►
From YouTube: Istio Community Meeting 20191017
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
A
So
the
first
equation
today
is
the
one
I
for
scheduled
know
what
output
is
coming
very,
very
soon
for
people
who
haven't
been
involved
in
any
of
the
COC
or
planning
meetings.
You
know
we
have
been
on
this
quarterly
release.
Schedule
which
has
been
doing
very
well
and
I'm
super
super
happy
with
it.
Our
last
release
was
in
September,
and
so,
if
we
release
a
quarter
after
that,
it
would
have
been
in
mid-december
and
we
decided.
A
Quarter
release
but
but
that's
what
December
least
so
we're
planning
to
release
in
the
middle
of
November,
that's
November,
14th,
I!
Think!
As
of
yesterday,
the
branch
occurred
and
the
community
Testing
Service
agent,
Ally
condors
on
either
the
working
groups
calendar
or
the
community
calendar.
You
will
see
community
testing
day.
One
is
the
details
are
not
in
the
calendar
event.
Yet
that
will
come
in
the
next
few
days,
but
it
can
head
to
a
London
and
table
and
according
on
arms,
up
at
this
October
28th
29th.
A
October
20
back
mid-second
can
do
testing
day.
So
if
you
have
been
part
of
one
of
those
before
it's
just
a
day,
some
people
actually
join
a
hangout.
Some
people,
don't
the
Hangout,
is
good
to
join.
If
you
want,
if
you
have
questions
and
and
we'll
we
don't
want
to
interact
with
somebody
in
real
time,
there's
usually
a
little
bit
of
a
war
room
going
on
so
I,
please
feel
free
to
join.
We
there's
a
lot
different
ways.
We
do
testing
one
common
ways.
A
That's
document
in
that
page
works
appropriately,
so
it's
fairly
easy
because
the
steps
units
do
are
well
documented.
They
are
the
documentation
and
if
you
have
any
problems,
you
can
bring
it
up
file
an
issue
fix
the
docs
whatever.
So
please
do
join
us.
There
were
a
bunch
of
event
that
happened
this
week.
I
happen
to
see
all
these
on.
Twitter
I
didn't
attend
any
of
them.
Unfortunately,
but
there
was
a
meet-up
all
things
open
men
and
Christian
were
there
with
the
Christian,
were
there
and
then
were
you
just
editing
it
off
in
the
meeting.
A
D
C
B
B
Even
said
this
will
be
the
last
ServiceMaster
discussion
of
the
day.
Apparently
there
were
a
lot
of
ServiceMaster
discussions
throughout
the
conference,
so
that
was
actually
really
good,
so
Linn
and
Christian
both
spoke
at
the
Meetup
Lynne
did
a
really
nice
overview
of
SEO
3,
and
that
was
great.
So
she
went.
She
went
through
all
of
the
the
new
features.
B
A
lot
of
the
usability
features
the
install
support,
but
what
was
really
cool
is
she
went
through
as
she
was
introducing
each
of
the
new
features
and
the
capabilities
she
actually
did
a
short
little
demo
of
it
live,
so
she
was
showing
some
of
the
utilities
and
showing
how
those
new
features
are
exposed
to
the
people
in
the
audience.
So
that
actually
went
over
really
well
and
everyone
got
a
really
good
understanding
of
what's
available
in
SDO
3
Christian's.
B
Session
was
right
afterwards,
it
was
about
30
minutes
as
well.
He
focused
on
more
about
adopting
a
service
measure
is
what
his
discussion
was
all
about,
and
it
focused
on
how
reasons
why
you'd
want
to
adopt
a
service
mesh
and
then
how
or
approaches
for
adopting
a
service
measure
and
the
the
main
outcome
of
that
or
the
takeaway
from
his
discussion
is
start
simple.
B
Agree
with
that
by
edge,
you
mean
ingress.
Well,
so,
yes,
ingress,
but
in
multiple
levels
of
ingress,
really
focusing
on
more
more
of
the
edge
and
API
management
capability.
That
solo
provides
and
then
driving
down
and
calling
into
his
do
gateways
within
applications
and
then,
as
you,
adopt
even
further
drive
down
within
the
namespace
or
within
a
cluster,
actually
implementing
the
full
mesh
with
the
sidecars.
So.
C
B
B
With
it
so
that
actually
went
over
well
and
then
he
opened
it
up
to
questions
because
he
had
a
bit
of
time
at
the
end
for
questions,
and
there
were
several
questions,
but
it
was
interesting.
A
lot
of
the
questions
from
the
audience
was
more
around
adopting
kubernetes
and
moving
workloads
from
a
hybrid
environment
from
on
premises
into
the
cloud
and
how
to
take
advantage
of
service
mesh
when
you're
doing
that.
So
there
were
a
lot
of
questions
and
that
kind
of
a
not
necessarily
directly
related
to
his
topic,
but
tangentially
related.
A
Yes,
I
think
that
I
think
that
that
incremental
adoption
is
really
important
for
almost
any
anybody
who's.
Taking
this
on
trying
to
do
all
of
it
at
the
same
time
is
yeah
more
than
you
need,
and
that's
not
because
of
this
do
by
the
way,
if
you
are
just
saying
hey,
you
adopt
a
new
networking
Zack
and,
at
the
same
time,
a
doc
I
anticipate
I'm,
implementing
new
significant
paradigm
from
three.
A
B
A
B
Not
then,
then
I'll
get
that
updated.
Another
thing
that
was
interesting
in
that
discussion,
so
obviously
Christian
was
focused
more
on
service
match,
not
necessarily
sto,
but
he
did
mention.
The
question
did
come
out.
Someone
asked
well
which
mesh
should
you
use
and
he
gave
a
good
answer
partly
politically
correct,
but
ultimately
it
came
down
to
that.
B
He
felt
that
sto
is
the
best
service
match
or
has
the
best
Community
and
Technology
basis,
and
he
does
believe
that,
while,
while
there's
no
single
one
service
mesh
to
rule
them
all,
he
does
believe
that
as
the
market
slims,
he
said
that
sto
is
highly
likely
to
be
one
of
the.
The
main
servers
meshes
that
are
going
to
continue
in
the
future
right.
A
Yes,
I'm
confident
the
duo
will
continue
in
the
future.
Meteor,
that's
good,
that's
good
and
I.
Think
it's
important
to
understand.
I
was
the
state
of
the
market.
You
know
how
the
project's
doing,
how
other
projects
are
doing
and
how
people
expand
event
out
there.
This
week,
Krista
I'm
busy
waiting
family.
He
completed
a
you.
A
With
Sunday
Parikh
on
will
be
tendon
ingress
that
there
was
a
low
volume,
but
we've
now
published
a
higher
quality
stream.
Yes,
everyone
look
at
that.
You
see
I
increasingly
see
kubernetes,
especially
in
on
Prem
use
cases
being
used
in
an
in
a
multi-ton
way.
In
the
cloud.
I
often
see
people
spinning
up
lots
of
kubernetes
clusters
every
brunt.
A
Every
team
that
wants
the
kubernetes
can
use
a
different
cluster
and
when
clouds
are
when
Koopa
dense
clusters
are
managed
for
you,
that
kind
of
makes
sense,
but
in
cases
that
you're
doing
a
lot
of
work
for
each
cluster,
then
people
are
tending
to
go
to
large
clusters
using
kind
of
namespace,
often
as
a
tendency
where
lots
of
different
teams
are
sharing
a
cluster
and
and
in
that
increasingly
common
use
case.
Multi-Tenant
ingress
is
a
key
part
of
it.
A
So
there's
a
video
everybody
can
check
out
I'm,
not
sure
if
we
post
that's,
the
SEO
stream
around
I'll,
make
sure
it's
on
the
sto
playlist
and
then
Megan
okie
foods.
Sometimes,
at
this
meeting
it
is
not
this
week
she
gave
a
talk
in
Boulder
and
it
is
an
intro
to
HBO
Talk.
It
is
not
a
deep
tour
through
the
features,
but
it's
a
really
good
intro
to
SEO
talk,
so
if
either
you're
joining
this
call
for
the
first
time
and
don't
really
know
everything,
he
would
like
to
know
that
it's
do
this.
A
A
E
A
We
yeah,
you
know,
I
actually
pinned
her
earlier
and
said:
hey.
Would
you
mind
coming
to
this
medium
and
maybe
spending
a
quarterly
meeting
on
a
politician.
She
gives
a
lot
of
these.
She
said
you'd
like
to
couldn't
this
time,
but
I
think
we
I
will
start
trying
to
heal
up
some
stuff.
In
fact,
Dan
Berg
is
going
to
be
hosting
the
one
up
for
tonight
from
now
and
wouldn't
surprise
me.
If
he's
got
some
demos,
he's
gonna
have
all
kinds
of
stuff
up
his
sleeve
because
it
is
Halloween
two
weeks
from
today.
A
A
Something
went
wrong:
monitoring,
Pablo
Montoya
with
the
Batmobile,
which
is
a
Spanish
telecom.
They
have
very
high
CPU
and
RAM
consumption
for
pilot
in
their
development
cluster.
This
is
not
in
their
production
clusters
in
their
development
cluster.
They
started
using
the
sidecar
object
and
reduced
their
CPU
consumption
by
90%,
and
then
they
upgraded
to
1:3
and
they
they
reduced
their
CPU
consumption
by
another
90
percent.
For
those
of
you
who
are
doing
the
math,
that's
a
big
reduction
in
so
I
wanted
to
point
out
a
couple
things.
A
There
first
is
that
dev
clusters
tend
to
have
lots
and
lots
of
deployments,
and
this
is
what
they
were
seeing
that
you
know
they
had
developers
who
are
repeatedly
repeatedly
deploying
in
there
and
every
time
you
know,
still
keep
track
of
the
changes
you
know
which
pods
come
up
and
down
and
where
they
are,
what
their
IP
addresses
are
and
when
you're
doing
lots
and
lots
of
deployments.
That
means
that
pilot
is
tracking
lots
and
lots
of
information
and
sending
all
of
that
information
or
default
to
all
of
the
proxies
that
are
out
there.
A
A
The
sidecar
is
perhaps
a
little
confusing
teenage
car
is
also
how
we
call
the
deployment
of
the
proxy,
but
the
sidecar
is
basically
a
resource
that
lets.
You
restrict
how
much
any
given
workload
knows
about
the
rest
of
the
network.
You
can
say:
hey
my
workload
only
calls
for
services
so
only
give
it
the
information
for
those
four
services.
In
doing
so,
you
drastically
reduce
the
amount
of
information
that
that
pilot
has
to
track
and
send
to
to
that
proxy,
and
so
that's
the
first
death
date.
A
It
is,
it
just
started
naming
their
dependencies
using
the
side
court
car
resource
it
brought
their
consumption
of
pilot,
as
I
said
down,
CPU
down
and
ninety
percent.
The
other
thing
that
happens
is
geo
is
getting
more
and
more
performance
over
time.
It's
much
more
efficient
working
hard
on
making
it
more
efficient,
so
upgrading
the
1.3
is,
is
also
really
useful.
I
just
wanted
to
call
that
out.
A
C
A
A
Their
their
usage
of
this
do
you
know,
and
he
said,
if
you're
using
it
in
this
scale,
which
may
also
are
you
using
the
using
the
sidecar
object
is
essentially
a
must
have
to
do
it.
So
I
wanted
to
call
that
everybody's
attention
rebuild
benefit
to
using
sidecar,
and
we
could
set
up
creating
two
1.83.
F
This
raining
I've
been
trying
to
play
around
with
the
MPLS
between
services
and
I'm,
not
sure
if,
if
it's
okay,
to
ask
the
query
here,
because
I
tried
posting
it
in
a
few
blah,
a
few
sides
and
I
did
not
get
any
response.
So
the
moment
I
establish
namespace
level,
MPLS
I'm
kind
of
getting
this
error.
The
upstream
connect
error
not
sure.
What's
going
on
so
the
moment:
I
disabled,
the
destination
rules
and
the
namespace
level
MPLS,
all
my
senses,
work
I
can
show
you
in
action.
American
share
the
desktop
and
I
can
show
you.
A
F
F
B
F
G
F
It's
for
all
type
of
services,
I
tried
it
with
another
or
namespace
with
where
I
had
zookeeper
and
detail
stateless
services.
There
also
I
ran
into
the
same
problem
and
I
tried
it
on
a
simple
namespace,
where
I
have
two
dotted
core
services:
one
UI
and
one
web
api
and
still
the
same,
maybe
I'm
doing
something
completely
wrong,
but
I'm
not
sure.
A
See
I
put
a
link
in
to
and
you
might
have
done
this
part
of
me
if
you
did
I
just
googled
a
the
error
and
found
where
someone
had
this
and
raised
an
issue
and-
and
there
are
some
debugging
tips
that
help
them
so
have
you
seen
it
then
feel
free
to
ignore
and
you
have
it
that's
one.
That's
one
place
to
look.
B
Other
the
other
thing
that
can
be
quite
helpful
is,
if
you
haven't
stumbled
upon
it,
yet
is
if
you're
using
if
you've
got
this
enabled
within
your
your
namespace,
the
mutual
TLS
go
ahead
and
use
the
sto
CTO,
often
command,
and
as
part
of
that
there's
a
TLS
check.
Okay,
if
you
haven't
used
that
I
highly
recommend
using
that
command
to
check,
because
what
it
will
do,
is
you
you
pass
in
basically
a
pod
in
the
namespace.
B
That
would
be
the
client
and
what
it
does
is
it
will
give
you
a
an
output
of
all
the
services
available
and
their
TLS
settings
from
its
point
of
view
as
a
client.
So
you
can
see
if
there's
any
what
it
expects,
what
its
configured
for,
what
type
of
communication
it's
going
to
traverse
to
the
server
what
the
server
is
expecting
and
whether
or
not
you
have
any
conflicts.
Hey.
Can
you
repeat
the
command?
B
B
Check
that
out,
it's
extremely
helpful.
It's
it's
a
nice
way
of
going
through
and
seeing
what
it
takes.
A
little
bit
definitely
read
the
instructions
around
in
the
help
around
it,
because
it
takes
a
little
bit
to
get
your
head
around
it,
but
keep
in
mind
that
it
takes.
It
provides
you,
the
TLS
and
the
security
information
from
the
point
of
view
of
a
pod
that
you
choose
within
the
mesh
and
you
can
narrow
it
down
to
a
target
service
as
well.
So
you
can
check
it
from
given
a
pod,
a
check.
C
F
And
the
next
issue
which
I
ran
into
is
usually
they
say,
as
I
mentioned
I
had
two
services
are,
you
are
an
API
when
I
have
the
I
have
graph
on
R
and
how
key
ally
everything
enabled
and
when
I
go
into
the
graph
and
strange?
You
know
if
I
have
all
those
headers
the
X
headers,
which
need
to
be
passed
in
in
the
request
response
enable
and
the
headers
are
being
asked
across.
F
B
B
B
By
the
way,
just
going
back
to
do
the
the
previous
error,
typically,
that
error,
the
where
you
get
the
connection,
failure
the
reset
headers
it's
because
the
server
is
expecting
mutual
TLS,
but
the
client
is
not
sending
it
and
therefore
the
the
server
side
is
not
permissive.
So
it's
not
allowing
clear
text
and
util
TLS
it's
requiring
mutual
TLS,
but
the
client
side,
the
pod,
your
originating
pod,
is
not
is
not
sending
the
traffic
encrypted.
B
F
F
F
Yeah
and
I
had
one
last
query
that
was
related
to
the
headless
services
right,
so
I
assume
the
main
branch,
the
1.4,
which
got
cut
a
few
days
ago.
I
suppose
headless
is
my
assumption.
True,
and
if,
for
the
supporting
do
we
need
do
we
need
any
say
it
still
llamo
to
be
configured
or
does
it
work
out
of
the
box?
The
headless.
B
F
A
B
F
A
A
Thank
you
for
bringing
your
questions,
appreciate
it,
and
just
so
everyone
knows
we
do
have
different
ways.
You
can
ask
questions.
It
is
absolutely
fine
to
come
here.
It
is
also
we
have
discussed
this
year
at
I/o
and
if
you're,
at
slack
calm
slack,
every
one
of
the
slack
is
discussed
is
more
like
a
message
board.
It's
durable
after
you
ask
a
question.
There.
C
A
C
A
B
A
A
bunch
of
stuff
in
there
and
and
in
addition
to
yeah
the
authentic
us
check,
if
there's
just
a
bunch
of
good
tools
and
if
you
look
at
experimental,
it's
getting
to
be
really
really
cool.
There's
a
bunch
of
stuff
going
on
it's
that
guy
OMI
I
saw
a
really
cool
demo
internally
this
week
and
I
might
see
if
they
can
get
them
to
come
to
demonstrate
some
of
these
features
here,
because
there's
some
there's
some
cool
stuff.
So.
B
F
A
A
C
C
A
This
is,
this
is
a
different
inventing
model
entirely.
So
I
would
say
it's
probably
a
compliment.
The
thing
that's
interesting
is
that
I
frequently
get
past
when
I'm
speaking
about
this
deal
about.
Well,
you
know
what
is
it
you
do
about
inventing?
How
do
you
deal
with
that
and
I
have
said,
I
think
that
we
as
the
service
mesh
industry,
that's
one
of
the
one
of
the
unsolved
problems
and
one
that
we
just
haven't
haven't
looked
at
yet
because
I
think
it
is,
you
know
if
you
are
have
systems
that
are
communicating.
A
You've
got
something
like
SEO
can
say:
hey
system,
a
it's
communicating
with
system
B.
If
some
of
that
communication
is
asynchronous,
you
likely
want
to
see
it
visualized
in
the
same
way,
of
course
I'm
thinking
visually.
But
this
means
all
the
tracing
everything
right,
so
I
think
it's
highly
related
and
in
fact,
well
we
haven't.
A
Deborah
is
essentially
a
sidecar,
for
you
know
for
these
asynchronous
events.
You
could
imagine
building
this
functionality
into
Envoy
and
saying
hey,
that's
couldn't
serve
the
same
thing.
I
haven't
looked
at
it
close.
We
have
no
firm
plans
right
now
to
do
that.
But
I
know
what
but
I
imagine
I
kind.
F
A
You
still
know
that
you're,
integrating
with
Kafka
it
seems
with
with
dapper
it
gives
you
kind
of
and
again
I.
Don't
know
this
is
just
from
Karina
blog
post.
It
seems
to
me
that,
without
integrating
with
a
message
queue
you
get
some
of
the
functionality,
but
if
it
again
I
I
think
that
if
someone
wants
to
really
dive
in
and
come
back
and
after
having
done
an
evaluation
and
kind
of
look
at
what
they
think
is
it
you
know,
the
difference
is
the
compliment.
I
think
it
would
be
a
great
day
for
the
community
sure.
F
A
Now
that
is
interesting
because
I
only
read
I
didn't
want
the
samples
I
just
read
the
blog
post,
which
seems
to
be
a
better
mechanism.
Yes,
so
I
guess
the
story
is
I,
don't
know,
but.
F
A
G
Then
I
have
a
quick
question.
I
can't
cite
right
now:
cuz
I'm
on
the
phone.
We
have
more
and
more
speakers
coming
to
present
about
SEO
at
the
meetups
for
kubernetes,
so
we
were
starting
to
examine
the
thing.
Maybe
we
should
start
they've
made
up
for
is
to
itself.
Do
you
guys
have
like
any
support
for
native
probe,
because
right
now,
I
mean
up
is
changing
their
policy?
There
will
be
charging
also
be
people
who
is
coming
to
the
meetups.
G
A
A
Up
the
steering
committee
I'll
put
it
on
the
agenda
for
next
week
and
we
will
we'll
look
at
and
decide
would
love
to
provide
support
for
people
who
want
to
do
sto
meetups.
There
is
definitely
interest
out
there
for
for
service
messy.
Looking
yep,
so
will
will
will
take
it
up
and
and
we'll
report
back
on
that.
Thank.
G
A
A
C
I
do
have
a
question
I'm
John
from
Cruz,
and
right
now
we
are
at
a
state
where
we
have
deployed
is
still
east.
You
are
there
clusters
and
the
next
step
is
just
starting
doing
some
load
testing
and
do
that.
We
have
gone
through
a
couple
of
documents
and
we
landed
on
using
fourth
year.
After
all,
and
the
first
question
will
be
like
it's
48
most
common
tool
that
community
is
using
for
load
testing.
What
are
the
other
alternatives
there?
People
that
scan
recommends
by
the
way.
C
The
reason
why
we
take
40
so
far
was
like
both.
It
is
a
good
tool
for
creating
generating
the
loads
accurately
and
also
it
provides
the
server
components
for
just
visualizing
the
results.
That
was
the
reason,
but
still
I'm
open
for
in
other
recommendations,
and
the
second
question
is
like
how
do
you
guys
automates
running
your
test
because,
like
okay
for
your
tool
for
just
running
them
for
a
single
time,
but
still
it
requires
some
extra
work
for
automating
gets.
A
B
We
yeah
we've
got
a
combination
of
that
and
and
other
homegrown
tools
for
internal,
especially
around
our
kubernetes
platform
for
testing.
That
I
was
just
thinking
Sura
to
see
what
tools
they're
using
for
again
from
the
environment
squad
for
SEO
performance
testing.
I
didn't
get
a
response
yet,
though,.
B
C
A
A
A
A
A
A
Okay.
Well,
once
again,
thank
you.
Everyone
for
attending
really
appreciate
it.
I
love
the
the
participation
in
the
community.
Dan
Byrd
will
be
hosting
next
time.
I'm
sure
he's
gonna
have
all
kinds
of
surprises
for
Halloween,
in
which
a
good
couple
weeks,
everybody
will
see
you
at
testing
day
on
the
twenty
seconds
all
right.
Thank
you.