►
From YouTube: Flatcar Container Linux community meeting 2021-12-17
Description
No description was provided for this meeting.
If this is YOUR meeting, an easy way to fix this is to add a description to your video, wherever mtngs.io found it (probably YouTube).
A
Last
meeting
this
year-
and
I
see
that
we
have
a
new
guest-
stefan,
you
wanna
feel
free
to
speak
up
and
say
hello
to
us.
There's
there's
an
official
introduction
introduction
later
on,
but
I
I
think
we
have
some
time
to
fill.
B
Yeah
sure,
if
you'd
like
to
yeah
hi,
I'm
stefan
I'm
first
time
here,
I'm
eager
to
hear
about
cluster
api
integration
at
the
time.
That's
my
main
driver
for
joining
today.
I
won't
be
there
every
time,
but
we
are
flat
car
users.
We
have
bare
metal
clusters
like
more
than
900
servers,
all
bare
metal,
big
beefy,
stuff
and
yeah
we're
running
flat
car
underneath.
A
Amazing
and
I've
been
using
your
services
quite
extensively
so
awesome
to
have
someone
join
great
nice
peter.
You
made
it
great
sorry
for
the
for
the
following:
you're
still
muted,
but
still
great.
C
To
have
you
yeah?
No
sorry,
sorry
about
that.
I
had
to
finish
up
another
meeting
first,
so
the
links
didn't
affect
me
until
like
seconds
ago,
so
perfect.
D
A
But
it's
all
good!
Thank
you
awesome
all
right.
I
think
we
just
get
started
and
have
everybody
else
who
joins
later
introduce
themselves
when
they're
joining
so
as
usual,
we've
prepared
some
slides.
Let
me
share
those
with
you
real,
quick.
A
Yeah,
I
know
I
know
here
we
go
anyways,
welcome
everybody
to
the
end
of
your
meeting
of
the
flakka
linux
maintainer
team.
We
have
a
very
special
meeting
for
you
today,
it'll
be
almost
entirely
armed.
64
focused
though
we
have.
We
have
a
few
news
regarding
cluster
api
as
well.
Let's
have
a
quick
look
at
the
agenda.
A
This
is
our
welcome
and
the
welcome
section
usually
has
all
of
you
saying,
quick,
hello,
that's
what
we're
gonna
do
next,
then
we
have
some
exciting
news
to
share
with
you
and
after
that,
it's
pretty
much
show
and
tell
mostly
tell-
and
this
is
about
your
personal
arm-
64
story
doesn't
need
to
be
flat
correlated
it's
awesome
if
it
is
threat
correlated,
but
it
does
not
need
to.
A
F
D
Hi
everyone,
here's
matthew
working
in
the
flat
color
team
too,
and
I
mainly
focus
on
testing
the
release
before
going
into
production.
Edward.
G
C
Thanks
ed
peter
puglia,
from
imperial
computing,
I
donated
a
bunch
of
armed
servers
to
the
flak
art
team
a
couple
years
ago
and
we
use
a
flat
car
on
top
of
well
within
the
context
of
both
virtual
machines
and
some
bare
metal
in
the
context
of
openstack.
C
Oh,
and
I
guess
I
can
hand
over
to
who
I
don't
know,
who
hasn't
gone
yet
how
about
dongsu.
H
I
J
Hi
hi,
I'm
joachim.
I
am
also
a
part
of
keen
fault
now
microsoft
and
I
actually
work
with
a
different
team
from
flat
car,
but
is
the
one
that
is
responsible,
among
other
things,
for
the
nebraska
project
that
that
deals
with
managing
the
updates
for
flight
car.
J
Oh
sorry,
who's!
Next,
let's
see.
K
Hi,
I'm
danielle
based
in
berlin,
also
part
of
the
flatcar
team
program
manager
for
arm
and
community,
so
very
excited
to
have
this
meeting
towards
the
end
of
the
year
and
what
a
nice
present
to
to
finish
the
year
with.
K
So,
I
think
scott
yeah
you're
next.
L
Sure
yeah,
hello,
everybody,
my
name
is
scott
malkey.
I
work
at
aws.
I've
been
there
about
six
years.
I
handle
graviton
and
super
excited
to
to
be
here
and
just
you
know,
bask
in
the
glory
of
all
things
arm,
64.,
arm64
and
help
out.
However,
we
can.
K
I
think
stefan's
still
an
official
introduction.
B
Yeah
and
hi
again,
I'm
stefan
I'm
first
timer
today.
Here
I
work
in
one
on
one
being
a
product
owner
for
our
kubernetes
clusters.
We
are
on
the
kubernetes
journey
since
2017
have
900
and
more
bare
metal
service.
By
now
running,
kubernetes
running
flat
car
on
them,
we
started
with
container
linux
and
then
we
transitioned
to
flatcar,
and
we
are
eager
eventually
to
go
in
the
direction
of
cluster
api,
and
this
is
why
I'm
here
today.
A
Awesome,
thank
you
everybody
for
introducing
yourselves.
It's
it's
very
a
great
loop
today.
I
guess
so.
Let's
go
to
our
next
section.
Just
share
a
few
of
our
latest
news
that
we
have
on
on
flatkar
linux.
A
Many
package
updates
and
security
fixes
in
the
beta
we've
bumped
up
to
open
ssl
3.0,
which
helps
us
and
our
users
towards
standards
compatibility.
A
I
need
to
admit
that
I
forgot
the
standard
phipps
thanks
thanks
for
helping
out
yeah,
so
you
can
now
use
flat
car
open,
ssl
in
fibs
compatible
mode,
and
we
have
a
blog
post
on
that.
It
also
includes,
as
usual,
for
major
release
many
other
packages
and
cve
fixes
the
new
alpha
is
kind
of
stuck
in
qa
right
now.
We
want
to
make
sure
it
gets
really
good,
so
we
spent
a
few
more
days
on
it.
A
It
will
ship
kernel,
515
and
it'll,
be
a
big
jump
in
kernel,
support
for
flat
car,
and
there
is
another
news
item
and
stefan
joined
just
because
of
that
which
is
amazing
yeah.
So
after
quite
a
bit
of
work,
we
finished
merging
a
foundational
pr
into
the
cluster
api
project.
A
It's
the
cube,
811
cube
adm
bootstrap
provider,
which
has
been
extended
to
now
also
cover
ignition
configuration
that
one
was
like
the
big
thing
that
we
needed
to
merge
it'll,
be
part
of
the
plus
api
v11
release
and
for
some
time
now,
image
builder,
which
is
used
by
class
api,
also
supports
flat
car.
So
that
work
is,
is
pretty
much
done.
We're
now
looking
into
the
aws
and
azure
providers,
which
are
almost
good
to
go
they're
pending
a
few
more
iterations
and,
of
course,
the
class
api
core
release
that
will
ship
ignition
support.
A
There's
a
tracking
issue
like
kind
of
a
road
map
milestone
thingy
that
we
maintain.
If
you
want
to
like
to
join
the
effort.
That
is
the
place
where
you
need
to
go
and
since
the
rest
of
the
meeting
is
on,
64
focused
I'll,
stop
right
here
and
hand
over
to
stefan
again
to
share
the
interests
in
class
api.
B
And
that's
not
so
much
to
say
so
far.
We
didn't
do
anything
with
cluster
api,
but
it
seems
cluster.
Api
is
the
right
thing
to
use
for
us
to
manage
our
clusters.
We
have
around
16
clusters
so
far
and
of
course
we
have
quite
some
elastic
workload
at
daytime.
Of
course,
there's
a
lot
more
load
on
the
clusters.
B
One
use
case
for
us.
Is
it
it's
important
for
us
that
every
week
we
are
currently
fully
redeploying
all
of
our
clusters,
so
every
node
gets
teared
down
and
gets
booted
up
and
redeployed
from
scratch,
and
this
is
the
thing
we
would
like
to
do
fully
automated
and
ideally
at
night,
so
like
elastically,
we
would
like
to
to
spin
down
the
clusters
at
night,
even
if
we
have
bare
metal
instances,
so
it
doesn't
from
a
cost
perspective.
K
B
Would
be
great
to
be
able
to
do
that
via
cluster
api.
We
could
dynamically
move
servers
then
from
one
cluster
to
the
other.
They
are
distributed
in
different
data,
centers
and
different
network
layers,
and
it's
it's
kind
of
volatile
where
we
need
most
power,
and
that
would
be
a
great
addition
to
our
setup
to
do
that.
Automatically.
A
That
sounds
like
a
really
interesting
use
case,
particularly
carrying
the
services
that
are
displaying
right
behind
you.
That's
amazing
powering
that
yeah,
let's
keep
in
contact
and,
let's
make
sure
we
do
a
cluster
api
edition
of
this
call,
as
soon
as
like
all
of
the
support
is
wrapped
up,
there's
a
number
of
other
providers,
which
may
be
more
interesting
for
you,
since
you're
working
on
bare
metal
but
they're,
also
in
the
queue
and
are
part
of
the
tracking
issue,
actually
perfect.
Awesome
thanks
for
joining,
and
thanks
for
the
information.
A
This
is
the
part
where
all
my
slide
design
skills
went
into
this
slide.
So
I'm
not
very
creative,
but
this
is
the
big
rm64
part
it's
more
more
or
less
open
mic
time
after
we
give
you
some
intro
to
how
flat
car
arrived
at
this
at
this
major
goal
that
we
had
and
then
it's
on
you
to
share
your
arm.
64
story
doesn't
necessarily
need
to
be
flat.
Car
related
yeah,
so
we're
in
stable
now
all
of
the
releases,
since
the
alpha
release,
of
course,
have
been
fully
released
tested.
A
We
have
a
number
of
release
tests
that
we
do
for
every
release
and
the
hardware
this
currently
runs
on
is
graviton
2,
that's
aws
virtualized.
We
have
requested
access
to
graviton
3,
and
that
is
a
process
and
it's
pending
as
soon
as
we
get
access
we'll
also
cover
the
latest
graviton
family.
There
we
have
been
using
for
quite
some
time
empire
emax
in
the
equinix
metal,
bare
metal
cloud.
So
that's
the
part
where
we
test
arm
support
on
actual
bare
metal
mp
alta
is
coming
soon.
That's
not
on
equinix
metal.
That's
on
us.
A
We
need
some
updates
to
our
test
environment
that
mafia
is
working
on
and
as
soon
as
that's
available,
we're
gonna
include
ultra
testing
in
our
release
test
as
well,
and
then,
since,
like
forever,
we've
been
testing
on
qemo
uevi.
I
Yeah,
so
maybe
to
give
some
background,
it
was
already
some
existing
work
that
we
started
with
so
core
s
had
an
arm
port
that
was
not
complete
and
they
decided
to
remove
it
again
when
this
end
of
life
decision
came
because
they
had
reduced
their
maintenance
and
yeah.
So
we
had
to
like
revive
these
old
patches,
put
them
to
the
newer
state
and
then
start
a
lot
of
fixing
and
yeah.
It
was
not
like
a
constant
focus
work.
We
had
some
different
business
model
at
that
time,
so
that
means
we
yeah.
I
We
had
some
kind
of
pieces
where
we
worked
on
this
part,
and
then
we
worked
on
aws
arm
support
and
then
later
on,
we
worked
on
that
part.
So
yeah
it
took
a
while
to
go
through
all
these
different
bumps
that
we
hit.
For
example,
there
are
some
cross
compilation
pitfalls
that
maybe
everyone
runs
into
at
some
point,
so
some
here
I
would
like
to
share.
For
example,
normally
you
have
the
standard
build
system
like
auto
tools
or
meson,
and
they
do
their
jobs.
I
They
know
what
to
do
and
they
will
pick
up
the
right
compiler
and
they
the
right
cross-compilation
options
but
yeah.
Of
course,
this
stops
working
when
people
customize
their
build
system,
and
that
seems
to
be
a
common
thing
that
everyone
hacks
their
own
things
into
auto
tools
and
then
cross
compilation
doesn't
work
anymore
and
that
is
even
worse
with
some
hand
rolled
make
files.
I
So
yeah
you
always
have
to
carry
some
down
pad
downstream
patch
for
these
kind
of
things
until
it
gets
fixed
upstream,
and
then
we
have
the
problem
that
some
build
systems
assume
they
can
make
a
test
compilation
to
check
for
some
compiler
feature
or
because
they
have
some
small
header
binary.
They
want
to
compile
first
and
then
they
want
to
run
this
binary.
But
now,
for
example,
you
are
on
an
amd64
host
and
you
are
compiling
this
arm.
64,
binary
and
yeah
running.
It
doesn't
work
unless
you
set
up
chemo
user
emulation.
I
I
For
example,
it
works
with
docker
containers
or
whatever
apartment
containers
at
that
point,
but
it
doesn't
work
when
you
have
yeah
this
build
script
cross
compiling
to
some
other
folder
where
the
new
root
file
system
should
be,
and
then
it
wants
to
start
it
without
change
rooting
it.
So
that's
yeah.
They
also
do
this
on
64
and
it
normally
works,
because
the
libraries
are
similar
to
the
whole
system,
but
it
normally
yeah
access,
no
guarantees.
This
would
actually
work
but
yeah
for
our
arm
64
purposes.
I
When
you
start
this
small
helper
binary
that
the
build
system
wanted
to
compile
and
then
another
issue
we
had
was
that,
while
rust
has
a
good
cross
compilation
setup
with
rust
up,
things
are
different
with
gen
2
and
we
had
to
make
some
changes
to
the
evil
files
that
we
use
and
yeah
like
finding
the
right
linker
and
using
the
right
link
arguments,
and
so
on
things
like
yeah.
You
don't
know
why
you
are
doing
this,
but
yeah
you
have
to
and
yeah
that's
also
related
to
our
build
system.
I
In
general,
we
have
this
sdk
from
chromium
os.
It's
some
workflow
for
building
our
image,
and
it
has
this
separation
between
regular
packages
and
tool
chain
packages
like,
for
example,
the
gcc
cross,
compiler
and
rust,
is
somehow
a
mix
of
both,
and
that
was
also
a
bit
yeah,
probably
some
and
often
a
recurring
scene
yeah,
and
then,
when
all
these
things
are
done,
we
have
our
fields
running.
I
We
have
this
image
and
then
realize.
Okay
now
some
applications.
They
don't
actually
work
as
they
should.
For
example,
there
was
this
pull
kit
crash
where
yeah
we
were
blocked,
because
we
didn't
have
updated
this
mods.js
library.
Yet
and
now
that's
done
and
that's
why
we
are
in
stable
now,
and
another
issue,
of
course,
is
on
even
higher
layer,
some
container
images
that
we
use
in
tests,
for
example,
they
are
not
multi-arc
images
and
yeah
in
general.
I
Thanks
to
this
chemouser
emulation,
it's
normally
really
easy
to
build
multi-arc
container
images
when
you
know
the
trick
how
to
do
it
by
specifying
on
this
arm,
64
based
image,
explicitly
yeah
but
yeah.
I
mean
it
takes
some
time
until
everybody
adopts
this
and
then
publishes
correct,
multi-arc
images.
A
And
if
I
recall
correctly,
even
just
in
the
last
bullet
point,
there's
like
six
months
hidden
behind
that
bullet
point,
that
was
quite
some
work.
Great
work
team,
william,
did
actually
quite
a
bit
of
the
pocket
work
he
can't
he
couldn't
have.
He
couldn't
make
it
today
he's
usually
our
go-to
person
for
m64.
A
Thank
you
for
jumping
in
kai
and
for
giving
the
details
and
the
upper
part.
Of
course
that
was
before
william
joined.
So
this
is
definitely
it
has
definitely
quite
a
bit
of
past.
I
see
that
one
more
guest
has
joined
and
we
usually
have
the
say
hello
round
at
the
very
beginning.
M
Sorry
I
was
just
listening
in,
I
was
expecting
to
say
hello,
hey
everyone,
I'm
george
castro,
big
flat
car
user
and
fan-
and
mostly
I
just
like
to
listen
in-
and
I
just
kind
of
wanted
to
be
here
for
the
arm.
64
stuff.
A
G
A
little
bit
so
my
arm,
64
and
flat
car
story
goes
back
all
the
way
to
core
os
when
it
was
when
it
briefly
had
arm
64
support
and
then
didn't
it
was.
G
It
was
clear
at
the
time
so
I
was,
I
was
working
at
the
time
for
packet
packet
got
acquired
metal,
so
I
didn't
didn't
really
change
jobs
just
got
absorbed,
but
it
was
clear
at
the
time
that
arm
64
support
for
a
system
as
complex
as
an
operating
system
that
wanted
to
run
all
the
containers
was
going
to
be
a
bunch
of
work.
G
I've
seen
firsthand
the
build
system
stuff
that
you
mentioned,
and
the
process
of
porting.
Before
that
there
were
things
like
making
sure
that
compilers
generated
the
right
code.
There
were
some
memorable
memorable
bugs
and
go
that
caused
crazy
crashes.
If
you
had
a
containers
that
had
too
many
layers,
I
mean
just
like
weird
stuff
that
after
you
fixed
it,
it
worked
just
fine.
But
if
you
were
the
first
person
to
try
it,
you
might
be
the
first
person
to
experience
a
bug.
G
So
you
know
my
I've
been
doing
internet
for
a
long
time
and
remember
the
days
when
people
were
porting
things
to
intel
because
they
were
reporting
them
from
sun,
3
or
sun
4.
G
Or
what
have
you
and
the
arm
story
reminds
me
very
much
of
the
early
intel
story
at
first
people
make
fun
of
you
because
they
can't
imagine
these
sort
of
things
being
used
seriously,
because
these
are
just
toy
computers
or
very
esoteric
things
and
then
momentum
starts
to
build
and
lots
of
people
start
to
have
these
systems
and
put
them
into
use
and
about
halfway
through
the
process.
G
You
realize
that
yeah,
you
know,
we've
run
through
a
lot
of
issues,
but
there's
a
finite
number
of
issues
and
and
people
will
help
us
fix
them
once
we
set
a
good
example
for
things,
so
I'm
I'm
very
much
looking
forward
to
the
multi-architecture
environment
of
the
you
know
of
this
decade,
the
the
work
that
aws
has
done
with
with
graviton
and
graviton
2
and
graviton
3
the
work
that
amp
here
has
done
with
emag
and
ultra
the
amazing
world
of
single
board
computers
that
are
out
there
that
people
are
itching
to
try
everything
out
on
and
and
do
you
know
completely
crazy
stuff
because
they
can,
because
the
cost
of
making
a
mistake
is
really
low.
G
A
C
So
you
know
I
started
out.
I
worked
in
scale-based
computing
for
for
a
long
time
on
openstack
from
very
early
on.
I
did
it,
you
know
on
on
windows
back
then,
but
it
introduced
me
to
scale
based
computing
and
essentially
you
know
for
anybody
who
works
in
technology.
You
collect
a
lot
of
systems
over
time
and
in
fact
you
know.
For
me,
the
birth
of
the
raspberry
pi
was
awesome
because
it
allowed
me
to
have
lots
of
compute
with
very
little.
You
know
electrical
surface
area
being
consumed.
C
So
essentially
you
know
it
opened
the
door
to
being
able
to
play
with
you
know
high-end
technologies
on
low
end
hardware,
and
it
you
know,
helps
helps
people
like
me
build
you
know
and
us
build
the
skill
sets
in
others.
You
know
build
the
skill
sets
necessary
to
work
in
this
in
this
environment
which
isn't
the
way
it
was
when
I
first
started
a
long
time
ago.
C
So,
from
my
perspective,
you
know
having
the
ability
you
know
like
I
I
got
to
play
with
all
the
as
I'd
said,
the
toys
for
a
long
time,
and
then
you
know
to
get
to
do
that
and
transition
it
into
a
career.
Where
now
I
get
to
play
on
some
of
the
most
cutting
edge
hardware
on
the
planet
and
redefine
what
it
means
to
put
scale
in
your
data
center
right
from
a
amount
of
cores,
and
you
know
per
per
socket
and
all
that
stuff.
C
So
it's
super
exciting
to
be
part
of
this.
You
know
part
of
the
fun
that
I
get
to
do
with
my
day.
Job
is
helping
enable
projects
like
this
with
resources
to
make
it
you
know
such
that
they
can
have
you
know
native
testing,
compilation,
binary,
artifacts
produced
right
and
doing
it.
You
know
the
the
mental
evolution
of
moving
from
a
raspberry
pi
to
a
server-based.
C
You
know
computing
model
from
arm
right,
because
most
everybody
has
had
experience
with
raspberry
pi's
the
level
of
excitement
that
operators
get
when
they
get
to
actually
see
you
know.
Armed
servers
in
their
data
center
is
usually
off
the
charts
so,
and
I
can,
I
can
give
some
references
for
those
who
would
be
glad
to
talk
about
it
who
work
specifically
in
the
open
source
space
so
so
yeah.
I'm
I'm
super
excited
super
glad
that
you
know
flat
cars.
You
know
evolving
to
have.
C
You
know
stable
support
in
the
distribution.
I've
been
a
long
time
user
from
back
in
my
microsoft
days
in
openstack
days,
so
yeah
really
like.
What's
going
on,
and
thanks
guys
for
inviting
me
to
participate.
L
Oh
yeah,
I
just
I
just
want
to
echo-
I
mean
peter
kind
of
stole
my
thunder
there,
like
it's
just
super
exciting
to
be
a
part
of
it
right,
like
you
know,
I
think,
just
seeing
the
community,
I
think
you
know
in
open
source
there's
always
this
sort
of
spirit
of
like
coopetition
right
like,
and
I
really
dig
that
right,
because
it
is
about
making
each
other
better
and
about
giving
the
best
experience
to
our
customers
right,
yeah
and-
and
I
think
that's
that's.
L
What
motivates
me
and
again
I'll
echo
peter
and
just
say
like
once,
you
sort
of
talk
through
a
customer
and
they're
sort
of
doing
it
in
this
sort
of
x86
kind
of
very
sort
of
traditional
legacy
way,
and
you
get
them
to
see
it
in
an
arm
64
like
in
in
a
a
data
center
in
a
a
scale
out
environment
right
in
aws,
wherever
right,
it's
just
really
cool,
because
they
you
sort
of
see
that
that
gear
turn
right.
That,
like
oh,
my
gosh,
this
is
this
is
real.
L
This
is
this
is
awesome.
I
can
do
what
I
want
to
do.
It's
faster,
better,
cheaper!
It's
more
sustainable
right.
That
was
a
big
thing
that
we
we
focused
on
when
doing
that.
Two
to
three
graviton
two
to
graph
on
three
is
really
sort
of
leaning
into
the
sustainability
aspect
of
it,
and
so
it's
just
it's
great.
It's
great
to
be
here,
it's
great
to
be
part
of
the
community
and
we're
happy
to
to
be
a
part
of
it
and
help
out.
L
However,
we
can
the
graviton
three
access
is
coming
tilo.
There
is
a
little
bit
more
approvals
necessary
given
the
company
you
for
in
the
company
I
work
for
so
you
know
just
know
that
we're
we're
working
on
it
we'll
get
you
in
as
soon
as
we
can
and
excited
to
see
what
you
can
do
with
it
so
yeah
great
to
be
here
and
yeah.
The
the
arm64
going
stable
is
just
it's
so
awesome,
it's
great
to
see
and
and
appreciate
all
the
work
of
the
team.
L
You
know
for
the
long
time
doing
it,
we
see
a
lot
of
customers
doing
the
same
thing
right.
The
the
whole
like
actually
getting
the
binary
to
compile
on
arm.
64
is
typically
the
easiest
step.
It's
usually
the
tool
chain:
it's
usually
the
unit
test.
The
smoke
test,
the
integration
test,
the
whole
thing
where,
like
hey,
the
the
compiler
or
the
the
build
chain,
really
wants
to
do
just
a
quick
compile
just
to
test
something
and
oops.
You
know
exec
format,
error.
L
You
know
like
yeah
that
totally
resonated,
because
we
run
into
that
all
the
time,
and
so
you
know
the
best
processor
in
the
world
is
not
terribly
useful.
If
the
software
you
want
to
run
doesn't
run
on
it.
So
yeah
again
just
appreciate
all
the
work
and
and
happy
to
be
here.
D
Okay,
so
maybe
I
can
go
with
my
short
story,
it's
more
a
personal
story.
It
was
my
first
meeting
with
raspberry
pi.
It
was
a
couple
of
years
ago,
maybe
six
or
seven
years
ago,
beginning
of
my
studies
and
with
a
couple
of
friends,
we
decided
to
participate
to
a
hackathon
so
kind
of
competition
to
bring
some
innovation
regarding
digital
stuff
and
most
of
the
participants
were
developing
website
and
and
mobile
application,
and
we
decided
to
go
with
raspberry
pi
to
do
some
iot
because,
as
everyone
said,
it's
super
easy
to
use
raspberry
pi.
D
I
think
it
was
my
first
meeting
with
raspberry
pi
and
when
I
just
booted
the
the
operating
system
at
this
time,
it
was
mentioning
that
I
it
was
required
to
you
know,
extend
the
root
partition
or
something
like
that,
and
I
just
ignore
this
message
and
I
continue
with
installation
and
so
on
and
a
couple
of
hours
after
after
installing
all
bunch
of
stuff
on
the
raspberry
pi,
like
packages
and
programming
stuff
like
node.js,
the
the
root
partition
was
just
full
and
my
respy
has
crashed
and
yeah.
D
It
was
just
10
minutes
before
going
on
stage
to
present
the
project
in
front
of
a
jury.
So
I
was
just
nothing
to
to
to
present
to
the
jury,
so
I
was
running
everywhere
in
the
room
of
a
hackathon
to
try
to
find
the
sd
card
to
try
to
flash
a
new
operating
system
and
to
reinstall
all
my
stuff
on
the
respite,
but
yeah.
I
ran
out
of
time,
so
I
just
been
in
front
of
a
jury
and
present
the
project
with
hands
in
my
pockets.
D
We
have
nothing
to
show
actually
no
no
proof
of
concept
and
yeah
plot
twist.
We
won
the
the
competition,
so
that
was
amazing
to
win
this
competition
but
yeah.
It
was
my
my
first
experience
with
raspberry
pi.
So
with
this
thing,
and
I'm
not
sure
now,
it's
required
to
extend
the
the
root
partition
when
you
first
boot
a
respite,
because
maybe
so
much
people
are
just
falling
into
this
trap,
so
yeah
so
yeah.
That
was
my
story
with
arm64.
A
Awesome
I'll
throw
one
in
because
I
have
some
history
of
arm
in
general
and
actually,
when
I
started
working
professionally
with
linux
back
in
the
day,
early
2000s.
The
very
first
project
I
worked
on
was
a
roller
on
linux,
on
on
an
arm
system
on
a
chip
like
this
was
my
first
contact
with
embedded
linux,
and
I
was
amazing
back
in
the
day,
we
were
using
strong
arm,
1110
cpus
and
later
switched
to
x
scale,
and
who
remembers
that
particular
period
of
time
both
cpus
were
produced
by
intel.
So
there
was.
A
There
was
a
time
where
intel
actually
built
our
cpus
very
exciting
and
two
things
stuck
in
mind
from
it's
more
than
20
years.
The
first
is
the
sheer
amount
of
software
workarounds
for
missing
or
odd
hardware
features
back
in
the
in
the
v4
rv
five
days
yeah.
You
only
had
four
byte
aligned
load
and
store
access
on
some
architectures.
So
what
the
the
little
kernel
did
back
in
the
day
was.
It
was
trapping
exceptions
that
did
misaligned
excesses.
A
Then
in
assembler
basically
fudged
the
axes
to
actually
make
sense
and
they
changed
it
into
two
loads
or
two
stores
and
then
shifted
it
and
handed
it
back
to
user
space
just
to
make
weird
applications
that
weren't
compiled
for
that
architecture
in
particular
just
work
and
the
second
is-
and
it's
even
more
fun,
even
in
the
2000s
most
of
the
arm
system
on
the
chips
did
not
have
floating
point
units
and
there
was.
There
were
multiple
floating
point
implementations
in
software
in
the
linux
kernel
that
did
the
same
thing.
A
They
just
they
just
caught
the
respective
exception.
When
the
floating
point
group
processor
was
not
found,
and
then
they
calculated
the
result
and
handed
it
back
to
user
space.
But
you
should
not
back
in
the
day,
you
could
not
use
reliably
floating
point
in
in
kernel
space.
So
that's
that's
amazing.
Like
that's
the
first.
The
first
thing
I
will
always
remember,
and
the
second
is
coming
from
x86
and
back
in
the
day
we
used
to
call
x86
advanced
accumulator.
A
A
C
I
guess
I've
got
one
more,
oh
yeah,
so
I
I
do
some
work
with
oregon
state
university,
they're,
historically
known
as
the
open
source
labs
there
and
their
hosts
start
to
be
known
for
for
hosting
a
lot
of
the
alternative
architectures,
for
you
know,
infrastructure
for
the
you
know,
community
projects,
you
know,
including
linux
and
all
that
stuff,
and
they
have
essentially
12
of
our
first
generation
servers
and
when
I
to
give
you
you
know
some
of
the
you
know
the
you
know
talking
about
the
excitement,
I
guess
or
the
you
know,
experience
of
what
happens
when
you
know
you
get
to
put
them
in
your
in
your
data
center,
especially
when
it's
surrounded
by
open
power
system
or
power
systems
and
lots
of
intel
systems.
C
So
anyway,
I
gave
him
these
12,
these
12
servers
and
when
he
first
racked
mounted
them
and
let
them
up.
For
the
first
time
the
the
12
systems
used
less
electricity
than
the
three
top
iraq
switches
and
he
had
to.
He
was
he
wasn't
sure
if
he
was
seeing
something
that
was
wrong,
so
he
had
to
go.
Get
someone
else
from
outside
the
room
to
come
and
verify
that
what
he
was
looking
at
was
real
right.
So
you
know
talking
about
the
sustainability
aspect
and
the
you
know
green
computing
aspect.
C
You
know,
let's,
let's
be
real,
like
especially
here
in
the
united
states.
We
have
a
lot
of
physical
infrastructure,
I'm
talking
buildings
that
were
created.
You
know
50,
60
years
ago
to
house
telecommunication
equipment.
Those
physical
buildings
are
still
owned
by
the
same
telecommunications
providers
and
they're
not
changing
right,
but
the
dynamics
of
the
workloads
and
aspects
of
of
you
know
the
technology
needed
for
them
to
do
their
job
is,
is
you
know,
changing
very
rapidly,
but
the
physics
of
their
physical
environment
where
that
equipment
is
housed,
hasn't
right?
C
As
a
byproduct
of
that
you
know
one
of
the
one
of
the
benefits
of
moving
to
arm
platforms.
Is
you
know
you
get
more
density
per
rack
unit
right?
You
can
reduce
the
amount
of
electricity
consumed.
You
know
in
some
cases
we've
seen
it
like
almost
two.
I
think
it's
almost
two-thirds
right
so
depending
upon
what
it
is.
So
that's
you
know.
That's
that's
really
really
crazy.
C
In
terms
of
you
know,
seeing
it
in
reality,
when
you
know
we
used
to
think
of
it
as
a
I
don't
know,
you
know
like
a
like
an
unattainable
goal
right
so
seeing
it
in
reality,
especially
since
we're
at
the
beginning
of
this
movement,
I
think
there's
an
opportunity
to
you
know
completely
change
the
way
we
think
about.
You
know
both
density
in
our
data
centers
and
how
we
decide
to
you,
know,
manage
and
and
consume
that
right
as
an
end
result.
C
So
so
yeah
that
that's
you
know
the
the
level
of
reality
of
of
what
happens
when
you
put
in
armed
servers
and
I'm
sure
our
friends
at
aws
can
can
easily
attach
a
test
to
some
of
that.
But
that's
to
me
that's
the
part.
That
is
really
you
know.
When
we
talk
about
sustainability,
that's
the
part
that
really
hits
home
and
becomes
really
exciting
from
you
know,
doing
your
park
perspective
so
so
yeah,
that's
my
other
thing.
I'd
like
to
share.
A
A
Okay,
folks,
I
thank
you
everybody
for
joining.
I
thank
the
flatcar
team
for
the
great
work
on
arm
and
I
thank
our
users
and
guests
for
their
input
and
their
their
use
of
the
stuff
that
we
build
it's
exciting
to
have.
All
of
you
see
you
in
the
next
year
and
yeah
have
a
great
new
year's
time.