►
From YouTube: Istio September 21 Community Meeting
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
All
right,
it
looks
like
got
most
of
the
people
on
now
so
which
probably
get
started
so
April
is
out
this
week
and
has
asked
us
to
kind
of
run
the
community
meeting
in
our
absence.
So
that's
why
you're
seeing
me
rather
than
April
and
the
link
to
the
agenda
is
in
the
chat
for
anyone
who
has
just
joined
and
we'll
get
started
so
first
off
just
a
couple
of
announcements
and
for
that
turn
it
over
to
Mike
snow
from.
B
Thanks,
like
I,
said
regarding
April
I've,
actually
known
any
broker
for
several
years.
Well,
nicest
people
ever
met
him,
so
I
know
she
really
wishes.
She
could
be
here
for
this,
but
which
I
had
a
family
situation
come
up,
definitely
miss
her,
but
she
will
be
back
next
week.
So
in
terms
of
announcements,
one
thing
we're
really
excited
to
announce
here
at
play
era.
Is
we're
actually
releasing
this
your
training
so
now
available
on
time
era?
Dial
is
a
introduction
to
service
mesh
everybody.
B
C
B
Somebody
wants
you
to
service
mesh
or,
if
any,
issues
with
the
bra
and
key
again,
this
train
is
available
on
Tiger
Tahoe.
Any
questions
about
the
training
can
be
directed
to
myself.
At
my
cat,
Tiger
dial
or
Tom
was
the
next
meet
here
at
t
be
so
letter,
T
hasn't
Tom
letter
B,
isn't
blossom
that
Tiger
dial
and
there's
only
a
second
all.
D
We
had
a
lot
of
activity
last
month
or
so
since
last,
community
coal
is
being
a
changes
to
the
configuration
model
now
we're
using
crts
instead
of
TBR's,
just
like
the
rest
of
his
show,
and
we
also
added
a
lot
of
support
for
alternative
service
discovery
services.
So
we
have
basic
integration
working
of
console
and
this
is
a
prototype
working
with
Utica
there's
a
demo
of
console
with
pilot
I,
think
by
Greg
he's
on
the
call
here
and
we
also
improved
the
model
itself.
D
D
E
F
And
it
is
a
custom
community
testing
on
Monday
that,
if
folks
are
interested
to
try
out
the
latest
master
branch,
the
release
candidate,
we
feel
very
happy
to
like
you
know,
have
folks
test
those
and
give
us
feedback
on
the
dogs
and
the
installation,
the
user
experience
and
things
that
we
can
actually
improve.
Maybe
not
major
features,
but
actually
at
least
minor
tweaks
count
around
limiter.
A
G
So
I
think
some
point
during
this
cycle.
We
make
sure
it's
sent
out
the
0.2
design
dock
for
the
adapters
and
we
have
followed
along
with
that
design,
and
we
are
almost
done
with
the
implementation
of
everything
in
that
design.
We've
transferred
all
of
the
existing
adapters
over
to
the
new
model,
and
that
includes
defining
templates
for
new
mesh
functions
to
match
the
old
hard-coded.
G
Will
be
gathering
in
Seattle
at
the
end
of
next
week
and
we're
hoping
to
get
a
community
event
organized
so
if
that's
something
that
you'd
be
interested
in
joining
or
finding
out
about,
please
let
me
know
here
in
the
chat
or
send
an
email
to
sto
integrations
and
we
will
try
and
make
sure
that
we
can
get
something
set
up
so
we'd
love
to
interact
with
people
and
talk
about
all
the
changes
that
happen
with
mixer
and
what
we
think
is
gonna
happen.
Next,
that's
sort
of
my
update.
H
H
Okay,
okay,
so
first
they
offer
up
several
options
of
route
base.
We
support
of
all
the
ends
and
the
way
we
we
distributed
team
specific.
Is
it
different
for
well-being?
We
use
a
note
agent,
so
basic
user
agent
is
only
on
each
of
all
diem.
You
create
a
so
there's
no
agent
creative,
probably
keep
up
a
pair,
and
the
interest
is
art
is
to
see
a
and
then
the
vsonic.
The
facility
is
sent
back
to
note
agents.
H
So
in
the
future
we
tend
to
do
this
for
the
net
as
well,
because
it's
kinda
we
using
scarier
than
a
secret
bottom
mount
for
heinous,
significant
provision
which
is
considering
not
secure
and
also
we
have
another
updated.
We
support
the
kiosk
traffic
parties
to
the
TTL
support
helper
T
to
be
traffic
and
also
achieved
like
we
have
another
PR
persist
in
the
CA,
a
canister
storage.
So
right
now
as
a
first
steps
on
the
PA,
the
key
and
instead
is
generating
memory.
H
So
it's
not
persistent,
so
the
ideas
that
do
by
the
key
in
13
to
the
city
decayed
so
mean
to
the
secrets.
So
whenever
the
CA
gotta
crash
or
restarted
so
was
you
to
see
the
same
tina
search
so
in
this
way
they
can't
even
get
it
in
the
node
agent
will
still
work,
that's
pretty
much
out
date,
her
point
to
and
the
weapon
would
be
a
lot
of
things
to
do.
4.3.
I.
A
F
Sure
so
we
had
changes
in
proxy.
We
ordered
up
a
lot
of
play
in
house
code
that
we
were
actually
using
board
thread,
pooling
and
or
GRCC
libraries,
and
we
started
using
his
malevolence
some
of
the
features
that
we
actually
added
the
ability
to
turn
on
metrics
but
turn
off
like
in
a
policy
checks
and
vice
versa,
so
that
you
can
actually
like
you
know
you
can
deploy
the
mixer
and
use
metrics
for
all
internal
services
and
or
for
and
turn
on,
the
policy
met
policy
checks
only
for
services.
F
So
that
is
one
of
the
key
things
and
then
we
have
also
added
support
for,
like
you
know,
in
Envoy
itself,
we've
added
support
for
like
talking
to
headless
services.
This
is
one
of
the
the
pain
point
that
was
there
from
zero
point.
One
point
six,
so
we've
added
some
features
the
non
way
and
we're
taking
advantage
of
those
features.
When
configuring
envoy
pilot
such
that
today,
you
can
go
and
deploy
your
stateful
sets
zookeeper,
you
name
recorded
and
it
would
work.
So
that's
that's
another
big
feature
and
of
course
there.
F
I
I
J
Has
become
famous
for
you,
have
a
product
page,
a
ratings
reviews
and
details
micro
services
and
different
versions
of
this
review
service,
product,
page
information
from
all
those
services
and
displays
it
in
one
spot
and
reviews.
Here
there
are
three
different
versions
and,
depending
on
the
version,
it'll
call
this
readings
lists
to
get
a
variety
of
stars
to
play
with
the
review
now
in
terms
of
getting
ISTE
over
running
on
backer.
J
J
J
And
sure
enough,
but
on
there's
white
stars
and
if
I
sign
out
again
I
go
back
to
red
and
if
you're
not
familiar
with
the
book
info
demo,
there
are
a
few
other
sample
rules
for
testing
default.
Our
thoughts
and
delays
and
traffic
split
and
those
are
all
available
out
there
on
the
SPO
dong-il
website,
and
that
can
be
all
so
that's
about
it.
For
me,
any
questions
so.
A
J
So
the
coconutty
api
server
is
just
one
component
of
kubernetes
and
it's
just
running
in
stand-alone
without
those
other
components,
and
so
that's
just
available
out
there
an
image
that
you
can
pull
from
dr
hub
and
yet
registrate
also
another
open
source
project.
It's
just
a
matter
of
running
the
desired
micro-service
container
with
the
SPO
pilot
process
and
inside
there
so
that
it
starts
up
a
proxy
so
that
you
can
reach
the
individual
micro
services
in
your
application.
K
J
K
F
K
A
L
F
This
is
just
the
purpose
of
the
demo,
actually
so
yeah.
So
in
reality,
like
you
may
I
may
not
be
running
Dockers
and
docker
compose
bed
consoles
I
mean
you
might
turn
in
consulting
your
VMs
or
wherever
you
want
to
right,
and
so
the
idea
is
that
you
cool
okay,
the
sidecar
with
the
application
process
so
and
then
the
IP
table
the
rules,
it's
just
another
convenience
thing
to
automatically
reroute
all
traffic
bhaiyyaji.
J
I
A
M
Hi,
this
is
Shannon,
hey
it's
more
of
a
brief
introduction
into
cloud
County's
journey.
So
far
with
regard
to
the
steel
project,
introducing
cloud
foundries
use
cases
to
the
projects
community
and
our
intentions
for
collaboration,
so
I
think
I
have
Gabriel
rosenhouse
who's,
an
engineer
on
the
cloud
foundry
project,
I'm,
a
product
manager
for
routing
and
load
balancing
in
cloud
foundry
I
know
that
this
project
was
born
out
of
kubernetes
use
cases,
but
to
provide
a
brief
summary
of
cloud
foundry
to
provide
context
for
our
use
cases.
M
M
M
We
support
tens
of
thousands,
hundreds
of
thousands
of
tenants,
and
so
most
of
what
we
do
is
targeting
that
side.
That
kind
of
scale
and
multi-tenancy
we've
had
several
discussions
so
far
with
folks
in
the
steel
project,
most
recently
Shriram
gave
us
some
of
his
time.
It
was
a
very
viable
conversation.
M
We're
targeting
pilot
initially
and
the
work
that
we've
already
done
actually
is
where
is
in
flight,
is
to
deploy
on
boys
with
every
container
for
the
purposes
of
a
service
mesh,
we're
interested
in
leveraging
sto
both
for
north
wet
north-south
and
east-west
traffic,
as
well
as
using
that
API
for
service
discovery
and
look
to
collaborating
with
the
community
in
coming
weeks.
I
just.
K
Want
to
add
one
thing,
which
is
that
may
be
the
biggest
architectural
difference
between
the
way
that
Cloud
Foundry
is
imagining
using,
is
do
and
the
way
that
it
seems
to
be
envisioned
with
kubernetes
is
that
with
Cloud
Foundry?
We
expect
that
a
single
SEO
control
plane
would
be
accessed
by
many
mutually
trusting
tenants
and
so
we're
interested
in
adding
multi-tenancy
support
into
the
control
plane
in
order
to
support
that
use
case.
K
A
E
K
A
N
We
we
do
move
it
into
the
normal
kind
of
go
to
go,
expect
things
to
be,
and
then
we
have
a
script
that
makes
siblings
from
vendors
into
Basil's
external
packages
and
that
way
both
of
them
coexist
and
then
it
even
yeah.
So
vim
works
pretty
similar
that
they
all
uses
the
external
tools
and
yeah
I
just
basically
used
right.
E
M
Have
one
for
Shriram
and
let
me
know
whether
I
should
take
this
to
the
Envoy
list,
but
we
have
infrastructure,
as
we've
discussed,
for
managing
and
rotating
certificates
that
are
unique
to
a
container
and
I.
Think
one
of
the
the
challenges
will
be
interested
in
resolving
is
being
able
to
hot
rotate
those
certificates
in
envoy.
F
Yeah
I
think
that
feature
is,
is
a
to
do
in
on
one.
So
today
what
we
do,
that's
the
reason
we
have
this
pilot
agent.
It's
basically
monitors
that
file
and
any
time
one
of
these
certificate
files
change.
It
does
a
hot
reload
upon
way,
but
the
the
ability
click
watch
a
particular
directory
and
whenever
the
certificate
files
change
automatically
those
them.
That
is
a
to
do
non
Hawaii,
and
it's
just
that.
Somebody
has
put
pic
of
the.
H
Theater
he's
working
on
this
feature
is
called
key
discovery,
service,
kts,
I'm,
not
sure
if
you've
done
a
call,
but
thinking
that
they
gave
more
context,
though
I
think
the
idea
is
to
provide
this
key
as
certificated
as
an
API.
So
every
time
we
don't
need
this
to
load
different
file
and
every
time
we
need
it,
we
just
call
it
so
we
can
automatically
achieve
this
rotation.
We
don't
have
to
like
have
a
pilot
agent
to
hary
star
boy,
but.
H
Yes,
yes,
yes,
not
because
then
the
the
provider,
the
API,
can
be
local
agent,
like
we
have
an
old
agent,
so
that
one
can
actually
be
the
host
of
the
API.
So
first
key
ideas.
So
it's
sort
of
the
local
call,
but
that's
the
ghost
in
this
the
goosing
is,
we
don't
have
to
install
it.
There's
a
proxy
to
reload
the
key
in
the
certificate.
I
M
F
Yeah
so
I
mean
the
key
discovery.
Mccann's
would
only
work
as
long
as
you
have
something
that
can
serve
out
certificates
over
a
localhost
connection,
so
the
the
need
for
monitoring
a
file
system
and
loading
certificates.
That
is
also
in
the
plan.
It's
a
to
do,
and
somebody
has
to
just
take
that
were
nothing
but.
M
F
That
is
pretty
simple,
I
mean
it's
like
PRI
welcome
on
the
are
my
front,
so
it's
much
more
smaller
and
focused
so
just
file
an
issue.
Now
probably
there
is
already
an
issue
and
then
mackerel.
You
implementation
skies.
If
you
wanted
one
or
if
you
didn't
already
know
you're
we're
on
the
code,
then
it
can
just
go
ahead
and
contribute.
If
not,
he
can
give
you
an
implementation
sketch,
and
you
know
we'll
be
there
like
del
Plata
people
who
can
actually
like
help
out
and
reviews
and
feedback
and
so
on.
That's.
M
K
O
F
K
In
the
current
sort
of
proof
of
concepts
that
we're
playing
with
it's
yes,
it
would
be
similar
to
that
demo
that
we
saw
earlier
where
the
Envoy
is
in
this
same
container
as
the
application.
But
we
are
also
looking
at
adding
a
native
concept
of
pods
to
Cloud
Foundry,
which
would
then
make
it
much
more
natural
to
map
the
could
the
kubernetes
approach
onto
Cloud
Foundry,
but
that's
adding
pods
to
Cloud
Foundry
is
not
on
the
it's.
It's
gonna
take
a
little
while
to
get
there.
So,
in
the
meantime,
we
injected
back,
like
we've.
N
K
K
That's
the
idea
we're
trying
to
make
it
as
statically
compiled
and
not
dependent
on
the
file
system
as
possible.
I
think
there
may
be
a
couple
of
small
dependencies
like
if
you,
if
you
Cloud,
Foundry,
push
a
docker
image.
That's
maybe
missing
some
basic
C
library,
something
might
not
work,
but
the
the
general
idea
is.
You
should
be
able
to
drop
this
onto
a
and
next
to
pretty
much
anything
and
regardless
of
what
built
packet
was
compiled
with
okay.
F
K
That's
a
really
good
question.
I
think
we
haven't
settled
on
whether
we
want
to
make
envoy
operate
as
a
transparent
proxy
in
Cloud
Foundry,
or
whether
we're
gonna
instead
have
a
different
mechanism
for
routing
telling
the
application
to
connect
through
the
Envoy.
So
foundry
already
has
an
idea
of
the
service
binding
in
which
an
application
looks
inside
of
its
environment,
variable
to
discover
how
to
connect
to
a
service
yeah.
That
would
be
a
natural
place
to
inject
something
like
a
1
to
7
address
for
the
Envoy
alternately.
K
N
F
K
F
Yes,
I
think
we
do
have
plans
for
supporting
the
non
transparent
proxy
mode,
and
so
we
are
halfway
through.
As
I
said
like
we
have
GT
proxy
support.
We
just
need
to
be
able
to
generate
the
unroyal
listeners
on
IP
table
equivalent
format.
It's
a
it's
a
small,
it's
a
very
small
set
of
pixels
that
we
have
to
do,
but
we've
actually
get
to
that
room,
and
this
really
is
the
is.