►
From YouTube: Kubernetes Community Meeting 20200416
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
Like what you see here? Continue the conversation on https://discuss.kubernetes.io
A
A
A
I'm
gonna,
ask
folks
to
please
review
that
policy
for
what's
allowed
in
terms
of
third
party
and
dual
source
content.
It's
hosted
on
the
website
right
now
in
our
contribution
guide.
There's
the
link
so
we're
going
to
see
a
whole
bunch
of
PRS
opened
with
relevant
SIG's
and
stakeholders,
tagged
in
to
help
decide
whether
to
refactor,
rehome
or
remove
affected
content,
and
my
request
is
to
be
cool
about
that.
A
A
lot
of
work
went
into
making
this
discussion
as
clear
and
as
open
handed
at
this
community-minded
as
possible,
so
deciding
to
respond
to
removal
PR
with.
Why
was
I
never
consulted?
How
did
this
possibly
become
policy?
Can
we
please
skip
past
that
and
get
straight
to
the
this
is
going
to
happen.
It's
not
a
question
of
whether
it's
going
to
happen.
A
That
was
a
five
months
discussion,
but
just
now
concluded
so
just
be
cool
about
it
figure
out
how
best
to
make
sure
there
are
docs
are
true
and
trustworthy
and
free
of
vendor
pitches
and
move
on
so
jumping
back
up
to
the
top
of
the
page,
so
upgrading
upgrading
the
site's
Hyuga
version
so
pages
that
have
HTML
content
mixed
with
markdown
content.
Those
are
going
to
break
so
we
will
announce
when
we
flip
that
switch.
A
But
if
you
see
broken
HTML
like
tables
that
aren't
rendering
properly,
please
do
file
an
issue
and
if
it's
specifically
a
broken
HTML
issue,
those
make
really
good
good
first
issues,
because
they're
entirely
self-contained
they're
self
describing
and
they're
great
for
first-time
contributors
to
things.
So,
if
you
see
broken
HTML
on
any
page,
please
open
an
issue
about
it
and
mark
it.
A
Good
first
issue
will
also
give
a
heads
up
before
we
flip
over
to
the
doc
CP,
because
that
will
almost
certainly
cause
some
formatting
and
display
issues
and
we
have
a
tracking,
an
umbrella
tracking
issue
for
doxy
related
issues
already
up
I'm.
Sorry,
that's
not
linked,
so
that
too
may
affect
your
experience.
At
the
website
and
again,
we
would
ask
that
you
open
issues
to
describe
that
and,
like
you
talked
about
by
changing
the
repost,
merge
strategy
make
sure
you're
making
clean
and
squash
commits
when
you're
submitting
to
the
Ducks.
Thank
you
next
slide.
A
A
You
know,
let's,
let's
uphold
the
code,
the
code
of
conducts
in
the
gentlest
ways
and
with
the
best
assumption
of
good
intent
as
possible
when
it
comes
to
removing
third-party
content
and
yes
open
issues
and
PRS
for
broken
HTML,
where,
wherever
you
find
it,
we're
going
to
be
will
occur
as
we
update
upgrade
the
site,
you
don't
have
to
wait.
If
you
see
it
now,
please
open
it
now.
A
B
Yeah,
absolutely
so
soapy
Arthur
always
welcome.
We
love.
We
I
know
that
in
the
marketing
team,
understood
contrived
X
were
really
focused
on
some
storytelling.
That
would
get
an
issue
open
an
umbrella
issue
open
for
a
lot
of
storytelling
stories
where
we're
walking
through
each
of
the
different
SIG's
and
telling
some
of
the
interesting
things
about
them,
what
they
do,
how
to
get
engaged
involved
and
just
really
really
fun.
B
A
That
kept
is
done,
makes
me
so
happy
if
you're
interested
in
the
deep
history
of
how
this
kept
came
to
be.
That
link
is
a
good
place
to
start
next
slide.
Please
right.
This
is,
if
you
need
to
reach
a
care
or
a
tech
lead
for
cig
docks.
This
is
who
we
are,
and
this
is
how
to
find
this
Jim
angel
is
super
awesome
and
amazing.
Katelyn
Barnard
same
our
repo
is
github.com
/
kubernetes
/
website,
please
come
join
us
in
hashtag,
Docs
I.
Think
that's
everything.
Thank
you.
B
Thank
you
so
much
thank
fantastical
and
scream
sure.
Okay,
I'm
moving
on
to
you
again,
sick
Apps
was
unable
to
join
us
this
week.
Unfortunately,
that
just
kind
of
wanted
to
give
them
a
wave,
and
you
know
thank
you
very
much
for
letting
us
know
on
that
front.
I
would
like
to
move
over
to
SIG's
Eli
next
tip.
C
Taylor,
can
I
interject
for
a
sec,
so
wait.
So
one
great
thing
that
was
real,
recently
updated
I
believe
was
some
guidance
on
actually
how
to
do
squashing
and
what
your
get
workflow
should
look
like
that
some
sig
Docs
contributors
worked
on.
So
maybe
if
we
can
get
a
link
out
for
that
as
well,
I
think
that
would
be
valuable
in
the
in
the
workflow.
B
D
D
Much
so
I'm
Sean
Sullivan
I'm,
the
co-chair
of
the
command-line
interface
interface,
cig
I'm,
joined
today
by
Monta
Sulak,
who
is
the
co-chair
as
well
as
the
co
tech
lead,
we'll
go
on
to
the
next
slide
and
and
get
into
it
right
away.
So,
let's,
let's
talk
about
what
we
did
our
last
cycle
and
then
we'll
talk
about
what
we're
gonna
work
on
and
maybe
there's
a
couple
questions
at
the
end.
So
what
did
we
do
last
cycle?
D
Now
that
it's
in
staging-
but
there
are
two
remaining
items
to
to
get
out
of
kubernetes
core
before
we're
out
completely,
and
that
is
the
good
control
converting
command
and
the
poop
control
our
threaten
style
command
once
once
so
those
two
dependencies
are
removed,
then
we
are
completely
out
of
humour
nettie
score,
but
we
still
have
those
cheating
to
do.
Another
thing
that
we
worked
on
was
server
side
apply.
D
Another
effort
we've
been
working
on
is
a
brand
new
command
called
the
debug,
which
is
leveraging
ephemeral
containers,
it's
an
alpha
command
and
it
should
help
significantly
with
with
anybody's
debugging
efforts.
Another
new
command
is
the
control
alpha
events,
so
we
recognize
that
could
control
again.
The
events
was
not
nearly
as
helpful
as
it
could
be
and
so
we're
under
their
way
for
an
effort
to
to
make
the
presentation
of
events
significantly
more
usable
and
readable
and
helpful
and
there's
a
cab.
D
So
so
what
are
we
gonna
work
on
so,
as
I
mentioned,
for
this
ongoing
effort
trying
to
get
coupe
control
into
its
own
repository,
we're
going
to
continue
the
small
remaining
items
to
to
get
to
coupe
control
out
of
the
kubernetes
core
repository
and
into
staging
initially,
and
we're
going
to
continue
working
on
the
to
alpha
commands
that
I
just
described
the
alpha
debug
and
the
Alpha
events
and
a
another
effort
is
server
side
described.
So
this
has
to
do
with
control
and
describe
for
coop
control.
D
Yet
that
is
already
performed
on
the
server
side
and
we
intend
and
these
serve.
The
could
control
get,
however,
is
already
described,
is
not
still
client
side
we're
going
to
move
that
to
to
the
server
side,
we're
going
to
work
on
metrics
to
be
able
to
basically
adding
X
headers
to
our
requests
of
the
API
server
so
that
we
can
go
through
logs
and
get
information
and
metrics
on
which
could
control
commands
are
being
used.
Obviously,
there's
some
there's
some
issues.
D
We
want
to
make
sure
that
we're
not
collecting
any
personally
identifiable
information,
but
just
the
sub
command
from
coop
control.
That
is
being
that's
being
invoked
should
be,
you
know,
pretty
easy
to
to
send
as
a
header
and
to
be
able
to
pull
through
our
logs
to
find
out
just
what-what
subcommands
people
are
using
we're
going
to
add
a
ski
warning.
So
a
very
common
problem
is
that
some
of
our
our
users
will
invoke
a
poop
control
that
is
out
of
the
support
window
for
the
API
server.
D
So
if,
if
the
version
of
the
code
control,
for
instance,
is
118
and
the
API
server,
we've
got
a
116
cluster,
then
that
actually
is
not
supported.
We
support
plus
and
minus
1
version
for
the
difference
in
control
in
the
API
server
or
the
cluster,
and
so
we're
going
to
start
printing
a
warning
to
to
let
people
know.
D
We've
we've
seen
many
issues
and
bug
reports
that
are
are
exactly
this
issue
and
so
alerting
our
users
or
food
control
users
to
the
fact
that
they
may
be
using
a
client
which
is
not
supported
by
the
API
server
they're
talking
to
who
should
help
that
situation.
It
will
also
be
a
fairly
intrusive,
so
heads-up.
D
D
D
If
you're
not
familiar
with
it,
please
have
a
click
on
customize,
there's
been
a
three
point,
five
point
four
release,
and
so
one
of
the
one
of
the
efforts
that
customized
has
been
working
on
customized
is
actually
available
through
coop
control
with
the
minus
K
flag.
However,
for
dependency
issues,
there
has
been
a
problem
incrementing
the
version
of
customized
within
good
control
and
I
think
we
are
at
a
at
a
stage
where
we
have
solved
those
dependency
snafus
and
are
and
we'll
be
able
to
finally
update
the
version
of
customize.
That
is
in
control.
D
So
crew
is
our
plug-in
index
and
vomit
has
been
very
ably
running
and
it's
getting
quite
a
bit
of
adoption.
Please
have
a
look:
click
through
the
link
on
crew
to
find
out
what
some
amazing
plugins
for
food
control
that
are
available
and
they've
come
and
they've
have
a
their
version.
0.38
point
for
release,
so
please
have
a
look
and
have
a
small,
quick
search
for
some
really
cool
plugins
that
are
available
through
your
next
slide.
D
D
But
if
you're
interested,
please
have
a
look
at
a
GUI
really
excited
to
have
the
steel
project.
Okay,
so
here's
some
links
and
some
info
along
with
with
Mattei
there's
myself
and
not
on
this
call
I
believe,
is
still
wet
rock
and
we've
got
to
check
out
some
of
the
some
of
the
links.
If
you'd
like
to
dig
in
even
further
and
we'd
be
willing
to
take
questions
now,.
B
F
F
F
F
So
first
time
we
call
it
was
a
thermo
containers
as
a
feature.
That's
been
long
under
discussion
and
is
slowly
making
its
way
through
the
alpha
beta
GI
evolution.
The
ephemeral
container
support
was
added
for
the
docker
shim
instantiation
of
the
cubelets.
So
if
you
wanted
to
explore
that
feature,
you
can
take
a
look
and
then
the
actual
support
for
the
other
CRI
providers
that
are
managed
out
of
Rene's
is
gonna,
be
dependent
on
that
see.
All
right,
topology
manager
is
probably
the
largest
stand
out
feature
that
we
moved
to
beta
in
the
last
cycle.
F
Alignment
between
the
CPU
and
any
devices
it
uses
are
critical
to
performance,
so
a
classic
example
is
you're
doing
an
AI
or
ml
type
of
workload,
and
you
want
your
CPU
to
be
close
to
your
GPU,
because
it
makes
a
huge
performance
impact.
The
feature
was
moved
to
beta.
There
are
other
enhancements
planned
in
the
future
and
a
lot
of
those
are
highlighted
in
the
excellent
blog
posts
that
our
friends
from
Nvidia
Intel
and
read
helped
author.
F
So
if
you're
interested
in
this
feature,
welcome
you
to
check
out,
in
addition,
start
up
probes
moved
to
beta
and
taint
based
evictions
moved
to
GA,
and
so
it
was
a
good
exercise
doing
that.
With
six
scheduling
on
the
stability
and
debugging
front,
we've
been
doing
a
lot
of
work
really
trying
to
I
think
we
have
a
goal.
F
Let's
say
one
out
of
n
pods
might
have
a
flake
and
trying
to
root
cause
that
course
of
that
flake,
and
so
a
number
of
improvements
were
made
around
pod
status
and
other
education
areas
that
cause
challenges.
I
wouldn't
say
we're
done
on
that,
but
as
always
we're
trying
to
improve
that
the
error
rate
of
the
cubelet,
so
that
these
things
happen
way
higher
values
of
n.
F
Here,
we've
had
a
overarching
goal
to
figure
out
a
path
forward
on
C
groups,
B
to
enablement.
The
prerequisite
work
was
done
in
run,
C
project
and
so
last
cycle.
We
kind
of
did
some
work
on
reviewing
an
initial
kept
and
this
cycle,
and
have
it
says
some
code
has
already
been
merged,
shout
out
to
Giuseppe,
there's
a
new
signal
contributor
to
allow
a
cubelet
to
tolerate
running
on
a
secret
sweetooth
host,
and
this
is
important
because
certain
Linux
distributions
have
started
defaulting
to
secrets.
F
F
F
F
We
want
to
really
make
sure
that
we
work
with
community
to
figure
out
how
to
handle
a
node,
power-down
event,
gracefully
and
so
I
know
some
folks
across
the
community
were
starting
to
explore
that,
and
so
it's
awesome
a
code
cleanup
of
the
qubit
is
perpetual
challenge
and
it's
one
that
some
new
contributors
have
started
to
offer
their
assistance
swords.
So
big.
F
Thank
you
and
then
finally,
I
think
folks
are
aware
that
cig
architecture
had
recommended
a
new
course
of
action
around
how
to
avoid
api's
being
in
like
a
permanent
beta
state
and
so
runtime
class
is
the
one
that's
impacted
from
that
work,
which
basically
is
going
to
require
us
to
figure
out
our
plans
for
either
GA
in
the
API
or
doing
an
extravert
revision
of
the
beta
by
122,
I
believe,
and
so.
If
there's
any
users
who
have
been
using
runtime
class
and
provide
feedback
to
the
sig
I
think
we
would
love
to
hear
it.
F
So
we
can
address
that
feedback
next
up
next
slide.
So
yeah,
like
I,
said
runtime
class
is
an
area
that
we
need
to
figure
out
how
to
proceed
on
and
then
there's
some
interest
in
some
parts
of
the
coroners
community
to
see.
If
we
can
make
the
scheduler
more
aware
of
node
topology,
and
so
it's
possible
that
across
six
scheduling
and
sig
note,
we
might
want
to
explore
how
to
address
these
types
of
problems.
F
The
blog
post
that
I
referenced
has
some
thoughts
from
various
members
of
community
on
ways
that
we
could
or
couldn't
choose
to
explore
and
so
I
think
just
general
feedback
on.
If
this
is
something
we
should
or
shouldn't
do
as
a
community
is
probably
the
first
thing
to
figure
out
and
then
kind
of
prototype
on
passed
forward.
So
next
slide
things
we
need
from
you
basically
I
would
say
just
participation
is,
is
really
welcome.
So
for
those
who
join
recent
sigmoid
calls
big.
Thank
you
a
pain
point.
F
I
would
say
that
we
have
a
little
bit
right
now
in
Sigma
ism,
as
a
maintainer
community.
We're
trying
to
be
cognizant
of
stability
first
and
sometimes
I
would
say
we
we
get
PRS
that
make
it
difficult
to
understand
the
root
cause
of
the
issue
that
was
being
fixed
or
if
the
issue
that
was
purported
to
be
fixed
is
actually
still
a
real
issue
or
not,
and
so
I'll
touch
on
this
a
little
bit
later,
but
in
general
for
folks
who
do
want
to
contribute
to
the
sig.
F
F
As
I
said,
we
have
a
well
yeah
and
Jonathan.
You
talked
to
these
things.
The
next
slide.
If
that's
okay,
the
CRI
API
is
they
talked
about.
There
is
a
kept
of
interest
where
we
are
exploring
kind
of
three
different
options
on
how
to
handle
errors
in
the
CRI,
and
so,
if
you
are
a
CRI
author
or
interested
in
becoming
one,
your
feedback
is
helpful.
I
would
say
right
now.
Our
approach
on
this
cap
has
been
the
least
change
for
the
biggest
impact.
F
But
if
you
read
the
enhancement,
I
think
we'd
love
your
feedback
next
slide.
Cri
tools
is
the
other.
Is
another
sub
projects
once
been
sick
somewhere
to
see
our
API
in
general?
We're
always
doing
improvements
in
that
project
to
do
testing
enhancements
and
then
kind
of
pending
the
outcome
of
what
we
do
around
CRI
or
handling
would
anticipate
us
trying
to
make
updates
and
CRI
tools
to
provide
better
guidance
to
CRI
authors
or
implementers
I'm.
Sorry
on
how
to
handle
these
error
handling
scenarios
best
next
slide.
F
F
An
incremental
rate
of
PRS
that
come
in
every
now
and
then
to
either
improve
testing
on
particular
hardware
or
to
make
more
Hardware
discoverable
by
the
component
and
I
would
say
recently.
There's
been
a
lot
of
work
to
do
stuff
around
SR
iov
discovery,
typically
associated
with
them,
knowing
how
to
direct
workloads
to
hardware
these
special
devices
so
probably
from
about
a
16-point.
This
project
is
getting
decent
traction
and
the
user
community,
and
so,
if
there
are
other
hardware
that
we'd
love
for
a
nice
feel
to
expose
that
are
within
reason.
F
Of
course,
please
reach
out
to
that
sub
project
and
see
what
we
can
do
next
up
related
caps,
yeah
kind
of
called
out
a
couple,
just
CRI
handling,
given
its
impact.
The
code
cleanup
one
is
one
that
I
like
calling
out.
We
don't
typically
do
a
cap
for
these
things,
but
I
think
the
depth
of
the
cubelet
is
broad
enough
that
this
wasn't
a
an
enhancement
that
I
found
actually
really
very
helpful
as
a
qubit
maintainer.
F
F
It
was
following
that
model
and
then
secrets
we,
if
you're
interested
in
in
the
plan
that
were
returning
towards
the
cap
that
was
linked.
There
recently
emerged
and
we'd
love
help.
Next
up
just
to
be
clear.
The
resource
management
workgroup
had
effectively
archived,
but
we
didn't
complete
the
paperwork
last
year
that
the
paperwork
is
now
completed.
So
we
do
not
have
any
existing
working
groups
being
sponsored
by
the
cig.
So
next
slide
how
you
can
contribute.
F
This
is
like
a
real
thing:
pain
point
that
I
think
Don
and
I
would
love
to
get
help
with,
which
is.
We
would
love
folks
to
contribute
with
more
testing
of
our
features.
So
a
lot
of
our
recent
cig
discussion
has
been
around
what
we're
actually
covering
and
no
dte
testing
and
what
we
can
do
to
to
in
excite
the
community.
Maybe
add
more
one
of
the
areas
where
this
has
come
up
is
oftentimes.
F
We
have
a
new
contributor
that
wants
to
contribute
a
bug-fix,
but
there
wasn't
either
an
existing
test
that
demonstrated
that
the
bug
was
occurring,
and
so
it
sometimes
makes
it
hard
to
understand
if
the
bug
or
flake
that
was
fixed,
wasn't
reproduced
in
tests,
so
I
think
as
a
Sikh
community.
We're
kind
of
encouraging
those
authors
who
wanted
to
fix
those
bugs
to
either
first
create
the
test.
So
we
know
that
we
don't
regress
later
or
to
make
that
issue
actually
discoverable
to
the
community
in
a
more
manifested
way.
F
Fixes
is
very
welcomed,
given
the
complexity,
other
can
feel
it
and
then
finally
I
think
it's
fair
to
say
that
as
I
say,
we
could
use
help
around
monitoring
that
the
health
of
our
test
infrastructure,
and
so,
if
folks,
are
particularly
good
at
that
or
have
the
time
to
help
support
the
cig
in
that
regard,
I
think
we
would
love
your
input
and
assistance.
I
do
want
to
thank.
F
The
contributors
from
Nvidia
and
Intel
and
Red
Hat,
who
did
a
lot
of
work
to
ensure
that
the
at
least
that
we
had
good
test
for
the
topology
management,
enhancement
and
so
there's.
Another
thought
here
is
as
more
complex
features,
get
nuclear
burn
any
as
we
could
use
help
in
even
just
getting
the
test
harness
to
test
it,
and
so
that's
also
been
an
area
of
recent
discussion
next
slide
so
where
to
find
us,
you
can
reach
out
to
myself
or
dawn
on
github
like
many
senior
members
of
the
project.
F
Github
notifications
are
a
difficult
firehose,
but
you
can
also
find
us
on
slack
insig
note
as
well
as,
of
course
joining
any
of
our
signal
meetings
or
reaching
out
on
the
mailing
list
in
general.
We're
from
the
group
and
we'd
love
to
have
people
help
above
significant
provide
their
assistance
to
serve
the
community.
So
thank
you.
What's
it.
B
Thank
you
so
much
dear,
so
we're
just
about
at
the
top
of
the
hour
I've,
seen
final
announcements
and
a
little
bit
housekeeping
to
do
and
then
I
wish
you
all
well,
we
got
a
notice
from
CNCs
that
they
are
going
to
be
launching
the
community
bridge
mentorships
for
q2
of
2020.
Now
we
have
the
link
to
that
in
the
notes.
B
If
you
want
to
read
a
little
bit
more
on
that
front,
it's
also
on
the
CNCs
blog
next
month,
record
madryn
is
going
to
be
hosting
the
community
meeting,
so
it
should
be
fun
to
see
him
in
the
hot
seat.
He
is
a
software
developer
at
loot,
C
project,
maintainer
of
Cupid
Korn
and
many
more
projects
as
well
and
then
I
had
one
shout
out.
I
went
through
shoutouts
on
our
kubernetes
slack
and
was
actually
me
as
like
looking
in
the
mirror,
and
it
was
a
shout
out
to
mr.
Bobby
tables.
B
Bob
Dylan
he's
always
just
very
helpful
in
the
community.
If
you
have
any
questions,
he's
always
very
responsive
and
that
helped
me
with
I
think
like
three
or
four
things
right,
so
another
shout-out
to
you.
Mr.
Bobby
tables-
and
that
is
time
thank
you
so
much
everyone
for
joining
us
for
this
community
meeting,
kubernetes
I
hope
you
have
a
fantastic
rest
of
the
week
and
an
incredible
weekend.
Thank
you
very
much
everyone
you,
you.