►
From YouTube: Meshery Development Meeting - March 25th, 2020
Description
Welcome Shalini Bhawsingka and Hasan Baig!
B
B
A
B
B
B
B
Thank
you,
new
poor
man,
Wow
I
can
I'm
kind
of
embarrassed
to
admit
that
this
is
actually
the
second
cup
of
coffee.
So
you
would
think
that
would
be
over
the
that
kind
of
a
mistake,
but
anyway
good
we're
six.
After
now,
let's,
let's
get
the
meeting
rolling
a
couple
of
folks
just
joined
the
call,
so
I'm
gonna
I'm
gonna
paste
the
meeting
minutes
into
the
chat.
B
Please
jump
into
the
meeting
minutes.
Please
take
notes,
put
in
things
you
want
to
talk
about,
and
hopefully
Hassan
will
come
back.
I
think
we
just
lost
him,
which
means
we
only
have
one
individual
that
hasn't
been
on
this
call
before,
and
so
it's
worth
taking
just
a
moment
to
say,
welcome,
Shalini
very
nice
to
have
you.
D
Yes,
I'm
currently
looking
as
it
will
be
engaged
to
the
implicit
affinity,
school
I
interest,
basically
isn't
Dancing
most
of
the
time.
We
regret
it.
I
love
to
dance
I.
Think
great
opportunities
are
moving
here
as
I'm
able
to
meet
new
people
and
them
such
bunch
of
interesting
people.
Much
more
to
know
it
will
be.
B
B
B
For
a
moment,
so
I'm
hoping
he'll
come
back
a
couple
of
things
and
we've
got
mr.
Jane.
We've
got
a
couple
of
topics
that
I
just
listed
down
as
things
to
potentially
talk
about.
But
really
this
meetings
call
is
to
go
around
the
room
with
each
of
you
and
talk
about
active
areas
and
focus
or
if
you've
been
working
on
something
and
if
you
haven't
and
we'll
talk
about
something
you
might
jump
into
and
so-and-so.
B
And
actually,
maybe
Naveen,
just
you
knowing
and
having
reviewed
some
of
the
mesh
sync
work
that
you've
been
doing.
It
might
be
something
worth
demoing
to
folks
on.
The
call
I
was
actually
talking
with
Shalini
a
little
earlier
this
week,
showing
her
some
of
the
that
work
and
how
it
is
that
some
of
the
user
experience
can
be
improved
by
the
fact
that
measure
II
is
now
cognizant
of,
or
will
very
shortly
be
cognizant
of
Prometheus
and
Ravana
deployments
and
being
able
to
autodiscover
those
through
mesh
sync.
E
Hello
yep:
there
we
go
oh,
so
we
were
discussing
about
it.
Basically,
what
it
will
be
able
to
do
is
so
there
are
various
release,
measures
that
we
support
currently,
but
not
all
of
them
needs
to
be
installed
using
our
detectors
about
not
all
of
them
are
installed.
Some
of
them
may
be
in
for
report
and
by
anyone.
So
what
mesh
think
will
do?
It
will
basically
fetch
all
the
information
and
keep
the
state
updated
like
we
have
also
written
a
little
bit
of
design
dog
architecture.
E
Design
is
still
pending
and
related
to
this
I
had
already
sent
TR
we're.
Currently,
what
we
do
is
we
just
scan
the
cluster,
whether
there
is
there
is
sto
present
or
not.
There
is
like
30
present
about
basically
the
service
matters
and
how
it
is
done
is
standing
by
images.
So
it
is
the
sort
of
light
of
method
to
scan
the
images
to
get
the
deployments
and
to
get
in
information
about
the
service
message.
E
So,
but
this
we
do
you
planning
to
have
two
level
of
discovery,
service
methods
of
discovery,
so
how
we
are
trying
to
do
is
like
first,
there
will
be
separate
command
and
adapter
in
every
a
raptor
in
every
representative
come
on
its
present.
We
will
call
put
that
comment
and
like
that
adaptable
itself,
standards
and
and
deploy
mention.
It
will
be
done
that
what
I
will
diversion
and
deployment
are
present
and
if
it
is
not
present
or
the
adapter.
E
If
it
does
not
support
that
operation,
then
we
will
manually
scan
the
images
like
we
are
doing
currently.
So
if
we
guards
to
this
I
send
the
PR
without
bear
instead
of
just
showing
the
word
is
a
number
of
the
service
mesh.
We
will
also
like
we
have
complete
knowledge
of
deployment,
and
so
what
is
the
namespace
that
is
deployed
in
so
if
there
are
which
multiple
deployments
of
a
service
night
week,
and
so
we
can
get
to
know
that
and
what
are
the
name
hitches?
We
can
go.
B
B
B
B
Know,
and
so
actually
I'll
jump
in
and
say
this.
That
Naveen
is
highlighting
that
the
fact
that
when
Mishra
becomes
much
more
aware
of
the
environment
and
is
able
to
discover
various
objects,
deployment
services,
various
kubernetes
constructs
retrieve
that
information
and
pull
pull
that
info
back
and
by
the
way
the
info
that
it
would
be
pulling
back
is
things
that
you
would
see
if
you
did
a
cutesy
TL
get
pods
for
example,
or
your
cube
cube
Cpl
describe.
B
B
Through
provisioning
of
the
service
mash,
which
is
great
like
in
to
the
extent
that
there
isn't
a
mesh
present
but
italy,
it
makes
for
a
poor
user
experience
when
you
might
go
to
deploy
mystery
and
you
already
have
a
mesh
deployed
and
configured
or
you
go
to
use
ma
cherie
and
you
use
it
to
deploy
and
service
mass
grade.
And
then
in
the
UI
is
there's
two
date.
B
And
so
this
area
enables
brown
fielding,
which
is
a
term
to
describe
the
fact
that
that
you
bring
measure
in
after
the
fact
after
a
mesh
is
already
up
and
the
reality
is
it's
not
just
the
green
fielding
and
the
brown
fielding
is
it's
kind
of
the
ongoing
synchronization.
It's
the
it's.
The
fact
that
when
someone
is
deployed
a
service
much
that
they're
not
always
going
to
use
measure
as
the
tool
to
make
changes
to
that
mesh,
things
will
just
happen
to
it.
B
Services
will
be
onboard
and
off-board
operators
will
interface
with
the
service
mesh
separately
from
measure
e
kubernetes
will
reschedule
pods
and
things
move
stuff
around
and
so
messy
as
a
management
plan
needs
to
be
in
continual
needs
to
continually
understand.
What's
going
on
in
the
environment,
and
so
one
of
the
mechanisms
to
do
that
is
to
interrogate
actively
interrogate
kubernetes
and
go
out
and
query
its
API
and
ask
it.
B
For
current
details,
you
know
current
current
information
about
the
environment,
the
other
way,
that's
a
bit
more
efficient,
that's
good!
That's
sort
of
an
active
polling
mechanism
and
active
scraping
of
information
from
kubernetes
the
other
way
of
synchronizing
the
environment.
That's
a
little
more
slightly
more
real-time,
slightly
more
efficient
is.
B
B
So
rather,
you'd
like
to
do
do
this,
this
synchronization
this
mesh
sync
efficiently,
there's
a
couple
of
ways
to
make
that
more
efficient
one
is
what
we
were
just
describing,
which
is
efficient,
is
to
subscribe
for
notifications
from
kubernetes,
so
semestre
just
listens
and
receives
events.
That's
a
little
more
efficient,
that's
good!
The
second
thing
that.
B
Two
phases
discovery
so
so,
even
when
mesh
think
is
going
out
and
scraping
kubernetes
for
information,
it
might
do
that
in
a
cursory
way
as
a
phase.
One
Peck
discovery
just
very
quickly
and
very
go
retrieve
a
light
amount
of
information
and
based
on
the
objects
that
are
retrieved
do
a
secondary
set
of
discovery.
That's
a
bit
more
in
depth,
but
only
doing
that
on
objects
that
are
of
interest
to
massery
I
mean
that
way.
You
can
begin
to
build
in
a
hierarchy
of
the
the
intensity
of
scanning
of
discovering
and
do
it.
B
Someone
officially
said
that
this
type
of
a
mechanism,
this
type
of
a
paradigm
of
discovering
and
syncing
synchronizing
the
infrastructure,
very,
very
common
to
management
software
in
general
and
that's
what
my
Cherie
is-
is
a
management
plane.
So
this
part
of
the
reason
that
you're,
seeing
that
within
the
project,
we're
calling
something
by
kind
of
a
catchy
phrase
if
you
will
is
because
this
algorithm,
this
routine
of
synchronizing
information
is
ultimately
quite
should-
should
really
be
a
part
of
at
the
heart
of
the
measure
itself,
because
it
should
just
be
this
continual
process.
B
If
you
can,
if
you
get
some
time
like
one
of
the
ways
I
would
encourage
each
of
you
to
contribute
is
to
you
know,
code
or
not
it's
just
in
the
thought
process
of
what
this
means
and
but
what
these
various
design
designs
mean
and
give
your
input
or
just
ask
questions,
asked
really
stupid
questions
I,
encourage
you
to
ask
the
the
simplest
and
silliest
questions
you
can,
because
when
you
do,
everyone
else
will
learn,
including
me.
So
so
that's
mesh
sink
I'm
excited
about
that!
That's
that's
great!
B
B
B
F
We
have
a
discussion
about
investing.
That's
really
interesting.
Do
you
want
to
figure
out
what
is
it
for
yes
endures
as
you
asking
about
so
some
I
just
I,
just
think
in
that
sink
is
doing
that
from
the
service
mesh
and
you'd
apply
it.
Then
they
had
a
be
introduced
with
you,
get
it
custom
CR
is
some
service
mesh
tool,
so
what
mess
but
mess
Inc
does
is
what
the
service
mesh
do
is.
There's
events
out
here
having
successfully
of
Spain.
That's
when
you
talking
about.
B
Being
why
us
how
us
mesh
mesh
sync
would
improve
the
user
experience.
One
of
those
that
actually
Shalini
and
I
were
talking
about
is
the
notion.
When
you
start
up
measuring
today,
you
will
often
find
that
it
discovers
your
kubernetes,
your
kubernetes
cluster,
and
that
makes
can
make
some.
You
know
using
mesh
way
very
convenient
it'll,
also
automatically
identify
that
you've
got
any
number
of
adapters
out
there.
B
Those
are
actually
the
Garifuna
and
prometheus
instances
that
are
inside
of
the
sto
control
plane,
so
just
as
much
think
is
going
out
and
interrogating
kubernetes
and
interrogating
isseo
for
how
its
configured
and
where
its
various
components
are
part
of
that
component
tree
is
Agri
fauna
instance.
In
a
Prometheus
instance,
today
the
user
has
to
go
over
and
manually
wire
up
a.
B
Connectivity
to
those
instances
now
that
Naveen
has
been
working
on
mesh
sync,
this
enables
the
ability
to
bring
in
a
better
user
experience
to
just
you
know,
autodiscover
the
fact
that
Prometheus
is
there
or
maybe
multiple
instances,
let
the
user
pick
which
one
that
want
to
enter.
They
want
Mary
to
speak
to
and
so
yeah,
and
so
also
the
same
thing.
If
I've
got
sto
deployed,
it
should
detect
that.
So
when
I
go
here,
it
should
detect
the
fact
that
there's
an
sto
deployed.
This
is
what
it
looks
like.
B
F
Sense,
yeah
I
got
to
dance
what
you
are
already
discussing
as
a
creator.
I
think
that's
a
very
good
feature,
but
one
thing
this
as
I
lost.
My
Nana
is
stewing
my
laptop
I'm
missing
some
time.
We
could
do
it.
We
could
do
at
the
primary
still
and
forget
to
forget
to
do
the
deployment
for
ingress
controller,
so
I'm
figuring
out,
what's
happening
inside
I.
Think
if
we
in
leveraging
us
could
all
kind
of
thing
nice
English,
we
are
nice
to
messenger.
F
B
That's
a
great
suggestion,
your
your
spot
on
that
particular
issue
that
you
just
mentioned,
like
is
maybe
one
of
the
first
points
of
frustration
that
that
people
who
are
adopting
a
service
match
or
trying
to
operate
a
service
map
or
sto
specifically
run
into
frustration
about
like
the
fact
that
they
know
they're
running
their
workloads
on
the
mesh.
Why
can't
they
access
them?
B
Is
pretty
poorly
done
is
pretty
done
in
a
very
cursory
way
that
not
only
should
it
be
showing
you
the
fact
that
you've
got
these,
you
know
six
or
seven
sto
control,
plane
components
deployed
like
whether
or
not
you're
running
an
ingress
gateway
or
not.
It
should
also,
then,
give
you
a
bunch
of
more
details
about
the
fact
that
you
do
have
an
ingress
gateway
deployed,
but
it
and
then
here's
its
configuration,
which
may
or
may
not
serve
the
needs
of
your
freshly
on-boarded
services.
B
You
know
workloads
anyway
yeah,
yes,
so
I'm,
just
reinforcing
your
point
that
there
are
any
number
of
changes
enabled
through
something
like
mesh
syncing,
and
it
has-
and
that's
in
part
why
I
was
saying
we're
kind
of
giving
a
special
name
and
identifying
that
it's
one
of
those
sort
of
heartbeats
of
how
it
is
messy
and
any
management
where
needs
to
work,
because
it'll
really
rely
on
that
capability.
Quite
a
bit.
F
Person
even
I,
also
the
check
out
the
new
budget
of
Steel
1.5,
so
doing
kind
of
working
in
kind
of
every
sample
is
quite
everything
they
introduce.
Some
kind
of
optimism
has
introduced
some
new
things,
so
I
think
we
could
do
is
we
could?
We
could
have
have
a
plan
for
that
and
I
think
is
well
that's
some.
This
was
a
great
way
in
thinking's
information
from
getting
in.
So
we
could
do
that
implementation
in
the
same
way
that
the
students
for
their
deployments.
B
Yeah,
yes,
awesome
and
envoy
filters
have
been
on
my
backlog
for
two
and
a
half
years,
you're.
So
right
that,
like
that's
such
an
interesting
area
to
focus
on
in
part
in
large
part
because
of
how
efficiently
the
visibility
of
the
traffic
that
you
can
see
through
there,
the
efficiency
by
which
you
might
be
able
to
manipulate
it
or
do
things
yeah,
you're,
you're,
spot-on,
they're,
only
and
the
frustrating
bit
for
me
after,
like
two
and
a
half
years
of
wanting
to
do
it
is,
is
I.
B
Contributors
that
are
here
now
or
enough
of
them
and
the
project
is
mature
enough,
and
some
of
them
are
capable
enough
that
that
we
I
we
what
we
should
get
to
that
this
year
or
like
if
anyone's
ready
to
do
it.
Let's
start
like
hey
I,
agree,
there's
a
bunch
of
different
use
cases
for
things
that
we
might
do
inside
of
an
envoy
filter
instead
of
Lawson.
B
B
C
Hi
everyone
I
just
like,
went
through
the
issue
right
now
and
I.
Don't
have
a
lot
of
new
about
the
issue
to
be
on
to
being
first
good
in
the
first
place,
but
when
I
was
going
through
they're
just
two
points
mentioned
in
the
implementation,
one
is
moved
the
folder
or
I,
which
is
I,
guess
already
done
by
Ashish,
and
the
other
implementation
point
is
Nestle.
Docker
app
deployment
instruction
will
need
to
be
updated
with
the
new
location.
So
what
exactly?
Does
that
mean?
C
B
B
New
poor
had
recently
written
I'm
sorry
Nuno
had
recently
written
a
docker
app
for
measure
II
and
had
he
had
only
done
it
in
a
cursory
way.
It
was
which
is
fantastic.
He's
done
a
good
job.
I
just
mean
he
hasn't
been
able
to
spend
all
the
time
that
he
wanted
to
on
it,
and
so,
no
doubt
it
needs
the
way
his
approach
needs
some
improvement
likely
and
also
yeah.
B
Learning
darker
app
and
trying
to
take
the
docker
compose
file
that
we
have
today
from
ESRI
and
making
it
into
the
format
that
docker
app
will
use
and
yeah
kind
of
learning
dr.
app
going
through
there
documenting
the
fact
that
we
would
support
that
right
now
it
so
from
the
poor
and
everyone
else.
There's
been
a
lot
of
updates
on
the
main
readme
for
massery
feel
free
to
express
opinion
or
or
improve
upon
it.
B
One
of
those
enhancements
had
just
been
calling
out
fairly
prominently:
how
for
people
to
get
started
with
measure
II
and
depending
upon
what
tooling
they
really
like,
or
what
platforms
are
running
on,
I'm
giving
them
links
to
ghost
get
started
for
docker
app.
I
think
it
points
to
docs,
slash
installation,
slash
docker,
which
is
good,
except
it's
actually
missing
how
to
do
this
with
dr.
app.
C
So
I
will
soon
side
the
other
documenting
and
will
styling
before
that
I'll
start
working
on
it
and
learning
the
dr.
app
first
of
all,
then
in
the
process
of
that
I
will
work
on
how
my
Shri
composed
well,
the
current
image
of
my
chicken
was
how
it
is
working.
I
need
to
make
compatibility
with
the
doc.
No
panic,
maybe
and
then
I
will
document
the
whole
process,
and
that
is
the
main
task.
The
final
step,
if
I
am
correct,
yeah.
B
B
B
B
B
Initiate
a
performance
test
that
you're
looking
to
initiate
you'd,
be
able
to
load
that
up
in
a
Yammer
file
or
use
this
flag
to
point
to
a
file,
as
opposed
to
having
to
type
out
all
those
commands
and
all
those
flags
on
the
command
line
that
that
llamó
file
is
really
the
service
mash.
The
smps
service
match
performance
spec.
So
it's
really
this
a
llamó
file
in
this
format,
and
you
can
get
an
example
of
this
format,
one
here
in
this
repo,
but
two
when
you're
using
mystery
anymore.
G
G
B
B
B
This
is
no
perfect
in
its
most
basic
form.
I
think
you
need
to
give
it
I
think,
maybe
just
one
flag,
so
URL
in
this
case
I'm
just
gonna
hit
something
like
github
calm,
because
because
me,
you
know
mesh
the
performance
testing.
That
measure
does
can
be
used
against
workloads
on
the
mesh
and
against
workloads
off
the
mesh
I.
Think
if
I,
okay,
I,
don't
know
that
that
actually
worked.
B
Okay,
yes,
this
is
another
area.
That
is
a
good
reminder
for
us
that
we
need
to
go
back
and
document,
but
so
master
ect
I
will
let
you
run
a
master,
see
Taylor
powerful,
that
you
run
a
performance
test,
whether
you're
authenticated
through
the
measure,
II
cloud
provider
or
whether
you're
not
using
that
cloud
provider
you're
just
using
a
local
provider
and
in
both
cases
measure.
'I
will
facilitate
the
collection
of
that.
B
And
that
is
that
basically
you'd
give
it
I,
think
the
name
of
your
provider,
which
we
recently
renamed
the
local
one
to
none
I.
Think
actually,
if
you
don't,
even
if
you
don't
specify
that
cookie,
if
you're,
not
if
you
don't
specify
the
cookie
is
gonna
assume
that
you're
not
using
an
authentication
provider.
B
B
G
B
I
The
part
the
whole
port
I
don't
know
who
actually,
oh
sure,
I
making
of
that
function,
to
provide
the
description
part
not
able
to
understand
how
to
return
by
seeing
the
previous
got
you
taking
a
bit
okay,
so
that's
done,
then
I
guess
it
will
not
take
much
longer.
The
PR
will
be
ready
by
cooking.
If
that
this
is
imparted.
B
B
Okay,
so
good,
and
then
so
can
each
car
and
Naveen
between
the
the
between
the
two
of
you
guys
been
working
on
some
significant
new
features,
new
enhancements,
both
of
which
would
be
really
nice.
If
you
were
in
a
position
to
demo
those
or
to
present
them
on
Friday's
call
I
think
that
that
would
feel
good
to
get
some
feedback.
People
would
understand
the
importance
of
those
foundational
things.
B
I
think
we
I
think
that
that
was
conveyed
pretty
well
today
from
mesh
sync
and
of
the
authorization
framework
that
can
each
cars
working
on
it's
a
little
bit
behind
the
scenes
and
I.
Don't
know
that
everyone
realizes
is
aware
of
it
and
also
if
they
are
realizes
quite
how
meaningful
that
is
to
the
project
and
how
it
will
enable
a
few
things.
So
I.
H
Think
once
the
author
I
mean
once
you're
able
to
authenticate,
make
sure
ii
CPL
using
trip
the
tokens,
then
there'll
be
some
demonstratable
work
like
otherwise
he'll.
Just
look
like
normal
login
I.
Think:
okay,
yeah
yeah,
that's
you
yeah
yeah.
Maybe
the
use
case
diagram
might
something
like
that
might
clear,
or
what
we're
trying
to
do.
Yeah.
F
Okay,
yeah
I,
just
checking
on
the
energy
diagram,
I,
see
yeah,
so
yeah
I,
just
seen
with
some
kind
of
update.
We
have
some
showing
three
icons
they
have
cleared
ever
from
using
the
machine
recipient,
so
we
I've
seen
that
I
think
is
rather
to
shown
the
icon.
We
just
should
just
show
the
magista
tier
come
on
there,
so
I
check
other
included
such
an
that
is,
but
other
projects
are
doing.
F
If
we,
if
I
talk
about
to
burn
it,
is
cube,
City
oh,
come
on,
it's
just
just
show
that
you
CDL
come
on
so,
regardless
of
their
average,
we
are
I,
think
yes
and
in
architecture
we
should
so
the
mesh
recipient,
regardless
of
we
are
intense
lending
people.
If
we
showed
some
icons
like
the
laptop
on
the
cloud
and
everything
might
be
in
in
future,
in
some
other
cases
might
do
some
integration
there
so
kind
of
a
night.
You
know
I
think
we
should
one
commands
and
then
get
the
feeling
that
is.
B
Okay,
yeah,
if
I
can
play
that
feedback
back,
it
was
the
current.
The
current
architectural
diagram
that
we
have
doesn't
include
these
bits
to
the
left.
It
doesn't
include
the
fact
that
Mestre
has
a
rest
api
and
that
you
that
any
any
client
can
go
interface
with
that
rest
api,
that
there
are
two
clients
that
we
include
in
the
project.
One
is
mastery
CTL
and
the
other
one
is
the
Mestre
web
interface,
the
mesh
or
UI.
B
And
I
think
you
know
one
of
the
comments
that
you've
made
before
was
like:
hey
it's
not
and
based
on
the
current
version
of
this
diagram.
It's
not
entirely
clear
where
ma
cherie
CTL
as
a
tool
fits
into
the
architecture,
and
so,
and
so
this
is
trying
to
help
inform
people
of
the
fact
that
it
is
a
tool
and
where
it
fits
and
I
think
part
of
what
you're
saying
right
now
is
that
in
general
this
is
an
improvement.
F
Please
go
send
so
right
now
we
in
the
new
version
of
the
ground
si
si.
We
builded
three
icons
there
to
show
that
we
can
interact
with
the
machine
with
my
Cpl,
mentioning
UI
and
the
script
we
are
on
it.
So
my
required
is
tagged
to
be
colorless
of
we
are
showing
the
difference,
which
is
the
UI
and
the
client
and
the
script
I
think
we
should
do
the
sample
icon.
That
is
Mishra
CTL
come
on
so
behind
that
PR
you're
running
from
is
not
is
not
a
big
deal,
but
there
is
a
come
on.
F
You
need
to
interact
with
the
machine
and
behind
that
we
have
the
mobile
UI
scraped
or
that's
that's
generic
to
us,
and
you
can
do
that
easily.
I
think
the
point
that
I'm
making
here
this
I'm
talking
about
the
cumulative
or
other
they
just
showing
that
client
face
and
a
server
face,
API
server
and
to
interact
with
this
service.
You
have
a
good
city
income,
regardless
colony
in
the
image
Amazon
Cloud
or.
B
B
Feel
free
to
jump
in
and
try
as
well
that
there's
two
versions
of
this
diagram:
this
one
is
an
SVG
that
I
was
just
editing
the
SVG
itself.
The
other
version
of
the
diagram
is
in
Google,
slides
in
the
Mestre
architecture,
Google
slide,
and
so
it's
it's
also
downloadable
as
an
SVG,
and
so
the
latest
update
that
I
did
I
for
better
or
worse
didn't.
Do
it
inside
of
Google
slides,
it's
worse,
that
I
didn't
do
it
there,
because
everyone
can
go
use
Google,
slides
and
do
that.