►
From YouTube: Kubernetes sig-aws 20180309
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
But
hello,
everyone,
it
is
Friday
March
9th.
This
is
the
Sikh
AWS
bi-weekly
meeting.
We
have
a
fairly
light.
Well,
we
had
a
fairly
light
agenda
and
now
there's
lots
of
great
stuff
appearing.
Thank
you.
Arun
primarily
looks
like
so
looks
like
we
have
a
couple
of
things
and
we'll
get
right
into
it.
There
is
one
Publix
in
responsib
which
I
wanted
to
make,
which
is
that
what
Ted
is
coming
soon
and
I
think
in
this
particular
release
they
are
encouraging.
People
to
you
know,
try
it
out
on
your
list
about
give
them
earlier
feedback.
A
If
you
are
inclined
to
do
that,
that
would
obviously
be
appreciated,
but
one
tenth
should
be
landing
fairly
soon.
I'm,
not
aware
of
any
huge
AWS
specific
features
that
are
landing
in
there.
It
looks
like
there's
some
mock
support
for
for
using
eight
of
us
EBS
volumes
as
block
devices
rather
than
as
file
systems
on
that.
The
next
item
on
the
agenda
is
a
discussion
point
actually
gonna
hold
a
kind
over
to
Bob
here,
because
I
think
Bob.
You
have
a
broader
view
of
what
what's
happening
here,
and
maybe
you
couldn't
share
that
too
sure.
B
Trying
to
get
the
SIG's
to
be
I
would
say
a
bit
more
formal,
but
I
may
even
say
a
lot
more
formal
in
terms
of
what
they're
responsible
for
and
how
decisions
are
made.
One
of
the
things
that
starts
to
become
apparent.
If
you
read
through
the
templates
and
start
thinking
about
how
to
actually
adapt.
The
templates
is
the
notion
of
more
formal,
sig
membership,
and
this
is
something
that
Joe
bata
addressed
a
little
bit
at
the
community
meeting.
B
But
I've
had
this
conversation
with
a
couple
of
the
others
during
string
20
folks
around
the
notion
that
we
really
do
need
to
start
giving
a
bit
more
formal
about
how
like
who
who
and
how
we
vote
on
things,
especially
when
they're
contentious.
So
it's
one
of
those
things
that
doesn't
matter
until
it
matters
and
a
lot
of
the
times.
The
informal
process
has
been
working
fine.
B
B
So
for
us
that
means
we're
gonna
have
to
figure
out
how
how
and
how
how
we
figure
out
who
the
members
are
and
how
we
actually
record
that
and
make
that
a
more
formal
thing,
I
think
that
the
B
and
I'm
I'm
I'm,
let's
say
I,
haven't
heard.
There's
nothing
formal
here,
but
in
just
in
terms
of
discussions
I
think
it's
pretty
clear
that
the-
and
this
should
not
be
a
surprise
that
the
steering
committee
view
is
that
voting
on
the
project
should
really
be
a
contribution
contribution
based,
in
other
words,
a
contributor.
B
It's
a
dev
contributor
vote,
not
an
end-user
vote
and
I.
Think
you
and
Justin
you
and
I
talked
a
little
bit
about
this
earlier
in
the
week.
I
think
you
know:
sig
AWS
is
probably
a
lot
heavier
in
terms
of
end
users
relative
to
some
of
the
other
things,
certainly
relative
to
a
couple
of
the
other
six
that
I
attend.
B
American
involvement
regularly
very
contributor
happy
and
it
would
probably
never
even
come
up
so
I
think
you
know
one
of
the
things
we
want
to
make
sure
we
do
is
not
have
not
let
there
be
any
surprises,
make
sure
everybody
has
a
chance
to
talk
about
it.
We
can
always
take
an
alternative
back
to
the
steering
committee.
We
get.
We
get
some
self-determination
here
comments
I.
A
Mean
I:
have
it
obvious
how
much
as
I
love
that
our
sig
is
like?
Has
a
broad
community,
Ike
I
feel
like
we
develop
based
on
like
the
like?
Even
if
you
don't
code
telling
us
what
works
and
what
doesn't
is
a
valuable
contribution
right,
it's
almost
harder
to
know
what
to
code
than
it
is
to
code
it
up.
Something
like
the
coding
is
the
easy
pay
at
the
end
of
the
discussion
as
it
were.
Oh
I,
don't
want
to
have
a
communities
community
I
think
we
should
have
community
community.
A
B
So
a
good
good
one.
One
of
the
important
topics
which
I
think
is
next
on
our
agenda,
is
to
talk
about
sub
projects.
So
how
do
we?
How
do
we
approve
and
agree
that
the
sig
is
going
to
take
on
responsibilities
of
a
sub
project
or
create
new
sub
projects
based
on
discussion?
So
that's
clearly
one
that
the
member
the
member
group
would
vote
on.
Let
me
let
me
and
Doris
your
point
of
view.
B
There
adjust
them
that
it's
really
important
for
us
to
be
working
with
users
and
be
getting
good
feedback,
and
we
really
want
to
make
sure
that
happens.
I
I
would
be
completely
I,
wouldn't
I
would
endorse
the
notion
that,
if
someone's
showing
up
and
giving
us
lots
of
feedback
that
that
is
about
a
highly
valuable
contribution,
I
think
that
the
the
place
I
would
be
a
little
uncomfortable
is
like
just
you're
just
showing
up
and
lurking
on
the
sig.
B
Every
week
is
probably
not
quite
probably
doesn't
wouldn't
for
me
passed
across
the
bar,
so
I
don't
know
how
I
don't
know
how
we
quite
do
that.
But
you
know,
I
think,
I
think,
a
informal
way
to
just
say:
hey
yeah,
great
you've
been
showing
up
and
giving
us
feedback
and
participating
it
can
be.
We
can
be
very
liberal
in
terms
of
how
we
grant
membership.
C
B
So
so,
at
the
last
community
meeting
I
asked
this
question
and
a
Brian
grant
was
very
specific
about
it,
and
you
know
the
degree
to
which
his
opinion
is
shared
by
the
steering
committee
on
sort
of
an
official
basis.
I
couldn't
I
couldn't
say,
but
he
clearly
is
one
of
the
you
know
really
key
people
on
the
project
and
his
his
view
was
that,
like
a
two-level
way
to
view
this,
like
kind
of
membership
which
is
sort
of
everybody,
that's
interested,
and
then
you
know,
contributors
loosely
defined,
I
would
say,
I
would
add.
B
C
Everyone's
reminds
specifically
I
mean
I,
don't
I,
don't
feel
the
need
to
vote
on
most
things
I'm
here
to
understand
where
things
are
going.
You
know
there
may
be
a
point
where
I
do
have
a
strong
opinion,
but
that
hasn't
happened
yet
and
I
haven't
been
here
all
that
long.
But
I
think
that
sounds
like
a
good
balance.
C
D
I
understand
that
with
I
understand
that
perspective
and
I
understand
where
Brian's
coming
from
police
said,
the
I
would
assume
that,
like
city
a
juror
and
the
other
user
groups
that
are
more
cloud
to
find,
you
know
bare
metal
as
well.
You
know
this
is
more
a
user
group
that
is
of
users,
none
of
developers
to
Justin's
point
so
I'm
wondering
if,
if
we'd
be
a
little
bit
more
unique
in
that
manner,
I
I.
B
Think
that's
exactly
the
point
Justin
was
making.
You
know,
I
I,
here's
the
thing
that
I
think
would
not
be
good
for
us.
Yes,
someone
shows
up
and
lurks
and
then
a
critical
vote
comes
up
and
suddenly
they're
voting
and
they've
never
said
anything
or
participated.
Just
because
they're
showing
up
they
get
a
say.
You
know,
I,
don't
think
that's
really
the
I
don't
think
that's
really
the
basis
on
which
projects
like
kubernetes
kind
of
survived
and
do
well.
B
B
B
A
A
B
A
B
B
A
Name,
yeah
I,
think
that
is
I
think
that's
definitely
ground
silver
to
proceed
on
and
certainly
like
I
I.
Don't
want
to
put
words
to
do
what
smells
out.
We
started
with
the
project
that
I,
don't
I,
hope
no
one
would
object
to,
but
like
I
would
hope
that
most
of
our
projects
are
going
to
be
fairly
out
contentious,
and
so
yes,
if
people
are
the
people,
have
objections
and
it's
interesting
to
know
what
those
objections
are
at
least
to
me.
A
So
they
I
just
as
IB.
We
talked
about
this
briefly
I
think
two
weeks
ago.
The
a
proposed
first
sub
project
is
an
implementation
of
the
API
server.
Api
server
can
now
support
encryption
so
that
your
data
Annette
CD,
is
encrypted
with
a
secret.
There
is
a
plug-in,
a
GRDC
API,
to
provide
the
encryption
keys
using
four
different
mechanisms.
There's
going
to
be
a
GCE
kms
provider,
we
should
have
a
the
proposal,
is
to
have
an
AWS
kms
provider
so
that
your
keys
would
live
in
native
OS
kms,
and
we
talked
about
that
briefly.
A
Two
weeks
ago,
Robin
Percy
stepped
up
as
well
to
be
my
proposed
C
good,
proposed
co-owner.
Anyone
else
is
also
very
welcome
to
contribute,
of
course,
in
any
level
they
want
to,
but
we
are
proceeding
with
that
and
if
anyone
objects
that
I
would
like
to
or
if
anyone
has
any
reasons
or
concerns,
then
you
know
I'd
love
to
hear
those
did.
B
A
I'm
happy
to
do
that.
There
is
a
doc
that
I
very
brief
talk
that
I
circulated
to
like
Robin
and
I.
Think
Chris
must
happy
to
share
it
with
anyone.
I
but
yeah
pull
request
would
be
good.
I,
don't
see,
we're
still
figuring
out
the
process,
so
I,
don't
I
the
I
think
the
biggest
we
have
to
figure
out
a
name.
A
A
A
Yes,
I
think
I
think
if
so
these
situations
are,
if
anyone
is,
has
an
objection
to
this
notion
of
this
being
our
for
sub-project
and
please
say
something
or
if
you're
not
comfortable,
saying
something
ear
then
feel
free
to
be
used.
We
judge
me
personally,
and
otherwise
we
will
be
progressing
with
the
assumed
lazy
consensus
of
the
sake
and
if
it
turns
out
we
need
formal
consensus,
then
we
will
put
it
to
a
vote
of
our
undetermined
membership
rate.
B
I
I
think
we
had
to
do
a
pull
request
to
get
I
think
the
last
time
we
ran
across
this
air
and
actually
when
and
did
the
created
something
for
us,
so
I
think
you
may
have
to
actually
have
to
do
the
pull
request
anyway.
To
get
the
get
someone
on
the
steering
committee
to
actually
create
the
stuff
bits
we
need
the.
A
A
Okay,
that's
good.
Alright
I
didn't
I,
you
I
was
hoping
it
was
gonna,
be
too
contentious,
so
that
is
good
and
yes,
we
are
still
figuring
out
the
processing
it
may.
It
may
come
back
to
haunt
us
for
another
another
week,
but
hopefully
hopefully
we
are.
We
have
consensus
on
that.
So
shall
we
move
on
I
think
a
run
has
an
item
which
is
the
chaos
toolkit
as
we
need
more
chaos,
a.
F
Run
at
you:
yes,
sir
chaos
is
always
good.
Isn't
it
and
I
guess
brings
some
normalcy
to
our
lifestyle
right,
fantastic.
So
there
is
this
I
think
a
non-profit
or
I
think
a
company.
You
know
that
builds
I've
not
looked
into
how
this
project
was
started,
but
there's
this
project
called
as
chaos.
Toolkit
I
had
a
chat
with
them
earlier
this
week
and
they
are
very
interested
in
actually
working
with
AWS
to
figure
out
how
chaos
toolkit
how
chaos
engineering
could
particularly
be
applied
in
a
kubernetes
land.
F
So
I
had
my
first
discussion
with
them:
I
think
Monday
or
Tuesday,
I
guess
and
now
there
is
a
issue
open
in
the
kubernetes
workshop
that
we
have
built
and
delivered
at
cube,
corn
and
raiment,
and
that
pull
request
should
be
coming
out
soon,
hopefully
sometime
next
week.
So
the
idea
is
that
will
show.
Essentially
once
you
have
deployed
once
you
have
done
a
deployment.
F
F
What
I
can
do
is
I
can
actually
share
the
PR
that
the
issue
that
was
filed
this
morning
we
talked
about
it,
a
while
back
I
mean
a
few
days
back,
but
I
can
drop
the
link
in
the
chat
window.
So
take
a
look
at
this
way,
and
this
at
least
shows
what's
happening.
So
I
would
say,
follow
this
issue.
The
PR
would
be
sent
against
this
issue,
and
once
this
comes
along,
then
I
think
this
would
be
a
good
piece
of
information,
forcing
AWS
to
be
aware
of
at
least.
B
Quick
question
there:
there
are
two
ways
to
I
think
fairly
distinct
ways
to
apply
chaos.
Engineering
in
a
kubernetes
system,
one
is
to
apply
it
at
the
application
level.
I
know
killing
off
pods
would
be
an
example
of
doing
something
like
that.
The
other
would
be
to
apply
it
at
the
kubernetes
control,
plane
level,
killing
off
at
CD
cluster
bits,
or
you
know,
API
servers
and
so
forth.
Is
there
a?
Is
there
a
kind
of
direction
that
the
swings?
In
other
words,
is
it
viewed
as
kind
of
application,
level,
tooling
or
infrastructure
level?
F
This
is
more
application
level
tooling,
although
I
will
wait
for
the
PR
to
come
through
to
see
what
it
looks
like
really
I
mean
I
didn't
read
through
their
website
and
I
just
had
a
like
a
30
minute
discussion
with
these
guys,
but
once
the
PR
comes
along,
no
I
want
to
always
start
with
a
simple
hello.
World
C
was
possible,
and
once
we
understand
that,
then
the
idea
would
be
to
kind
of
let
this
evolve
in
a
little
bit
broader
chapter
by
itself
and
of
course
encompass.
F
You
know
how
can
we
introduce
start
with
a
simple
deployment,
for
example,
but
then
start
killing
the
pods
left
and
right
and
then
you
know
of
course,
start
killing
nodes
start
killing
at
CD,
because
I
think
that
will
only
help
and
then
get
something
very
AWS
as
if
I
can
see
you
know
if
we
can
kill
some
artifacts
over
there.
So
I
think
there
are
different
levels,
just
not
just
the
application
and
the
infrastructure,
but
because
we
are
talking
about
AWS.
Maybe
look
at
some
AWS
aspects
as
well
over
there.
F
Yeah-
and
this
will
I
mean
the
way
I'm
looking
to
design.
This
is
definitely
first
at
the
app
level,
because
that
is
more
critical
to
begin
with,
so
that
you
know
we
can
figure
out
what.
Hopefully,
some
recommendations
will
come
out
of
this
exercise
that
okay,
when
you
are
designing
your
app
or
when
you're,
creating
your
kubernetes
manifest?
A
B
But
that
is
a
great
call
out,
especially
around
aj
testing.
We
need
a
lot
more
of
this.
You
know
my
my
kind
of
inclination
is
the
likely
as
soon
as
you
get
into
anything
the
infrastructure
layer.
It
has
all
the
cloud
provider
issues
like
if
you're
gonna
kill
a
VM,
that's
gonna,
be
as
part
of
your
chaos
toolkit.
That's
clearly
gonna
be
a
clowder
cloud,
specific
kind
of
operation.
F
Agree
and
the
way
I'm
looking
at
it
is,
let's
start
when
we
have
already
started
working
on
this
effort.
Let's
see,
you
know
how
it
evolves
and
if
the
discussion
needs
to
enough
different
seg
or
at
a
different
level,
I'm
completely
open
for
that,
but
I
think
let's
put
a
stake
in
the
ground
and
see
what
needs
to
be
done
and
have
a
formal
chaos
testing
as
part
of
kubernetes.
A
There's
there's
some
great
yeah.
There
definitely
are
some
things
that
we
definitely
do
in
specific
state
of
us,
like
I,
think
you'll
be
hilarious,
like
no
one,
six,
nine
to
five
four
and
maybe
don't
root
PC
to
API,
were
introduced
like
packet
loss
of
their
I.
There
definitely
some
things
we
don't.
Any
research
will
be
funny.
F
Right
and
so,
if
you
guys
have
any
ideas
in
the
in
the
sense
of
you
know,
just
in
your
ideas
that
are
brilliant
and
bump,
your
classification
is
also
very
relevant.
So
if
you
guys
have
any
specific
ideas
on
what
you
would
like
to
see
as
part
of
this
chaos
toolkit,
you
have
the
issue
in
the
chat
window.
I
would
definitely
say,
add
any
scenarios
over
there.
F
My
first
PR
is
going
to
be
relevant
to
just
a
simple
deployment
and
adding
chaos
with
it,
because
I
like
to
start
simple
and
no
curl
walk
and
run
so
once
I
crawl,
then
I'll
be
able
to
say
okay
now
add
more
scenarios
but
feel
free
to
add
your
scenarios
or
use
cases
or
thought
processes
that
we
should
cover
this
as
part
of
the
chaos
to
hook
it
all
up
for
it.
So.
D
A
F
So
here's
the
project
that
I've
been
working
on
for
the
past
few
weeks,
essentially
in
this
project.
What
I
have
is
I,
have
three
services
and
I'm
talking
about
a
web
app
service
that
receives
a
request
from
the
client
which
then
talks
to
a
greeting
and
a
named
service
and
returns.
A
response
back
so
essentially
shows
three
simple
micro
services
and
how
they
work
with
each
other
to
return
a
response
back
to
the
client.
F
Now
these
services,
if
you
look
at
it
or
deployed
using
docker,
just
obtained
docker,
compose
darkest
form,
ECS,
kubernetes
and
lambda,
and
if
I
get
into
and
the
idea
is
to
deploy
the
exact
same
application,
three
micro
services
using
different
compute
options
on
AWS
if
I
get
into
Kate's.
Basically
now
in
kubernetes,
I
have
a
standalone
manifest,
of
course,
so
I
can
click
on,
say,
standalone
here,
and
here
is
my
manifesto
Chambal
and
it
just
says:
ok,
deployment
and
service.
You
know
pretty
straightforward,
then
the
second
deployment
model
is
helm.
F
So,
ok,
we
need
to
create
the
help
charts
so
now,
I
have
the
handout
short
Hamill
values
templates.
You
know
I
got
stuff
here,
not
using
some
templating,
but
not
a
whole
lot,
but
I'm
sure
it
could
be
improved
and
made
much
better,
but
this
is
essentially
a
hand
chart
and
then
last,
but
not
the
least
I
have
a
kisana
application
as
well.
This
is
pretty
brand
new
literally
got
it
working
yesterday,
but
what
I'm
looking
at
here
is
from
kubernetes
perspec
we
got
standalone,
manifests,
help,
charts
and
kisana
applications.
F
B
I
think,
in
terms
of
application
packaging,
this
looks
pretty
good.
I.
Think
the
the
the
rough
edges
are
the
which
I
don't
think
as
a
community
we've
sorted
out,
but
it's
the
the
notion
of
operators
and
they,
you
can
have
a
helm
chart
to
install
a
database
like
my
sequel
well
or
something
just
installing.
It
is
not
really
quite
enough.
You
actually
need
to
automate
all
the
run
books
around
it
and
trying
to
understand
like
like,
but
operators
are
also
like.
B
Presently
this
sort
of
custom
thing
they
you
know,
may
use
helm
like
an
operator
might
use
home
as
part
of
what
it
does.
But
there's
this
I
don't
know
sort
of
ambiguous
packaging
about
what
is
it
really
and
how
does
it
work?
So
it's
not
really
it's
just
more
a
general
commentary,
because
this
is
something
like
that
topic
is
something
I've
been
doing
a
lot
of
head-scratching
on
lately
and
it
seemed
like
a
good
time
to
mention
it,
but.
B
Think
yeah
there's!
Yes,
you
might
because
you
know
generally.
The
idea
is
that
you
know
an
operator
probably
needs
to
do.
Do
some
deployment
check
on
something
it's
maybe
gonna
be
doing
like
managing
the
upgrade
to
deployments
and
so
forth.
I
I
don't
know
I.
Maybe
this
is
a
rat
hole,
but
it's
definitely
something
I've
been
like
scratching
my
head
around
ahead
or
around,
which
is
I.
F
B
Depends
on
on
within
this
context,
how
things
like
upgrades
are
managed
and
I
suppose
in
a
lot
of
cases,
it's
really
being
managed
by
the
effectively
that
management
is
done
by
whatever
CI
or
CD
system
good.
However,
the
CDs
system
is
being
used,
whereas
operators
kind
of
interfere
a
little
bit
with
that
model,
maybe
in
a
positive
way,
maybe
not
anyway,
I'll
sort
of
withdraw
the
comment,
because
I
don't
think,
there's
something
anything
actionable
here:
okay,
front:
okay,
this
one!
That's.
F
A
F
F
A
Image
with
the
get
shot,
for
example,
is
what
I
do
personally
and
you
then
have
to
tag
it
right.
So
you
don't
know
sleep
but
like
a
sauna,
is
a
tool
to
address
that
there
are
a
plethora
of
others
that
were
talking
about
ones,
ways
to
make
that
more
in
the
standard
you
know
like
Brian
grants.
The
clarity
of
application
management
is
is
an
ongoing
proposal,
which
is
very,
very
good,
but
I
think
I
think
the
three
you've
got
there
are
the
right
three
to
start
with.
Okay,.
F
All
right
that
sounds
good,
so
now
one
of
the
things
and
as
we
look
at
different
deployment
pipelines,
we'll
definitely
have
one
of
the
options
is
as
far
as
the
deployment
pipeline
you
have
to
patch
the
manifest
so
that'll
automatically
be
taken
care
of.
So
definitely
looking
at
that
and
on
the
deployment
pipeline
side
we
are
looking
at
Jenkins
as
the
one
options
Cote
pipeline
is
another
one
and
possibly
spinnaker
so
before
I
get
into
the
deployment
pipeline,
so
my
ask
would
be
if
people
can
take
a
look
at
this
see.
F
If
these
help
charts
are
these
Kisan,
it
charts
or
standalone
manifests
you
know:
do
they
look
good
or
do
they
need
to
be
improved,
or
you
know
worked
upon
in
any
sense,
I'm
always
happy
to
get
feedback
or
you
know,
make
any
changes
over
there.
So
that's
one
I
think
that
was
my
second
point.
Actually
so
I
just
wanted
to
kind
of
share
that
you
know
I've
been
looking
at
these
essentially
taking
these
applications,
deploying
them
as
a
kubernetes,
deploying
them
in
a
kubernetes
cluster
running
on
AWS.
A
Yeah
I
think
this
is
great
I.
Think
it's
I
think
it's
like
having
this
view
of
how
you
know
suppose
you
have
a
committees
cluster.
What
now
right,
that's
a
that's
an
important
thing
and
I
think
this
is
a
great
start
like
I.
Don't
think
this
will
be
the
end
like
I.
Don't
think
it'll
be
like
you
write
this,
so
you
never
revisit
it
because
I
imagine
this
will
be
something
that
we
iterate
a
lot
over
the
next
like
years,
but
yeah,
it's
great,
oh
yeah!
Oh.
F
Yeah
yeah
absolutely
I
mean,
as
a
matter
of
fact,
the
way
I'm
looking
at
this
is.
This
is
scheduled
as
a
three-hour
tutorial
at
an
event
in
Paris
in
next
month,
so
essentially
in
those
three
hours,
I
need
to
talk
about
how
my
applications
are
designed,
how
I
create
the
docker
image
and
the
full
table
of
contents
is
here,
so
it
talks
about
how
do
I
build
the
docker
image.
How
do
I
deploy
the
application
to
kubernetes?
F
You
know
the
different
ways:
how
do
I
do
for
gate
or
ECS
in
lambda
and
for
each
one
of
those
essentially
I'm
talking
about
my
local
dev
test,
debug
environment,
my
monitoring
environment
and
that's
sort
of
my
next
topic
as
well.
But
how
do
I
do
my
pipelines
so
I'm,
looking
at
Jenkins,
spinnaker
and
Cote
pipeline
sort
of
the
three
common
options,
so
one
of
the
big
gaps
that
I
see
for
kubernetes
in
general
is
very
heavy
op
centric
and
we
don't
do
a
lot
of
coaching
around
app
development.
F
A
F
I
think
I
think
that's
a
good
point,
though.
So
what
I'm
gonna
do
is
I'm
gonna
take
a
look
at.
Maybe
that
is
another
application.
Packaging
format
for
deploying
kubernetes
application,
so
standalone,
helm,
hasanat
and
docker
compose.
Maybe
I've
not
seen
very
many
people
using
docker
compose
to
deploy
to
a
company's
cluster,
but
I
think
it's
about
education
and,
being
you
know,
a
stack
or
application
framework
agnostic,
saying:
okay,
what
here
are
four
choices:
we're
laying
out
all
the
cards
and
you
pick
what
works.
I
was.
F
A
F
Sure,
okay,
so
before
I
get
here,
I'll
find
this
issue
later,
but
a
new
tab
here,
so
I
spent
a
significant
amount
of
time
yesterday
with
the
AWS
x-ray
team,
because
one
of
the
tools
one
of
the
good
parts
of
x-ray,
is
it
really
allows
you
to
track
across
your
multiple
applications
or
multiple
micro
services.
So
to
say
on
allows
you
to
trace
the
call
you
know
and
if
you
can
dig
into
like
it
shows
you
a
very
nice
visual
diagram.
So
let
me
actually
bring
up
something
in
a
second
for.
F
Open
tracing
open
tracing
the
format
is
not
that
rich,
but
we
are
working
on
making
it
to
be
open,
tracing
compliant
because
the
payload
that
goes
exchange
between
different
you
know,
x-ray,
daemon
and
the
x-ray
agents
is
far
richer.
So
we've
been
working
with
the
open
tracing
team
to
make
it
compliant.
Is
it
the
same
space
as
open
tracing
very
much
very
much
yeah
yeah,
but
it's
a
managed
service
by
AWS.
F
So
essentially,
what
you
do
is
and
I'll
walk
through
the
programming
model
as
well,
but
once
you
have
configured
it,
then
what
you
see
on
the
screen
is.
This
is
a
client.
This
goes
to
my
web
app,
and
this
goes
to
my
name
container
and
a
greeter
container
and
is
showing
the
service
and
at
any
point
of
time,
what
I'm,
showing
you
is
a
slide
deck.
But
if
I
double
click
on
a
particular
circle,
it
will
dig
deeper.
F
F
Yes,
it
is
on
how
to
get
this
up
and
running
on
kubernetes
cluster
I
already
worked
with
somebody
from
the
X
one
of
the
solution:
architects,
who
basically
showed
how
do
you
do
application
tracing
on
kubernetes
with
AWS
X
ray
polymers
again,
I
can
finish
this
logo,
so
then
the
blog
shows.
Basically,
how
do
you
set
this
up
so
I'm
trying
to
mimic
all
of
that
in
the
repo
here?
So
if
I
go
to
my
tab
here,
hims
a
docker
file,
the
taco
file
says:
how
do
you
create
a
docker
image
for
X
ray?
F
Then
it
says
how
do
you
deploy
it
as
a
daemon
set?
You
know
I
got
the
daemon
set
here
then
we
can
actually
go
to
the
application,
say
my
services
and
greeting
and
my
source
file
and
if
I
look
at
my
actual
source
code
here
and
if
I
look
at
my
greeting
endpoint
there,
it
says:
okay,
how
do
you
add
X
ray
imports
and
how
do
you
can
begin
an
end
segment?
So
I'm
hoping
I
want
a
fully
functional
example.
F
There
is
one
which
exists
for
nodejs,
but
I
want
to
make
something
for
Java
as
well.
So
once
that
sample
is
up
and
running,
then
I'll
be
again
happy
to
share
the
link
with
rest
of
the
folks
makers.
Now,
essentially,
you
can
deploy
your
kubernetes
application
on
AWS,
just
integrated
with
x-ray
and
then
maybe
in
a
future
meeting,
and
hopefully
in
the
next
week
week
or
so,
I
should
be
able
to
show
a
live
working
example
on
how
these
things
work
and
show
actually
complete
drill
down.
A
F
So
there
is
no
sidecar
or
anything
running
like
that.
Essentially,
x-ray
runs
as
a
daemon
set,
which
is
sort
of
so
now
all
I'm
doing
is
in
my
application.
I
add
you
know:
x-ray
begins
segment
and
sub
segment,
so
I
have
to
instrument
my
application
once
you
instrument
your
application
and
then
depending
upon
the
language
that
you
are
using.
F
So
if
I
then
I,
look
at
say,
start
of
being,
I
need
to
have
one
singleton
beam,
which
will
say:
okay
go
talk
to
the
today's
ECS
plug
in
and
I'm,
giving
a
lot
of
feedback
to
the
x-ray
team
that
guys
I
want
I
mean
we
are
piggybacking
on
ECS
plug-in
today,
but
I'm
giving
I've
given
feedback
to
the
x-ray
team.
Already
that
I
want
a
first-class
support
for
kubernetes
in
x-ray.
So
I
would
like
this
to
be
a
kubernetes
plugin
instead,
so
essentially
you
can
set
up
your
x-ray
as
part
of
your
application
say.
F
Apps
and
if
I
look
at
my
kubernetes
and
my
help
chart
and
I
look
at
my
templates
and
I
look
at
my
greeting
deployment.
All
I
have
is
just
a
AWS
x-ray
daemon
address
here,
which
points
to
the
x-ray
service
default,
which
is
basically
how
my
x-ray
daemon
is
set
up.
So
you
need
to
specify
this
environment
variable
because
that's
what
x-ray
daemon
you
know
is
where
it
started,
and
then
your
services
started,
which
is
the
x-ray
service,
our
default
name,
and
then
you
need
to
instrument
your
part
and
that's
about
it.
F
F
F
A
G
F
I
showed
this
one
here,
I
would
love
to
submit
this
talk
with
anybody
in
CIWS.
Basically,
it
says
kubernetes
eyes.
Your
java
application
is
take
the
same
application
that
I
showed
earlier,
but
it
walks
you
through
different
phases
of
how
do
you
take
your
java
application
converted
into
a
kubernetes
application?
You
know
manage
it,
monitor
it
deployment
pipeline
its
service
visibility,
all
of
those,
so
this
is
purely
from
kubernetes
running
on
an
AWS
cluster,
so
I
think
it'll
be
a
fun
abstract
to
collaborate
on
this.