►
From YouTube: Meshery Community Meeting - April 12th, 2019
Description
Demo of the new @linkerd adapter in this week's Meshery community meeting.
See https://layer5.io/meshery for more Meshery project details and other recorded meetings.
A
B
C
I've
been
doing
some
of
the
test
trying
to
make
some
low-level
assessments
of
the
numbers
that
we
are
seeing
yesterday
marguerite
pushed
in
the
liquidy
stuff,
I
haven't
had
a
chance
to
go
through
the
liquidy
stuff,
but
I
think
when
I
have
some
numbers
from
sto
and
comparing
them
against
our
different
service
mesh.
That
would
be
good
way
to
put
put
things
into
perspective.
C
B
C
Like
I
was
no
testing
and
down
with
still
and
whatever
setup,
that
I
have
eight
core
machine
and
like
that,
the
number
that
ups,
the
maximum
our
queries
per
second,
does
not
seem
to
go
beyond
like
about
580
600
for
a
simple
200
kilo,
byte
300
kilo,
byte
response
data,
so
similar
results.
How
does
it
work
against
other
service
meshes?
And
oh,
it's
a
8,
node
machine.
A
B
B
A
B
B
You
know
contributors
update,
so
one
from
fatigue,
I've
got
a
topic
on
google
Summer
of
Code
I've
got
a
little
bit
of
an
update,
I
think
on
hashey
corpse
behalf
I
know
the
Reach's
got
and
adapters
update
as
well
as
we've
got
an
installation
and
update
on
installation
in
terms
of
just
some
of
the
code
changes
that
have
happened
since
last.
We
met
just
trying
to
think
of
the
other
topics.
Real
quick
I
think
last
time
we
already
spoke
to
to
the
places
that
measure
has
been
presented.
B
But
I
guess,
which
is
a
reminder
if
you
guys
you
pratik
and
pranaam,
if
you
guys
don't
like
the
pictures
that
have
been
used
in
the
slide
deck
of
you,
you
know
now's
your
chance
to
change
it
because
we're
gonna
use
it
a
container
world
docker
Kong
after
that,
and
then
cube
con
tu
after
that
I
think.
There's
one
thing:
another
thing
to
talk
about
with
respect
to
CN
CF
traffic
sig.
B
B
Okay,
fair
enough,
we
might
have
all
of
the
topics
listed
out
so
to
go
through
those
I'll
say
that
so
with
Google
summer
summer
coat
that
program
&
pratik.
Thank
you
for
sending
that
around
and
panov
I
think
between
the
two
of
you
guys
so
passing
that
along
I
pass
it
over
to
not
not
the
ECE
students.
I
think
you
guys
did
that
well
enough,
but
I
pass
it
over
to
the
computer
science
school
as
well.
I
think
they
just
sent
over
a
link
to
google
Summer
of
Code
and
maybe
not
to
measure
e.
B
B
It
may
just
come
to
hands
and
knees
separately
so
anyway
about
three
folks
and
they've
come
out,
so
a
nod
whom
you
might
have
been
seeing
some
of
his
questions
and
him
getting
familiar
with
the
product
project
in
the
channel.
He
had
heard
about
the
project
through
google
Summer
of
Code.
He
didn't
get
an
application
in
before
the
deadline
which
was
on
Tuesday,
but
he
still
wants
to
contribute
anyway,
and
so
that's
great
I
would
have
it
would've
been
great.
B
Had
he
gotten
an
application
in
so
that
he
could
get
a
stipend
at
the
end
of
the
summer,
but
either
way
it's
great
Sacco,
whose
last
name
I
can't
pronounce
or
spell
I
was
half
anticipated.
He
might
be
on
the
call
today
he
is,
he
probably
he
will
be
going
through
the
google
Summer
of
Code
program
and
so
hopefully
we'll
get
his
submission
accepted.
B
He
hadn't
previously
worked
on
engine
mesh,
which
is
a
project
that
that
I
had
some
doings
with
engine
mesh
was
its
deal
with
the
envoys
citecar
proxy
displaced
by
nginx,
and
that
was
the
engine
mesh
project.
That
project
has
been
set
down
and
nginx
has
gone
a
different
way,
I
think
like
engine
unit
and
and
some
other
things,
but
he
worked
on
that
so
has
familiarity
with
part
of
what
we're
trying
to
do
here
site
that
you
want
to
warmly
welcome
Sacco
when
anybody
comes
I
mean
John.
B
Is
another
interested
party
I'm,
not
sure
if
he'll
come
to
bear
and
the
project
or
not,
is
he
doesn't
have
a
lot
of
familiarity
with
this
space?
So
then,
honestly,
I
can't
remember
who
else
we've
been
talking
to,
but
enough
people
that
we
end
up.
Writing
a
welcome
newcomers
guide,
so
I'll
link
it
in
our
meeting
minutes,
but
give
a
give
a
read
to
it.
B
B
B
D
Yes,
so
he
was
still
working
on
it.
I
think
they
recently
started,
but
he
did
actually
share
his
plan
with
us,
which
is
actually
to
use
their
shared
control
plane.
We
just
I
mean
like
at
the
control
plane
on
the
SAS,
which
will
not
be
hosted
by
the
individuals,
but
rather
it'll,
be
you
know,
or
of
the
oculi
SAS,
but
they
also
have
this
controller
component
along
with
the
data
plane
that
will
be
deployed
as
part
of
the
adapter
is,
is
kind
of
like
Nevada
he
was
sharing.
D
B
C
Sorry
I
wasn't
me:
I
had
a
look
at
it,
I
haven't,
given
it
a
shot.
I
will
say:
I
have
been
like
thinking
of
what
are
the
good
metrics
to
measure
Michonne
like
stress
testing.
Looking
up
at
the
state,
startup
Luton
sees
what
are
the
maximum
QP
is
that
we
are
getting.
There
are
some
threads
like,
even
though
I
am
requesting
for
a
particular
amount
of
queries
per
second,
it
maxes
out
at
some
limit,
and
that
limit
is
not
independent
of
threads.
It's
how
many
threads
are
running?
C
How
many
cores
are
on
the
machine
and
things
like
that
so
kind
of
looking
into
micro,
benchmarking,
the
different
components
of
which
component
adds
on
to
how
much
latency
and
those
were
the
kind
of
slimes
where
I
have
been
working
on
for
the
past
week.
I
saw
the
liquidy
stuff,
but
I
haven't
really
gone
and
tested
it
out.
Yet.
B
C
There
are
some
of
their
boats.
There
are
some
of
the
data
that
I'm
collecting
from
the
Missouri.
The
some
of
them
are
from
the
CLI
logs
that
we
have
collected
I
haven't
like
there
are
some
numbers
that
pop
up
I
am
looking
at
those
numbers
and
trying
to
put
up
making
sense
to
it.
Maybe
I
can
put
number
in
excel
sheet,
and
then
we
can
figure
out
if
we
can
combine
these
results
and
create
sort
of
some
sort
of
reports
out
of
it.
B
Great
there's
a
shared
Google
folder
that,
if
you
link
to,
if
you
don't
have
access
to,
will
make
sure
that
you
get
it
so
that
you
can
create
a
sheet
to
begin
iterating
there,
the
further
along
we
get
there.
The
earlier
we'll
be
able
to
begin
to
speak
to
some
of
those
or
maybe
even
next
week
next
week
for
sure,
we're
gonna
want
to
have
at
least
a
couple
of
example,
results
yeah.
B
So
that's
great!
That's
alright,
an
insightful
though
so
so
one
of
the.
Let
me
regurgitate
one
of
the
points
that
you've
said
and
that
was
then
it
looked
like
news
you
can
tell
right
now
one
of
the
constraints
of
the
tests
that
you're
running
is
not
particularly
tied
to
the
number
of
concurrent
threads.
C
So,
basically,
if
we
have
some
number
of
cores
and
we
are
creating
threads-
which
are
larger
than
the
number
of
course-
then
the
course,
our
time
shared
between
the
threads
that
we
have
initiated.
So
they
cube
mother
queries
which
each
each
thread
gets
to
execute
depends
on
how
how
the
scheduling
of
those
thread
works
on
the
core.
So
that
is
one
thing
that
I
have
figured
out
and
that's
kind
of
intuitive
as
well.
C
B
C
Basically,
like
the
Envoy
deploys
and
starts
up
a
particular
function,
so
there
is
a
latency
from
where,
on
the
actual
code
that
we
want
to
execute,
starts
working
out.
So
that
is
the
link.
I
have
not
seen
a
response
in
that
amount
of
time
and
after
that,
I
see
continuous
responses
back,
which
which
matches
the
actual
function
time
that
I
see
from
the
CLI
reports,
so
that
that
needs
to
be
dealt
deeper
to
see
what
exactly
the
startup
time
latency
is
kanaeva.
C
B
B
C
I
think
a
lot
of
these
lambda
kind
of
environments
are
do
that
as
the
demand
grows.
But
if
a
particular
customer
has
a
set
up,
vm
set
up
with
particular
number
of
course,
and
we
can
definitely
say
that
there
for
a
targeted,
NQ
peers
based
on
these
based
on
your
other
set
of
environment
set
up
and
the
latency
is,
we
can
say
we
can
tell
them
that.
Okay,
you
might
need
to
increase
your
core
by
two
four
or
something.
Oh,
that's
one.
B
Understood:
okay,
yeah-
that
would
be
super
interesting.
You
know,
did
to
be
able
to
as
a
result
of
this
research
to
be
able
to
come
out
with
a
formula
that
yeah
that
that
Tunes,
your
mesh
based
on
your
underlying
infrastructures,
yeah
all
right.
We're
going
back
up
real,
quick,
just
to
say,
I,
think
we
will
try
to
do
it.
B
Do
a
demo
and
walk
through
it,
but
the
the
notion
that
measure
E
as
a
project
to
help
provide
very
detailed
and
granular
levels
of
knobs
that
people
can
tweak
and
tune
for
what
different
types
of
tests
that
they
can
run.
Hopefully
measure.
He
ultimately
will
be
a
tool
like
that,
but
not
having
all
of
those
it
at
least,
and
minimally
initially
needs
to
have
a
great
user
experience,
be
very
easy
to
use
and
let
really
facilitate
adoption.
Let
the
layman
pick
it
up
and
go
to
answer
some
of
their
basic
questions.
B
You
know
what
is
the
overhead
just
at
a
high
level
and
so
to
that
end
and
hopefully
to
benefit
Pranav
and
Prateek
and
and
others
through
a
focus
on
part
of
how
you
do
the
QuickStart.
You
know
you
do
the
QuickStart
install,
hopefully
that
will
be
improved
so
so,
let's
take
a
look
at
that
here
in
a
second
last
update
was
that,
just
as
the
the
scene
CF
we
had
presented
to
the
CN
CF
and
networking
working
group,
it
was
a
very
small
working
group
that
had
really
more
or
less
faded
over
time.
B
We're
looking
at
reinvigorating
that
and
really
reincarnating
it
into
a
new
special
interest
group
called
traffic
sig
traffic
special
interest
group,
which
is
going
to
encompass
different
new
different
types
of
projects
like
pub
sub
models,
G
RPC,
servus,
proxies
service
messages
and
I
expect
we'll
get
a
lot
more
activity
going
on
in
there.
So
it
would
be
something
that
hopefully
we're
announcing
it
at
Q
come
okay
with
that
it
will
go
down
to
gireesh.
If
you
want
to
go
next.
D
D
We
have
it
functional
up
to
a
certain
point
of
the
servants
right
now,
installation
and
our
installation
of
lincoln
itself
and
two
of
its
demo
applications.
One
is
called
the
emoji
photo
and
the
other
one
was
called
the
books
app
installing
from
from
mastery
and
also
forgive
silicon
or
installing
also
works.
D
D
Well,
see
you
later
so,
this
adapter
was
just
pushed
like
an
hour
to
a
fully
functional
state.
Pretty
much
like
you
know,
and
the
Wednesday
I
would
say
early
yesterday.
There
are
some
fine
killings
that
were
done,
but
but
this
adapter
is
now
functional
like
a
lead.
Actually,
that's
three
three
functions
pretty
well
and
of
course,
like
you
know
the
option
for
applying
custom
yam
bowls
on
the
cluster
I've
tried
to
retain
them
so
that
across
the
board
any
attacker
at
least
like
another
ones
which
are
being
developed.
D
You
know
yeah
by
the
team
here,
but
you
know
has
that
functionality
so
they're
not
just
limited
to
the
ones.
Sorry
to
the
operations
we
are
providing
them.
So
that
is
there
then
other
that
this
console
a
number
is,
is
also
created,
but
it's
very
bearable,
but
I
mean,
as
of
today,
like
a
leper,
you
know
it'll
compile
a
docker
images
ready,
but
then
it
doesn't
really
do
anything.
D
All
we
have
done
is
like
no
kind
of
Korea's
stopped
public
and
all
the
service
methods
have
been
needed
and
I
made
it
compatible
and
ready
to
go
like
you
know.
So
hopefully,
just
like
to
know
the
update
that
I
gave
a
few
minutes
back
pavani
and
her
team
will
be
able
to
data
from
you
kind
of
start
adding
functionalities.
This
is
kind
of
like
where
the
console
adapter
is
the
other
update
that
was
done
was
actually
you
guys
can
actually
see
the
two
fields
here.
D
So
as
far
as
the
test
one,
this
has
been
long
pending.
I
always
wanted
to
come
back
and
add
this,
but
I
since
I
was
busy
with
adding
the
other
functionalities.
You
know
just
not
being
able
to
would,
but
this
we,
like,
you,
know,
try
to
squeezes
in
so
every
time
a
test.
This
one
you
know
I'm
collecting
a
friendly
name
for
the
test.
It
doesn't
have
to
be
a
unique
name,
but
a
friendly
name
like
it.
D
D
But
of
course,
looking
out
there
are
times
when
people
just
want
to
run
a
test
for
10
seconds
or
maybe,
like
you
know,
if
you
want
to
run
it
for
like
an
hour
or
you
know
something
like
that,
so
so
yeah,
if
you
hover
over
it
like
I,
also
have
a
tooltip
which
actually
says
you
know
you
can
actually
use
an
H
M
or
s
suffix
for
our
minute
or
seconds
so
right
now
we
get
the
default
value.
I
have
said
there
is
30
seconds,
but
you
can
actually
change
it
to
like
5
minutes.
D
For
example,
it
does
have
validation.
So
again,
if
you
don't
have,
if
it
doesn't,
have
a
value-
or
you
know,
if
you
have
an
invalid
value
with
a
little
used
to
be
show
it
by
not
gonna,
buy
an
error
by
a
red
marker
of
the
field,
so
that
was
actually
one
thing
that
was
recently
added
other
than
that,
like
you
know,
I
think.
If
we
go
to
the
results
tab,
you
can
see
that
the
name
now
appears
all
the
old
test
did
not
have
a
name.
D
D
So
all,
but
but
all
the
searching
and
sorting
functionalities
were
actually
implemented
in
LA
I
can
ask
part
of
the
migration
to
post
sequel,
which
was
executed
later
last
week.
So,
but
this
is
what
I'm
just
reiterating
on
that.
So
now,
coming
back
to
the
adapter
itself,
just
before
I
show
that
just
a
second.
D
D
Now
the
mission
hunt
which
I'm
actually
running
this
is
obviously
not
a
powerful,
so
I'm
gonna
actually
connect
back
to
our
sto
adapter
for
just
one
second
difference
and
I
see
there
is
the
book
info
out.
That's
deployed
someone
remove
that
for
a
few
minutes,
but
I'm
also
going
to
remove
the
ISTE
of
installation.
Now
I
know
you
guys
are
actually
seeing
my
screen,
but
let
you
know
let
me
see
if
I
can
actually
also
share
my
other
one.
D
D
D
Namespace
we're
removing
and
now
they're
all
done,
which
is
fantastic,
so
switching
adapters,
so
I'm
just
going
to
actually
again
choose
my
config
file
again,
like
it'll,
be
a
working
on
improving
those
experience
are
probably
very
soon
we're
gonna
actually
have
this
change,
so
most
of
the
poor
guys
are
on
this
call.
They
can
probably
know
this
TV
adapter
was
running
on
four
10,000
by
default.
Now,
in
this.
A
D
At
ten
thousand
one
man
for
console
lectures
at
ten
thousand,
two
and
and
again
let
you
know
right
now:
I
have
all
the
services
running
on
the
local,
hence
as
localhost,
but
if
you're
running
using
docker
compose
are
using
the
CLI
tool,
which
is
which
is
almost
ready,
you
will
actually
be
entering
the
service
name,
which
is
specially
sto
or
actually
liquor,
D,
so
yeah,
please
don't
get
confused
by
the
localhost
I'm
using
here
so
now.
This
will
actually
connect
it
to
the
linker,
the
adapter.
D
So
if
I
go
to
the
play,
you
can
see
the
options
have
changed.
So
again,
all
these
options
are
again
coming
from
the
adapter
code,
not
hard-coded
in
memory.
Someone
first
try
installing
liquor.
Do
you
know
what
this
does
is
liquor
DS
way
of
installing
the
mesh
or
the
easy
way
to
install
liquid
is
actually
using
a
CLIA?
They
have.
What
this
tool
is
actually
does
is.
Does
exactly
that,
like
you
know
it
actually,
first
stalks
to
get
up,
gets
the
latest
words
of
information
and
then
uses
the
URL
information
from
the
metadata.
D
It
actually
goes
twice
to
download.
The
CLI
puts
it
in
temp,
folder
and
then
from
then
on
and
I'm
sorry
tries
to
execute
appropriate
commands
using
dot
CLI
to
actually
install
liquidy.
Now
the
difference
between
linker
DNS
TOS
are
sto
is
actually
something
which
usually
gets
deployed
at
the
cluster
level.
It
still
will
be
deployed
in
our
namespace,
but
in
order
to
keep
it
simple
like
that
for
SD,
adapter
I'm
totally
boring
the
namespace
field,
or
they
still
is
installation,
but
the
booking
for
installation,
for
example.
D
D
It
is
known
to
be
a
kind
of
fun
evening,
its
installation
to
a
particular
namespace,
so
I'm
gonna
actually
pick
a
namespace
called
liquor
DNS
like
liquor,
namespace
and
then
my
more
I
should
choose
it's
all,
but
before
that,
I
just
want
to
ensure
that
like
I,
don't
have
such
a
namespace
in
the
cluster,
you
can
see.
I
don't
have
a
namespace
of
this
is
again
a
ruminant
from
the
previous
install
but
I'm
just
going
to
leave
it
there.
D
So
there
is
no
namespace
called
linker
DNS,
so
I'm
just
going
to
use
that
and
I'm
gonna
try
to
install
liquor,
D
and
if
you
guys
have
been
using
my
sure,
you
probably
know
that,
like
you
know
the
installation
of
all
these,
like
don't
take
some
time
so
I
try
to
actually
push
them
to
the
background,
which
means
what
you
know
it'll
happen
or
synchronously,
and
you
can
see
that
the
linker
D
is
now
deployed
again.
If
you
click
on
it
like
there
is
nothing
nothing
fancier
here.
So
laters
were
something
Cadiz.
D
D
So
you
can
see
a
new
namespace
did
come
up
so
now
I'm
gonna
actually
see
if
there
is
anything
in
the
namespace
so
incur
dns,
yet
also
to
show
what
all
it
actually
installed
or
installing,
and
you
can
see
that
it
actually
installed
a
whole
lot
of
things
in
microseconds,
which
is
very
nice,
like
a
whole
lot
of
POD
services,
department
sever
up
the
gussets,
so
pretty
much
equities
installed,
and
what
other
thing
is
that
I'm?
Actually
installing
linker
D
with
the
proxy
ejector?
D
So
that's
one
thing
to
note
here
and
apart
from
that,
like
you
can
see,
link
ready
actually
is
also
deploying
Prometheus
and
Cortana
for
its
use,
and
we
can
actually
go
and
take
a
look
at
a
resume.
Thank
you
so
so
this
is
the
installation
of
liquidy
and,
like
you
can
see,
it
was
very
quick
now
I'm
going
to
actually
switch
and
install.
D
D
So
once
you
do
that,
like
all
the
applications
are
always
accessed
through
the
arrest
gateway,
which
is
kind
of
the
recommended
way
to
access
services
on
sto,
but
that's
probably
not
the
case
with
liquidity,
let
you
know,
because
link
Rd
does
not
really
have
the
concept
of
request
gateway.
So
essentially,
that's
all
it
is
I
actually
did
a
mistake,
like
you
know,
so
you
can
actually
see
that
I
choose
liquor
DNS
for
the
namespace,
which
is
actually
a
bad
thing,
because
what
it
did
was
it
actually
did
deploy.
D
The
apology
would
have
like
you
can
see
the
emoji
stuff.
Anyone
see
stuff
are
all
these
four
paws
which
are
sitting
right
next,
but
this
is
one
thing
to
be
cautious
about,
like
you
know,
so,
which
is
actually
a
mistake
which
I
did
so
I'm
just
going
to
delete
it.
I
probably
should
do
its
own
namespace
so
that
we
can
visually
see
the
the
stuff.
Is
this
kind
of
like
work
like.
B
To
say,
coming
back
sorry,
Grisha,
actually
on
that,
that's
maybe
a
good
thing
to
highlight
to
pratik
and
print
off
is
then
you're
paying
attention
to
and
separating
out
the
namespaces
in
which
you're
deploying
the
control
plane,
meaning
the
things
that
run,
link
or
D,
or
the
things
that
run
sto
and
the
the
sample
app
or
the
the
real
lab
itself,
you're
doing
it
just
for
cleanliness
and
hygiene
and
the
ability
to
look
at
things
separately,
but
RT,
Kanpur
knob.
This
is
a
good
test
case
for
us
to
capture
and
the
design.
Doc
is.
B
There's
a
bit
of
understanding
whether
or
not
there's
a
difference
when
you,
you
do
put
your
application
into
the
same
network,
namespace
I'm
starting
out
in
the
network
name
space,
but
the
same
namespace
or
separate
it
out
or
put
it
put
it
the
same
namespace
as
the
controller
separate
it
out
or
separate
it
into
multiple
namespaces
at
scale.
The
number
of
namespaces,
or
rather
at
scale
name
spacing-
and
this
is
new
in
1.1
sto,
when
at
one
takes
into
account.
B
B
D
Okay,
so
I
just
I'm,
not
sure
if
you
observe
like
I
just
uninstalled
and
reinstalled
incurred
a
matter
of
seconds
which
is
very
nice
okay,
so
so
now
I'm
just
going
to
actually
create
a
new
name.
Speer
is
called
emoji.
So
the
other
thing
is
that
if
the
namespace
given
here
doesn't
exist,
make
sure
you
will
create
the
namespace,
which
is
kind
of
nice
thing,
so
I'm
going
to
deploy
the
emoji
photo
app
in
the
emoji
NS
0
1
namespace,
so
I
just
want
to
hit
submit,
let's
see
if
it
actually
comes
up.
D
D
It's
gonna
check
what's
deployed
and
you
can
see
like
an
awful
lot
of
things
are
deployed
in
that
namespace
now.
The
other
thing
is,
you
know
it's
obviously
now
like
I
said
liquor
deal,
I
can
subsist
apply,
but
you
know
we
don't
know
how
to
access
it
yet.
So
what
I'm
trying
to
do,
let
me
see
if
it
actually
works,
is
that
if
you
click
on
the
details
here,
you
can
see
that,
towards
the
end,
I'm
also
actually
trying
to
suggest
that
the
service.
D
Four:
three:
zero
one:
nine
three
on
the
on
the
you
know
on
the
kubernetes
host,
so
I
want
to
try
if
I
can
actually
answer
since
35
with
the
API,
and
you
can
see
they
go
so
the
app
comes
up.
So
that's
what
this
was.
One
of
the
things
like
I've
tried
to
mention
me:
the
slack
channel
that
for
liquor,
D
demo,
applications
I'm,
also
trying
to
capture
the
port
number
on
which
the
application
will
be
exposed.
Also,
this.
D
D
Two
groups
are
zero,
one
as
a
namespace
name
and
then
I'm
going
to
try
to
deploy
them
boy
it,
and
it
says
that
it
is
deployed
so
I'm
just
going
to
confirm
that
it
is
deployed
just
to
be
sure
so
books,
AB,
0
1,
that's
the
namespace
name,
so
I'm
gonna
go
inside
it
and
see.
What's
all
deep,
why
sure
things
are
indeed
functional?
D
You
can
see
that
all
the
pods
are
up
in
my
face,
like
you
know,
the
pods
started
running
very
quickly
like
matter
of
seconds,
because
you
know,
since
I've
been
using
this
hose
for
all
my
tests.
You
know
the
container
images
already
existed
on
the
host,
but
if
you
are
doing
it
like
a
little
late,
they'll
take
a
little
more
time,
because
you
know
it
will
also
take
time
or
pulling
in
the
images.
So
please
be
a
bit
patient
now.
D
The
other
thing
is
like
I
said
earlier
or
like
for
the
boots:
AB
example
you
can
see
if
you
click
on
the
details.
It
actually
saves
that
now
this
application
is
exposed
on
a
different
port,
which
is
three
one
triple
seven.
Oh,
you
can
just
going
back
to
the
emoji
water
app.
You
can
see
that's
deployed
on
a
different
host,
sorry
different
port
on
the
same
host,
so
I'm
going
to
go
here
and
just
to
confirm.
The
port
number
is
three
one
triple
seven.
D
Let's
see,
if
that's
functional
three
one,
seven
and
voila,
you
can
see
the
books
info.
Sorry,
the
books
app
as
part
of
the
liquor,
DS
liquor,
D
Department,
is
up
and
running
now.
This
is
actually
like.
I
said
I.
Can
it's
all
on
the
mesh?
No,
how
do
you
know
that
so
as
part
of
the
pots
you
can
see
in
the
second
column,
but
you
actually
see
a
12-2
that
actually
says
the
number
of
containers
in
part.
D
B
Grease
time
you
interrupt
for
a
second
say:
well,
you've
got
13
minutes
left
I
want
to
make
sure
that
we
both
wrap
up
the
you
cover
any
of
the
other
stuff
outside
of
the
linker
D
adaptor
that
you
have
also
catch
up
with
Pavan
on
the
August
at
I/o.
So
maybe
this
just
flip
quickly
to
Pavan
sure
smells
good
and
it
will
come
back
because
I
think
you
could
go
all
day
absolutely.
B
B
D
I'm,
actually,
you
know
taking
notes,
as
you
were
speaking,
that's
this
computer.
What
I
was
trying
to
do
or
you
don't
want
to
get
up
on
like
you
know,
feel
free
to
go
inside
the
you
know
the
community
meeting
minutes
and
you
know
record
your
update.
Excuse
me
I'm,
making
notes
ensure
that
well,
you
know
what
I'm
typing
is.
Actually
there
I
think.
D
So
do
thank
you
yeah,
like
I'm.
Having
should
be
lobster
lagdi
like
it
would
be
very
helpful,
like
you
know,
because
I
think
they
give
like
an
hour
maximum
of
three
gigs
a
month
which
is
kind
of
like
in
a
way
more
than
what
paper
trail
offers
plus
I.
Think
the
limit
with
paper
trail
is
kind
of
hard
to
the
metalic
and
I.
D
Think
that's
what
we
confirmed
after
you
know,
talking
with
them,
which
means
I
think
the
first
one
they
give
you
some
some
more
room,
but
then,
from
the
next
month
onwards,
it's
a
maximum
of
15
eggs
a
month
which
I'm
pretty
sure,
like
you
know,
our
app
will
actually
kind
of
run
out
like
an
often
matter
of
like
in
our
days,
so,
which
is
a
reason
why
everything
we
thought
logs,
you
might
actually
be
a
better
candidate,
yeah
awesome
Leah.
Can
you
hear
me
yeah?
D
If
I
can
hear
you
grease?
Okay,
yeah,
okay,
cool
yeah,
so
I'm
not
sure
if
you've
heard
foam
on,
but
for
once
updates
I
mean
he
gave
us
a
piece.
I've
tried
to
record
them
in
the
minutes
and
that's
pretty
much
like
I
was
just
trying
to
explain
it
to
pull
on
we're
gonna
ye
logs
Ian
on
paper
trail.
You
know
we
just
we
haven't
want
to
delimit
from
paper
clip
which
is
a
hard
limit
of
50
Meg's,
which
means
they
don't
accept
logs
beyond
50
Meg's,
probably
following.
D
Maybe
that's
the
reason
why
you
were
seeing
some
logs
there
and
some
logs
I'm
just
guessing
but
but
I
think
with
logs
II.
It
might
be
better
and
also
I.
Think
logs
II
has
has
a
Cabana
like
interface,
which
I'm
not
sure
if
you
guys
have
used,
but
if
you
guys
have
used
elasticsearch
like
I,
think
you,
you
are
more
at
home
than
the
paper
trails,
so
user.
D
B
Very
good,
so
I'm
gonna
ask
her
Nava,
pratik
and
Vaughn
having
them
sort
of
haven't
gone
through
this
much
earlier
on.
What
do
you
think
of
the
newcomers?
Welcome
guy,
because
we're
gonna
have
a
couple
of
folks
coming
for
sure.
Take
a
look.
If
you
get
a
moment,
that'd
be
helpful.
The
other
thing
that
I
know
that
we
wanted
to.
B
We
put
some
effort
into
and
is
on
the
cusp
of
being
done,
isn't
quite
there
yet
is
a
refactored
set
of
initial
QuickStart
instructions
so
out
on
the
project
page
as
people
come
to
it
hopefully
they'll
understand
the
its
goals.
Maybe
they
look
at
its
architecture,
maybe
look
at
screen
stuff
name
or
not,
but
they
go
to
run
it
that
the
really
the
quick
start
way
of
doing
this
is
and
the
way
in
which
I
think.
B
Outside
of
that
pratik
I
guess
unit
it'd
be
good.
If
I'll
try
to
remind
myself
to
poke
at
you
early
next
week
and
just
collaborate
a
bit
on
or
see
what
I
can
do
to
help
with
respect
to
the
spreadsheet
that
you're
starting
to
form
the
results
are
some
cold-ass,
some
of
the
insights
you're
starting
to
identify
and,
yes
sure.
B
And
put
TD
you
what's
your
sense
of
well,
just
you
know
as
part
of
the
direction
you've
gotten
from
mojito
or
or
from
him
or
or
not.
He
is
maybe.
This
is
a
premature
question,
I'm
just
kind
of
I'm
thinking
toward
you
toward
the
end
of
end
of
the
project
or
at
least
or
newer
and
end
of
interest
in
the
project.
Hopefully
that's
not
soon,
but
how
do
you
imagine
the
results
coming
out
like?
Is
there
you
know
from
your
perspective?
C
From
a
conference
from
academic
conference
I
not
sure
because
these
are
empirical
results
so
like
at
this
point,
academic,
innovative
point
of
view-
I
am
not
sure
where
that
would
go,
but
from
a
point
of
view
of
presenting
two
places
as
you've
suggested,
like
UConn
and
Nancy
of
meetings.
That's
definitely
one
aspect
to
it.
C
There
is
tentative
and
cool
where,
after
doing
some
kind
of
micro,
benchmarking
or
benchmarking
of
service
measures
to
bring
in
accelerated
designs
and
things
like
that
and
seeing
if
those
help
in
in
our
service
mesh
at
all
or
how
unfair
and
things
like
that,
so
those
are
some
of
the
lines
that
me
and
Moffat
have
been
discussing
from
an
academic
perspective
nice,
but
that
that
is
a
bit
down
the
line.
First
of
all,
I
want
to
yeah.
B
B
Makes
perfect
sense
sounds
good,
but
by
the
way
there
is
the
possibility
of
hosting
a
meet
up
on
the
25th
of
April
here
in
Austin
I'm,
going
to
plant
that
I
think
about
whether
or
not
you'd
like
to
come
and
either
present
outright
or
be
part
of
a
presentation
of
measure
e
at
a
local
meetup.
Here
in
town,
the
25th
a
Thursday
few
weeks
from
now
think
about
it.