►
From YouTube: Kubernetes Community Meeting 20180117
Description
We have PUBLIC and RECORDED weekly video meetings every Thursday at 10am US Pacific Time.
(Sound missing for first 30 seconds, but kicks in right after, sorry about that!)
Notes: https://docs.google.com/document/d/1VQDIAB0OqiSjIHI8AWMvSdceWhnz56jNpZrLs6o7NJY
A
Welcome
everybody
to
the
kubernetes
community
call
I
am
Tim
pepper
and
open
source
developer
at
VMware.
In
today's
meeting
moderator.
This
meeting
has
a
rotating
set
of
volunteers,
moderating
taking
notes
as
well
as
presenting
and
if
you
want
to
contribute,
I
suggest
reaching
out
to
Jorge
Castro
or
Jorge
rather
Jorge
on
slack
to
volunteer
the
meeting
is
being
recorded.
Please
mute
your
microphone
if
you're,
not
speaking
as
a
courtesy
to
the
other
speakers
for
those
presenting
and
asking
questions,
just
be
mindful
that
we
have
a
full
agenda
and
time
slots.
A
We
do
want
a
collaborative
meeting.
Obviously,
but
at
times
we
might
have
to
take
something
off
line
just
to
keep
the
conversation
there
going
and
keep
the
meeting
going
as
well.
The
meeting
minutes
are
being
kept
in
a
Google
Doc
by
Jace
and
the
meeting
recording
is
being
run
by
George
and
will
be
posted
to
the
kubernetes
YouTube
channel
following
the
meeting
so
first
off
today
we
have
a
kubernetes
on
docker
for
Mac
demo,
which
I
think
should
be
pretty
exciting
for
folks
and
take
it
away.
B
C
B
B
D
B
And
so
in
advanced
here
you
can
set
the
memory
for
for
doctor.
You
can
set
proxies.
You
can
set
some
flags
on
the
demons,
but
the
new
preference
that
you
see
here
is
kubernetes.
So
if
you
enable
kubernetes,
you
will
have
a
one-note
kubernetes
cluster
installed
with
docker
and
here
I
say
show
system
containers
because
I
want
to
see
the
kubernetes
containers
running
in
there.
B
B
Home
does
cube
config.
You
see
that
when
you
install
docker
it
has
created
a
configuration
for
your
local
kubernetes
cluster
for
coop
CTL
to
discover
it.
This
configuration
is
added
to
the
one
that
you
have
actually
had.
So
if
you
need
to
switch
context,
the
name
of
the
cluster
is
called
docker
for
desktop,
and
this
is
all
described
in
in
the
documentation.
B
B
B
B
B
So
that's
it
for
how
to
get
started
with
kubernetes
on
on
docker.
For
mac,
then
there's
one
one
thing
that's
super
interesting
is
that
kubernetes
comes
with
comes
with
the
dashboard
and
the
dashboard
is
not
pre-installed,
so
I'm
just
going
to
follow
the
can
and
install
the
kubernetes
dashboard
to
to
my
local
cluster
in
order
for
the
security
of
it
to
work,
I
need
to
create
an
admin
user
I
created
that
EML
file,
where
I
create
a
role
on
a
service
account
for
the
kubernetes
dashboard.
B
There
you
go
it's
proxy
to
this
and
then,
if
I
go
to
or
have
I
think
I
have
it
in
there
and
then
I'm
going
through
the
proxy
to
the
dashboard
that's
running
inside
of
my
cook
cluster
and
there
you
can
see
that
I
can
I
can
see
everything
in
a
graphical
user
interface.
You
can
see
my
three
nginx
parts
running.
My
replica
sets
my
services,
so
I
can
do
everything
from
the
dashboard
and
that's
pretty
convenient.
B
So
another
thing
that
you
may
want
to
do
is
there
are
lots
of
lots
of
people
are
packaging,
their
application
containing
several
services
using
a
system
called
helm,
so
I
have
installed
the
hem
client
on
my
magnet
for
and
there
in
order
to
install
helm
on
the
could
cluster
I
need
to
create
a
role
for
tailor
a
rolecall
tiller
that
helm
is
going
to
use
to
deploy
my
applications
so
I
have
created
a
yellow
file.
There.
A
B
Maybe
if
we
are
lacking
on
time
I'm
going
to
write
a
blog
post
explaining
how
to
install
monocular
and
how
to
show
it
and
then
once
you've
done
all
that
you're
able
to
install
things
on
the
command
line
or
just
clicking.
Why,
with
monocular,
to
install
the
packages,
you
can
also
use
Prometheus
on
that
lab.
B
That
shows
how
to
do
GTE
application
and
that
lab
has
a
docker
compose
file
and
because
the
kubernetes
cluster
is
using
the
docker
demon
that
ships
with
docker
format
when
you're
building
containers
on
that
docker
engine,
they
are
available
to
schedule
with
Banaras
and
with
composing
and
kubernetes
integration.
You
can
do
a
darker
stack,
deploy
with
a
compose
file
and
that
will
deploy
that
your
proven
NS
cluster
as
kubernetes
resources,
as
we
demoed
at
dr.
pol.
B
A
B
A
B
So
could
control
we
install
it
as
a
convenience
on
the
Mac
when
you're
installing
docker
for
Mac,
but
you
can
get
your
own
version
of
it
and
that
will
work.
So
you
get
the
control
version
that
for
Mac
and
then
the
configuration
is
the
one
that
I
showed
that's
in
home,
dot
cube
config,
so
you
can
have
your
other
kubernetes
clusters
in
there
and
you
can
configure
the
parameters
in
there
the
traditional
way.
B
B
B
Is
it
open
source
I?
Don't
know
I,
don't
think
so.
No,
because
the
team
is
still
working
on
that
I
think
this
judge,
but
I
I,
don't
think
so.
You
may
find
the
client
part
on
the
docker
clients
codebase,
but
the
server
part
I,
don't
think
so,
and
this
year
these
also
use
in
docker
Enterprise
Edition
with
kubernetes
install.
A
So
I
think
we
need
to
start
moving
on
for
time,
but
obviously
a
lot
of
interest.
Is
you
draft
up
that
blog
post
that
you
mentioned?
Please
share
out
the
link
and
I
I
can
imagine,
there's
gonna
be
a
lot
of
interest
in
this
at
UConn
and
other
forums
in
the
future,
so
keep
keep
sharing
the
information
I
think
people
are
really
appreciating
it.
Thank
you.
Thank.
C
C
That's
that's
something
that
he
would
do
so
he's
gonna
focus
on
getting
the
privileges
set
up
for
that
and
whatnot
and
we'll
we'll
announce
when
the
alpha
the
first
alpha
comes
out
again.
The
the
Alpha
is
really
intended
to
capture
the
steps
necessary
to
cut
it,
make
sure
we
have
the
privileges
in
place
and
whatnot.
So
this
is
exactly
the
the
circumstances
that
we
had
planned
for
by
having
an
L
for
that
early.
E
He
was
here
and
to
do
this
so
yeah
feature
feature.
Freeze
deadline
is
next
Monday,
June
or
22nd,
and
I
strongly
ask
everybody
who
is
developing
some
new
feature
for
communities,
one
for
them
to
add
and
to
chain
to
add
this
feature
to
Cuban
artists
features
repo
or
if
it's
Oh
dd-dear,
please
update
the
milestone
to
110.
Today
we
have
terribly
low
number
of
features
for
one
thing,
I
can
expect.
The
real
number
is
much
bigger,
I,
also
don't
get
about
restricting
spreadsheet
yep.
That's
all
from
me
all
the
links
in
the
middle
aisles.
C
A
Also
give
a
shout
out
to
your
Twitter,
you,
you
kind
of
talked
about
your
origin
story
and
I.
Think
in
this
context
of
the
the
community
call
it's
really
cool
to
see
people
sharing
stories
like
that.
Thank
you.
Thank.
F
So
I'll
be
really
fast.
I
was
looking
through
the
CN,
CF
graphs
and
I
was
kind
of
wondering
how
the
holidays
affected
everyone.
So
these
are
the
approves
that
were
done
in
December,
and
if
you
see
this
line
here,
that's
where
there
was
the
obvious
dip
and
then
one
thing
that
I
thought
was
also
interesting
was
there's
a
approvers
histogram.
So,
as
I
went
through
the
months,
I've
noticed
that,
depending
on
the
month,
different
people
are
different
or
active
at
different
times.
F
So
if
you're,
one
of
the
persons
is
kind
of
hurting
for
a
review,
you
can
come
to
this
histogram
and
go
either
by
day
or
week
and
see
who's
been
really
active.
Lately,
if
you're
fishing
for
a
review,
which
I
thought
was
cool
and
the
last
bit
is,
I
did
think
that
this
specific
graph
was
showing
the
total
number
of
approvers,
and
it
did
not
do
that.
A
G
Okay,
so
this
is
gonna
function
a
little
bit
as
sort
of
the
state
of
kubernetes
Docs
currently
and
where
we've,
why
we've
made
the
decisions
that
we
have
as
sig
Docs
from
version
one
or
two
today
version
one
nine
circle
one
for
around
that
time.
Everybody
knew
that
the
documentation
was
a
problem,
but
nobody
could
really
articulate
what
about.
It
was
a
problem,
so
sig
Docs
underwent
an
investigation,
and
what
we
found
is
that
largely
structural
issues
were
really
hurting
us.
G
G
G
Most
of
the
documentation
is
what
we
call
feature
centric
documentation,
as
in
everybody
wrote
guides.
That
said,
this
is
my
feature,
and
this
is
what
it
does
and
what
that
relies
on
is
for
users
to
actually
know
what
your
feature
is
and
be
able
to
identify
it
by
name
in
order
to
find
any
content
about
it
so
like.
If
anybody
wants
to
learn
to
use
cluster
Federation,
for
example,
they
need
to
know
that
Federation.
G
Associates
multiple
clusters
together
and
actually
know
what
Federation
is
before.
They
know
that
they
can
use
it
to
associate
multiple
clusters
together.
Most
of
the
time
we
found
that
users
really
only
know
their
task
ahead
of
time,
rather
than
what
feature
name
they
use
to
accomplish
it.
Another
related
structural
issue
was
differentiating
between
content
for
set
up
core
kubernetes,
primitives
and
environment
specific
tools,
all
of
it
was
just
kind
of
mushed
together,
and
each
of
these
guides
topics
had
a
lot
of
variants.
The
structure
was
ad
hoc,
some
of
them
by
step
instruction.
G
Some
of
them
had
design
background.
Some
of
them
had
troubleshooting
sections,
and
all
of
them
were
some
of
them
had
all
of
those
things
and
the
more
you
put
in
a
single
topic,
the
more
you
risk
overloading
the
reader
and
giving
them
a
bunch
of
information
that
isn't
relevant
to
them
and
making
them
go.
I
can't
find
anything
and
just
giving
up
before
they
find
what
they
need.
G
If
users
trying
to
accomplish
the
task,
they
usually
don't
need
design
background
or
any
discussion
of
implementation,
trade-offs
or
background
on
how
the
universe
used
to
be
before
this
feature
existed.
All
of
those
things
were
included
in
1.5
s
guides.
We
found
that
if
users
want
that
information,
they
tend
to
explicitly
search
for
it.
G
So
what
did
we
do?
We
set
up
documentation
to
be
organized
around
a
user's
operational
context.
We
decided
on
these
four
core
things.
Actually,
five,
because
setup
falls
into
them.
They
are
tasks.
Tutorials
concepts
and
reference
tasks.
Are
generic
instructions.
I
want
to
learn
how
to
do
something.
I
need
instructions
for
it.
I
will
sort
of
insert
my
own
context.
I
will
insert
my
own
cluster
name.
A
tutorial
is
a
real
world
example
of
like
say
how
do
I
set
up
WordPress
or
something
on
my
cluster
concepts.
G
This
basically
means
that
topics
our
guides
get
broken
up
into
multiple
topics
of
these
types,
but
the
way
we
hold
everything
together
is
by
linking
them
together.
Tasks
can
link
out
two
concepts,
so
in
case
the
user
actually
wants
and
background
information
when
doing
their
tasks.
They
can
follow
the
link,
but
it
doesn't
lengthen
the
documentation
or
mess
with
the
order
by
putting
it
in
the
middle
of
it.
G
So
how
do
document
a
feature
now
in
an
upcoming
release?
You
may
need
to
write
multiple
topics
unlike
before,
but
they
are
shorter
topics,
tasks
organized
by
youth
case,
concepts
for
background
and
tutorials
for
basically
demos,
sig
docs
maintains,
can
provide
guidance.
We
have
an
editors
at
core,
listen
from
CN
CF
and
Andrew
Chen
and
Steve
Perry
had
been
instrumental
in
installing
all
this
a
new
structure
and
migrating
the
existing
content
over
it
and
they
are
a
great
resource
for
you.
G
What
are
we
doing?
Next?
We
need
to
work
on
better
entry
points
for
the
documentation
which
we're
trying
to
do
with
critical
user
journeys,
automation
for
automatically
generated
reference
Docs.
That
process
is
still
extremely
labor-intensive
and
error-prone,
support
for
localization
and
translation,
organizing
and
standardizing
contributor,
centric
documentation
and
finally,
continuing
to
support
our
quarterly
launches.
A
Thank
you
for
that
I
for
one
is
a
relative
newcomer
into
the
kubernetes
and
finding
the
three
organization
there
at
simplification,
very
good
and
also
from
a
contrived
X
perspective,
even
though
it's
having
to
contribute
multiple
docs
in
multiple
places,
potentially
now
the
on
the
split
between
those
and
that
they're,
smaller
I,
think
makes
it
less
intimidating
to
somebody
and
it
makes
it
easier
to
spread
the
load
across
contributors.
So
I
think
this
is
awesome
stuff.
Thank
you
thanks.
Guys.
Next
up
is
cig
network
with
Dan
Williams.
D
Hello,
everybody
I've
put
some
notes
into
the
agenda
doc,
so
you
can
follow
along
there.
If
you
want
I'll,
just
do
a
really
quick
overview
of
some
of
the
features
and
work
that
we've
identified
for
the
1:10
cycle,
one
of
the
more
important
ones
of
those
and
the
one
that's
been
getting
a
lot
of
attention
is
ipv6,
single
and
dual
stack.
D
If
you're
not
aware
single
means,
you
can
run
the
cluster
and
data
plane
in
either
ipv6
or
ipv4,
but
you
have
to
pick
one
of
those
and
dual
stack
means
that
you
can
run
in
either
of
those
or
both
at
the
same
time,
so
that,
for
example,
pods
could
have
ipv4
addresses
and
ipv6
addresses.
You
could
have
nodes
with
ipv4
and
ipv6
addresses.
So
that's
dual
stack
and
we
hope
to
get
there
soon.
Next,
one
is
IP,
vs,
proxy
I.
D
D
One
of
the
things
that
is
currently
being
worked
on
is
bandwidth,
shaping
quality
of
service,
which
the
cube
net
plug-in
currently
does
by
itself,
but
we
do
have
a
plug
in
PR
for
CNI
that
implements
that,
and
once
you
know
so,
that's
one
part
of
that
there's
a
few
other
little
things
that
need
to
happen.
They're
also
windows,
code
and
compatibility
is
has
been
a
big
focus
for
the
last
couple
of
months.
We
have
windows,
support
for
kube
proxies
IP
tables
type
mode.
It's
not
viously,
not
IP
tables,
since
IP
tables
as
linux.
Only.
D
There's
also
some
implications
there
around
health
checking,
because
we've
found
a
number
of
use
cases,
while
implementing
network
plugins
that
work
with
kubernetes
to
have
health
checking
on
a
separate
IP
address
from
the
pods
actual
data
plane
itself.
So
we're
continuing
to
work
through
some
of
that.
There's.
A
couple
of
different
use
cases
here
for
multiple
pod,
IP
addresses
and
part
of
the
effort
is
to
find
a
solution.
H
D
Works
for
all
of
them,
or
at
least
many
of
them,
but
doesn't
impose
too
much
complexity
on
the
cube
API.
Also
coming
out
of
some
of
the
cube
con
discussions.
I
nothing
specific
here
is
going
to
hit
110,
but
there
is
some
dissatisfaction
with
the
current
service
API,
it's
not
quite
as
flexible
for
some
things
as
we'd
like
and
so
we're
going
to
continue
to
explore
what
might
happen
there.
D
Obviously
this
is
a
big
big
topic,
because
making
any
changes
to
this
API
needs
either
be
backwards
compatible
or
we
need
to
start
brocation
cycle
for
some
of
this.
So
that's
just
something
that
we're
gonna
continue
having
keeping
under
discussion
and
then
lastly,
topology,
where
ingress
is
another
thing
that
sort
of
ties
in
with
services.
This
is
related
to
the
kubernetes
feature
that
landed
I,
believe
in
1:7
about
local
backends
for
the
proxy,
so
that
when
traffic
comes
in
to
a
node,
that
node
prefers
a
pod
backing
that
service.
D
A
D
A
Thanks
Stan,
a
number
of
really
complex
technical
concepts.
They're
in
progress
describe
very
succinctly.
Thank
you.
So
our
next
cig
update
was
going
to
be
a
cig
service
catalog,
but
Paul
is
out
sick.
So
that
will
probably
happen
next
week
and
then
next
is
the
steering
committee
governance
update
from
Phillip.
I
I
Structure
for
them
to
take
advantage
of
what
we've
done
so
far
as
we've
reached
out
to
save,
leads
and
had
a
long-form
questionnaire.
That
asks
a
bunch
of
questions
that
we
were
grappling
with
and
got
pretty
detailed
insight
from
many
leaders
within
the
community
about
how
their
sakes
function
and
found
really
diverse,
set
of
results,
and
that,
unsurprisingly,
different
SIG's
have
very
different
needs
and
different
amount
of
desired
structure.
I
And
so
we're
currently
trying
to
are
working
on
developing
a
template
which
defines
the
structure
of
forsakes
to
then
fill
out
customized
for
their
specific
needs
and,
as
part
of
that,
we'd
like
to
get
the
insights
and
experiences
that
the
communities
had
within
their
sakes
and
and
how
they
feel
about
the
functioning.
And
so
we
put
together
a
small
questionnaire.
It's
about
seven
questions
that
are
all
multiple
choice
or
checkbox,
that
you
can
give
us
feedback
and
say
I,
really
like
the
way.
A
That
looks
like
nothing
in
the
chat
and
no
one
online.
So
thank
you
for
that
and
yeah
fill
that
out.
Everybody
on
the
call
and
share
it
in
your
circles,
get
more
people
more
contributors,
filling
it
out.
This
brings
us
to
announcements.
For
the
week
we've
got
some
google
Summer
of
Code
updates
from
a
I
think
next.
E
A
F
F
What
we're
gonna
do
is
actually
set
a
schedule
for
all
the
SIG's
for
the
entire
cycle
for
your
sig
updates,
so
they
will
know
exactly
how
many
sig
updates
you'll
have
we'll
have
exactly
three
per
meeting
instead
of
one
sometimes
and
for
the
next
on,
and
it
will
allow
the
SIG's
basically
to
plan
ahead
of
time
and
and
to
make
things
a
little
bit
more
tightened
up
in
that
regard.
We're
also
gonna
make
some
demo
changes.
F
We
can
just
pluck
that
person
do
them
later
and
move
on
to
somebody
else,
so
that
we're
more
efficient
and
more
professional
in
that
regard,
and
also
just
a
quick
call
that
we
always
need
note
takers,
Jase
kinda,
always
gets
stuck
with
it
and
that's
that's
a
real
shame,
considering
that
the
amount
of
people
that
come
to
the
meeting.
So
please
please
consider
volunteering
for
a
note-taker.
We
can
rotate
and
things
like
that
and
that's
all
I've
got
Paris.
H
Hey
near
me,
alright
go
yes,
we
are
going
to
continue
I
meet
our
contributors.
Ask
us
anything
type
of
office
hours
on
a
monthly
basis.
Our
next
one
is
February.
7Th
I
will
put
the
link
in
the
doc
here
shortly,
but
will
do
I
will
actually
do
two
times
we'll
do
a
8:30
a.m.
Pacific
time
as
well
as
an
afternoon
time
TBA,
but
this
is
intended
to
help
out
either
new
contributors
or
current
contributors
with
questions
that
they
and
they
want
to
talk
to
a
human,
just
sort
of
learning
on
demand.
H
A
Thank
You
Paris,
a
couple
of
people
in
the
chat,
are
asking
hey.
How
do
I
how
to
actually
sign
up
to
take
notes?
First,
obviously,
it's
a
Google
Doc,
so
type
pull
it
up
and
collaboratively
make
it
happen,
but
it
is
also
useful
to
have
somebody
who
is
specifically
tagged
with
making
sure
the
bulk
of
it
is
getting
done.
George.
You
have
a
preferred
process
for
somebody
to
reach
out
to
you
on
that.
I
will.