►
From YouTube: OKD Working Group 2020 09 15 Full Meeting Recording
Description
OKD Working Group 2020 09 15 Full Meeting Recording
co-chairs: Diane Mueller, Christian Glombek, Vadim Rutkovsky
https://okd.io
A
I
think
we
have.
Let
me
just
double
check.
You
know
come
back
here,
who's
who's
on
the
call
right
now
we
have
christian
who
I
will
unmute
and
unmute
all
christian
and
charo.
So
I'm
wondering
I
know
vadim
couldn't
come,
maybe
christian.
If
you
could
give
us
a
little
update
on
any
issues
or
engineering
bits
that
have
gone
over
the
past
two
weeks
have
come
up.
C
A
C
A
C
D
E
C
C
E
C
C
Packer
group
I'll
just
drop
a
link
into
that
into
the
chat,
and
then
it's
a
question
of
figuring
out
how
we
want
that
to
distribute
like
there's,
we
yeah
so
like,
I
think
putting
it
somewhere
is
the
easy
part
figuring
out
how
people
can
how
it
will
be
replicated
so
that
it
is
accessible
for
everyone
is
the
harder
part.
Technically.
C
C
Itself,
oh
well,
the
binary
is
already
fully
self-contained,
so
it
could
just
be
mirrored
as
a
blob
yeah.
No,
no,
no,
oh
yeah,
so
we
can
probably
put
it
on
the
mirror
network,
along
with
the
rest
of
the
evcos
stuff.
It's
a
question
of
just
working
out
the
logistics
of
how
we
upload
and
how
we
sync
and
things
like
that.
Most
of
that
is
tied
into
bodhi
and
like
obviously,
this
has
worked
that
way.
C
This
will
probably
wind
up
being
something
similar
to
what
we
do
with
fedora
media
writer,
which
is,
I
don't
remember,
exactly
how
that's
distributed,
but
there's
some
precedent
for
it:
file
a
ticket
there
and
let
and
we
can
work
on
figuring
out
how
to
get
that
rolled
out.
I
don't
expect
there
to
be
a
problem
other
than
like
figuring
out
the
logistics
of
how
this
is
going
to
work.
C
A
A
Here
you
should
see
it
again
all
right,
and
I
know
that
charo
and
I
have
made
no
progress
on
cookbook
and
recipes.
So
we
we
still.
We
still
have
that
the
all
of
the
all
of
the
videos
that
we
did
on
the
17th
are
now
up
on
the
youtube
channel
and
that
that's
the
content
that
we're
going
to
use
to
do
this
for
or
at
least
kick
it
off
and
yeah.
A
A
Well,
it
was,
I
think
it
was
the
person
that
we
who
was
going
to
do
the
alibaba
one
is
now
just
getting
to
it,
and
I
don't
know
that
person
is
on
this
call,
so
they
are
speak
up
and
let
let
me
know,
did
we
get
any
further
with
the
fedora
containers?
Folks,
I
know
you
were
gonna.
Talk
to
his
name
is
escaping
me.
Christians
did
we
figure
out
anything
to
do
with
them
jointly.
C
A
F
C
A
C
A
C
B
C
B
So
one
problem
we
have
here
is
that
our
pro
ci
only
runs
on
x86
and
we
need
to
pull
our
releases,
our
payloads
from
that
ci.
So
right
now
we
don't
have
any
yeah.
We
we
can't
build
any
other
architectures
right
now
and
we'll
have
to
figure
out
a
way
how
to
either
use
a
second
and
third
build
system
for
each
architecture
or
how
to
extend
pro
to
actually
have
some
worker
nodes
that
run.
B
It
might
be
an
open
shift
one,
but
I
haven't
heard
of
anybody
running
a
mixed
kubernetes
cluster
either.
Okay,
well
as
openshift
is,
you
know,
has
a
strong
support,
commercially
from
red
hat.
I
don't
think
we'll
allow
that
for
the
time
being
how
that
would
work
with
upstream
kubernetes.
I'm
not
sure
I
mean
I.
I
don't
see
a
reason
why
it
wouldn't
work,
but
we'll
definitely
have
to
do
something
about
that
in
the
future.
I
think
once
there's
more
interest
from
the
ibm
side.
C
We
can
probably
finagle
something
from
aws
with
their
graviton
stuff.
I
mean
I
saw
tom
callaway
posting
numbers
about
how
how
fast
bass
it
took
to
build
a
full
chromium,
build
on
the
graviton
systems
in
like
what
is
it
a
little
over
an
hour
which
is
20
times
faster
than
our
build
system
in
fedora
can
do
right.
Now,
though,
I
think
I'm
being
conservative
with
saying
20
times
pretty
sure
it
takes
like
more
than
a
day.
So
so
like
there's
a
potential
thingy
there.
D
Yeah,
I
was
just
gonna
say
for
the
people
for
people
who
are
interested
in
kind
of
these
hybrid
cluster,
topologies
and
and
multi-architecture
clusters.
There
have
been
some
discussions
coming
out
of
the
cluster
api
community
about
like
how
they
can
improve
tooling
to
allow
for
more
high
years.
You
know
that,
obviously
this
started
as
a
discussion
about
how
to
have
a
mixed
cloud
cluster.
B
Yeah
that'd
be
awesome.
I
don't
think
we
have
investigated
this
enough,
but
yeah.
I
think
we
should.
We
should
do
that
now
on
the
on
the
operator
side,
maybe
just
to
give
another
quick
status,
update
on
that,
there
has
been
a
new
bundle
format
that
was
introduced,
but
the
operator
hub
hasn't
actually
migrated
to
that
once
the
operator
hub
does
that
I
will
be
able
to
kind
of
release
the
same
operators
that
we
built
for
openshift,
also
for
the
community
for
okd.
B
It's
a
little
bit
we're
kind
of
in
yeah.
We
kind
of
have
to
wait
a
little
bit
for
that
here,
just
maybe
as
a
as
a
little
teaser
we'll
be
releasing
the
windows
machine,
config
operator
to
operator
hub
very
soon.
It's
already
the
pr
is
already
open
and
we
definitely
want
to
aim
at
supporting
okd
there
as
well.
I'm
currently
helping
out
that
team
for
a
little
bit
so
working
on
that,
mostly
right
now
and
yeah.
There
should
be
something
soon
for
that.
E
B
C
B
B
B
D
Just
to
follow
up
on
what
christian's
saying
there
are
a
couple
pr's
out
right
now
adjust
the
templates.
I
think
this
is
on
azure
that
you
can
use
to
mint
windows
machines
from
so
I
know,
there's
an
ability
to
change
the
template
that
a
windows
node
gets
minted
from
so
like
that
it
doesn't
allow
you
to
actually
change
the
configuration
at
the
mco
level,
but
there
there
is
a
way
you
could
do
it
through
the
templates.
I
think
that's
on
azure.
A
C
We
don't
have
anything
actually
for
arm.
I
I'm
pretty
sure
metal.
Three
will
get
us
to
the
point
where,
like
once,
we
have
binaries
it'll,
be
fairly
straightforward
for
us
to
go
and
say
here's
the
thing
go,
do
the
thing
now
you
have,
you
now
have
a
node
or
a
master
or
whatever
I
feel
like
that'll
be
fine,
but
we
we
need
to
have
something
to
put
on
there
first,
and
I
think,
as
mike
was
mentioning
earlier.
C
I
think
we're
gonna
have
to
get
some
more
focus
in
the
it
sooner
rather
than
later
on
this
idea
of
hybrid
clusters
across
architectures,
because
I
don't
know
about
you,
but
I
am
not
all
that
enthusiastic
about
running
an
open,
okd
master
node
on
on
a
raspberry
pi.
I
I
feel,
like
that's,
going
to
end
very
badly
for
all
parties
involved,
so
so
I
think
it
it's.
I
think.
That's
definitely
something
that
we're
gonna
have
to
as
part
of
doing
like
serious
arm
bring
up
or
any
alternate
architecture
bring
up.
B
So
and
one
more
thing
to
add
there,
I
think
with
the
especially
with
arm
hardware,
if
it's
not
a
full-on
server
thing,
we
probably
won't
be
able
to
use
the
machine
api
for
that
for
for
bringing
up
new
nodes,
because
that
needs
an
automatic
management
part
which,
for
example,
raspberry
pi,
doesn't
have
yeah.
But.
B
E
C
Even
make
the
raspberry
pi
pretend
to
be
a
uefi
capable
device
like
it's
gotten
to
the
point
now
that,
basically,
as
long
as
you're
working
with
a
raspberry,
pi
or
real
computer,
you're-
probably
fine
when
it
comes
to
when
it
comes
to
arm
stuff.
But
if
you're
working
with
anything
else,
you're
just
toast,
so
which
is
actually
quite
unfortunate,
because
this
is
mostly
the
fault
of
the
arm
ecosystem.
And
I
don't
expect
this
to
get
any
better.
But
hey
whatever.
B
Yeah,
having
them
move
towards
uefi
would
be
would
be
nice
having
at
least
that
on
all
the
devices
anyways.
I
think
we
should
focus
on
getting
a
upi
user
provision
provision
infrastructure
here,
working
first
with
arm
devices,
for
example,
and
then
we
can
get
the
machine
api
operator
for
that
later,
but
yeah
right
now.
The
most
important
thing
is
that
we
don't
have
any
binaries
built
any
containers
built
for
other
architectures
than
x86
64.
D
I
would
I
would
just
say
to
you-
and
this
is
kind
of-
I
guess
a
little
bit-
you
know
from
where
I'm
sitting
the
metal
three
stuff.
You
know
it's
going
to
be
landing
in
4.6,
but
I
think
we'll
probably
see
big
growth
for
it
in
4.7,
they're,
doing
a
lot
of
work
on
kind
of
the
internal
infrastructure
for
how
those
resources
are
managed.
So
I
you
know
back
to
kind
of
diane's
point.
C
Fine,
we
just
have
a
plan
of
attack
if
we're
going
to.
Even
if
we're
going
to
look
at
doing
arm,
we
just
need
to
have
a
plan
of
how
we're
going
to
get
there,
and
I
think,
starting
from
binaries
and
images
and
then
going
to
upi
is
a
good
starting
point,
because
by
this
point
we
have
enough
upi-based
documentation
thanks
to
charo,
that
it
should
be
relatively
straightforward
for
someone
to
manually
online,
a
raspberry
pi
based
note
if
they
wanted
to.
A
A
So
if
you
are,
if
you
want
to
come
with
your
questions,
that's
I
just
thought
that
it
might
be
a
a
way
to
open
up
that
conversation
and
just
get
it
on
their
radar
and
I'll
post
the
link
to
that
in
into
the
into
here
as
well.
So
I
I
just-
I
know
that
it
always
feels
the
arm.
Stuff
always
feels
to
me
like
this
outlier,
because
we
got
so
much
other
stuff
to
do.
But
it's
on
the
radar,
the
I.
B
I
just
then,
I
just
posted
in
the
chat
if
you're
gonna
meet
with
the
metal
three
guys.
You
could
please
remind
them
to
have
a
look
at
that
issue,
which
is
the
list
vadim
made
with
all
the
images
we're
missing
and
they
should
yeah
in
an
optimal
case.
They
would,
they
would
add
them
at
those
images
themselves.
B
A
E
I
think,
because
ci
things
are
still
linked
in
that
where
I
don't
have
access
to,
and
I
think
it
would
be
great
to
get
more
acceptance
from
the
community
if
they
can
build
everything
on
their
own.
I
know
that
this
is
a
big
task,
but
it's
the
only
chance
to
get
more
people
involved
without
yeah,
leaving
the
stuff
uncompilable
behind
and
getting
frustrated.
E
B
A
E
E
F
E
E
B
What
I
have
done
in
the
past
is
just
replacing
some
of
those
images
in
there
which
you
can
use
within
oc
command,
but
I
haven't
actually
composed
my
very
own
payload
from
the
from
the
start,
so
I'll
have
to
look.
We
do
that
in
ci,
somewhere
and
I'll
I'll.
I'd
have
to
check
where
that
is
done
and
how
exactly
that
is
done,
but
essentially
the
payload
container
only
contains
a
manifest
of
all
the
other
containers
yeah.
It
just
contains
references
to
all
the
containers
that
are
used
by
that
payload.
C
B
Yeah,
that's
definitely
a
thing
that
is
possible
just
with
with
the
oc
command
and
then
our
open
is
it
openshift,
install
yeah.
You
can
definitely
replace
images
kind
of
define
your
yeah
an
override
for
a
specific
images.
That's
kind
of
the
development
flow
you
use
when
you
develop
for
one
specific
operator.
You
just
want
to
test
a
devil,
build
of
a
specific
operator
image.
You
can
kind
of
take
the
standard
payload
and
then
just
replace
that
one
image,
so
the
image
version.
D
I
just
wanted
to
add
another
comment
here
about
building
the
individual
components
from
the
docker
files
that
are
in
those
repos.
We
just
had
a.
We
just
had
an
internal
effort
where
we
were
trying
to
get
these
docker
files
kind
of
all
in
the
using
the
ubi8
images,
all
building
off,
relate
or
whatnot,
and
my
impression
from
that
is
that
not
all
of
the
projects
have
the
same.
I
don't
know
consistency
with
how
they're
creating
their
docker
files,
our
team.
D
In
specific,
like
we
tried
to
make
very
careful
to
have
dockerfile.rel
as
the
builds
that
use
you
know,
relay
subscription
base,
and
then
we
have
a
docker
file,
which
is
where
we're
trying
to
keep
all
the
community-based
builds
in
place.
So
you
could
use
just
publicly
downloadable
images,
but
the
problem
I
noticed
is
that
not
every
project
has
done
that
and
there
seems
to
be
an
unevenness,
so
I
think
joseph
what
you
might
have
run
into
I'm
not
sure.
If
that's
an
open
shift
project
you
were
building
from,
but
it
could
be.
D
Some
of
our
docker
files
are
not
up
to
community
standards.
At
this
point,
in
my
opinion,
I
think
we
probably
need
to
do
another
round
of
making
sure
that
our
build
ci
for
our
product
is
building
off
of
these
rel
images
and
that
we
we
leave
the
community
with
a
good
set
of
artifacts
to
build
from
so
that
this
is
just
a
problem-
it's
not
in
every
repo,
but
it
there's
an
unevenness
to
the
consistency
of
these
things.
I
just
it's
a
warning
I'll,
throw
that
out
there.
E
B
Okay,
yeah
the
same
way
we
built
those
images
on
the
ci
dci
can
actually
replace
that
from
directive
within
the
docker
file.
So
if
you
try
to
build
it
locally,
you
might
get
a
different
base
there
and
what
even
one
that
might
not
be
publicly
publicly
accessible
or
even
out
yeah
out
there
anywhere.
So
you
may
have
to
adapt
that
if
you
run
it
locally.
D
E
G
A
A
D
A
That,
okay
cool,
let
me
I'll
start
a
thread
with
you
and
sally
and
see
which
of
you
has
the
time
to
do
it,
and
that
would
be
great
so
and
then
we
can
edit
down.
We
can
fast
forward
through
the
cluster
up
part
of
it,
because
that
was
in
on
the
day
of
that
was
the
thing
that
took
the
longest.
I
have
a
little
bit
of
bad
news.
The
t-shirts
are
not
coming
and
I'm
trying
to
get.
A
I
I
tried
I
this
is
me,
I'm
just
not
a
t-shirts
and
stickers
kind
of
gal
and
now
we're
trying
to
get
them
into
the
red
hat
cool
stuff
store
so
that
I
can
ship
from
there.
I
apologize
for
that.
It
turned
out
in
order
to
ship
them
to
everybody.
It
was
going
to
cost
75
each
t-shirt,
so
I
decided
that
we
might
want
something
better
than
a
t-shirt
for
75
bucks
like
a
meal
with
a
drink
together.
C
A
A
A
A
A
Yeah,
I
was
very
embarrassed
because
here
was
they
managed
to
get
this
whole
little
thing
off,
but
when
you
start
thinking
about
the
cost
of
fedexing
that
little
bag
of
stickers
to
everybody
who
attended
nest,
my
cheap
new
englander
mine
just
said
no.
So
even
when
it's
not
my
money,
it
said
no
so
anyways
we
did
it.
C
A
Yeah,
it's
it's
swag
in
the
virtual
era
has
is,
is,
is
not
an
easy
thing,
so
it's
a
good
thing.
I
got
all
the
swag
that
I
everyone
needed
over
the
past
few
years,
so
that's
that
so
next
meeting,
hopefully
we'll
have
clement,
come
and
talk
to
us
about
what's
going
on
in
fedora
containers,
world
and
other
than
that,
I
think
I'm
hearing
a
little
bit
of
closure
for
today
ending
20
minutes
early.
Unless
someone
else
has
another
top.
A
And
going
going
on
so
charo,
hopefully
we
can
connect
with
the
fedora
folks
and
get
that
image
up
there
and
update
the
landing
page
and
then
once
we
get
that,
then
we'll
move
on
to
the
next
task
of
cookbooks
and
recipes
and
figuring
out
what
to
do
with
all
how
to
get
the
videos
more
prominent.
On
the
pages
of
okd.io
and
in
the
world
and
working.
C
A
So
all
right
is
charo
taking
care
about
the
dev
community
together.
Yes,
charo
is
like
on
the
hook
for
everything.
Now
too
so
joseph
is
asking
oh
yeah.
Char
charo
was
supposed
to
be
my
external
person
on
like
co-cherry
thing,
and
then
we
hired
him
so
yeah.
So
now
we're
going
to
have
to
like
find
another
externally
external
body,
so
who's
willing
to
do
that.
It
might
be
an
on-boarding
process
for
a
red
heart
red
hat
higher.
So
just
be
careful.
A
Shows
up
for
meetings
and
answers,
questions,
and
you
know
so:
that's
there.
There
is
no,
it
yeah,
there's
not
really
anything
super
official
about
the
co-chair
role,
but
I
I
like
having
myself
as
a
community
person
a
couple
of
engineers
from
the
red
hat
team
and
at
least
one
external
community
representative,
as
a
co-chair,
so
too
bad
old
dominion,
transport
was
good,
but
that
lasted
about
15,
nanoseconds
and
then
and
how
he
got
himself.
A
C
We're
working
on
it,
our
infrastructure,
is
being
revamped,
is
the
nice
way.
I
could
do
it.
The
the
the
more
accurate
way
is
it's
being
set
on
fire
and
and
and
then
rebuilt
from
the
ashes
to
support
things
like
okd
at
scale.
So
our
our
production
deployment
strategy,
hopefully
should
be
kicking
in
fairly
soon
now
things
have
been
sluggish
because
of
well
world
events
have
not
exactly
made
things
easy,
you're,
muted,
diane,.
A
A
So,
let's,
let's,
let's
see
how
that
all
rolls
out
and
and
tell
me
again
how
what
the
name
of
your
company
is
joseph
schwartz
eurodian,
schwartz.
Okay,
I
knew
I
was
gonna
say
it
wrong
if
I
said
it
out
loud,
so
that
that
might
be
another
possibility
too.
So,
let's
talk
about
it
on
the
email
thread
and
yeah
bcit.
We
love
you
bcit
british
columbia,
institute
of
technology.
A
A
F
A
Englander
so
yeah,
but-
and
I
don't
think
they
live
in
vancouver-
I
know
the
long
story,
so
that
would
be
good
luck,
joseph
and,
and
maybe
what
we
can
do.
Joseph
is
showcase
you
at
the
gathering
in
november
for
that
deployment
and
what
it
took
if
you,
if
you're
willing
and
your
company
is
willing
to.
Let
us
do
that.
A
That
would
be
great,
because
I
really
want
to
figure
out
how
to
showcase
some
of
the
okd
work
and
bruce
yeah.
Ask
your
boss,
joseph
that
would
be
great,
and
maybe
the
other
co-chair
like
I
I
you
can
never
have
too
many
co-chairs
is
in
my
book
is,
is
the
fedora
liaison,
so
maybe
that
we
can
put
neil
in
in
that
role
in
something
I
love.
I
love.