►
From YouTube: Kubernetes Community Meeting 20190829
Description
The Kubernetes community meeting is intended to provide a holistic overview of community activities, critical release information, and governance updates. It also provides a forum for discussion of project-level concerns that might need a wider audience than a single special interest group (SIG).
See this page for more information! https://github.com/kubernetes/community/blob/master/events/community-meeting.md
Liking what you see here? Continue the discussion at https://discuss.k8s.io
A
Welcome
everyone
to
the
August
29th
kubernetes
community
meeting.
This
is
a
community
meeting.
That's
live
streamed
and
will
be
posted
publicly
on
YouTube.
So
please
be
mindful
that
what
you
say
is
recorded.
We
also
have
a
code
of
conduct,
so
please
be
excellent
to
each
other.
I'm
Don
Foster
I
work
at
pivotal
on
our
open
source,
kubernetes
contributors
strategy
and
I
participate
as
part
of
the
contributor
experience.
A
Sync
and
I
will
be
your
host
for
the
day
and
I
will
remind
everyone
to
please
mute
when
you're
not
talking
as
a
courtesy
to
the
rest
of
us,
because
this
tends
to
be
a
rather
large
meeting
and
with
that
I
think
we
have
the
they
could
have
the
administrative
stuff
out
of
the
way,
with
one
one
exception.
So
is
there
somebody
who'd
like
to
volunteer
to
be
a
note-taker.
A
A
A
D
Is
okay
dawn?
Hello?
Can
everybody
hear
me?
Okay,
fantastic
okay,
my
name
is
Lachlan
Evensen
I'm.
The
116
release
lead
few
release
updates
for
the
week,
so
in
116,
specifically,
some
milestones
for
this
week
today,
which
is
August
29th,
is
code
freeze.
So
if
you
are
working
on
any
enhancements
and
have
poor
requests
in
the
kubernetes
kubernetes
repository,
please
make
sure
they
are
appropriately
labeled
and
put
in
the
queue
hey
Joe.
D
There
is
a
large
backlog
in
the
queue
that
we're
trying
to
garner
so
make
sure
they're
in
the
right
state
so
that
we
can
get
them
in
and
you
do
not
get
frozen
out
of
the
release.
Also,
those
enhancements
must
be
that
must
be
documented
need
to
have
the
PRS
for
the
docs
ready
for
a
review
on
Tuesday
September
3rd.
That's
next
Tuesday
and
September
4th
next
Wednesday
we're
cutting
the
second
beat
a
release
of
116,
so
they're.
D
D
All
the
patch
releases
are
the
same
as
they
were
last
week
and
if
you're
interested
in
getting
this
information
hot
off,
the
press
feel
free
to
subscribe
to
the
Cuban
Eddie's
day
of
mailing
list,
where
you
can
see
that
and
I've
also
placed
the
links
for
the
release
schedule
for
all
supported
kubernetes
releases
in
the
agenda
with
that
I
will
pass
it
back
to
dawn.
If
there
is
any
questions,
I
will
be
available
on
chat
to
answer
them
and
Thank
You
Don.
A
E
E
Yes,
thanks!
So
the
problem
with
the
slides.
What
was
that
George
is
sharing
my
screen
cool,
so
I
was
in
the
military
with
with
a
couple
of
well,
we
in
Finland
we
have
finished
mandatory
conscription
and
at
the
end
of
military,
we
we
had
been
doing
some
really
interesting,
interesting
stuff
and
I
got
an
idea
of
how
to
simplify
the
firecracker
user
experience
which
led
to
this
wheel,
ignite
project
and
the
10th
of
July.
E
Yes-
and
next
thanks
so
unfortunately
due
to
well
the
technical
problems,
I
won't
be
able
to
to
do
the
demo
because
of
zoom
on
Linux
didn't
enthrall
a
work
today.
Sorry,
but
anyways.
We
big
night
is
an
open
source,
virtual
machine
manager,
manager
with
container
user
experience
and
built
up
built-in
get-ups
management,
the
one
line
sentence.
E
What
this
actually
means
is
that
we
we
take
the
container
experience
and
what
we
learn
from
what
is
good
with
containers
and
try
to
go
a
layer
down
so
to
say
and
do
virtual
machines
with
this
and
we
utilize
firecracker,
which
is
insanely
fast.
It
can
put
in
1
or
8
fraction
of
a
second.
So
getting
that
kind
of
speed
and
minimalistic
design
is
really
really
key
to
this
project,
but
also
taking
some
some
things.
We
learn
about
the
container.
E
E
So
we
started
looking
into
firecracker
and
I
mean
we
thought
we
thought
this
was
really
impressive
tech,
but
it
took
us
a
pretty
long
time
to
like
understand
how
we
how
we
could
actually
use
it
for
for
something
other
than
the
hello
world
thing,
and
there
was
a
lot
of
manual
work.
I
mean
what
we
had
in
the
military
to
get
VMs
running
was
a
500
line.
Bash
script,
kind
of,
but
but
I'm
not
now
upon
upon
leaving,
we
formalized
this
as
a
cleanroom
implementation.
So
it's
something
breathing
going
and
then
called
we
big
night.
E
So,
if
you're,
not
a
KBM
developer,
this
will
help
you
use
psy
cracker
in
a
friendly
way,
cool
and
yeah.
So
what's
different
between
containers
and
VMs
conceptually
is
that
with
containers
we
reuse
well
on
Linux,
see
groups
and
namespaces
mainly
to
isolate
one
workload
from
the
other,
and
this
listen
well
by
design
shares
the
same
kernel
and
has
a
lot
of
good
things.
E
So,
for
example,
container
networking
is
handled
seamlessly,
it's
integrated
with
CNI
it
integrates
with
docker
and
or
parameters.
Sorry,
docker
and
or
container
D
also
Prometheus
integrations
are
in
there
and
then
the
key
thing
is
is
to
be
able
to
manage
your
fleet
declarative
know
of
these
true
through
checking
in
files,
VM
specifications
actually
using
the
convenience,
API
machinery
ingot,
and
if
I
have
now
the
possibility
to
boot.
E
A
lot
of
different
VMs,
like
just
less
than
a
second
I,
can
run
a
lot
of
VMs
locally
on
machine
on
my
machine,
for
example,
running
kubernetes
testing
kubernetes
clusters
without
having
to
do
doctor
in
docker
stuff
like
like
with
kind
which
works
pretty
well
but
has,
as
obviously
technical
limitations
too.
Next
one.
E
One
too
much
I
think:
yes,
there
we
go
so
there
are
at
the
moment
two
ways
to
use
ignite
in
parrot,
liver
or
declaratively.
So
if
you
do
use
ignite
imperatively,
you
can
use
the
CLI
which
has
the
same
UX
a
stalker,
so
it
has
docker
run
docker
PS,
sir.
If
you
guys
run
ignites
PS
unite
logs,
attach
images
and
kill
exec
all
of
those,
but
also
you
can
use
it
declaratively,
using
a
specification
that
looks
like
a
CRD
or
like
a
normal
convenience.
E
So
we
use
the
OS
noci
image
pushed
registry,
so
docker
hub,
for
example,
as
a
way
of
distributing
image
content,
and
we
also
use
like
an
actual
container
runtime
to
isolate
the
firecracker
process.
So
that's
what
I
said
about
container
D
or
docker
is
that
that
we
actually
run
the
firecracker
process
itself
in
a
container
for
well
for
using
integrating
with
T
and
but
also
for
isolating
the
the
thing.
I
mean
containers
proved
to
be
a
very
good
way
to
isolate
stuff.
Next.
E
So,
oh.
E
C
E
This
is
just
a
small
I,
don't
know
if
you
can
see
that
I
had
well
hope
to
do
so.
Demo
demo,
my
laptop
anyways,
you
see
that
Iran
ignite
run
and
then
give
it
a
docker
image
or
OCI
image.
In
this
case,
we
works
ignite
upon
2,
which
is
just
talkable
two
from
docker
hub
and
which
systemd
added
basically
and
that's
an
SSH
daemon.
E
Yeah
there
we
go
so
with
this.
You
can
see
that
you
just
do
from
Ubuntu.
Add
systemd
add
like
four
because
system
deep,
because,
if
you're
on
a
VM,
you
need
to
have
some
kind
of
init
system
running
something
that
that
can
be
paid
one
and
supports
that
which
is
different
from
containers
where
you
can
be
where,
for
example,
your
application
like
nginx,
can
be
paid
one
yep
and
finally,
the
pretty
much
last
slide.
I
had
so.
E
This
only
works
on
Linux,
obviously
because
it
uses
KBM,
but
also
that
we
have
a
cubed
MH
a
sample
guide.
So
you
run
three
VMs
and
into
them
from
from
there.
You
can
do
cube
at
a
minute.
You
don't
join
footloose.
Another
way
works
project
is
like
you
can
think
of
like
docker
compose
for
docker
Footloose
is
to
ignite,
so
you
can
run
a
couple
of
many
different
ignite,
VMs
or
other
container
machines,
and
then
lastly,
cluster
API
integrations
we're
lucky.
So
last
slide
we
have
documentation
on
ignite
that
dot,
read
the
docs
dot
org.
E
E
Yes,
so
so
the
virtual
cubelet
integration,
where
we're
thinking
that
you
could
register
an
ignite,
so
we
run
ignite
on
a
physical
machine
and
then
the
just
all
that
as
a
virtual
qubit
and
now
every
time
you
run
a
port
assigned
to
this
virtual
cubelet,
which
is
then
ignite
you,
it
will
run
it
as
a
real
VM.
So
the
contents
of
that
container
is
container
images
specify.
It
will
run
that
as
a
real
VM.
But
but
you
see
a
right
way,
we
also
considered,
but
the
problem
is
that
or
like
we.
E
E
Compatible
on
that
kind
of
layer
that
we
very
well
it's
a
bit
harder
and
also
also
maybe
not
as
user
friendly
or
like
clear-
to
use
what
happens
as
compared
to
using
a
CRD
and
then
the
virtual
cubelet
qubit
instance.
But
I
mean
we
haven't.
We
haven't
done
all
of
that.
Yet
so
would
be,
would
be
happy
to
get
out
of
feedback.
F
E
Hence
we
couldn't
like
have
the
dependency
on
cube
on
kubernetes
itself,
but
chose
to
use
the
git
basically
use
check
in
stuff
to
get
as
our
convenience
API
server
when
we
don't
have
a
convenience,
API
server,
and
so
that's
that's
a
difference
between
you
know,
I,
think
Hubert
and
that
Cubert
uses
like
this
VM
VDI
VMDK,
whatever
Keuka
formats,
and
we
use
pure
pure
OCI
images.
So
it's
like
it's
not
a
doctor
like
file.
E
E
We're
ignite
your
your
reflect
of
the
unit
if
Liam's
so
you're
running
in
the
VM
when
it
starts
up
it
runs,
the
kernel
runs,
has
been
in
it
and
then
you
can't
you
can't
like
run
nginx
as
your
first
thing.
That
is
sorry.
It's
like
fundamentally
two
different,
two
different
things
ignite
gives
you
real
VMs,
whereas
kind
of
wraps
a
container
with
an
extra
shell.
That
is
the
VM.
What's
that
make
sense,
yeah.
F
E
A
A
Okay,
let's,
let's
go
sick
p.m.
Steven
you're
up.
B
Hey
friends,
give
me
a
second.
B
Alright,
so
sick
came
up
to
eight.
This
will
be
a
quick
one,
I
like
quick
ones,
for
the
community
updates.
So
what's
happened
recently,
we've
had
some
chair
changes,
so
Igor
has
moved
to
emeritus
chair
for
sig
p.m.
so
thank
you
eat
or
for
all
your
work
and
we've
also
added
lucky,
so
that
happened
right
after
keep
going
I
think
maybe
a
little
less
than
a
month
after
cube
come
so
on
the
kept
side,
there
has
been
some
work.
Did
you
kept
metadata
validation?
B
So
thank
you
check
this
that's
kind
of
a
big
deal,
because
one
of
the
things
blocking
being
able
to
put
up
a
cap
website
was
making
sure
that
all
of
the
metadata
was
salad.
So
Chuck
Chuck
also
wanted
to
view
caps
on
the
command
line.
So
he
wrote
a
tool
for
that
and
part
of
that
tool
was
a
metadata
validator.
So
we've
brought
that
into
the
enhancements
repo.
Recently
Tim
Claire
worked
on
a
table
of
context
table
of
contents
validation,
so
now
caps
are
now
the
pre
submits
will
scream
at
you
on
enhancements.
B
If
you
have
an
incorruptible
of
contents,
which
is
again
good
for
the
overall
metadata
for
the
website
and
kept
website
right,
so
this
is
still
in
progress,
but
I
was
talking
with
Brian
a
Lyles
during
cube
con
and
he
was
like
we
should
just
do
this
I'm
gonna
hack
on
it
and
after
a
cube
con.
So
by
the
way
chair
of
you
like
chair
for
cube
god,
the
weekend
of
as
he
was
flying
back,
he
hacked
on
this,
and
so
we've
got
a
lot
of
content
there.
B
B
Whatever
framework
we
have
for
standing
up
websites
plug
it,
plug
it
into
that.
So
some
things
we
still
need
to
work
on.
I
think
you
know
between
the
summer
and
columns
happening.
It's
been
more
of
a
quiet,
more
of
a
quiet
quarter
for
sig
diem,
but
things
that
we
want
to
get
to
you
are
the
kept
one
sheet
or
kept
101
are
for
the
community.
B
Do
you
a
kept
retrospective
I
think
the
kept
for
respective
gets
trickier
every
time,
because
it's
kind
of
something
that
you
need
to
schedule
is
a
cycle
ascending
or
right
as
a
cycles
beginning
so
I
think
we
haven't
gotten
to
timing
down,
but
that's
still
something
that
we
want
to
do
getting
a
cup
office
hours
in
in
place.
Maybe
this
is
not
something
that
happens.
Maybe
this
is
hat,
there's
something
that
happens
during
the
sig
p.m.
meeting.
B
Maybe
it's
a
separate
meeting,
still
kind
of
noodling
on
that
keep
pushing
on
adding
kept
reviewers,
so
I
see
that
SIG's
have
started
to
to
do
this.
You're
kept
reviewers
do
not
necessarily
need
to
just
be
your
chairs
and
technical
leads.
If
you
have
people
in
your
sig
that
are
qualified
to
review
and
approve
architectural
documentation
and
decisions
for
your
sig,
please
consider
adding
them
to
two
kept
reviewers,
so
that's
kubernetes,
enhancements,
keps
and
then
your
SIG's
folder
and
the
owners
files
within
those
holders.
B
So
if
you
have
additional
reviewers
approvers
for
EPS,
please
please
please
add
them.
There
are
I
want
to
say
at
this
point.
Probably
a
hundred
in
50-plus
open
PRS
for
caps,
which
is
massive
to
many
there's,
still
a
lot
of
capture,
and
this
would
be
easier
if
we
had
more
people
able
to
review
those.
B
So
please
please,
please,
please,
please
I
know
I
said
it
already,
but
please
add
murderers
and
rivers
for
your
caps
and
then
finally,
we
want
to
start
to
wind
down
the
KK,
the
non
KK
targeted,
kubernetes
enhancement
issues.
So
there
are
a
few
things
that
are
open.
Some
of
them
actually
are
kept
improvements,
kept
and
proven
issues,
but
some
of
the
enhancements
open
are
targeting
improvements
for
non
KK
repos.
So
in
that
case
we
want
to
kind
of
shift
those
out
of
kubernetes
enhancements.
Those
are
enhancements
for
non
kubernetes.
B
Kubernetes
projects
can
be
handled
within
the
project,
so
some
aspirations
for
next
cycle
I
want
to
do
a
lot
more
grooming
of
the
open,
sig
PM
issues.
What
I'm
noticing
is
probably
75%
of
those
are
assigned
to
me
right
now,
and
that
shouldn't
be
the
case
so
starting
to
starting
to
divest
from
some
of
those
issues
and
making
sure
that
we
can
can
pass
them
off
to
interested
contributors,
just
making
sure
that
they
have
the
right
content
in
them.
B
If
you
have
have
had
the
opportunity
to
read
the
Tim
Hawkins
deadlines
are
horrible
email
and
the
the
resulting
Fred
from
that
one
of
the
big
issues
that
we're
running
into
is:
how
do
we
triage
the
issues
in
PRS
that
are
coming
into
urban
Nettie's
Durban
at
ease
all
right?
So
if
you
check
out
those
links
and
the
slide
deck
is
in
the
meeting
notes,
if
you
check
out
those
links,
there's
some
discussion
about
the
the
work
flow
potential
proposed
workflow
and
some
additional
labels
that
we
can
add.
B
So
you
can
see
we're
talking
about
like
what
the
workflow
for
this
looks
like,
potentially,
and
also
some
of
the
like
what
the
state
machine
for
for
issues
coming
in
and
out
of
kubernetes
kubernetes
so
check
some
of
that
out.
This
needs
to
still
be
cleaned
up
a
little
bit
and
more
discussion
has
to
happen,
but
this
issue
is
very
meaty
and
I
think
it
warrants
some
larger
discussion
since
we're
talking
about
changing
process
for
multiple
SIG's.
Essentially,
this
would
be
a
project.
Why
change
so
and
then?
Finally,
the
project
board
improvements.
B
There
was
an
open
PR
or
there
is
an
open
PR
around
writing
a
proud
plugin
that
can
do
some
fancy
things
like
add
issues
and
PRS
to
project
boards
automatically.
This
is
one
of
my
probably
largest
complaints
that
that
we
don't
have
that
functionality
out
of
the
box,
so
so
prowl
is
so
someone's
been
working
on
a
proud,
plugin
I
want
to
see
where
that's
at
in
the
next
few
weeks
and
see
if
we
can
keep
carrying
that
work
forward.
B
There's
also
the
project
bot,
which
I
think
is
a
pro
bot
plug-in
or
something
I
was
chatting
with
one
about
it
a
little
bit
and
we
want
to
see
which
route
is
the
right
route.
Ideally,
we
could
get
something
native
on
on
github,
if
not,
of
course,
we'll
we've
been
known
to
write
things
ourselves,
so
I'm
doing
some
investigation
over
that
over
the
next
quarter,
and
if
you
want
to
reach
us,
this
is
who.
A
B
A
B
B
A
G
G
G
A
Survey
Monkey
form
that
you
can
fill
out
just
tell
us
what
you
worked
on
and
will
determine
whether
you
can
vote
or
not,
and
that's
a
take
into
account
if
you're
contributing
something
that
isn't
exactly
a
measure
to
github,
maybe
you're
doing
some
p.m.
angst
of
or
or
you
know,
a
lot
of
things.
I
do
things
like
that.
G
Also
reminder
you
need
to
ensure
that
you're
in
those
voters
MB
by
September
11th,
so
if
you
haven't
filled
out
the
form
you
need
to
do
that
this
week
and
next
week
and
please
check
the
list
because
the
same
deadline
is
for
whether
you
are
going
to
run
or
not.
So
if
you
are
thinking
about
running
the
deadline
is
also
September.
11Th,
we've
gotten
two
new
people
today
who
have
submitted
to
run
so
the
next
major
dates
are
September.
11Th
I've
posted
all
this
information
near
the
current
IDs
dev
list,
I'm
just
repeating
it
here.
A
Okay,
cool
with
that,
we
will
proceed
to
shoutouts,
so
there
was
only
one
shout
out
in
slack
this
week
and
that
was
from
Antonio
shouting
out
to
bend
the
elder
and
Katherine
for
all
of
their
great
work
on
blocking
the
queue
for
merge.
So
thank
you
to
Ben
and
Katherine
any
other
quick,
shout
outs.
I
want.
H
A
B
Wait
sorry
Steven
I
was
just
gonna.
Do
a
quick
shout
out
for
I'm
the
wait.
We
don't
go
yet
so
quick
shout
out
for
I.
Really
everyone
in
cig
release
the
release
team
I
think
has
been
doing
a
phenomenal
job.
It's
it's
nice
to
see
a
lot
of
so
much
of
the
the
improvement
around
process
and
automation
that
we've
been
doing.
Spinning
up
the
release,
engineering
team,
everyone
who's
been
working
on
that
stuff,
actually
digging
into
some
of
the
scary
bits
of
like
the
tools
that
we
used
to
actually
release
kubernetes.