►
Description
Join us for Kubernetes Forums Seoul, Sydney, Bengaluru and Delhi - learn more at kubecon.io
Don't miss KubeCon + CloudNativeCon 2020 events in Amsterdam March 30 - April 2, Shanghai July 28-30 and Boston November 17-20! Learn more at kubecon.io. The conference features presentations from developers and end users of Kubernetes, Prometheus, Envoy, and all of the other CNCF-hosted projects
Deep Dive: CNCF CI - Lucina Stricko & Denver Williams, Vulk Coop & CNCF
The CNCF CI status dashboard -- cncf.ci -- provides a third party validation of builds, deployments and end-to-end testing for CNCF’s Graduated and Incubating projects. CNCF welcomed Arm as Gold member in February 2019. The cncf.ci dashboard helps ensure that Kubernetes and other CNCF-hosted projects run well on Arm-based platforms. Attendees will leave this deep dive with an understanding of how the cncf.ci dashboard builds and tests projects on Arm and x86 architectures.
https://sched.co/MPj8
A
Good
morning,
thanks
for
joining
us
today
for
the
deep
dive
into
CNC
FCI,
my
name
is
Lucena
strike.
Oh
I'm,
here
with
Denver
Williams,
we're
with
voc
cooperative,
doing
this
project
for
the
CNC
F
in
this
35
minute
session,
we'll
go
through
a
quick
intro
of
the
CNC
FCI
project
will
introduce
the
CI
team,
the
goals,
key
features
and
I'll
walk
you
through
the
dashboard
and
then
I'll
hand
it
over
to
Denver
for
a
deep
dive
into
how
we
added
armed
support
to
the
sea
ice
testing
system.
A
A
So,
a
little
bit
about
vote
cooperative,
we
are
a
worker
owned
software
consultancy.
We've
got
four
worker
owners
and
five
to
ten
associates
and
we're
located
all
around
the
world
we're
a
fully
remote
team.
We
are
based
in
Austin
that
some
of
us
are
based
in
Austin
Texas,
where
we
host
to
meetups
monthly
meetups,
one
on
Austin
software
cooperatives
and
the
other
called
open
source
axis
where
we
meet
up
with
other
open
source
folks
in
Austin
and
throw
axes
at
a
target
pretty
fun.
A
A
A
A
The
first
feature:
the
first
goal
is
that
the
CN
CF
CI
project
is
project
centric.
We
want
to
put
the
graduated
and
incubating
projects
of
the
CN
CF
front
and
center,
so
we're
validating
the
stable
and
head
releases
of
those
projects.
We
can
reuse
the
build
containers
from
the
project
CI
system
we
use
upstream
helm,
charts
for
the
deploys
and
we
can
reuse
end-to-end
tests
provided
by
the
project
maintainer
x'.
A
Secondly,
collaboration.
We
want
to
increase
engagement
with
the
sand.
Cf
project
maintainer
x'
we'd,
like
the
projects
to
be
able
to
maintain
the
project
and
release
details
through
a
pull
request.
We'd
also
like
to
use
maintainer
provided
helm,
charts
and
smoke
tests
for
the
app
deploy
phase
for
the
end
the
end
test.
We
welcome
contributions
from
the
project
maintainer
x'
on
those,
and
we
wish
to
integrate
the
CNC
FCI
testing
system
with
the
project's
testing
systems
on
Jenkins
Circle
CI,
whatever
they
happen
to
be
using.
A
Thirdly,
the
scene
CF
CI
project
is
agnostic
and
provides
agnostic
testing
we'd
like
to
validate
projects,
build
and
deploys
in
a
configurable
test
environment.
So
we
support
per
kubernetes
release.
We
currently
do
stable
in
head
per
architecture,
which
is
currently
x86
and
arm
64,
and
we're
on
bare
metal
using
packet
net.
A
A
A
B
B
B
So
there's
skewing
you
static
or
qemu
binary,
and
you
can
configure
that
so
when
you're
on
an
AMD
64
system-
and
it
sees
that
you're
executing
an
armed
binary,
it
can
process
it
hand
it
off
to
QEMU
and
process
it
in
the
user
space
which
allows
you
to
build
an
arm.
64
binary
on
AMD
64
machine
envoy
envoy
was
probably
one
of
the
most
challenging
projects
we
added
support
for
because
they
themselves
had
only
just
got
I'm
support
working,
probably
in
the
month
before
us.
So
the
first
problem
we
had
with
that
was
basil.
B
It's
written
in
Java
and
no
one
from
the
basil
release
team
is
currently
publishing
on
64
out
of
odd
facts.
So
we
had
to
build
that
from
source
before
we
could
do
anything
else,
and
it's
written
in
Java
so
doing
the
emulation
to
build
it
on
my
AMD
64
box,
not
gonna
happen,
there's
a
massive
amount
of
bugs
with
that
and
I
mean
I'm
sure
you
can
get
it
working.
B
If
you
wanted
to
spend
a
couple
of
days,
debugging
QA,
Mia
logs,
so
we
instead
had
to
build
on
a
native
arm,
64
machine
and
then
now
we
could
build
Envoy.
But
there's
no
documentation
is
about
two
issues
that
they've
tracked
it
in
and
that's
about
it.
We've
got
to
go
off,
but
what
was
some
of
the
benefits?
What
were
some
of
the
projects
that
we
four
items
that
we
found
surprisingly
easy
and
we
were
adding
home
support?
B
Well,
good
news.
Is
this
Cuban?
It
just
doesn't
require
any
configuration
changes
when
you're
deploying
with
arm
so
our
provisioning
tools
that
we
were
using
all
they
need
to
be
told
to
do
was
use
our
own
binaries
that
we'd
built
and
everything
from
their
out-of-the-box
projects
written
go
are
very
easy
to
deal
with
because
go
has
a
really
nice
cross
compiling
support.
So
it's
generally
the
process
of
adding
a
flag
and
you've
got
arm.
60
and
Cordia.
Nelson
Prometheus's
makes
groups,
and
the
repositories
already
had
already
had
support
for
that.
B
So
you
can
make
and
you'd
have
wineries
to
deploy
so
I'm
gonna
do
a
bit
of
a
card
review
now
and
go
for
each
project
and
show
some
of
the
pots
that
were
easy
to
implement
and
some
of
the
pots
that
were
really
difficult,
so
cube
entities.
We
now
have
a
menu
here
that
shows
both
stable
arm
and
hit
arm
and
we're
actually
currently
working
on
an
update
to
this.
That's
what's
this
out,
so
we
have
an
independent
column
for
the
release
and
in
a
separate
column
for
the
architecture
and
then
the.
B
B
So
what
we
we
head
to
our
configuration
files,
we're
having
to
support
both
builds
and
the
same,
get
load
configuration
file
which
is
a
bit
of
a
restriction
on
get
websites.
So
we
had
to
do
conditionals
if
the
architecture
is
and
do
some
different
steps,
and
can't
everyone
see
that
alright
then
alright
cool
all
right
so-
and
this
is
where
I
was
talking
about-
we
had
to
move
away
from
the
Cuban.
B
It
is
really
scripts
to
actually
do
this
on
line
43
here
you
can
see
we're
using
their
their
or
their
libraries
to
do
the
build
and
then
appending
the
platform.
We
want
to
use
one
exam
64,
and
then
we
write
they
come
figure
out
generate
at
one
release,
type
script
and
I'm
line
52
we
run
the
build
and
then
on
the
for
the
container
job.
This
is
where
we
take
those
binaries
and
we
package
it
into
a
dog
or
container
for
Cuban.
B
B
So
we
pretty
much
disabled
I
could
take
she's
except
arm,
so
we
can
do
an
arm
build,
and
then
we
generate
the
script
necessary
to
do
that,
to
publish
the
images
and
then
on
one
22.
We
run
that
and
that
will
go
through
and
take
the
binaries
built
in
the
previous
step
and
then
publish
them
to
a
docker
registry.
B
So
we
have
a
similar
conditional
statement
here
and
that's
pretty
much
the
magic
there
system
we
want
to
do
it
for
Linux
and
the
architecture
we
want
to
do
it
for
home
64
and
then,
when
we
publish
there
the
artifacts
or
packaged
them
into
a
docker
container,
we
reuse
their
existing
scripts
to
do
the
Manx.
We
actually
even
share
the
same
commands
down
here
for
the
build
for
both
arm
and
MD
64.
B
B
So
here
we
have
a
similar
conditional
statement.
Is
it
armed
and
if
it's
armed,
we
want
to
enable
enable
a
map
here
which
tells
when
you
run
this
on
an
AMD
64
system.
It
enables
configuration
on
that
system.
So
when
it
receives,
when
it's
someone
tries
to
execute
an
armed,
64,
binary
or
I'm
64
executable
code,
it
realizes
what
it
is
and
then
passes
it
off
to
queueing
you.
So
it
can
do
an
emulation
and
process
it,
and
then
we
just
use
their
normal
make
scripts.
B
B
B
And
then
the
most
difficult
project
was
byte
was
definitely
envoy
and
on
the
production
version
of
dashboard
it
still
says
na
the
work
is
actually
done
and
it's
something
that
we'll
be
working
to
promote
there
and
the
next
week.
So
if
you
take
a
look
at
the
dashboard
in
the
next
week
or
so,
you
should
be
able
to
see
on
the
way
running
for
arm,
and
this
is
the
configuration
that
I'm
least
impressed
about
it's
full
of
said
and
not
too
wonderful.
B
So
for
this
one
thing,
you'll
notice,
that's
a
little
bit
different
is
we've
got
completely
independent
jobs
for
arm,
because
we
have
to
do
this
on
line
three
on
line.
33
we've
had
to
set
a
tag
and
tell
it
to
build
on
a
physical,
our
machine,
then
the
build
scripts
have
absolutely
no
support
for
arms,
so
we
take
their
existing
scripts
and
we
told
to
use
a
base
image
that
has
an
ARM
version
of
basil
installed
and
then
remove
a
whole
bunch
of
packages
and
commands
that
don't
exist
in
our
repositories.
B
B
B
What's
this
line
here,
so
we
need
to
use
G
lib
CL,
one
for
the
base
image
on
home
and
our
plane
doesn't
publish
any
G
Lipsy
binaries
packages
for
I'm
64.
So
you
have
to
go:
build
G,
love,
C
from
source
and
package
it
and
install
it,
which
is
a
real
nightmare,
but
we've
done
that
and
the
works
open
sourced
upstream.
So
if
anyone
wants
an
armed
version
of
G,
Lib
C
on
Alpine
and
you're,
welcome
to
take
a
look.
B
So
how
to
get
involved
if
you're
interested
in
collaborating
we're
still
needing
to
add
support
for
a
lot
of
incubating
projects
and
what,
if
you're
an
incubating
project
here
from
an
in
keeping
project,
you
like
to
contribute
we'd
love
to
work
with
you,
so
we're
currently
in
the
progress
process
of
refactoring
ciencia
CI,
to
make
it
easier
for
external
parties
to
contribute,
especially
projects.
So
those
configuration
files
we
were
looking
for,
so
that
we
want
to
make
those
more
public
and
ideally
fully
maintainable
by
projects
themselves.
B
So
they
can
directly
manage
that
themselves
and
projects
have
their
own
CI
systems.
So
but
the
perfect
world
for
us,
we
were
aiming
to
be
in
the
next
half
a
year
is
leveraging
existing
CI
system.
So,
instead
of
managing
a
configuration
file
that
and
be
able
to
integrate
your
existing
circle,
CI
or
your
jenkins
builds
with
the
dashboard
and
show
status
there,
and
then
we
can
augment
their
data
and
then
do
a
deploy
using
the
artifacts
you
produced
from
your
CI
system.
B
We
have
a
working
group
on
the
fourth
Tuesday
of
every
month,
so
if
you're
wanting
to
get
involved
there
to
be
a
great
place
to
show
up-
and
the
next
meeting
is
on
June
the
25th
and
we
also
have
a
public
mailing
list.
So
if
you
have
any
questions,
feel
free
to
ask
there
or
create
an
initial
and
github
and
also
on
slack.