►
From YouTube: Kubernetes WG IoT Edge 0220323
Description
March 23, 2022 meeting of the Kubernetes IoT Edge Working Group. Presentation on new release of ioFog project and recent updates by Kilton Hopkins
A
Hi
welcome
to
the
March
23rd
meeting
of
the
kubernetes
iot
edge
working
group
on
today's
agenda.
We've
got
kilpin
Hopkins,
who
will
give
us
a
briefing
on
the
latest
updates
to
the
io
fog,
open
source
project
and
how
things
going
on
there
might
relate
to
kubernetes.
A
I
haven't
had
a
chance
to
look
at
the
agenda
for
the
last
hour,
but
that
was
the
only
item
on
the
agenda
last
time.
Look
but,
as
usual,
we're
open
to
people
nominating
topics
or
an
ad
hoc
discussion
or
just
free
form,
birds
of
a
feather.
If
we
have
time
so
with
that
said,
kilton,
let
me
make
you
co-host
here:
cool.
B
I'll
just
do
some
screen
share,
so
I'll
have,
unfortunately,
I'll
have
to
drop
at
9
30,
my
time
so
in
in
23
minutes.
But
it
should
be
plenty
of
time
for
me
to
go
through
just
some
updates
of
the
the
eclipse,
Iowa
fog
project,
and
so
thanks
for
for
having
me
go
first
and
and
thanks
for
putting
this
on
the
agenda.
B
So
the
the
eclipse,
IO
fog
project
is,
is
doing
a
run
of
version
three
betas
and,
as
of
very
recently
we're
on
we're
on
beta
7
for
some
of
the
components
beta
6
beta
7,
so
that
we
can
start
to
get
close
here
to
a
an
official
release
of
of
eclipse,
IO
fog.
Three
and
what
I
want
to
do
is
I'll.
Just
I'll.
B
Just
basically
look
at
look
at
and
talk
through
the
release
notes,
but
to
give
the
update
of
what's
what's
new
and
what's
on
the
horizon
there.
But
when
talking
the
context
of
of
you
know
what
we're
learning
as
we
get
3.0
out
the
door
yeah
until
why
yeah.
A
B
Oh
yeah
very
good
point
right:
it's
I'm,
assuming
I'm,
assuming
that
that
yeah,
that
it's
a
certain
audience
and
it
could
actually
be
that
the
folks
want
to
know
the
gist.
It
was
great.
Thank
you
so
much
for
that.
Okay!
B
Well,
in
that
case,
what
and
I
why
don't
I
share
my
screen
and
land
right
on
the
the
the
root
the
root
site
for
documentation
and
and
for
the
open
source
project
and
stuff
yeah
very
good
call,
so
Eclipse
IO
fog
is
has
been
around
for
for
quite
a
while,
and
it
was
originally
it
actually,
the
origins
of
eclipse
iofog
were
they
predate
kubernetes,
and
so
at
that
time,
I
thought
boy.
B
Somebody
should
be
orchestrating
microservices
and
that
you
know,
obviously
what
an
undertaking
so
but
I
do
remember.
B
I
do
remember
someone
here
in
silica,
Kelly
said:
if
you're
going
to
orchestrate
microservices,
you
should
hear
about
this
new
project
called
kubernetes,
so
they
all
go
check
it
out,
but
the
the
the
purpose
of
of
eclipse
IO
fog
is
to
is
to
set
an
agent
down
on
all
devices
that
are
capable
of
running
the
type
of
microservices
you
that
you
want
the
agent
being
very
similar
to
cubelet,
but
designed
to
run
on
devices
that
are
regularly
disconnected
or
under
constrained
bandwidth
and
constrained
resources
and
so
on.
B
So
in
other
words,
it's
an
edge
Computing
agent
and
the
purpose
is
to
to
do
your
bidding
in
a
way
that
that
complies
with
your
control
plane.
So
for
those
who
are
already
looking
into
Edge
Computing,
it
should
be
a
really
familiar
Concept.
In
fact,
it's
the
architecture
pattern
that
that
that
we
now
know
dominates
for
distributed
computing
systems.
B
You
really
want
to
have
an
agent
on
your
nodes,
and
then
you
want
to
have
some
some
method
of
of
issuing
command,
whether
it's
centralized
or
decentralized
and
so
on,
and
so
Eclipse
IO
folks
started
with
this
agent
and
then,
of
course,
had
to
add
a
control
plane
to
it.
B
As
the
years
went
on
we
and
we
saw
that
kubernetes
was
in
a
lot
of
ways
taking
over
the
world
or
at
least
taking
over
the
cloud.
It
became
obvious
that
integrating
the
two
was
a
was
a
really
good
idea.
B
So,
but
we
still
have
this,
this
iofog
control
plane
that's
its
own
control
plane,
which
which
might
have
some
value
in
your
kubernetes
turbine
system,
because
you
may
need
all
of
the
specific
controls
for
the
edge,
and
you
really
want
to
interface
between
the
two
versus
having
all
of
your
Edge
nodes,
directly
connect
to
your
kubernetes
infrastructure.
B
The
the
main
theme
Here
is:
there
are
options
right
and
so
I'll
look
at
some
of
the
the
integration
Parts.
So
you've
got
it's
not
a
mono
repo.
The
eclipse,
IO
fog,
I,
GitHub
organization
is,
is
a
set
of
repositories
for
the
different
components
open
source.
It's
always
a
tough
choice.
Do
I
do
model
repo
or
not,
and
we
decided
not.
B
You
can
see
some
of
the
pieces
here,
so
we've
got
the
the
agent
I
talked
about,
but
then,
when
we
get
into
looking
at
how
we
interconnect
with
kubernetes
back-end
systems,
we
made
a
choice
to
use
the
the
Scupper
project,
which
is
a
project
that
that
came
out
of
red
hat,
which
is
really
good
for
cloud
to
cloud
cloud.
Edge,
Etc,
et
cetera
communication,
and
so
we
have
here,
router
and
Scupper
proxy
are
part
of
what
we
call
the
the
the
layer.
B
Seven
or
virtual
application
network
overlay
and
I'll
talk
a
little
bit
about
that
when
I
go
through
some
of
the
updates
here.
So
that
bringing
it
back
to
like
why
you
would
use
it
or
what
it's
for
Eclipse,
IO
fog
is,
is
Edge
application
infrastructure
and
so
designed
to
give
you
the
ability
to
stitch
together
a
bunch
of
different
Edge
devices
such
that
you
can
deploy
and
manage
applications
across
them
and
have
them
intercommunicate.
B
And
so
that
means
that
this
layer,
7
Network,
gets
laid
down
along
with
the
the
agent
or
it's
rather
the
first
thing
that
gets
fired
up.
And
then
you
can
pass
data
between
all
of
these
Edge
devices,
whether
they're
on
the
same
same
network
or
not
right,
because
we
have
this
overlay
and
so
that's
a
concept
that
we
call
an
edge,
compute,
Network
or
an
ecn.
And
so
you
bring
these
themes
together
and
you
realize
that
edge
applications
require
the
hardware.
B
But
on
top
of
the
hardware,
you've
got
the
pieces
that
make
the
hardware
come
alive
and
host
those
applications.
So
of
all
of
the
pieces
here.
Controller
is
that
is
that
part
that
I
I
mentioned?
That
is
the
control
plane,
but
it
has
a
set
of
apis,
and
so
it
was
designed
to
be
interfaced
with
other
systems,
and
so
obviously
interfacing
with
with
kubernetes
is
is
the
topic
of
of
me
presenting
in
here,
and
so
we
have
a.
B
We
have
an
operator
and
a
way
to
interface
with
with
kubernetes
systems.
We
are
evaluating,
of
course,
the
extent
to
which
we
want
to
re
revise
that
iofog
3.0
is
not
centered
around,
like
a
complete
revision
to
how
we
interface
with
kubernetes,
but
is
rather
an
exploration
into
some
particular
features
that
had
been
requested,
which
I'm
going
to
now
go
into,
which
is
kind
of
the
the
theme
Here
of
like
what's
new
in
Iowa
fog.
Three,
and
what
should
you
be
paying
attention
to
so?
B
The
first
beta
was
all
the
way
back
in
the
fall,
and
let's
start
there
to
talk
about
what
3.0
has
that
the
the
other
versions
didn't
well.
We
made,
of
course,
some
updates
that
that
go
along
with
the
usual
rolling
of
of
operating
systems,
but
a
key
feature
here
is
this
concept:
Edge
resources
and
it's
actually
a
really
great
day
for
you
to
be
on
Kate,
because
this
is.
B
This
is
exactly
the
area
where
you
and
your
team
are
doing
your
work
to
to
identify
and
and
map
up
right
things
that
can
be
found
at
the
edge
there's
so
much
room
for
for
collaboration
here.
We're
taking
a
whack
at
it
with
this
thing
called
that
resources,
because
we
need
to
have
some
data
definition
into
our
system,
but
ideally
I,
think
the
world
would
benefit
from
having
a
universal
approach
to
detecting
and
identifying.
B
Edge
devices
objects
attached,
sensors
Etc,
but
so
that's
in
for
iofog,
3
and
and
it
and
it's
stabilized
after
just
a
few
beta
versions.
B
B
That,
for
example,
a
Docker
container
or
Docker
compatible
container
was
was
pulling,
but
you
didn't
know
how
much
so
you
might,
you
might
see
it
in
the
state
of
pulling
down
for
an
hour
and
wonder
is
something
broken,
but
so
we
added
in,
of
course,
the
ability
to
to
see
through
the
control
plane
the
the
detailed
state
of
things.
B
Additionally,
a
lot
of
folks
were
mounting
volumes
into
their
their
containers
and
then
experiencing
failures
because,
for
example,
the
the
the
volume
that
they
wanted
to
mount
never
got
created
on
that
edge
device,
and-
and
there
was
a
lot
of
digging
required
to
know
what
the
problem
was
so
now,
of
course,
if
we
pass
back
some
more
information
about
what
happened
when
trying
to
start
up
that
microservice
made
some
updates
to
the
the
UI
that
that
is
part
of
the
ioflog
project.
B
It's
called
ecn
viewer
and
this
is
a
and
it
is
a
UI
that
just
hits
those
apis
that
I
mentioned
and
gives
you
a
visual
display
play
of
what
the
what
your
Edge
environment
looks
like
so
made
some
updates
there
and
now
include
the
The
Edge
resources
capability.
B
Let's
see,
we've
got
some
bug
fixes
and
then
there's
also
I.
Guess
it's
worth
noting,
because
I
haven't
said
it
yet:
there's
there's
command
line
tool
called
iofog
CTL,
which
is
really
close
to
cube
control
in
terms
of
the
way
that
it's
structured
and
it's,
of
course,
for
dealing
with
all
of
the
the
iof
fog
pieces
and
so
just
added
some
capabilities
to
that.
And
that's
where
we
came
out
with
the
first
with
beta
1
of
iofog
3..
B
So
we
proceeded
and
I
added
some
more
bug
fixes
and
some
more
features
to
to
Iowa,
folks,
ETL
and
so
on.
And
but
then
we
began
to
upgrade
some
information
in
the
in
the
controller.
Some
features
in
the
controller.
So
here's
where
I
want
to
introduce
the
ever
maturing
feature
that
we
have
historically
called
public
ports,
but
now
are
calling
Public
Services,
and
so
we've
talked
in
this
very
working
group
before
a
lot
about
like
service
mesh
at
the
edge.
B
And
how
do
you
expose
something
like
a
an
Ever
running
database?
That's
on
an
edge
node
and
then
it
maybe
you
have
some
vehicle
wants,
is
out
and
about
and
wants
to
deposit
information
in,
but
there
are
completely
different
networks,
this
one's
over
cellular,
this
one's
behind
that
layers,
and
so
what
you
want
to
do
is
find
a
way
to
expose
services
that
are
running
as
microservices
somewhere
at
Edge
or
cloud
or
whatever,
and
have
those
services
available,
but
in
a
safe
way.
B
So
this
is
using
that
that
that
that
Scupper
project
layer,
7
Network
overlay
on
what
we
do
is
we
expose.
If,
if
you
put
the
definition
into
your,
your
infrastructure,
spec
into
your
yaml,
will
open
a
publicly
available
URI
that
directly
talks
down
to
that
port.
On
that
microservice,
with
with
a
proxy
the
proxy
is
high
performance.
B
It's
really
expensive,
so
you
say:
I'd
really,
rather
only
see
a
little
snippet
when
there's
something
important
well,
what
if
you
actually
wanted
to
go
and
monitor
video
stream,
but
just
pipe
it
directly
from
the
edge
right
into
your
browser
window
or
right
into
your
rtsp
stream
viewer?
This
is
a
way
to
do
that
where
you
never
store
the
data
anywhere
in
the
cloud
it
just
moves
directly,
and
so
what
gets
exposed
is
a
public
service.
Endpoint
is
related
to
where
you're
hosting
your
control
plane.
Well,
that's
perfect!
B
If
you're
hosting
the
alpha
control
plane
as
part
of
your
it's,
it's
like
part
of
your
kubernetes
deploy,
because
you
can
open
up
all
kinds
of
networking
there
and-
and
you
know,
open
up
the
all
of
the
pathways
that
you
would
prefer
and
then,
when
you
open
up
the
public
service,
it's
just
one
more
URI
in
your
infrastructure,
the
difference
here
being
that
when
you
stand
it
down,
there
is
no
connectivity
that
remains
to
that
edge
node.
B
And
so
it's
a
way
that
you
can
host
anything
from
anywhere
within
a
an
endpoint
that
exists
like
up
in
the
internet
at
large
and
then
take
it
away
when
you
don't
want
it
anymore.
So
it's
also
really
good
for
temporarily
offering
a
service
and
then
shutting
it
down.
So
that's
getting
more
mature
as
we
as
we
go
and
then,
as
we
entered
into
this
year,
we
started
to
address
some
security
issues
that
were
identified
at
large
in
some
of
the
hosted
kubernetes
infrastructure
of
the
world.
B
In
other
words,
there
was
kind
of
a
rash
for
a
minute
there.
At
the
the
big
turn
of
the
year
of
things
hosted
on
various
and
I'm,
not
going
to
mention
any
details,
because
that's
the
sort
of
thing
that
you'll
know
about,
if
you
are
meant
to
know
about
it
on
some
hosted
kubernetes
systems
out
there,
the
major
cloud
system
stuff.
B
There
were
some
openings
that
were
that
were
there
by
default,
and
so
we
added
some
protections
against
that
and
then
got
that
out
the
door
and
then
with
the
very
latest
with
which,
which
just
came
out
in
the
last
30
days,
with
the
beta
7
on
our
folk
CTL
beta
6
on
the
agent,
we
finally
moved
away
from
OS
specific
packages
right,
different
ones
for
or
Debian
systems,
a
different
one
for
RPM
based
systems,
and
now
it
doesn't
really
matter
you
can
pull
from
anywhere
and
all
that
move
required
was
that
times
had
changed
regarding
the
way
that
you
install
and
configure
Docker
and
things
like
that.
B
It
no
longer
required
special
treatment
on
every
system
and
so
on,
every
OS,
and
so
we
finally
were
able
to
roll
to
a
universal
approach
there.
And
that
was
just
an
nice
little
Improvement.
B
You
can
see
that
the
changes
from
beta
to
Beta
are
getting
smaller
and
smaller,
and
that's
because
we're
rapidly
approaching
the
point
of
where
we
say
Well,
when
do
we
want
to
push
this
out
as
as
a
full
iofog
3
is
now
the
The,
Standard,
Version
and
so
on,
and
we're
busy
at
edgeworks
and
we've
got
some
some
a
couple
of
outside
contributors
who
who
can
help
as
well,
but
we
figure
we'll
give
it
enough
enough
time
to
settle
in
on
this
latest
and
if
we
don't
need
any
other
changes.
B
Probably
this
summer
is
what
we're
planning
for
this,
and
so
that's
really
the
the
gist
of
the
what's
going
in
and,
of
course,
you'll
see
here,
things
will
change
when
We've
added
the
third
version,
you'll
get
all
of
the
official
documentation
will
change
here,
and
so
the
beta
is
for
those
who
are
working
directly
off
the
repo
or
who
are
using
the
Iowa
folks
slack
to
to
keep
up
with
the
project
and
and
what's
going
on
there.
B
So
that's
the
gist
of
what
we've
got
in
iofog,
3
and
and
some
reasons
behind
what
we
why
we
added
what
we
added.
It's
got
some
time
for
questions
if
anybody's
got
them.
A
Yeah
thanks
help.
That
was
a
good
overview
and
I
like
the
introduction
too,
because
I
think
I
had
forgotten
some
of
the
aspects
of
it
over
the
over
the
your
appearances
over
the
last
few
years.
B
It's
been
a
long
time.
Hasn't
it
right.
It's
this
this
this
working
group's
early
days
and
which
was
just
a
couple
of
years
into
kubernetes
itself
right,
it's
I,
I
have
eclipse.
Iofog
is
an
old,
it's
an
old
project
in
terms
of
the
in
terms
of
the
foundational
Tech,
and
it's
not
something
I
had
realized
until
all
of
a
sudden
it
was
2022
and
I
was
like.
A
A
And
I
think
that
people
are
realizing
that
there
are
a
lot
of
Open
Source
projects
tied
into
Docker
in
some
way
that
maybe
can
be
reassembled
in
a
modular
fashion
to
use
things
going
on
in
other
in
other
areas.
To
oh,
yes,
benefit
of
users.
So
yes,
not
surprising!
Well,.
B
So
I'll
make
I'll
make
a
brief
in
my
last
couple
of
minutes
that
I
I
have
to
be
with
you
guys
today.
I'll
make
a
brief
like
like
seriously
I,
think
it's
a
soapbox
here
in
my
kitchen.
So
let
me
let
me
let
me
stand
on
it
for
a
moment
and
and
and
kind
of
make
a
make
a
pitch
for
where
I
think
things
can
and
maybe
should
go.
B
You're
right,
Steve
things
have
grown
and
changed,
and
we
now
know
that
microservices
architecture
exists
regardless
of
what
like
run
time
and
what
form
of
encapsulation
of
the
of
the
software
you
want
to
use,
I
would
bet
and
I
I
would
and
I
would
be
willing
to
even
like
lead
the
charge
on
this
I
would
bet
that
a
universal
agent
that
has
the
extensibility
to
interface,
with
whatever
software
encapsulation
you
want
web
assembly.
You
know
Linux
kernel
containers
that
would
be
something
and
and
also
can
talk
to
whatever
command
system
you
want.
B
Whatever
control
plane
you
want.
That
would
be
something
that
would
probably
be
worth
making
once
and
solid
for
everyone
if
it
has
all
the
right
extensibilities
in
it.
Because
of
how
difficult
it
is
to
get
it
perfect,
how
difficult
it
is
to
get
it
so
stable,
so
you
know
secure
that
that
you
can
rely
on
it
forever.
I
I,
don't
think
that
it
should
be
done
11
times
over
and
as
kind
of
like
the
the
ringleader
of
one
particular
implementation,
I'm.
B
Looking
and
thinking
what
we
really
want
is
we
want
Edge
Computing
components
that
are
usable
in
various
circumstances
that
that
are
really
designed
to
be.
You
know,
switched
out
or
extended
in
the
way
that
you
say
start
something,
and
then
you
have
a
modular
piece
of
code
that
knows
how
to
interface
with
the
runtime
that
you
want
to
start
well.
You're.
A
A
It
has
to
be
a
lot
better
exactly
maybe
this
movement
towards
web
assembly
is
what
would
provoke
something
like
that
to
make
people
realize
a
need
to
completely
do
something
in
a
new
direction
that
yeah
I
think
there
are
signs
that
people
will
want
to
do
both
and
you
could
do
each
with
their
own
kind
of
control,
plane
or
plug-in
interface
point,
but
maybe
a
common
one
that
could
manage
to
pull
off
both.
It
is
something
that
would
be
worthwhile.
B
In
in
the
country,
well,
I
I
think
the
form
is
probably
going
to
be
the
eclipse,
Edge
native
working
group,
because
it's
it's
bad,
it's
completely
back-end
agnostic,
and
that
and-
and
here
we
want
to
stay
focused
on
things
related
to
kubernetes
and
we
will
but
which
is
going
to
be
it's
going
to
comprise
a
huge
portion
of
the
back
ends
that
people
are
using,
but
it
won't
always
be
a
hundred
percent
right.
So
I'll
probably
take
up
the
you
know.
The
charge
of
talking
about
I
would
call
it
like.
A
Well,
well,
here's
something
I
I
take
a
step
back
when
you
say
that
that
I
think
there's
an
opportunity
for
this
unification.
You
know
web
assembly
to
Docker.
It's
not
just
Edge,
so
absolutely
a
reason
for
why
it
wouldn't
go
into
the
eclipse
Edge
Foundation,
but
I've
I've.
It
might
but
I
think
it's
far
broader
than
that
in
potential
I.
B
B
We
need
to
do
weird
stuff
on
these
little
devices
in
a
way
that
the
they're
not
even
networked
properly
and
that's
great,
because
maybe
the
middle
of
the
OSI
stack
is
not
where
we
want
to
be
configuring
networks
anymore,
maybe
yeah
everything
should
be
done
at
the
layer
where
you
have
control
over
data
flow
and
it
like
at
runtime,
love
layer,
7
networks,
I
love
them.
What,
if
that
plus
distributed
computing
controls
that
we
get
right
for
these
things?
B
What
if
that
actually
changes
the
way
that
we
do
cloud
and
all
of
the
systems
that
are,
you
know
basically
hosted
out
in
the
open
that
would
be
fantastic,
I
have
to
go
to
my
next
meeting.
This
has
been,
it's
been
really
nice
to
just
drop
an
update
in
here,
I
appreciate
it
and
I'd
like
to
talk
a
lot
more
about
all
of
this.
Things
have
been
super
busy,
but
super
great
to
be
back
in
here.
Well,.
A
B
B
There's
pretty
pretty
typical
for
this
working
group
is
to
these.
These
things
get
started
right
and
we
could
easily
go
for
four
hours
on,
and
hopefully
people
find
this
stuff
valuable.
Let's
put
a
pin
in
that
then,
and
then
continue
the
conversation,
because
I
think
yeah
I
would
definitely
put
effort
behind
in
the
architecture.
A
So
Diane
and
Kate-
maybe
a
few
things
we
can
talk
about.
One
is
Kate,
perhaps
moving
into
Cindy's
role
as
a
co-chair.
I
have
a
personal
email
for
Cindy
and
center
and
ask
on
you
know
what
her
thoughts
were
on
moving
to
the
America
status,
but
I
got
no
response
at
all,
so
I
don't
know
what
is
going
on
there.
Maybe
if
we've
gone,
you
know
with
attempts
to
contact
her
by
all
the
means
we
have.
We
can
just
do
it
unilaterally,
but
I'd
rather
somehow
get
verification
that
she's
informed.
C
Yeah
I
can
I
haven't
taken
that
step
to
reach
out,
but
I'll
do
that
right
after
this
and
see,
and
at
least
give
that
that
attempt
one
try
as
well.
A
Yeah,
because
we
noticed
on
her
Lincoln
Pro
profile
that
it
looks
like
she
left
Microsoft
and
now
works
at
Facebook
and
yeah,
so
yeah,
we'll
we'll
move
forward
with
that
plan
to
to
update
the
leadership
of
this
working
group.
A
The
other
thing
we've
got
going
on
is
the
the
kubecon
presentation,
I
actually
right
now,
I'm
on
a
vacation
but
on
the
plane,
I
just
put
together
a
rough
PowerPoint
outline
using
the
template
from
the
from
the
conference
and
really
all
it
has,
is
a
skeleton
of
some
agenda
bullets
and
a
closed,
but
it
needs
to
be
filled
out
and
I
didn't
even
get
a
chance
to
go
through
an
review
pass.
Otherwise
I
would
have
sent
out
the
link
share
to
that,
but
we
need
to
get
on
to
that.
A
We
still
have
six
weeks
assuming
that
we're
gonna
do
it
physical
I,
looked
in
by
the
way
to
the
requirements
for
physical,
they
seem
to
be
changing
and
the
cncf
itself
is
just
deferring
to
the
requirements
for
entry
to
Spain.
A
So
talking
to
the
cncf
leadership,
it's
a
process
where,
in
theory
the
EU
is
Unified,
but
they
slowly
get
to
a
point
where
there's
some
kind
of
Consolidated
policy
there
and
like
everywhere
else
in
the
world,
that
kind
of
thing
is
Shifting,
but
at
least
for
now
it
looks
like
you
need
proof
of
vaccination
and
the
definition
of
fully
vaccinated
that's
in
operation
now
is
that
you
also
have
a
booster
or
well
I.
Take
that
back!
It's
if
you
had
received
your
two
vaccinations
very
recently.
A
You
don't
need
a
booster,
but
if
they
were
done
like
a
year
ago
or
whatever
I,
don't
remember
the
exact
number
of
days,
then
you
need
a
booster
and
I
believe.
There's
a
proposal
to
relax
the
booster,
maybe
even
the
vaccination
requirement
other
than
that
I.
Don't
think
there
was
anything
on
that
list
that
would
prevent
travel,
I
think
that
they
they
are
thinking
the
venue
would
require
masking,
but
I'm,
not
sure
on
that
Steve.
Sorry,.
C
Go
ahead:
Keith
no
I'm,
just
going
to
Echo
I,
think
we'll
be
able
to
or
I
think
I
personally
will
be
able
to
make
it
in
with
those
requirements.
We
might
want
to
double
check
with
kilton,
but
yeah
I
think
it's
fine
to
assume
in
person.
If
we
all
feel
like,
we
can
make
those
requirements
now
and
then,
of
course,
things
change
and
the
requirements,
change.
I'm
sure
kubecon
will
also
be
understanding
if
that
needed
to
switch
to
Virtual
yeah.
A
For
me,
I
think
it's
a
90
chance
that
I'll
be
there
physically.
It
got
approved
for
travel
by
my
employer,
and
my
only
issue
now
is
I
haven't
been
boosted,
but
I
might
be
willing
to
if
it
looks
like
I
have
to
but
I
sort
of
as
a
personal
decision
here
didn't
I
think
I've
got
nothing
against
it,
but
it
just
didn't
seem
like
based
on
people.
I
know
who
got
it
and
still
got
covered
afterwards.
A
Convinced
me,
it
was
completely
worth
it,
but
you
know:
I
still
have
time
to
decide
if
I'm
just
going
to
wait
it
out
until
maybe
three
weeks
before,
because
it
looks
like
I
could
just
get
it
before.
A
I
go
and
I'd
be
okay,
so
I'm
I'm
gonna
just
keep
an
eye
on
it
for
the
next
few
weeks
to
see
whether
the
policy
changes
but
I'd
expect
that
I'll
be
able
to
go,
and
in
fact,
I've
got
a
bunch
of
other
things
going
on
there
too
I'm
there
is
that
kubernetes
on
the
edge
day
and
I
might
be
opening
and
closing
that
conference.
So
I'd
have
to
be
there
for
that,
because
that's
anticipated
to
be
physical
and
then
I've
got
a
couple
other
speaking
sessions
and
kubecon
too
so
anything
else.
C
So
I'll,
just
on
this
topic,
I've
been
I've
been
a
little
bit
out
of
the
loop,
with
the
cube
compensation
and
and
I
I
yeah
I
was
too
busy
with
other
stuff.
So.
A
Device
management
and
discovery
on
the
edge
I
think
it
was
called
device,
onboarding
and
management.
Something
like
that
I,
don't
don't
have
it
right
in
front
of
me,
I
think
if
you
go
back
in
the
slack
Channel
I
shared
the
link
to
a
to
Google
doc,
and
it
has
whatever
we
proposed
to
the
cncf
that
you've
been
through
this.
So
you
know
that
you
know
you
kind
of
have
a
rough
agenda,
but
you
know
it
was
drafted
in
a
way
that
we
have
some
level
of
flexibility
into
where
we
go
with
that.
A
It's
just
under
the
broad
umbrella
of
this
was
covering
devices
and
I.
Think
the
idea
was
devices
could
be
things
like
sensors
cameras
as
and
devices
the
way
it
was
word.
It
could
even
be
interpreted
to
include
the
hardware
that
hosts
your
kubernetes
notes
if
you're
running
kubernetes
at
the
edge
or,
if
you're
at
in
a
context
where,
what's
that
edge,
is
not
a
kubernetes
note
but
still
is
managed
in
some
way
by
kubernetes
it
the
umbrella
of
the
agenda
we
put
out.
There
is
big
enough
to
talk
about
that
as
well.
C
A
And
I
think
that
we
wanted
to
do
some
coverage.
I
I
would
like
to
get
this
to
cover
the
secure
device
onboard
process.
You
know
this
idea
that
in
reality
for
Edge
at
scale,
a
lot
of
people
expect
to
buy
Hardware
devices.
You
know
through
a
distribution
chain
yet
somehow
end
up
with
it
successfully
deployed
with
either
nobody
or
somebody
with
very
little
training
and
skill
set
out
at
that
edge
location.
A
Just
kind
of
somebody
who
can
plug
in
plug-in
power
plug
in
a
network
connection,
maybe
look
at
whatever
physical
status
might
be
there,
which
could
just
be
as
little
as
some
flashing
lights
and
somehow
get
that
device
to
come
boot
itself
up
bringing
itself
up
to
the
point
where
it's
remotely
manageable
and
it's
kind
of
the
same
story.
If
you
have
a
device
failure
and
we'll
just
do
a
broad
setting
of
the
terrain
of
what
open
source
projects
are
out
there.
C
I've
done
some
work
with
sdo
as
well:
I've
used
it
a
little,
so
I
can
talk
about
that.
If
we
need
yeah,
we
explored
it
a
little
bit
so
the
whole
setting
up
the
Rendezvous
service
and
getting
that
flow
working,
I.
A
Started
an
attempt
at
doing
it
and
I
found
that
that
has
been
really
in
transition
and
a
lot
of
it
is
by
Design,
where
that
project's
made
a
conscious
decision
that,
rather
than
be
one
of
these
things,
that
goes
spec
first
and
then
they
do
an
implementation
that
in
a
way
they
seem
to
be
operating
in
a
mode
where
often
the
reality
turns
out
to
be
this.
That
the
first
implementation
is
the
spec
so
and
they've
been
doing
drops,
maybe
as
often
as
monthly.
A
So
the
the
thing
is
just
completely
moving
around
and
they
had
a
demonstration
thing
that
I
started
down
the
path
in
January
of
trying
to
get
one
done.
But
I
had
my
day
job,
so
I
wasn't
giving
it
full
time,
but
I
I
never
was
successful
at
actually
synthesizing
an
actual
deployment,
but
I
think
it
was
just
on
me
not
so
much
on
the
project.
A
But
I'm
meant
to
go
back
to
that,
and
maybe
even
the
story
of
what's
involved
in
that
I
I
think
that
diversion
numbering
is
less
than
a
1.0
I
believe
and
applying
that.
Maybe
it
isn't,
you
know
in
its
final
form,
but
just
recounting
what
the
status
of
that
is
for
those
interested
I
think
would
be
useful
to
the
community
yeah.
C
I
mean
I
would
like
to
know
the
status
I
was
working
with
it
two
years
ago.
I
don't
think
there
was
really
even
much
public
documentation
so
and
what
you
were
saying
about
it
not
being
stuck
first
I
couldn't
really
put
the
words.
What
was
off
about
it
sometimes,
but
that's
definitely
how
it
felt
a
lot
of
reverse
engineering.
A
What's
happening
there
and
like
you,
when
you
go
that
thing
that
direction
the
last
thing
to
get
updated
is
the
docs,
but
if
you're
trying
to
adopt
it,
that's
kind
of
the
first
thing
you
want
to
go
to
so
yeah
it
was.
It
was
out
there.
I
didn't
try
to
Avail
myself
with
the
slack
Forum
to
ask
questions
because
I've
seen
other
projects
that
got
into
an
operating
mode
like
that,
and
that's
really
half
how
you
have
to
go
about.
It
is
just
go
out
there
and
ask
questions
but
yeah.
A
That
should
be
one
thing
we
should
cover,
because
that's
certainly
within
the
scope
of
kind
of
the
idea
of
getting
these
Edge
locations
booted
up
and
to
the
point
where
you
can
be
using
kubernetes
to
manage
them
and
operate
services
that
are
consuming
things
generated
out
at
the
edge
so
beyond.
A
If
that
answered
your
question
as
to
what
the
scope
of
this
is
and
any
ideas
you
might
have
for
shaping
the
agenda,
Hilton
was
unclear
whether
he'd
be
there
but
I
think
he
said
he
wanted
a
role
of
maybe
doing
a
demo
at
the
end
so
because
it
would
give
them
the
flexibility.
To
be
honest,
when
you
do
a
demo,
I
found
that
often
you
want
those
to
be
recorded
regardless,
particularly
with
this
Edge
stuff,
because
counting
on
conference
Wi-Fi
and
things
like
that
is
kind
of
a
little
optimistic.
A
So
maybe
that
maybe
Kate
and
I
could
talk
set
the
table,
for
you
know
the
broad
overviews
and
maybe
dig
a
little
deeper
into
selected
projects.
You
know
if
you
want
to
put
in
accurate
of
what
you're
going
working
on
personally,
that
that
would
be
good
to
have
a
subject
matter
expert
go
into
some
of
these,
and
we
just
we
don't
want
to
call
Favorites,
but
just
pointing
them
out
as
a
deeper
dive
into
one
example.
A
I
think
is
perfectly
Fair
as
long
as
you
mentioned
that
there's
a
couple
others
out
there
if
there
are
and
we'll
shape
it.
That
way,.
C
A
C
Nothing
particular
I
just
wanted
to
close
out
that
previous
just
should
we
you
Steve
me
and
kilton,
maybe
start
a
channel
and
kind
of
organize
when
we
want
to
bring
this
deck
together.
A
Yeah
I
think
so
let
me
I'm
on
vacation
through
this
week,
so
I'll
Monday,
I'll
I'll,
put
out
a
link
to
the
document.
Are
you
traditionally
when
we've
done
this?
We've
done
this
through
Google
Docs,
because
everybody
can
get
to
those
but
I
actually
prefer
to
work
on
the
Microsoft
stack
and
I.
Think
Hilton's,
okay,
with
it
too
so
and
I
assume.
A
Maybe
you
are
in
that
situation
too
I'm
just
more
experienced
in
PowerPoint
and
have
found
that
I
I
think
I'm
capable
of
using
features
that
are
in
PowerPoint
that
aren't
in
the
Google
Slides
so
but
I'm
open
to
whatever
you
want
to
use
for
sharing,
but
I'd
propose
that
we
put
a
deck
out
there
in
some
shared
form,
and
then
we
can
use
slack
VMS
with
a
three-person
group
or
baby
will
add
Beyond.
If
he's
volunteering
to
be
a
reviewer.
A
That
would
be
really
useful
yeah
and
we
can
use
slack
so
that
we
can
do
it
asynchronously
right
and
we
we
might
end
up
wanting
to
do
some
Zoom
meeting
as
we
get
closer.
But
you
know
that
that.
C
A
The
other
thing
we've
done
these
before
at
physical
conferences
and
we
usually
end
up
doing
a
final
rehearsal
in
person
at
the
conference
venue,
because
sometimes
when
you
go
through
kind
of
that
actual
run
with
live
people
trying
to
do
it.
You
you
find
a
few
flaws
and
I've
usually
found
a
a
guinea
pig
audience
member
so
that,
if
we're
there
in
person,
we
can
maybe
meet
the
day
before
and
just
do
a
dry
run
of
presenting
it
before
we
go
on
stage.
So
I'd
like
to
do
that.
A
Okay,
if
that's
it,
maybe
we'll
end
this
meeting
16
minutes
early
last
call
if
anybody's
got
anything
else,
they
want
to
chat
about.
Okay.
Well,
thanks
everybody
for
attending
great
meeting,
even
if
it
was
shortened
see
you
next
time.