►
From YouTube: OKD Working Group 2020 09 29 Full Meeting Recording
Description
OKD Working Group 2020 09 29 Full Meeting Recording
co-chairs: Diane Mueller and Christian Glombek
https://okd.io
A
B
A
A
B
B
B
B
A
No,
maybe
a
short
sentence
about
that.
As
this,
we
have
organized
the
training
over
two
weeks.
It
was,
I
decided
to
use
okd,
but
not
openshift,
and
the
thing
is
we
like
to
see
showcases
for
a
service
mesh
operator,
openshift
pipeline
operator
and
the
serverless
operator
I
was
able
it
was
very
easy
to
deploy
the
pipelines
operator
yeah.
It
works
like
a
charm
on
okd.
A
We
also
almost
managed
today
to
get
a
serverless
operator
running
on
okd.
It's
also
very
possible
even
to
build
all
images
on
our
own.
It's
okay,
but
the
self's
mesh
yeah.
It
was
a
little
bit
frustrating
because
it
it
tries
to
pull
images
from
docker
hub
that
are
not
even
there
in
this
version.
In
the
current
version
of
the
service
mesh
and
yeah,
it's
hard
to
find
the
repositories
to
build
it
on
your
own.
It's
a
lot
of
reverse
engineering.
I
want.
I
don't
want
to
complain
too
much.
B
B
On
that
I
know
I
will
endeavor
to
get
that
up
over
the
weekend
and
do
all
my
uploads
so
other
than
that,
let's
rock
and
roll
a
little
bit
here
today
there's
a
couple
of
new
people
here,
and
so
I'm
gonna
just
show
where
just
sort
of
how
we
add
it.
If
you
can
add
your
names
into
the
working
group
meeting
notes
so
that
we
can
just
keep
track
of
attendance.
B
B
Yeah,
no,
we
can
definitely
do
that.
So
this
this
just
one
column,
was
the
things
that
I
used
to
drive
through
the
stuff.
So
if
there's
things
that
we
can
take
off
today,
let's
let's
do
that
and
then
these
are
just
notes
and
things
that
we
were
doing
so
we
can.
We
could
hide
those
boards
as
well
so
yeah.
That's
I'm
happy
to
do
that
joseph.
C
I
don't
really
have
anything
concrete
to
show.
I
think
vadim
has
cut
a
new
release
and
yeah
other
than
that.
There's
no,
no,
no
measurable
progress
on
any
of
our
stories,
so
we're
still
having
some
some
problems
with
delivering
the
windows
machine
config
operator,
which
we
kind
of
want
to
use
as
a
template
for
all
the
other
operators
to
deliver
to
yeah
to
an
open
source
or
community
catalog
of
operators.
C
D
C
Yeah
there's
one
issue:
the
operator
hub
still
uses
the
older
format
for
for
catalogs,
while
we
internally
already
use
the
new
one,
and
there
is
an
open
issue,
I'm
kind
of
tracking,
because
I
think
it'll
be
easier
for
the
teams
that
work
on
those
operators
to
just
have
one
one
way
of
distributing
them.
One
kind
of
release,
tooling
they
use,
instead
of
doing
it
the
old
way
and
then
migrating
that
to
the
new
way
once
operator
hub.
Does
it
too
it's
been
in
yeah.
C
C
B
All
right,
so
you
so
just
to
finish
up
that
one.
If
you
can
find
the
link
to
the
hub.io
request
there,
there
aren't
as
many
eyeballs
on
that
as
they
were
previously,
because
I
think
they're,
mostly
focusing
on
the
cncf
side
of
things
sdk
and
olm,
and
shoot
that
to
me,
and
I
will
try
and
get
that
wrangled
in
the
next
week
or
so.
I
didn't
see
that
one.
So
apologies
all
right,
so
welcome
clement.
B
We
have
been
talking
about
you,
so
your
ears
must
be
burning
and
we've
been
talking
about
you
off
and
on
for
a
month
or
so.
So
we
wanted
to
give
you
or
maybe
more
than
that
we
we
haven't
been
able
to
connect
as
a
community
with
the
fedora
container
sig
in
a
real
way.
So
I
kind
of
wanted
to
have
a
little
take
a
little
time
today
to
talk
about
what
the
goals
and
the
mission
of
the
fedora
container
sig
is,
and
you
know
where,
where
you
see
us
helping
out.
E
Yeah,
so
I
see
bruce
that
was
there
when
we
had
pretty
much
just
after
nest,
so
I
think
end
of
august,
or
on
that
time
we
we
met
to
discuss
kind
of
restarting
the
container
sig
because
it
was
started
like
two
years
ago
or
so,
and
we
are
like
a
good
start.
But
still
things
fell
off
a
bit
like
people
got
over
priorities
and
we
did
have
like
a
strong,
a
strong
commitment
of
people
that
could
dedicate
some
time
to
to
help
there
and
pretty
much.
We
came
up
when
we
met.
E
E
E
B
E
Yeah
active,
I
think
we
pretty
much
play
like
three
3
or
4
people,
and
we
not
like
you
know
it's
just.
I
guess
like
most
of
you.
It's
like
kind
of
part-time
kind
of
activity,
so
me
and
pull
really
dedicate
a
lot
of
our
time
to
the
base
image,
the
fedora
based
image,
and
we
have
a
couple
of
folks
so
jakub
zaika
that
worked
a
lot
on
trying
to
have
like
multi-arch
builds
and
like
enabling.
E
B
So
it
it
looks
like
there's
at
least
one
volunteer
from
this
group
to
join,
and
I
don't
know
why
james
is
constantly
I'm
gonna
unmute
everybody
and
james.
You
can
put
your
camera
on
if
you
don't
have
bed
head
like
me,
and
so
it
looks
like
that
that
that
james
castle
who's
on
the
call
right
now
is
interested
in
joining
and
helping
him.
Maybe
if
we
could
get
one
person
from
this
group
and
neil's
raising
his
hands
and
mike.
F
Yeah,
I
just
have
no
idea
what
I'm
supposed
to
do.
I
mean
I
tried
to
join
the
earlier
incarnation
of
the
container
stick
a
couple
years
ago.
It
was
first
being
started
up,
but-
and
I
did
a
lot
of
the
early
reviews
for
most
of
the
initial
images
but
like
I
have
no
idea
how
to
do
anything
and
that's
sort
of
that's
sort
of
been
the
problem.
E
Yeah,
I
think
that
steve
a
lot
of
the
so
a
lot
of
the
fundamental
issues
we
have
are
still
things
that
are.
That
needs
a
lot
of
specific
knowledge,
for
example,
to
improve
the
the
experience
that
will
require
some
development
in
osbs
or
enabling
some
tools
in
the
federal
infrastructures.
Anything
like
that,
and
it's
honestly
not
something
that
is
even
someone
that
is
very
dedicated
and
very
interested.
That
goes
through
the
effort
of
learning
the
tools
and
like
trying
to
understand
things.
E
E
E
Yeah
another
way
to
kind
of
also
maybe
quickly,
join
and
like
help
is
to
look
for
the
image
that
is
already
available
in
fedora.
That
is,
that
is
not
maintained
anymore
and
try
to
maybe
pick
it
up
and
try
to
to
build
it
and
like
see
if
it
builds.
If
you
like,
you
know
just
kind
of
pick
up
the
old
image.
E
E
E
G
E
Yeah,
we
were
everything
to
acknowledge
the
fact
that
we
we
didn't
have
time
to
work
on
anything.
So
we
revived
the
mailing
list
all
in
august,
and
this
is
the
this-
should
be
the
main
base
to
have
communication,
and
if
we
feel
that
we
have
a
subject
that
is
important
enough
to
meet
another
video
call
or
an
irc
thing
that
should
be
bought
on
the
on
the
mailing
list.
E
B
So
I
think
we
have
done
done
our
our
duty
here.
We've
connected
some
dots,
we've
gotten
you
some
volunteers,
and
so,
if
we
can
maybe
ask
jamie
and
who
else
is
jamie's
saying
mike
and
others
to
look
at
that
and
then
just
let
us
know
back
here
in
this
working
group
when
and
if
we
need
to
do
something
or
get
our
acting
gear.
I
think
that'll
that'll
help
flavor
the
whole
collaboration
and
yeah.
So.
E
F
F
Python
stack
all
that
sort
of
stuff,
like
all
the
goodies
that
people
tend
to
expect
for
this
sort
of
thing
and
get
it
to
a
point
where
the
packer
project
can
say:
hey.
You
want
a
fully
supported,
containerized
deployment
for
production
uses.
Here
you
go.
This
is
the
way
because,
right
now
we
don't.
E
G
H
Yeah
hey
this
is
this
is
craig.
I
worked
on
a
couple
of
the
s2i
images
of
the
mariadb
recently
and
the
php
s2i
image,
so
I've
gone
through
and
made
some
commits
to
those
to
get
them
on
the
latest
versions,
and
I
was
just
I
was
really
just
planning
to
do
pretty
much
everything
that's
in
the
openshift
catalog
and
just
go
down
through
through
the
list
as
my
current
plan.
Unless
something
else
comes
out
of
this
call
that's
a
better
a
better
way
of
doing
it.
H
C
Yeah
definitely
so
right
now,
with
the
with
the
containers
we
built
on
our
pro
ci,
which
we
then
release
as
the
okay
payload
they're,
all
uv
i8
based.
So
we
don't
have
any
fedora
based
containers
there.
Yet
we
might
change
that,
but
that's
not
our
focus
right
now.
So
for
the
like
the
non-core
images,
though,
we
definitely
want
to
push
for
using
fedora
based
images.
C
I
think,
having
all
these
already
existing
containers
updated,
would
be
great
and
then,
as
kind
of
a
second
step,
looking
into
how
we
deliver
operators
which
are
kind
of
meter
containers.
They
have
this
bundle
container
now
that
references
other
containers
so
yeah.
If
we
could
leverage
the
the
fedora
diskette
for
versioning,
those
as
well
and
then
really
having
yeah
fedora
based
operators,
which
are
just
made
of
of
fedora
based
containers.
C
H
Okay,
hey
the
issue,
the
issue
I
had
and
I
can't
remember
the
php
container
or
the
maria
db
container.
I
think
it
was
mariadb,
but
I
was
able
to
get
it
updated
and
I
think
the
code
was
was
committed
to
the
repo
and
the
test
field
worked,
but
I
don't
know
how
to
get
it
to
the
fedora
registry
after
that
or
how
does
that
process
work
once
you
get
a
container,
that's
been
updated
to,
you
know,
say
the
latest
version.
How
does
it
get
into
the
fedora
registry?
E
E
E
C
All
right,
I
just
pasted
a
link
to
the
the
pigour
instance,
which
is
our
addis
kit
and
there's
the
namespace
container
in
there,
which
holds
all
the
repositories
for
containers
that
are
currently
in
the
registry
and
yeah.
If
you
want
to
own
a
container
or
help
building
one,
it's
probably
best
to
if
you
want
to
get
involved,
to
reach
out
to
the
current
maintainers,
especially
if
they
haven't
been
actively
pushing
updates
and
ask
for
for
you
to
become
a
co-maintainer
on
that
repository.
So
you
can
push
your
own
updates.
C
There
right
so
now
one
ask
I
I
have,
and
I
will
go
ahead
and
also
contribute
to
this
with
some
containers,
but
if
we
can
get
these
steps
needed
a
documented
really
for
dummies
for
people
that
had
that
have
absolutely
no
prior
knowledge.
That'd
be
great
if
you
could
kind
of.
If
you
do
that
jot
down
the
steps
you're
taking
here
and
then
we
can
kind
of
hopefully
build
up
a
a
documentation
around
this
to
make
it
easier
for
new
people
to
join
us
here.
B
C
E
Yeah,
so
we
have
some
documentation.
I
will.
I
will
review
this
and
see
yeah.
I
think,
last
time
when
we
talked
with
bruce,
I
don't
think
we
have
a
easy
like
just
just
getting
started
with
how
to
maintain
a
container
image
of
internet.
So
that's
definitely
something
we.
We
need
happy
to
to
have
a
look
at
that.
C
G
E
B
Well,
at
least
there'll
be
more
of
you
and
more
of
us
to
work
on
it
and
figure
it
out,
but
I
think
the
the
documentation-
and
I,
if
you
can
put
a
link
to
what
exists
now,
would
be
great
in
chat,
and
maybe
we
can
figure
out
how
to
socialize
this
and
an
easier
version
or
and
maybe
bruce.
If,
if
bruce,
you
can
take
a
look
at
the
documentation
and
add
in
some
thoughts
about
what's
missing,
to
make
it
simple.
B
That
would
be
helpful
as
well
sure
I
know
he's
he's
a
teacher,
so
teachers
usually
can
figure
these
things
out
and
make
it
simple
and
break
it
down
for
us
and
see
what's
missing,
you
know,
there's
we
assume
that
people
know
everything.
It
is
about
rpm
and
packaging
and
fedora
land,
but
and
don't
rewrite
anything
that
there's
tons
of
content
on
that
but
yeah,
maybe
that's
that's
something
that
we
need
some
better
documentation
and
and
how-to
guides
on
some,
because
we're
going
to
rely
on
you
guys
for
some
stuff
here
going
forward.
B
D
D
So
my
background
is
that
I've
been
working
a
lot
with
containers
on
power
for
probably
the
past
four
years,
and
so
this
is
sort
of
the.
I
got
a
lot
of
the
docker
stuff
done
and
did
some
like
the
docker
ci
work
and
some
other
things
before
now:
we've
sort
of
shifted
between
docker
and
podman
and
now
openshift,
and
it's
just
like
everything
in
the
sun.
So
that's,
I
might
guess
my
background
is
just
continue.
D
C
So
for
the
normal
ocp
openshift
release,
those
containers
will
actually
all
be
rebuilt
on
another
system
internally.
That
is
then
x86.
We
have
yeah
powerpc
and
all
those
architectures
there,
but
we
don't
have
those
architectures
in
pro.
So
the
real
blocker
is
really
that
we
don't
have
a
system
to
build
okd
releases
for
other
architectures.
C
H
D
I
am
now
a
part
of
the
multi-arch
team
inside
of
red
hat
and
I'm
trying
to
hand
off
the
ci
stuff
there
to
somebody
else,
but
it's
still
currently
kind
of
on
our
cia
mission.
But
we
do
have
right
now
and
this
isn't
going
to
help
it's
just
that
we
have
the
yaw's
x86
dei,
calls
out
to
this
really
hacky
liver
ci,
so
that
the
testing
is
done
on
power.
D
But
none
of
the
building
is
done
in
prow
and
the
systems
that
you
use
are
for
that
for
ci
or
ibm
internal,
but
the
ones
you
use
for
building
are
redhead
internal.
So
this
is
all
very
much
a
mess,
as
you
said,
as
you're
aware,
and
one
of
the
things
we've
talked
about
with
our
multi-urge
team-
is
that
we
are
now
beta-ing
something
we're
supposed
to
have
a
different
cloud
with
power.
D
But
now
we're
going
to
have
this
sort
of
our
vs
cloud
with
power
and
we're
baiting
that
this
week
with
some
customers
and
the
hope
is
that
we'll
have
like
a
build
farm
one
day
so
that
it
just
makes
it
easier
off
the
board
that
we
can
kick
out
and
like
have
some
pro
stuff.
We
have
another
bra
cluster
running
in
power
bias
right
now,
but
it's
not
gonna
help
you
again.
D
B
C
Yeah
me
neither
because
it's
not
we,
we
have
pro
already,
but
it's
just
it's
on
the
wrong
architecture,
and
we
will
need
another
public
prow
instance.
Essentially,
that
is
non-internal,
which
we
can
then
use
to
build
okd
and
release
from
there.
It
has
to
be
public,
can't
be
internal
and
it
has
to
run
on
on
that
architecture
on
powerpc.
F
Is
it
out
of
the
question
to
do
something
like
kind
of
halfway
with
maybe
quemo
user
and
doing
having
it
pretend
to
be
power
so
that
all
the
stuff
can
run
because
I've
used
cuemo
user
with
podman?
And
that
works
question
mark?
And
I
would
maybe
expect
that
that
might
potentially
a
little
bit
possibly
work.
I
don't.
D
D
G
D
F
H
D
B
So
I
I
think,
I'm
I'm
just
trying
to
figure
out
how
to
par
not
to
park
this
and
go
away
from
it,
but
where
we
can
put
an
issue
into
some
trello
board
or
something
christian
that
we
can
track
it
for
him,
because
I
know
it's
a
it's.
Above
probably
all
of
our
pay
grades.
The
conversation
resourcing
another
prow
incident
somewhere
so
I'll
have
to
think
about
that
and
talk
to
maybe
clayton
or
tracy
or
something
or
how
that.
C
C
It
could
be
an
ocp
cluster,
it
doesn't
matter,
it
just
has
to
be
public
and
then
pro
on
top
of
that
cluster,
and
then
we
kind
of
have
to
tell
it
to
rebuild
all
the
things
we
built
for
a
x86
release
also
built
that
for
yeah
for
power,
pc
and
ideally
we'd-
also
get
that
for
arm
somehow
but
yeah
one
step
at
a
time.
I
guess.
B
B
Yes,
the
the
power
z
folks
want
some
time
at
an
upcoming
on
the
stage
on
the
upcoming
openshift
commons,
gathering
that
I'm
trying
to
curate
a
an
agenda
and
schedule
for
so
I
was
trying
to
figure
out
how
to
fit
a
talk
of
some
ilk
around
power,
z
and
openshift
in
there.
So
maybe
you
could
help
collaborate
on
that
with
me
in
some
small
yeah
for
sure,
like
get
on
stage
and
yeah.
B
I
saw
that,
oh
sure,
thank
you,
okay,
all
right,
so
yeah.
So
that's
great
and
thank
you
for
joining
and
please
come
back.
You
know,
and
you
know
we
we've
got
lots
of
chores.
I
think
really.
The
focus
that
we're
really
working
on
right
now
is
getting
those
operators
working
for
okd.
First,
that's
really
where
we're
focusing
trying
to
get
that
going.
So
we
can
keep
people
like
joseph
happy
and
also
on
the
the
crc,
the
code
ready
container
that
charo.
B
We
talked
about
at
the
very
beginning
of
this
that
that
I
cheered
mightily
for
when
I
saw
it
so
enabling
people
to
start
playing
with
okd
locally
and
get
it
up
and
running
quickly.
So
that's
really
been.
I
don't
know
char
if
you
want
to
talk
a
little
bit
about
that
before
we
go
on
anywhere
else
and
the
work
that
you've
done.
I
think
there
was
one
last
fix
to
it
around
secrets.
F
Oh
yeah:
well
I
realized
that
it's
when
you
run
crc
start
it's
going
to
ask
you
for
a
pull
secret,
but
you
don't
actually
have
to
have
a
pull
secret.
You
use
a
fake
pull
secret,
but
it
it
doesn't
work
with
the
fake
pull
secret
that
we're
used
to
using
it
needs
a
base64
encoded
and-
and
so
I
I
provided
that
in
the
documentation.
B
B
I
was
thinking
that
we
might
have
to
do
more
than
just
diane
tweeting
it
to
get
some
people
from
it.
Diane
tweeted
it
everybody
loved
it.
I
loved
that
everybody
loved
it,
but
perhaps
we
could
do
a
bit
of
a
blog
post
or
something
about
make
about
its
availability
and
and
get
that
out
there
too,
so,
which.
H
B
I'll
talk,
I'll
talk
with
you
and
really
all
it
is.
Is
I
mean
500
words
here.
It
is
announcing
blah
blah
blah
and
we
can
do
that
as
a
google
doc
and
then
hand
it
off
to
a
guy
whose
name
is
alex
handy
he's
used
to
being
handed
things
off
so
and
he
will
get.
That
was
a
bad
joke
and
I'm
not
allowed
to
make
jokes.
F
B
Well,
challenge
accepted,
I
hope,
and
go
go
break
it.
So
there
you
go.
The
other
thing
I
mentioned,
and
then
jamie
I'll
ask
you
here.
I
just
got
a
buff
of
birds
of
a
feather
session
for
kubecon
north
america
for
okd
working
group.
I
haven't
got
all
the
details,
but
I
made
red
hat
pay
for
it.
So
we
will.
We
will
have
a
space
in
the
kubecon
pantheon
of
hundreds
of
other
talks
to
do
to
to
have
a
basically
lunch
and
learn
session.
B
B
B
B
A
C
So
if
you
don't
see
a
pr
on
on
the
repositories
you're
working
on
from
our
site,
then
please
do
open
one,
and
I
would
because
we
don't
want
to
branch
or
fork
much
here.
I
would
prefer
it
live
on
on
the
master
branch
of
the
respective
release
branch-
probably
best,
starting
with
the
master
branch,
and
then,
if
we
can
kind
of
make,
make
those
scripts
item
potent
somehow
maybe
an
argument
or
something
for
it
to
know
what
to
build.
C
C
A
And
yes,
okay
and
a
second
question:
if
some
current
images
are
missing
like,
for
example,
the
service
mesh,
1.8
1.1.8,
I
think
on
dockerhub.
How
do
we
get
them
there,
because
for
sure
we
could
create
our
own
repository
somewhere
in
quay,
but
this
yeah
rather
heck
here.
It
would
be
so
great
if
someone
could,
because
there
are
images
already
for
older
versions
who
one
is
in
charge
of
who
one
does
create
them.
I
don't
understand
the
process
yeah.
C
So
yeah
each
team
kind
of
is
in
charge
of
keeping
those
up
to
date,
but
a
lot
of
those
community
images
haven't
been
updated
in
a
long
time.
So
what
we
really
want
to
do
is
introduce
that
process
that
that
is
going
to
happen
kind
of
as
part
of
the
usual
work,
but
we're
not
there
yet,
and
until
we
can
kind
of
provide
the
team
with
the
template,
here's
how
you're
going
to
do
it
very
easily.
C
I
would
suggest
you
actually
create
your
own,
your
own
registry
or
repository
on
a
registry
and
push
them
there,
because
you'll
be
able
to
to
build
them
yourselves
push
them
there
and
yeah.
You
won't
have
yeah.
We
just
don't
have
a
way
to
tell
all
the
teams.
You
know
you
have
to
keep
them
up
to
date
and
push
them
to
the
to
the
registry.
C
C
So
yeah,
that's
what
we
use
internally
now
is
the
new
bundle
format
and
I
pasted
the
link
earlier
here.
The
operator
hub
doesn't
accept
that
form.
It
yet
doesn't
work
that
way.
I
haven't
really
looked
into
how
the
old
yeah
how
it
was
done
before,
but
you
can
probably
just
do
it
the
same
way
it
wouldn't
land
on
operator
hub,
but
you
you
could
still
pull
it
in
manually
or
create
a
new
catalog
that
references
it.
Okay,
thank.
C
You
this
is
as
a
temporary
solution
in
in
the
you
know,
in
the
long
run.
We
definitely
want
all
of
this
to
be
on
operator
hub,
easily
available
right
now,
yeah,
we
don't
really
have
the
it
has
to
be
done
automatically
to
like
the
new
version
has
to
be
built
and
pushed
there
automatically,
and
we
don't
have
that
in
place.
Yet.
C
F
No,
no,
it's
all
good.
What
what
I
was
talking
about
was
in
answer
to
bruce's
question
about
where
the
code
was
the.
If
you
want
to
build
crc
yourself,
the
the
the
biggest
piece.
The
hardest
piece
is
the
single
node
cluster.
That
code
has
been
merged,
and
so
it's
just
part
of
code
ready,
snc
now
and
I
added
some
documentation
to
the
readme
for
anybody
that
wants
to
build
their
own
snc
for
okd.
B
F
B
B
B
C
B
E
B
C
B
F
C
B
F
B
B
So
cool
well
christy,
it's
great
to
see
you
and
to
have
the
power
z,
folks,
represented
and
also
to
just
have
more
folks
here
and
I
already
pinged
pep
for
you
christian
and
I
will
book
you
charo
for
a
couple
of
hours
on
monday,
so
make
your
dinner
and
come
to
it
because
I
think
you're
on
east
coast
time.
So
it
might
be.