►
From YouTube: Kubernetes Community Meeting 20160707
Description
We have PUBLIC and RECORDED weekly video meetings every Thursday at 10am US Pacific Time.
https://docs.google.com/document/d/1VQDIAB0OqiSjIHI8AWMvSdceWhnz56jNpZrLs6o7NJY
Demo of Rancher; Demo of Kompose; SIG-API-Machinery; SIG-Testing; SIG-Cluster-Ops; SIG- Storage
A
So,
as
I
start
every
meeting,
this
is
a
public
and
recorded
meeting
of
the
Cooper
Nettie's
community.
So
please
welcome
everyone
and
we
will
get
started
with
a
demo.
Today
we
have
rancher,
Elena
is
going
to
give
us
a
demo
of
Rancher
and
then
sub
is
going
to
sub.
Goa
is
going
to
give
us
another
another
demo
of
compose
so
Elena.
Why
don't
we
get
started
with
you.
B
My
name
is
Alina
I'm
one
of
the
engineers
at
when
she
loves
I've,
been
working
on
the
cook.
Integrating
could
do
night
us
with
rancher
for
the
hofstra
for
the
past
half
a
year
and
I
would
like
to
share
this
work
with
you
today.
So
right
sure
is
the
management
platform
for
containers.
It
does
the
infrastructure
management,
access
control,
eight
foot.
B
B
B
So
yeah,
let
me
let
me
get
started
from
the
very
beginning,
so
my
name
is
Alina
and
I'm.
One
of
the
engineers
of
the
rental
house
and
for
the
half
past
year,
I've
been
working
on
the
integrating
could
earn
a
decent
run.
Cherry
and
I
would
like
to
share
this
work
with
you
today.
So
the
rideshare
is
a
management
platform
for
containers.
B
And
for
our
castration
platforms,
we
probably
be
support,
communities
swarm
and
masses
and
communities
has
been
the
most
photo
so
far.
Our
users
love
it.
So,
let's
get
started
and
an
environment
environment
in
rancher
is
the
way
to
share
the
resources
with
the
different
users
of
your
ganization
and
also
the
way
to
choose
the
orchestration
platform
for
your
environment.
I'm
going
to
put
the
lettuce
I'm
going
to
name
this
around
the
demo
and
I'm
going
to
invite
one
of
my
peers.
This
project.
B
A
B
D
B
I'm
going
to
pick
the
nearest
region
to
me,
San
Francisco
create
so
it
starts
in
an
urban
areas.
First,
we
are
waiting
for
the
hosts
to
be
active.
Then
we
are
going
to
create
a
cub
United
system
stack
and
then
we
are
ready
to
go
with
a
creative
serves
replication
controllers,
as
it
usually
takes
like
15
minutes.
B
B
So,
let's
start
creating
back
your
next
resources,
they
can
do
it
from
the
UI.
Let's
go
to
the
application
controller
and
a
dressy
okay.
So
we
wanted
to
provide
the
most
native
experience
for
communities
users
and
we
went
with
a
letting
user.
Add
the
replication
controller
service.
This
is
a
10-place,
so
I'll
just
create
a
copy.
B
B
B
We
also
allowed
to
use
that
cube
control,
so
you
can
either
do
it
from
the
UI
or
you
can
copy
the
cube,
convict
your
local
machine
and
wrong
cube
control
from
there
you're
going
to
ask
them
to
keep
the
user
and
we're
going
to
forward
it
to
the
Cougar
netis
cluster
that
belongs
to
this
user.
So
using
that
keeps
it
feel,
let's
create
the
ingress
controller.
B
B
Well,
yeah
you're
afforded
to
the
engine
app
to
see
the
details
of
the
load
balancer.
It
should
also
go
to
the
system,
help
and
here's
them
cougar,
lady
singer,
solbi
stack
and
here's
the
load
balancer
service,
where
I
can
see
the
details
it
for
the
traffic
to
engine
eggs
to
the
port
eating.
It
has
just
one
container
a
part
of
the
ingress
implementation.
French.
Are
we
allowed
to
scale
up
your
load
balancer
by
providing
the
custom
scale
in
your
ingress?
The
annotations.
B
In
addition
to
to
DUI
and
ingress,
we've
also
done
did
a
couple
for
the
specific
renter
features
like
we
provide
the
rancher
DNS.
It's
kinda
like
a
drop-in
replacement
for
Sky
D&S
is
available
by
default
and
we
also
provide
the
range
of
metadata.
So
metadata
allows
you
to
get
the
information
about
your
service.
Your
container,
I'm
going
to
show
you
in
a
minute,
say:
I,
haven't
12
education,
control,
running
I'm,
going
to
go
to
the
12
container
I'm
going
to
come
in.
B
B
There's
another
feature
that
Trenchard
provides
for
communities.
You
can
use
the
rancher
private
registries
here.
You
just
add
the
registry
of
either
dr.
Harbach,
where
your
custom
registry
just
provide
the
credentials
and
then,
if
this
registry
becomes
available
by
default,
the
communities,
so
we
can
try
it
our
own
credential
provider
on
the
next
side
to
support
that.
B
So
you
can,
you
can
create
the
services
from
the
UI
from
QC
TL,
but
also
we
have
something
called
catalog
where
user
can
share,
their
templates
could
do
native
templates
and
you
can
deploy
any
application
using
any
of
the
publicly
available
templates
and
you
can
share
your
own.
So,
let's
look,
for
example,
in
the
WordPress.
B
So
here
is
just
description
and
then
you
can.
You
can
preview
you
up
like
here's,
the
my
sequel,
llamo
WordPress,
yamel,
just
click
lunch
and
you
can
see
here
under
services.
You
see
the
WordPress
front
end.
If
you
go
to
replication
controllers,
you
can
not
look
yet,
but
you
can
see
the
water
press
in
my
sequel
as
well,
so
we
are
enhancing
our
you
guys.
You
can
clearly
see
like
what
application
got
deployed
from
one
catalog.
We're
also
going
to
implement
support
for
the
upgrading
of
the
application
to
the
latest.
B
B
B
So
I
want
to
showcase
something
that
is
going
to
be
available
in
the
next
tranche,
fruitless,
like
vocab
United
systems
tag.
We
also
deploy
it
from
the
catalog,
but
from
the
system
catalog,
and
we
currently
support
could
do
netis,
1124
virgin
and
we
are
going
to
enable
support
for
cornetist
1.3
version
in
the
next
release
and
once
this
version
once
this
template
of
this
version
is
available,
you
will
be
able
to
update
your
coordinated
system
stack
with
a
zero
disruption.
B
B
B
There
are
some
good
Unitas
1.3
features
that
we
are
particularly
interested
in
implementing
like
pad
set
and
brunette
is.
We
are
still
finalizing
our
design,
but
is
going
to
be
available
in
Rancher,
wonderful
version
as
well
and
I'm
done
with
my
dama.
If
anybody
has
any
questions,
you're
welcome
to
ask
them.
A
A
B
A
E
B
We're
on
our
ranch
or
actually
we're
on
our
ranch,
your
network
agent,
on
every
host.
This
is
a
system
dedicated
from
here
and
it
does
provide
the
dns
and
the
metadata
services.
So,
basically,
when
you
hit
the
range
of
metadata,
it
gets
redirected
to
the
to
the
air
to
the
network
asian
container.
My.
E
A
Take
that
there's
no
more
questions.
Thank
you.
So
very
much
Elena
have
a
yes.
This
is
super.
It
was
super
fun
having
you,
I
saw
wrench
or
get
demoed
and
was
like.
We
need
to
bring
this
in.
So
all
right.
So
we
have
another
demo
today,
because
we
had
a
pretty
light
agenda,
namely
no
specific
meaty
topics
to
discuss
unless
we
get
there.
But
let's
start
with
Sebastian,
showing
us
to
begin
they're
composed
sorry,
not
khoob,
test.
A
D
D
Yes,
okay
cool,
so
I'm
going
to
talk
about
compose
and
I
work
on
skip
box.
You
may
have
heard
of
us
a
little
bit
a
couple
months
ago.
We
demoed
k
machine,
but
here
we
going
to
talk
about
composed
with
a
que
tu
na
ng
yen
is
also
online.
Is
that
the
main
developer
composed
right
now
and
Chris
also
worked
on
compose?
You
can
reach
us
on
Twitter
and
so
on.
D
One
other
thing
we
we
did
earlier
early
on
was
to
work
on
a
community's
toolbox,
and
we
thought
that
we
needed
something
like
poker
machine,
but
for
Cuban
it
is
also
getting
a
development
environment
for
communities
single
node
and
that's
when
we
created
k
machine.
We
also
created
boot
to
KAS,
which
was
a
you
know
analogous
to
boot
to
docker.
Now
those
two
are
actually
superseded
by
mini
cube.
But
if
you
want
to
do
a
see
a
demo
of
chem
machine,
you
can
go
back
to
I.
D
Yeah
s'okay
machine
and
boot
to
care,
as
you
know,
we're
there
and
now
it's
superseded
by
by
mini
cube.
We
also
have
a
small
ski
box
UI,
which
is
analogous
to
tech
matic.
The
doctor,
you
I,
that
you
can
run
on
your
desktop.
It's
very
much
a
work
in
progress,
but
no,
maybe
in
a
couple
weeks,
I'll
demo
that
as
well,
but
if
you
think
about
the
doctor
tool
box,
now
you're
missing
dr.
Campos,
okay,
and
what
did
we
do?
Well,
we
created
composed
with
a
cake
and
what
does
compose
do
well.
D
It
takes
a
talker
compose
file
and
translate
it
into
criminate
these
primitives.
So
imagine
you
have
a
straight-up
talker
compose
that
works
with
a
regular
docker.
Now
you
can
easily
generate
replication
controller
services,
Damon
sets
replica,
said,
charts
and
so
on.
310
golang
available
on
github
open-source
apache
license
its
on
skip
box
/,
composed
to
the
two
there's
a
story
behind
it.
D
We
started
as
composed
without
the
number
and-
and
we
just
rebooted
the
project
last
week,
actually
because
there
is
some
interest
in
it,
so
we
cleaned
up
the
code
and
made
it
a
little
bit
better.
It's
based
on
lip
compose
from
docker,
okay,
which
has
a
gold
library
to
to
basically
parse
dr.
compost
pile.
Why
did
we
do
this?
Because
we
think
that
developers
use
docker
to
create
images
on
their
local
laptop,
that's
how
they
start,
and
then
they
have
multiple
images.
D
They
write
a
compose
file
to
to
write
a
full-blown
application,
potentially
they
targeted
at
swarm
or
rencher,
but
we
want
to
give
you
know,
developers
the
ability
to
take
that
compose
file
that
they
work
on
and
move
to
communities
in
a
very
simple
manner.
Okay,
so
that's
what
we
have
so,
let's
go
demo
I
only
have
five
minutes,
so
that
should
be.
It
should
be
quick
if
you
do
compose
and
you
look
at
the
help.
You'll
see
couple
comments.
D
Convert
up,
PS
delete
scale
I'm
only
going
to
demo
convert
a
PSD
beat
scale
work
in
the
previous
version
right
now
there
might
be
a
little
bit
broken
because
we
are
working
on
activity
and
they
all
those
command,
also
duplicate
what
you
can
do
with
cube
control,
but
let's
look
at
under
so
let's
drop
into
a
shell.
Hopefully,
you
can
see
the
shell,
so
Sarah
yell.
If
you
don't
see
anything
so,
let's
hook
it
here,
we
have
an
extremely
simple
daughter,
compose:
ok,
just
ya
mo.
D
It
says
there
is
a
ghost
service
image
ghost
and
now
what
you
do
is
that
you
do
blogger
compose
convert.
It
tells
me
Oh
ghost
dot
RC
created
so
God
goes
that
are
see
Jason
and
then
I
have
my
replication
controller
compatible.
You
know
for
sure
bandages.
So
that's
you
know
that's
great.
I
can
also
you
know
quite
give
it
to
standard
out
which
is
kind
of
nice,
and
if
I
want,
I
can
also
get
the
yellow
if
you
prefer,
yellow
or
to
json.
D
So
you
get
the
the
conversion
very
easy,
let's
clear
that
and
if
you
don't
want
to
generate
an
hour
see
so
we
you
know.
Of
course
we
have
to
make
some
assumptions
on
how
a
doctor
compose
maps
to
humanities
primitives
the
marketing
may
not
be
perfect,
but,
for
example,
let's
say
that
you
know
what
you
want
is
to
generate
a
demon
set.
D
You
know
now
you
get
a
demon
set
by
saying
that
the
HBS,
if
you
want
a
replica,
said
we
do
repeat,
I
said
just
dash
dash
RS
and
you
get
a
replica
set.
Okay.
So
that's
that's
quite
nice!
Now
what
what
you
want,
of
course
is
you
know
you
want
to
be
able
to
submit
that
to
a
relative,
so
here
I
have
a
simple
privileges
set
up.
Let's
look
at
my
reputation,
controller
and
services.
I
have
nothing.
So
let
me
convert
that
dr.
composed.
D
D
D
But
if
I,
if
I,
do
the
same
thing,
I
convert
this
and
instead
of
using
the
default
/
composed
file
I
said
take
my
engine
extra
llamo
Jeanette,
convert
it
to
start
out
pipe
it
to
keep
control
now
I
generated
an
edge
in
ex
Arce
and
a
service
I'm
going
quite
fast
here.
Hopefully
you
can
follow
an
automatic,
be
you
know,
I
generated
and
RC
and
also
a
service
at
the
same
time.
D
Right
now,
what
triggers
the
creation
of
a
service
is
actually
the
the
link
here
key
in
your
doctor
composed
and
we're
going
to
support
dr.
Campos,
b2
and
potentially
be
three
when
it
comes
out,
so
the
you
know
that
figure
to
generate
the
service
might
might
change
okay.
Obviously,
here
this
is
still
look
a
composed
version.
D
11
last
thing
that
I
wanted
to
show
you
is,
let's
do
a
conversion
into
a
chart
because
I
work
on
helm
a
little
bit.
So
let
me
convert
that
nginx,
dr.
compose
into
a
chart
and
simple
the
sheet
or
dash
dash
charts,
and
now
I
created
an
engine
ex-directory,
and
this
is
a
valid
chart,
so
it
created
a
skeleton.
Of
course.
Now
you
can
edit
that
to
make
it
better
and
in
the
templates
now
you
have
your
RC
and
service
definition
in
there.
D
A
D
Yes,
o
compose
composed
with
a
came
in
right
now,
it's
in
3d
a
transformation
tool
and
once
you've
done
the
transformation
you
know
you
can
definitely
fall
back
on
speed
control,
but
early
on,
we
did
support
that
compose
PS
composed
elite
compose
scale,
but
they
are
kind
of
redundant
with
you
control.
That
said,
you
know
if
you're
very
familiar
with
the
composed
common
line.
You
made
the
doctor
compose
command
line.
You
may
stick
with
using
composed
with
a
cake
for
doing
that.
Psd
eaten
from
it
was
the
question
it.
A
D
Its
base
to
take
developers
who
are
familiar
with
dr.
Campos
they've
developed
there,
dr.
Campos
file,
and
now
they
want
to
actually
move
to
Cuba
natives.
They
want
to
deploy
in
production
on
cumulative
and
they
can
still
use
their
doctor
composed
and
trade
up.
They
hit
a
community's
cluster
fantastic.
D
Now
that
we're
going
to
work
on
more
advanced
thing,
we've
been
talking
about
doing
annotation
of
the
doctor,
compost
pile
so
that
you
could
potentially
create
pods
with
multiple
containers.
You
know
by
by
taking
dr.
Campos
services
and
saying
hey
those
two
services:
I
want
them
in
the
same
pod,
so
we're
going
to
willing
to
work
on
more
features
so
that
you
can
you
can
control
what
you
meant
his
objects
you
generate.
F
A
G
A
You
Sebastian
all
right,
so
as
I
mentioned
at
the
top
of
the
meeting,
we
have
a
fairly
light
agenda
today.
So
up
next
would
be
any
cigs
special
interest
groups
that
wants
to
do
a
bit
of
reporting
and
you
can
take
notes
during
matt
and
alina
and
sebastian.
If
you
have
some
specific
notes,
you
want
to
add
in
to
the
to
the
working
document
for
this,
about
your
demos.
That
would
be
super
helpful
links
to
your
slides
and
the
specific
topics
or
points
you
want
to
make
in
there.
That
would
be
fantastic,
otherwise,
stick
reports.
H
Yeah
so
I
mean
we
have
a
sig
meeting
every
other
week
we
met
yesterday
we
had
a
lot
of
discussion
about
clients
and
what
we
want
to
do
going
forward.
So
we
talked
about
the
go
client.
We
talked
about
a
swagger,
we've
talked
about
making
other
clients.
If
that's
something
you
are
interested
in
helping
us
accomplish.
We'd
love
to
see
you
in
the
API
machinery,
sick.
A
D
A
A
F
H
Least
about
merging,
which
is
we're
still
working
on
getting
our
metrics
together
simac,
you
got
a
lot
better,
it's
in
deteriorating
a
little
bit
over
the
last
week.
Well,
that
seems
to
be
inside
Google
infrastructure
things
and
some
of
that
I'm
not
sure
I
haven't
looked
in
detail,
I'm
still
trying
to
get
one
change
into
ginkgo,
which
will
let
me
do
some
automatic
flaky
flaky
tests,
detection
in
PR,
PR
testing
I'm,
trying
to
eliminate
the
problem
where
you
have
to
like
go,
hunt
down
an
issue
and
ask
for
retest
manually,
okay,.
H
A
A
G
A
This
week
have
been,
of
course,
slower
as
we're
seeing
by
the
cig
saying:
nope
haven't
met,
oh
I've
been
on
vacation,
but
the
1.4
planning
has
been
going
on
and
I'm
going
to
push
on
everybody
who
is
in
six,
who
has
been
doing
their
own
cig
level,
planning
and
and
wants
to
send
that
up
to
the
coup
brunette
ESPN
group.
It
turns
out
that
the
two
people
who
have
been
reporting
out
from
the
sukkah
Brunetti
p.m.
group
this
week
are
also
on
vacation.
A
Although
I
think
some
of
them
are
lurking
here
on
bad
connections,
but
I
won't
put
them
on
the
spot.
So
if
you
have
work
that
you're
trying
to
commit
for
1.4
and
make
sure
that
they
get
into
the
list,
so
we
can
track
the
larger
features.
Please
make
sure
that
the
community's
p.m.
group
has
that
has
that
info
because
we're
starting
our
cover
Nittis
1.4
development.
This
week
we
collectively
all
of
us
and
I'm
sure
some
of
you
have
been
doing
it
already.
Oh
and
we've
got
two.
A
We've
got
one
updates
and
a
couple
of
questions.
So
let's
hit
the
1.4
because
I
was
just
talking
about
it.
Dims
asks.
Is
there
a
single
place
for
the
1.4
list
of
items
being
proposed,
and
the
answer
is
that
is
in
the
new
features,
repo
in
the
Cooper
Nettie's,
so
the
Cooper
Nettie's
organization
features
repo
and
that's
the
best
place
to
take
a
look
for
that
and
we'll
put
the
answers
in
the
notes.
Mike.
Thank
you.
A
I
So
last
time
I
checked,
there
is
about
five
or
six
people
working
on
the
document
and
essentially
we're
trying
to
get
a
little
headway
on
there
because
seems
to
be
a
precursor
for
a
lot
of
the
other
things
that
we
want
to
do
and
also
will
hopefully
help
inform
some
of
the
other
six.
So
that's
happening
today.
If
you
go
to
the
cluster
ops
slack
channel,
there
is
a
link
to
the
document.
Everybody
is
super,
encouraged
to
at
least
check
it
out
and
see
if
it's
something
you
can
help
with.
C
Six
storage
met
earlier
today.
There
was
an
agreement
to
do
a
retrospective
on
the
things
that
happened
in
storage
in
13,
because
storage
was
the
big
thing
that
caused
us
delays,
because
the
ridge
didn't
get
all
of
their
work
merged
quickly.
So
we're
going
to
try
to
talk
about
that
at
some
point
and
make
sure
we
don't
run
into
some
of
that
again.
The
big
feature
that
right
now
is
trying
to
be
discussed
in
storage
is
dynamic.
C
Provisioning
like
how
do
we
get
the
system
to
create
storage
for
you,
rather
than
you
having
to
do
it
yourself?
So
there's
a
big
push
right
now
across
the
whole
storage
community,
to
figure
out
how
that's
going
to
work.
There's
also
the
beginnings
of
rumors
one
day,
the
storage
sig
might
sometime
soon
get
together,
face-to-face
and
try
to
bang
through
things,
but
that
will
all
be
worked
out
in
mistake.
If
you're
interested
in
storage,
it's
an
active
saying,
please
don't
join
pay
attention.
Ask
questions
we're
all
here
to
help
so.
A
To
do
yes
more
things
into
the
features
PR
that
end
features.
Issues
specifically
is
where
to
look
for
the
one
point
for
planning
and
I
expect
that
we'll
have
a
big
agenda
item
next
week
for
1.4
planning,
as
well
as
as
well
as
I
hope
to
bring
forward
a
proposal
for
what
an
elders
group
would
look
like,
so
that
we
have
an
escalation
point
sort
of
the
Supreme
Court
of
the
kubernetes
community.
A
So,
in
the
event
that
there
are
any
sort
of
challenges
or
groups
can't
come
to
agreements,
you
know
sort
of
be
relative
consensus
on
their
own
there's
a
way
away
in
a
plan
to
escalate.
Oh
hey,
there's
a
Cassandra
demo
in
sig
apps
this
week,
wow
we've
got
lots
of
updates
from
people
and
there's
video
showing
that
says
see
real
time.
Alright,
I
have
nothing
more
than
a
couple
of
notices.
Does
anybody
else
have
any
more
topics
they
want.
A
No
all
right
so
that
doesn't
sound
like
we
have
a
lot
of
further
topics
to
discuss.
So
let
me
go
through
my
notices
quickly.
The
second
half
of
the
1.3
community
retrospective
is
going
to
be
a
week
from
tomorrow.
We
agreed
that
we
wanted
to
take
a
look
at
the
output
from
last
week's
first
half
and
add
more
notes
to
that
and
then
go
back
and
sort
of
do
the
the
drawing
up
of
the
conclusions
next
Friday.
A
If
any
of
you
are
planning
cooper,
Nettie's
meetups
in
july,
it
skipper
Nettie's,
first
birthday
and
so
the
last
week
of
July
we're
trying
to
have
a
bunch
of
Cooper
Nettie's
community
meetup,
community
birthday,
meetups
and
if
you're
interested
in
that,
then
please
send
an
email
to
myself
and
Cameron
zahedi.
Who,
let
me
also
introduce
Cameron
Cameron
is
another
new
person
here
at
Google
who
is
going
to
be
working
on
community
with
me?
So
I
don't
know
if
Cameron
you
have
video
handy
and
you
can
wave
at
the
friendly
face
on
video
awesome.
A
Let's
see
last
thing,
Oh
with
that,
of
course
come
streets,
so
we've
got
some
really
awesome
ideas
from
our
friends
over
at
core
OS
about
some
communities
cookies
that
will
go
out
to
people
who
are
having
meetups
and
party
hats,
because
what
is
a
birthday
party
without
party
hats
and
we'll
we'll
figure
out
some
other
swag
bits,
and
then
last
thing
is
almost
all
of
the
sig
mailing
lists.
Allow
public
posts
I
said
this
last
week,
but
feel
free
cross-post
share
information.
Try
to
spend
more
time
communicating
out
statuses
on
the
mailing
lists.
A
As
this
team
and
this
community
and
this
whole
crazy
ship
that
we
are
steering
keeps
getting
bigger,
communication
becomes
much
more
important
part
of
this,
because
you
can't
just
know
what
everybody's
doing
with
us
scattered
around
the
globe
across
a
bunch
of
companies
and
out
out
in
the
world
working
on
different
things,
with
different
priorities.
So
more,
we
can
communicate
clear,
the
better
that
that
will
be
and
the
more
helpful
it
will
be
for
everyone
stay
on
top
of
things,
because
nobody
that
I
know
can
keep
up
with
all
of
the
signaling
lists
directly.