maintainer
of
measuring
ctl,
as
such
he's
been
working
on
a
little
bit
of
high
level
roadmap,
for
it
he's
been
more
or
less.
What
he's
presenting
today
is
is
a
lot
a
lot
better
organization
around
that
client.
So
meshri,
as
a
system
has
a
number
of
components.
A
The
core
server
has
two
clients:
it
has
a
command
line,
client
and
a
ui
client.
They
both
use
the
same
meshry
server,
apis
to
interact
with
meshri
so
and
they
are
both
user
experiences.
I
think
a
lot
of
times
people
think
about.
Oh,
you
know
a
web-based
user
interface
as
like.
That's
where
the
ux
happens.
As
for
the
user
experiences,
yes,
that's
true.
Also,
there's
a
cli
client,
it's
a
client
just
like
the
the
ui
it.
It
also
has
a
user
experience.
A
It's
also
very
important,
if
you
think
about
docker
as
a
cli
and
making
containerized
things
easy.
That's
the
docker
user
experience
or,
or
was
for
many
years
so
mastery
ctl
same
thing
very
important
in
terms
of
the
user
experience
very
important
in
terms
of
exposing
all
the
capabilities
of
measuring
itself
into
into
a
cli
into
a
terminal.
A
A
lot
of
you've
been
putting
in
work
in
into
it
and
there's
some
you
know
along
the
way.
We've
we've
missed
a
few
times
on
documenting
a
command
or
two
or
a
flag
or
two
or
the
documentation
itself
like
navindi
was
pointing
out,
is,
is
not
the
most
legible
or
could
be
reorganized
a
little
bit
to
be
more
legible.
So
navindi
was
presenting
a
plan
going
forward
to
take
and
he's
got.
Links
here
was
presenting
a
plan
to
restructure
the
documentation,
such
that.
A
Not
only
is
there
a
central
reference,
a
central
command
reference,
but
also
that
there
is
an
individual
page
in
the
documentation,
one
page
for
each
command.
That
would
really
explain
the
behavior
of
the
command.
Give
a
couple
of
usage
usage.
Examples,
it
doesn't
have
to
be
very
long,
could
be
a
paragraph
on
the
page,
but
it
at
least
goes
a
little
bit
more
in
depth
than
just
a
simple
statement
of
fact
that
there
is
such
a
command
with
such
a
flag
and
here's
an
example
usage.
A
It
should
take
it
just
a
little
bit
further
and
so
there's
a
collection
of
contributors
about
six
or
seven
or
more
that
have
been
also
fairly
focused
on
mastery
ctl,
and
some
of
you
are
on
the
call
right
now.
Actually
a
few
of
you
are,
which
is
great.
Any
root
has
been
here
for
quite
some
time
helping
bring
forth
a
new
command.
That's
really
changed
the
way
measure
ctl
works
broadly.
That
command
is
called
context
and
we've
just
it
had
been
so
long
in
need
of
measury
mastery
ctl
system
context.
A
But
it's
not
all
about
documentation.
There's,
there's
much
more
to
do
in
terms
of
coding
than
there
is
documentation,
there's
just
a
little
bit
of
catch
up
on
some
on
a
few
docs
jumping
into
docs
great
way
to
get
familiar
with
the
community
to
get
the
contributor
flow
kind
of
I'm
comfortable
with
to
get
some
quick
wins,
get
a
bunch
of
commits
under
your
belt
each
component
that
we
have
here
and
abhishek
like.
I
don't
know
if
you
have
the
count,
there's
like
45
something
repositories.
A
A
So
what
I'm?
What
I'm
guess?
I
guess
the
gauntlet
that
I'm
laying
down
is
that
there's
there's
a
number
of
maintainership
roles
that
are
in
need
of
being
filled
and
they
can't
be
handed
out
willy-nilly
or
they
have
to
be
one
way
of
thinking
of
it
is
either
they
have
to
be
earned.
That's
one
way
of
thinking
of
it,
but
that's
really
not.
A
They
have
to
be
assigned
to
people
who
are
performing
that
role
and
then
it's
not
it's
not
about
earning
it.
It's
just
about
doing
the
role
and
then
and
then
being
recognized
in
that
role.
So
in
the
venue.
Hopefully,
I'm
not
repeating
too
much
of
what
you'd
said.
I'll
explain
a
couple
of
items
here
about
the
like
this
windows
support,
so
mesherie
itself
is
intended
to
you
know
to
allow
you,
okay,
so
mesherie
itself
will
deploy
either
inside
of
a
docker
host.
A
A
For
the
most
part,
a
lot
of
the
contributors
have
verified
that
new
features
function.
New
these
commands
function
well
on
mac
and
linux
and
kind
of
lean
into
file
path.
References
that
you
know
have
have
forward
slashes
or
backslashes.
You
know
aiming
one
way
or
the
next
or
or
make
certain
assumptions.
A
That's
a
it's
a
big
area
so
other
than
outside.
Of
that,
I
won't
repeat
the
rest
of
what
navenu
had
said.
He
was
outlining
a
better
contributor
process
for
individual
design
specs.
He
was
outlining
the
fact
that
there's
a
central
document
that
a
central
design
specification
that
tracks
all
of
the
commands
and
there's
a
new
design
spec
to
use
to
do
that
so
anyway.
Anyone
it's
a
lot
to
kind
of
take
in.
A
A
The
way
that
measury
interacts
with
that
specification
is
through
this
command
perf
mastery
ctl,
perf
node.
It
has
not
been
used
extremely
heavily
and
I
suspect
it
will
get
used
more
and
more
as
smp
picks
up
so
a
lot
to
do
there.
So
thanks
for
getting
us
organized
in
the
venue,
this
is
this
is
good.
There's
a
there's,
a
message,
ctl
channel
as
well
in
slack
that's
kind
of
dedicated
to
these
discussions,
so
last
thing
I'll
say
on
mastery
ctl
is
that
it's
a
great.
A
It's
often
been
the
place
where
people
who
are
either
contributors,
who
are
either
new
to
golang
or
or
who
are
just
looking
to
make
a
mark
on
measurey.
Initially
a
lot
of
times.
Contributors
will
get
started
there
because,
because
the
programming
is
a
little
more
functional,
a
little
more
procedural
because
you're
sort
of
dealing
with
just
hey,
what
does
this
particular
command
do
so,
and
so
the
code
base
is
relatively
smaller
as
well.
So
it's
a
good
place
to
get
started
before
we
switch
topics
last
comments
or
questions.
A
Nice
all
right
good!
Well,
you
know
who
to
disturb
if
you
do
have
comments
or
questions,
I'm
going
to
move
the
compatibility
matrix
down
a
little
bit
just
because
I've
been
talking
a
lot
wood
utkarsh,
if
you're
here,
I
think
we
heard
from
you
last
week
about
a
whole
new
feature,
the
performance
profiles
and
you
did
a
demonstration
of
what
was
a
pretty
slick
user
experience
and
you've
been
working
on
refactoring.
That
do
you
want
to
take
people
through
the
those
changes
yeah.
A
So
while
ukarsh
is
bringing
this
up,
he's,
probably
sick
of
introducing
the
topic
so
I'll
do
it
for
him,
which
is
we're
just
talking
about
performance,
so
one
of
the
things
that
meshri
does
as
a
service
manager
is,
it
helps
people
I'm
trying
to
think
about
where
to
other
than
manage
it
helps
people
characterize
their
the
performance
of
their
service
mesh.
The
performance
of
the
workloads
running
on
top
of
the
mesh
helps
them
gauge.
How
much
overhead
is
involved
and
helps
them?
A
You
know
squeeze
as
much
out
tune
the
the
mesh
tune,
their
workloads,
so
people
use
will
use
measure
to
run
performance
tests
generate
load
and
analyze.
How
quickly
the
mesh
responds,
the
people
will
re,
you
know,
adjust
and
configure
their
service
mesh
run
the
tests
again
and
it
has
become
kind
of
bothersome
over
and
actually
you
can
see
the
total
number
of
tests
that
people
have
run.
If
you
go
to
meshri.io
you'll
see
a
total
number
of
tests
run
and
that's
like
a
it's
a
real
time
ticker.
A
As
people
have
run
those
it's
become
bothersome.
As
a
user
experience
where
someone
will
go
in
they'll,
they'll
click
to
generate
a
to
run,
a
performance
test
and
they'll
enter
in
particular,
parameters
like
generate
load
against
this
endpoint
for
this
long
this
aggressively
and
send
pass
along
these
http,
headers
or
they'll.
D
Yeah,
so
the
interface
has
changed
quite
a
bit
since
our
last
order.
Now
we
have
two
different.
Actually,
this
isn't
like
it's
still
work
in
progress.
I'm
not
sure
what
thing
is
going
to
crash.
Let's
so
now
we
have
to
do
different
views,
but
this
is
incredible
where
you
can.
Actually,
I
know
this
is
the
kind
of
work
that
is.
The
tapes
gets
a
bit
smudgy
when
I'm
flipping
the
card
at
home,
and
that's
that's
something
I
think
soon.
I
hope.
D
D
So
right
now
we
have
only
one
and
one
the
generator
test
and
duration
and
last,
when
you
update
the
profile,
you
can
click
on
it
and
now
it
will
grow.
It
will
show
you
a
table
where
this
is
basically
one
of
the
performance-based
results
which
were
performance
tests
which
were
run
under
this
particular
profile.
You
can
click
on
details
and
you
would
get
the
graph
that
was
generated
through
that
test
and
you
can
click
up
again,
and
this
will
just
copy
this.
This
is.
D
Now
this
is
the
second
view.
This
is
similar
to
how
I
demoed
it
last
week,
except
now
we
have
like
we
get
these
drop
downs
in
this
drop-downs
you
have
miss
tables
first
table
was
performed
through
five
students.
Second,
one
is
the
same
table,
which
is
that
it's
being
presented
differently.
That
is
now
you
get
performance
test
results
which
were
run
under
this
profile
in
the
same
intro
that
you
were
getting
shown
in
that
review.
These
two
views
are
like
brightness,
supported
and
yeah.
This
is
basically
to
edit
the
form
or
yeah.
D
Now
another
page
that
is
like
this
is
still
under
work
here.
You
will
get
right
now.
This
this
right
portion
is
completely
function.
That
is,
if
you
would
have,
if
I
would
have
added
a
grafana
or
chromatizing
grafana
of
charts,
so
basically
metrics
not
not
exactly
the
one
and
right
not
only
the
final
communications,
but
in
general
metrics.
If
I
would
would
have
added
it
in
settings,
I
would
get
those
charts
in
here
in
the
left.
D
You
would
get
a
summary
of
all
of
the
tests
that
you've
run
or
a
number
of
profiles
that
you
have
people
created
which,
like
right
now,
it's
showing
like
not
really
mana
that
I've,
that's
true
but
ui
is
something
to
be
worked
on.
Other
details
will
also
be
added
in
here
others
of
what
the
last
step
like
when
the
last
tests
were
performed
along
with
the
dates.
So
this
is
this.
Is
this
question
is
in
progress?
D
One
thing
that
you
I
mean,
I
don't
know
if
someone
has
noticed
that,
okay,
where
do
I
create
profile
from
yeah?
There
is
no
button
for
getting
profiled
right
now,
because
again,
this
is,
I
was
just
working
on
it,
but
it
would
be
soon
added
here
also,
that
is
where
you
are
like
in
in
these
views.
You
would
have
a
button
directly.
You
can
create
a
new
profile
or
you
will
also
get
a
button
in
here.
That
is
like
just
below
the
summary
of
yours
or
something
or
we
can.
D
You
would
have
to
give
me
to
access
the
same
form
that
you
were
accessing
here.
Just
you
know.
Basically,
it
won't
be
people
for
you,
you'll
be
able
to
fill
it
for
yourself.
So
this
is,
I
am
represented.
What
is
the
current
state
as
well
as
what
I
am
planning
to
do?
At
least
we
are
planning
to
do
empty
in
a
day
or
two,
so
any
feedback
on
this
or
questions.
C
D
Yeah,
actually
it's
looking
clean
one
of
the
like
one
thing
that
is
missing
from
here,
which
would
be
added
once
we
have
maybe
our
engine,
and
that
is
adding
a
button
for
skinnies
but
yeah.
Definitely
this
is
a
bit
clean.
Oh
I
mean
on
the
back
side.
There
are
more
details.
We
can't
put
it
in
front
but
sure
I
think
we
can't
believe
it.
C
B
B
D
A
A
D
Okay,
this
thing
so
instead
of
presents
now
we
have
profiles
because
yeah,
if
someone
has,
I
mean
everyone
has
older
version
of
mystery,
so
you
would
see
here
instead
of
profiles,
you
would
see
results.
But
now,
as
all
of
the
performances
results
are
part
of
profiles,
that
is
all
of
the
tests
that
you
run
would
be
definitely
another
file
they
create
like
either
you
name
it
or
we
will
name
it
for
you,
but
definitely
there
would
be
a
part
of
profile.
D
So,
instead
of
results,
we
now
show
profiles,
and
you
get
this
view
through
multiview,
one
of
them
there's
a
preview
and
one
of
them
is
the
table
view
you
get
you
get
to
the
profiles
and
then
you
can
actually
click
on
your
profile
that
they
created
or
recreate
it.
And
then
you
can
expand
and
see
the
same
old
results
that
you
used
to
the
interface
has
quite
a
bit
changed.
D
So
the
only
thing,
basically,
the
only
thing
in
results
table
that's
changed
is
that
instead
of
getting
your
craft,
basically
getting
your
crop
in
a
expandable
room,
you
get
it
as
a
model.
So
as
I'm
actually,
I
will
be
we're
actually
trying
to
avoid
a
lot
of
nesting
on
the
tables,
so
that
is
where
now
you
get
your
crap
as
a
movie.
D
B
A
A
You
make
working
progress
and
this
is
the
all
right
so
great,
so
this
becomes
this
particular
view
like
today.
If
you
use
meshri
and
you
click
on
performance,
it
will
show
you
a
form
to
fill
in
to
run
a
performance
test
tomorrow
or
when
these
changes
land.
When
you
click
on
the
performance
menu
item,
it
will
be
it'll,
be
a
little
bit
more
of
a
dashboard
telling
you
about
how
many
profiles
you
have
how
many
tests
you've
run.
A
One
of
the
reasons
why
people
will
run
a
performance
test
is
to
well
is
to
initially
baseline
how
their
environment,
how,
how
well
how
how
well
they're
running
their
service
mesh
and
that
shifts
over
time
they
release
new
versions
of
their
apps
onto
the
mesh.
Maybe
things
get
slower,
maybe
things
get
faster?
They
upgrade
to
new
versions
of
that
service
mesh.
They
change
the
configuration
of
the
service
mesh.
You
know
they
add
new
nodes
to
their
cluster,
that
all
kinds
of
things
change.
A
A
So
and
that's
that's
something
that
we
facilitate.
That's
why
we
keep
track
of
the
total
number
of
performance
tests
that
have
run
been
run
on
mastery
data
hill
because
those
statistics
can
be
sent
in
anonymously
and
we're
engaged
with
a
couple
of
universities
about
studying
those
which
reminds
me
we're
meeting
with
nyu
tomorrow.
A
It's
a
good
reminder,
so
this
this
is
great.
One
small
item
of
feedback
on
karsh
is:
if
you
go
back
to
the
profiles
page,
one
thing
that
will
help
us
feel
less
nested
is
if
the
the
paper
on
the
background,
if
we,
if
we
remove
that
paper,
we'll
have
the
switcher
the
view
switchers
so
we're
sitting
on
the
background,
whatever
we
you
know,
sitting
back
there,
but
then,
as
you
go
to
look
at,
you
know,
drill
into
profile
a
and
it
or
or
rather
yeah.
A
Oh
by
the
way,
that's
awesome
I've
yet
to
see
someone
actually
do
a
anyway.
If
you
go,
if
you
nav
away
right
now,
oh
man,
that's
fantastic,
okay,
good,
good,
yeah
and
then
I
know
you're
working
toward
okay
good.
If
you
yeah
this
this.
So
this
is
what
I'm
saying
is
that
like
hey,
if
we
get
rid
of
that
background
paper,
we
will
feel
we
won't
feel
like
we're
three
layers.
Deep,
we'll
only
feel
like
we're
two
layers,
deep,
which
won't
be
as
many
nests
happening.
A
But
it
sounds
like
you
guys,
have
tied
it
off.
It's
we're
basically
either
ready
like
we're
ready
for
that
release
this
stuff,
that,
if
that's
the
case,
what
you're,
showing
here
karsh
is
probably
then
ends
up
landing
in
the
dots,
zero
six
release,
which
is
which
is
just
fine,
which
is
great
any
other
okay.
I
guess
given
the
time
if,
if
anyone
does
have
other
feedback
for
ood
cars
or
just
kind
of
questions
about
the
thought
process,
one
cu
cars.
A
A
A
A
F
F
F
You
know
install
your
service
mesh
in
this
case,
I'm
running
istio
and
you'll
see
this
pane
called
sample
app
applications,
so
one
of
them
would
be
imagehub.
So
imagehub
is
basically
a
demonstration
of
a
particular
capability
of
using
wars
and
filters
to
you
know:
control
traffic,
so
you'd
be
greeted
by
this
page.
Once
you
deployed
it
and
have
it
successfully
up
and
running
and
the
normal
thing
is
you
okay?
F
So
the
point
of
this
is
basically
this
is
a
just
a
central
repository
like
docker
hub,
but
this
is
a
repository.
You
know
mirroring
it
and
if
you
click
this,
you
can
sort
of,
you
know,
pull
an
image
and
how
you
would
go
about
doing
that
is
you
would
sign
up.
You
know,
create
your
account,
then
it
would
take
you
to
login.
F
Then
you
would
sign
in,
and
you
would
be
greeted
here,
and
so
one
of
the
important
things
is
that
you
have
this
plan
field
over
here
you
have
enterprise,
team
and
personal,
so
the
person
is
limited
to
10
pulls
is
limited
to,
I
think
100
enterprises
thousand,
so
that's
just
arbitrary
numbers,
just
for
the
sake
of
the
demo
and
essentially
right
now.
This
is
quite
broken
and
I'll
show
you
in
a
bit.
Why?
But
essentially
you
can
only
hit
this
pull
request
about
10
times.
F
If
you
are
signed
up
with
the
personal
thing
once
it
exceeds
10
it'll,
a
message
will
show
up
saying:
hey
you've
exceeded
it.
You
want
to
upgrade
your
plan
and
then
you'll
be
prompted
to
this
page
and
then
it'll
ask
you
for
your
credit
card
details.
Please
do
enter
it,
I'm
going
to
be
coding
that
in
quite
soon
that
was
a
little
joke
anyway.
F
F
So
this
is
taking
the
console
serves
mesh,
and
you
can
see
that
you
know
when
you
deploy
consoles.
You
have
this
thing
called
proxy
sidecar.
Essentially
what
it
does
is
that
whenever
you
have
a
pod
right,
the
sidecar
sits
here,
intercepting
all
your
traffic
and
you
can
do
all
sorts
of
cool
things,
but
the
one
we
are
after
is
that
envoy
specifically,
has
the
capability
of
you
know
having
a
filter
put
inside
it?
So
I'll
just
quickly
show
you
what
this
filter,
how
it
exactly
works.
F
So
traditionally,
in
this
container
service,
foo
right,
you
would
be
handling
you
would
have
coded
in
the
language.
Sorry,
the
feature
of
you
know
counting
the
number
of
requests
and
checking
whether
it's
valid
or
invalid,
depending
on
the
plan.
The
user
has
subscribed
to
now
this
the
cool
part
about
the
filters
you
can
literally
intercept
the
http,
packets
and
okay.
F
So,
as
I
mentioned
previously,
this
is
onward,
and
this
is
the
filter
that
you
know
we
insert
in
so
through
calls
happening
in
the
background.
Basically,
you
have
your
vosm
filter
inserted.
So
the
content
of
your
wasn't
filter
would
look
something
like
this,
so
it
looks
quite
daunting
and
complicated,
but
I
guess
once
you
understand
it
me,
so
you
would
be
able
to
you
know
easily
access
stuff.
So
the
core
logic
comes
over
here,
where
you
have
your
authent
authorization
header.
F
F
So
I'm
handling
the
time
updation
here
as
well
as
handling
the
count,
so
we're
incrementing
the
count
each
time
within
the
filter
itself,
saving
it
as
the
object
going
back
here.
We
see.
Okay,
fine,
has
this
guy
crossed
the
limit
or
not,
and
then
it
gives
a
limited
exceeded.
This
ideally
should
be
giving
that
message
and
that's
the
working
of
the
filter.
So,
instead
of
coding
it
all
into
this
application,
you've
now
shifted
it
and
you're.
You
know
handling
it
on
to
the
handling
that
responsibility.
F
F
You
would
have,
what
do
you
say
many
more
types
of
filters
so
right
now
I've
only
spoken
about
rate
limiting
right.
You
could
have
so
many
more
types
of
rules,
and
you
know
just
to
show
that
it's
actually
the
filter.
That's
intercepting
I'm
just
going
to
pull
up
this
inspector
thing,
and
so
each
time
I
click
pull.
You
can
see
that
you
get
powered
by
okay.
That's
also
not
that
great
right.
So
basically,
it's
istio
envoy,
that's
intercepting
it
and
it's
handling
this
verification
part.
F
So
it's
just
shifting
the
responsibility
onto
the
service
mesh.
That's
essentially
what
this
entire
application
is
about.
That's
essentially
what
this
demo
is
about,
and
I
think
that's
pretty
much
it
coming
to
where
I
am
proxy
on
the
library
and
how,
on
my
works,
it's
quite
finicky
to
say
the
least
and
currently
like
I
already
did
before
I'm
trying
to
add
on
many
more
types
of
filters.
So
I'm
generally
restructuring
this
entire
project
earlier
it
was
just
hard
coded,
so
you
see
these
paths
right
where
you
can
see.
F
Auth
pull
sign
up
upgrade,
so
these
were
basically
hard
coded
for
the
app
and
what
I'm
trying
to
do
is
no
matter
what
app
you
have.
What
app
you
deploy.
You
will
be
able
to
come
into
what
you
say:
measuring
ui
you'll,
hopefully
have
another
card
or
something
you're
saying
I'm
gonna
configure
json
and
you
can
then
add
your
path.
F
F
F
Okay
cool,
so
you
have
this
json.
So
you'll
have
the
name
of
the
path
sale,
saying
something
like
pull:
you'll
have
the
rule
type
saying
rate
limiter
and
then
you
can.
You
know
customize
your
plan
names.
You
can
customize
your
limits
for
each
plan
and
all
of
that
would
be
hopefully
in
the
future,
are
done
as
a
card
over
here.
So
you
just
enter
it's
just
a
form.
You
fill
click
hit
and
it'll
deploy
the
filter
automatically.
So
that's
the
vision.
F
A
F
A
F
When
you
hit
this
envoy
filter
for
image
hub,
so
when
you
apply
this
configuration
in
the
background,
essentially
it's
these
scripts,
this
one,
the
filter
and
the
patch.json
that's
triggered,
and
one
of
the
things
is.
This
is
what
you
tell.
What
do
you
say,
istio
saying:
hey,
listen!
I
have
this
filter.
I
want
you
to
apply
it,
so
this
is
basically
an
interface
I
guess
istio
provides.
I
need.
F
If
other
service
meshes
as
well
provide
something
like
this
or
like,
is
this
portable?
Probably
not
so,
as
you
can
see
here,
the
proxy
version
for
istio
is
hard
coded.
Ideally,
this
is
going
to
be
something
that
we
don't
want,
because
if
you
mess
up
the
version
number,
it
just
won't
load
up.
So
this
has
to
be
configured
manually
and
you
can
see
this
long
string,
which
is
base64,
that's
nothing,
but
this
filter.json
sort
of
you
know
squished
into
base
64
and
just
passed.
So
the
thing
is
how
wasm
works.
F
How
this
filter
is
run
is
that
there
is
a
virtual
machine
specifically
created
for
running
these
wasn't
filters.
So
this
is
it's
not
created
by
onward
it's
created
by
it's,
it's
there.
It
exists.
So
the
way
that
works
is
that
it
allows
very
few
interfaces
and
very
few
things.
So,
ideally,
I
would
like
to
just
directly
you
know
drop
in
this,
I'm
so
sorry
yeah.
F
I
would
ideally
like
to
just
draw
like
drop
in
this
json
file
directly
into
the
virtual
machine,
but
for
some
reason
they
don't
allow
it
and
they're
working
on
features,
and
you
know
even
this
filter
feature
is
quite
in
alpha.
They've
just
created
a
sig
committee
for
working
on
this
very
very
like
a
few
days
ago,
and
it's
just
a
huge
work
in
progress
and
so
working
with
the
jank.
F
A
A
Not
all
of
them
would
automatically
sidecar
a
proxy
next
to
your
application
container
and
that
the
the
notion
that
kubernetes
in
particular
could
do
such
a
thing
was
kind
of
felt
like
little
leprechauns
inside
of
kubernetes
doing
you
know
magic,
magic
things
now
that
that's
well
understood
sort
of.
I
think
this
is
another.
This
is
the
next
area
for
magic
sprinkles
to
be
happening.
That
just
are
happening
in
the
system
and
people.
Don't
you
can't
see
it?
A
It's
transparent
part
of
what
the
goal
that
pranav
was
articulating
here
is
to
expose
configuration
of
of
these
filters
so
that
a
product
owner
or
an
operator
or
developer
can
go
in
and
manipulate
the
configuration
well
manipulate,
which
filter
to
load,
maybe
when
to
load
it.
Maybe
it's
not
always
appropriate
to
have
loaded
and
then
to
configure
the
behavior
of
the
filter
or
the
filters
as
pranav
was
recently
investigating
sort
of
chaining
filters.
A
What
those
endpoints
are
such
that
when
someone
opens
up
mesherie's
ui
that
management
plane's
ui,
they
go
to
configure
the
filter.
Well,
that
that
can
there
can
be
a
pre-populated
list
of
potential
endpoints
that
they're
looking
to
filter
on,
because
traffic
has
already
been
directed
at
those
endpoints,
and
so
there's
a
lot.
If
image
hub
as
a
sample,
app
sounds
very
familiar
like
pranav
was
describing,
sounds
very
much
like
docker
hub.
It
is
it's
it's
a
ripoff
of
docker
hub
intentionally
as
a
sample
app.
A
There
was
a
contributor
here
previously
kenishgar
who
had
helped
in
the
creation
of
image
hub.
He
ended
up
demoing
it
at
this
last
dockercon
speaking
of
there's
another
dockercon
coming
up,
so
maybe
we
should
do
something
now,
not
all
75
000
of
the
attendees
went
to
that
talk,
but
there
were
quite
a
few
in
the
time
that
that
demo
was
given
that
talk
was
given.
A
I
don't
need
to
tell
a
couple
of
you
to
be
aggressive
about
getting
engaged,
because
I
guess
that
you
already
are,
which
is
great.
Do
it?
Do
it
be
aggressive
if
I'm
not
responsive,
that's
just
because
I'm
on
another
call
or
doing
something
else,
it's
not
because
I
don't
want
you
to
be
engaged
so
nice
with
that.
It
will
conclude
today
what
we've
got
the
community
meeting
on
friday.