►
From YouTube: OKD Working Group 2020 11 24 Full Meeting Recording
Description
OKD Working Group 2020 11 24 Full Meeting Recording
https://okd.io
https://groups.google.com/forum/#!forum/okd-wg
A
Hello,
everybody-
this
is
christian
here,
something
kind
of
came
up
for
me,
so
I
have
to
actually
run
right
now,
but
just
to
give
you
a
very
quick
update,
I've
summarized
the
current
state
of
the
release.
4.6
work
in
this
comment.
I
just
pasted
the
link
to
it
in
the
chat,
so
that
should
be
all
open
issues
where
we
have
right
now.
Unfortunately,
it's
still
taking
its
time
to
get
the
back
parts
in
so
we
can't
release
just
yet
and
there's
one
open
issue.
A
We
have
to
still
figure
out
with
with
regards
to
system
g
resolve
d
and
that
being
active
in
in
fedora
33,
which
we're
now
using
so
there's
still
some
ongoing
work
on
that
the
rest
are
really
just
back
parts
that
have
to
go
in.
So
that's
it
for
me,
I'm
gonna
have
to
drop
right
now.
Yeah.
Thank
you
all
and
have
a
great
meeting
bye,
yeah.
B
The
only
only
relevant
update
I've
got
is
that
there's
a
new
version
of
code,
ready
containers
that
I
pushed
up
to
the
fedora
site,
where
we're
hosting
the
okd
code
ready
containers
and
it
doesn't
require
a
pull
secret
anymore
at
all.
So
now,
when
you,
when
you
run
the
crc
start,
it
will
just
start
it's
not
going
to
prompt
you
to
paste
in
a
pull
secret,
real
faker,
otherwise,
and
obviously,
as
soon
as
we
cut
a
4.6
release,
we'll
cut
a
4.6
release
of
our
code,
ready
containers
too.
B
B
I
saw
in
the
in
the
chat-
hey
joseph,
I'm,
not
sure
how
quickly
those
issues
will
be
closed.
I
know
one
of
the
things
that
that
I
think
we're
waiting
on
is
is
vadim
is
is
out
on
pto
and
I
think
he's
the
one
that
would
close
some
of
those
and
then
cut
the
release.
So
we
might
literally
be
waiting
for
vadim
to
get
back
from
vacation.
C
B
B
D
B
E
E
But
you
know,
since
I'm
using
oh,
you
know,
you
know
open
source
stuff,
I
kind
of
want
to
get
involved
a
little
bit
more
than
I
had
been
in
the
past,
since
I
have
a
little
bit
more
time
now
so
kind
of
wanted
to
get
an
idea
of
you
know
where
to
get
started
with
you
guys
and
and
what
to
do
I
mean
you
know,
there's
always
well.
You
can
do
documentation
and
stuff,
but
I'm
really
more
into
the
testing
and
implementation
and
trying
to
to
figure
out.
E
What's
going
on
type
of
thing,
you
know
I've
got
a
couple
of
open
tracking
issues
going
on
and
stuff
and
starting
to
try
to
help
out
with
some
installation
issues.
You
know,
as
I
can
but,
like
I
said
I
just
kind
of
want
to
get
involved
a
little
bit
more
and
and
figure
out.
You
know
where
to
go
forward
from
from
here.
E
We're
running
vsphere
ipi,
so
my
my
test
cluster
is
three
masters
and
three
worker
nodes
had
lots
of
fun
over
the
last
month
or
so
trying
to
basically
figure
out.
You
know
the
little
idiosyncrasies
of
doing
the
installations
and
stuff.
I
think
that
the
minimum
requirements
may
be
a
hair
low
for
opens
for
okd
4.5
ran
into
issues
were
like
working
nodes,
wouldn't
start,
or
only
one
would
start
the
other
two
wouldn't
start
and
would
run
out
of
resources.
E
You
know
start
the
second
worker
and
then
it
would
fix
itself.
So
you
know
it
ran
into
a
bunch
of
video
synchronies
there.
You
know
also
things
where
they're
trying
to
get
ignite
files
for
the
masters
and
and
the
workers
and
getting
timeouts
from
you
know
this
or
that,
from
our
point
of
view,
that
seems
to
be
related
more
of
the
vmware
utilization
once
we
moved
things
around
response.
E
Time
dropped
from
30
to
60
seconds,
to
5
to
10
seconds
everything
installed,
so
those
are
kind
of
things
that
we're
kind
of
working
through
and,
like
I
said,
I've
kind
of
opened
up,
you
know
one
or
two.
Well,
I
guess
one
right
now:
installation
ticket
and
you
know
with
shoot.
What's
his
name
victor
yeah
hold
on,
I
was
just
looking
at
it
and.
E
Operators
are
are
one
of
those
things
that
are
still
sort
of
baffling
the
back
of
my
mind,
but
you
know
I
got
things
work
like
I
got
rook
sef
to
work.
You
know
some
you
know
got
you
know
stuff
working
on
my
back
in
storage,
so
that's
been
good,
but
you
know
that
one
of
the
things
I've
been
working
on
today
was
ingress.
E
E
Or
even
something
like
you
know,
the
the
the
newest,
you
know
h
a
you
know:
h
a
ingress
controllers
and
stuff
aj
proxy
ingress
controllers.
So
those
are,
you
know:
we're
not
really
going
into
production
with
4.5
or
4.6,
yet
we're
still
doing
a
lot
of
learnings
and
I'm
not
I'm
not
comfortable
going
to
production
with
it
at
this
point
but
yeah.
That's
that's
why
I'm
here
yeah
the.
B
B
B
Some
of
your
experiences
with
with
the
other
folks
and-
and
this
is
this-
is
a
normal
bi-weekly
meeting
last
week
or
this
meeting
and
the
last
one
have
been
a
little
bit
truncated,
partly
because
of
holiday
schedules,
partly
because,
but
but
I
will
say
this,
this
group
is
is
pretty
regular
that
this
this
meeting
here
on
on
bi-weekly
tuesdays,
can
have
anywhere
from
from
a
dozen
to
25-30
people
on
it.
So
there
is
a
core
group
of
people
bruce
and
joseph
and
neil
and
there's.
G
E
E
E
Yeah
I
like
it.
I
have
a
question,
though
I
saw
somebody
mentioned
something
about
the
slack
channel
for
openshift.
I
tried
to
join
the
slack
channel
and
I
got
like
into
a
circular
thing
where
it
said
you
had
to
have
an
email
to
get
ready
to
get
a
to
to
get
an
invite,
but
it
was
like
it's
horrible
slack
is
hard
yeah.
E
G
G
G
G
There
she
pre-populated
the
okd
working
group
up
front
in
the
very
beginning.
That's
why
okay
yeah
so
like
with
I.
This
is
one
of
those
things
where
it's
like.
We
are
supposed
to
be
an
open
group
where
people
can
come
freely
to
us
and
stuff,
but
we're
on
a
platform.
That's
actually
literally
designed
to
be
closed
off
to
the
world
like
this
makes
no
sense
at
all.
B
G
G
So
charo
what
I
was
saying
before
you
let
your
dog
go
bark
at
the
ups
guy.
The
indications
are
good,
that
fedora
will
be
getting
a
matrix
home
server
and
we
will
be
re-homing
all
of
the
irc
channels
for
fedora
onto
there.
That
might
also
be
a
good
opportunity
for
us
to
add
a
matrix
channel,
a
matrix
room
for
open
shift
and
okay
stuff
and
start
bringing
people
in,
as
that's
the
as
the
open
default
platform
for
people
to
collaborate
with
us.
H
E
E
Yeah
so
I
get
joined
kubernetes
on
slack
and
then
it
says
continue
with
google
tried
that,
but
it
also
says
or
that's
your
email,
but
it's
your
email
add
get
an
invite
at
slack.kubernetes.blah
blah
blah.
So
none
of
that
works.
So
when
I
try
to
go
in
with
google,
it
says
that
I'm
not
a
member.
When
I
try
to
get
an
invite,
I
don't
have
an
email
with
to
get
an
invite.
That's
slack
that
kubernetes,
whatever.
B
B
B
F
F
E
G
E
G
H
So
maybe
it
might
be
helpful.
Well,
it's
fresh
in
our
heads
to
do
a
debrief
of
last
friday.
Does
anything
stand
out
for
folks
that
were
there
friday
does
anything
stand
out
in
your
mind
from
the
buff
event,
I
was
there.
There
was
a
handful
of
other
folks.
Anything
stand
out,
anything
that
you
that
you're
thinking.
G
G
G
I
went
to
a
couple,
but
like
a
lot
of
the
events
I
tried
to
go
into,
despite
the
fact
I
had
like
a
full
pass,
it
kept
saying
I
didn't:
have
the
right
pass
to
actually
join
them,
and
so
I
was
like
after
a
certain
point.
I
was
like
all
right.
Clearly
this
isn't
going
to
work
and
I
don't
need
to
spend
my
day
like
hitting
a
broken
platform.
H
So
I
think
one
of
the
things
that
stuck
out
in
my
mind
is
that
it
seems
like
in
a
lot
of
discussions
and
also
in
a
lot
of
the
videos.
The
question
comes
up
repeatedly:
the
relationship
between
our
costs,
f
costs
and
openshift
and
ocp
and
okd.
It
might
be
helpful
in
the
documentation
to
actually
write
out
something
for
the
okd
website
or
for
somewhere
near
the
recipes,
or
something
like
that,
so
that
we
can
always
just
point
people
to
a
document
that
says:
hey.
Here's,
the
relationship
yeah.
B
B
B
B
F
B
G
B
G
G
B
B
C
C
B
B
F
C
Last
I
looked
at
the
okd
docks,
they
were
full
of.
You
know,
rh
cost
references,
so
I
think
I
think
I
think,
if
you're
talking
about
confusion,
that's
a
good
place
to
start
too
is
there's.
You
know
like
I
haven't
gone
through
and
you
know
put
in
a
a
ticket
on
it,
but
just
because
there's
so
many
of
them-
and
you
know,
life
is
short
anyway
right.
B
F
F
B
So
for
for
every
one
of
those
we
we're
at
the
mercy
of
that
specific
operator
team
to
to
start
conforming
to
that
new
format.
So
the
good
news
there
is
is
that,
as
they
start
doing,
that
we'll
start
seeing
those
operators
showing
up
in
an
operator
hub.
So
we
won't
have
to
wait
for
a
big
bang
where
they
all
get
done
and
then
suddenly,
oh,
we
got
all
the
operators
the
bad
news
there
is
that
any
stragglers
will
have
to
keep
pressure
on
on
those
individual
teams
to
get
those
operators
into
place.
E
B
C
B
So
if
you've
installed,
the
rook
ceph
operator
that
you
got
from
the
github
page
you've
effectively
installed
the
exact
same
operator.
The
difference
is
you
didn't
you
didn't
get
to
go
to
operator
hub
and
click
on
the
subscription
and
click
install,
so
it
it
really
is
more
of
a
free,
as
in
beer
thing
right
it
it's
freely
available
it.
It's
just
not
as.
C
F
And
general
question
we
always
have
is
if
a
the
code
base
is
really
exactly
the
same
for
the
community
versions
and
also
see
red
hat
versions,
because
there
are
some
hot
fixes,
I
think
red
openshift
users
will
get
first
before
the
community
can
get
them
or
is.
Als
are
also
the
hot
fixes
being
pushed
to
the
public
ripples.
C
B
B
You
know
if
you
follow.
If
you
follow
the
issues
like
say
you
come
across
an
issue
in
your
cube
operator,
right
and,
and
you
need
a
fix
for
it
if
you're
monitoring
the
issues
on
github
as
soon
as
those
pull
requests
hit
and
the
ci
system
builds
off
of
those
pull
requests
again,
it's
it's.
The
whole
free
is
in
beer
thing.
You
can
grab
those
and-
and
I
used
to
do
that
back
in
in
the
day
with
jboss.
B
When
I
was
enterprise
architect
for
advanced
auto
parts,
we
were
actually
running
community
jboss
instead
of
the
subscription
based
product,
because
the
subscription
based
product
was
so
far
behind
that
it
didn't
it
didn't
support
the
newer
jwe
features
that
we
were
trying
to
implement,
but
so
we
for
years
we
ran
community.
Before
we
we
started
paying
red
hat
subscriptions
for
eap
and
that's
exactly
what
I
did.
I
E
Is
that
a
is
that
a
valid
place
to
start
poking
and
prodding
to
add
some
help
to
or
something
I
one,
because
I
need
to
learn
more
about
operators
anyway,
but,
and
you
know
one
learning
how
to
build
them
and
how
to
install
them.
You
know
would
probably
be
a
good
thing
before
I
start
getting
deep
down
into
the
code
and
creating
my
own.
F
You
should
start
with
the
operator
sdk.
We
did
the
same
and
it's
very
it's
not
hard
to
to
create
your
own
operator
with
that
sdk.
It's
it's
much
more
easier
than
two
years
ago,
where
people
told
us
that
this
is
the
absolutely
greatest
king
size
thing
you
can
do
on
kubernetes.
It's
it's
rather
easy.
Now,
yeah!
If
you
do
it
with
operator
sdk
well,.
E
H
F
F
E
F
E
F
Be
out
there
sdk,
I
I
almost
everything
I
saw
that
is
related
to
openshift
was
written
with
the
operator
sdk.
You
see
that
if
you
go
into,
if
you
see
a
package
directory
package
controller-
and
there
is
a
go
file
with
this
function-
called
reconcile,
then
it's
created
with
the
operator
sdk
because
and
there
you
can
find
what
it
does
yeah.
That's
that's
the
beauty
of
the
operator
sdk.
We
have
also
reverse
engineered
some
openshift
operators,
where
the
documentation
was
a
little
bit
behind
what
it's
already
doing
and.
E
K
E
What
actual
git
commit
you
know
they
that
built
it
was
so
I
can
go
look
at
the
right
source
code.
I
did
that
today
with
the
aha
yeah
with
the
h
a
proxy
thingy,
because
I
was
like
there's
no
documentation
to
it.
You
know
as
to
what
you
can
do
with
it.
You
know
what
annotations
there
was
nothing
that
I
could
find.
I
had
to
go
into
that's
very.
B
K
K
D
B
K
K
B
I
C
G
F
B
F
F
F
G
G
G
K
G
To
be
clear,
the
reason
why
this
is
a
problem
is
because
rel8
and
rel
core
os
used
the
nft
implementation
of
iptables,
and
so
all
the
software
has
been
written
around
it
and
the
legacy
iptables
that
doesn't
use
nft
is
doesn't
support
enough
features
for
everything
to
work,
so
some
clusters
will
come
up,
but
others
won't.
So
I've
seen
some
issues
with
ips.
Some
people
have
complained
about
ipi
deployments,
failing
because
of
this,
whereas
upi
deployments
all
work
because
we
don't
do
any
network
configuration
in
a
upi
setup.
K
K
B
K
D
That
we
have
that's
perfect.
Thank
you
diana,
so
hi
everyone.
My
name
is
mike
yang
and
I'm
a
software
engineer
with
wet
hat,
focusing
on
the
product
called
advanced
cluster
management
for
kubernetes
for
acm.
Today
I
want
to
present
to
the
okd
community
what
we're
doing
in
the
upstream
project
called
the
open
cluster
management.
D
So
there's
a
lot
of
talk
about,
I'm,
probably
condensing
and
now
talking
to
quick,
five
to
ten
minutes
overview
and
quick
demonstration.
So
let's
get
started.
So
what
is
open
cluster
management
in
wet
hat?
We
have
the
enterprise
offering
of
advanced
cluster
management
for
kubernetes
and
ocm
open
cluster
management
is
the
open
source
backbone
of
that
project.
D
The
project
is
really
about
kubernetes
and
how
to
manage
multiple
kubernetes
clusters
in
a
consistent
way.
So
multi-cloud
obviously,
is
becoming
more
and
more
of
a
reality
today,
and
we
want
open
cluster
management
to
be
the
go-to
solution
on
solving
some
of
the
challenges
that
multi-cloud
management
raises.
D
So
how
does
this
all
ties
in
with
okd
well
through
the
operator
hub
users,
can
install
our
ocm
solution,
cluster
management
operators
and
multi
multi-cluster
subscription
operator
onto
the
control
plane?
Cluster,
which
we
call
it
the
hub
cluster
and
then
the
ocm
agent
cluster
led,
can
be
installed
on
t
onto
the
clusters
that
they
want
to
manage,
which
we
call,
which
we
label
the
manage
cluster.
G
D
Right
we're
working
on
launching
our
website.
We
have
the
we
have
the
community
here
that
talks
about
all
the
goals
that
we're
trying
to
accomplish,
and
then
we
also
have
we're
working
on
launching
our
that's
gonna
be
launched
real
soon
and
then,
which
will
give
like
overview
of.
Basically,
we
phrases
some
of
the
community
mission
statements
that
we
have
here,
so
it.
K
D
So
I
have
the
okd
cluster
here
on
the
left
representing
the
hub
cluster,
and
then
I
have
the
manage
cluster
on
the
right
representing
the
managed
cluster.
So
I,
in
the
interest
of
time
I
will
register
the
managed
cluster
to
the
hub
cluster.
So
now
I
can
do
things
on
the
hub
and
then
it
will
propagate
down
to
the
manage
cluster.
So
the
first
thing
I'm
going
to
do
is
a
quick.
D
D
So
when
I
apply
this
manage
this
work,
then,
on
the
on
the
manage
cluster
side,
I'll
get
the
and
in
this
workload
I'm
just
deploying
a
config,
a
quick
config,
a
small
config
map,
so
I'll
see
the
config
map
shows
up,
and
so
this
is.
This
is
a
really
quick
example,
and
now
I'm
going
to
show
off
a
little
bit
more
complicated
scenarios
where
we
want
to
deploy
a
helm
chart
from
the
hub
and
then
and
then
to
have
it
propagate
down
to
the
manage
cluster.
D
D
D
So
we
can
see
that
it's
placed
onto
the
status
say.
Oh,
I
found
the
placement.
Who
is
this
placement?
Will
sorry
there's
so
much
there's
so
much
details,
I'm
skipping
some
details,
so
placement
will.
This
is
placing
where
the
application
should
be
deployed
into
which
clusters.
So
in
this
case
I
found
the
cluster
one
which
now
says
is
propagated.
D
And
how
do
you
we
deal
with
the
policy
and
configuration
across
multiple
clusters?
How
do
you
deal
with
the
cluster
health
and
all
these
four
areas
are
the
open?
Cluster
management
is
fine
to
address.
So
please
take
a
look
and
join
the
conversation
in
the
community
repo
and
we
love
to
get
your
direct
involvement
or
bring
your
organization
and
get
more
directly
involved.
We'd
love
to
have
that
type
of
participation
and
that's
it
for
the
quick
overview.
K
Yeah,
that
was
very
quick
and
thank
you.
Can
you
pop
over?
Can
you
share
the
screen
just
for
for
the
record
and
go
over
to
operatorhub.io
and
show
everybody
which
one
of
the
operators?
It
is?
Yes,
there's
a
lot
of
naming
conventions
in
this.
The
product
side
of
this
is
acm.
The
open
source
side
is
open,
cluster
management
and-
and
I
think,
we've
named
or
you've
named
the.
F
D
D
L
B
It's
kind
of
like
core
os
acm
was
a
proprietary
product
under
the
under
the
ibm
umbrella,
and
it
is,
it
has
been
open
sourced,
but
they're
still
going
through
the
process
of
getting
all
the
code
out
there,
and
so
that's
that.
That's.
Why
that's
why
you
see
some
pieces
of
it
that
that
find
the
code
for
yet.
K
All
right
so
there-
and
I
know
when
I
talked
to
juliana
and
mike
earlier
they're,
also
looking
for
people
who
want
to
participate
in
their
community
as
well.
So
that's
going
to
be
an
interesting
group
of
folks
to
work
with
too.
I
think
what
I'd
love
to
see
is
is
to
do
once.
It
gets
a
little
bit
further
along.
Maybe
we
could
record
a
video
of
it
running
on
okd
and
do
do
some
demos
and
do
a
briefing
on
it
again
and
whatever
you
know,
whatever
you
need
really
is
we're
here.
K
B
G
G
K
K
G
K
G
K
D
L
K
Yeah,
so
I
think
that
the
other
thing
that
I
would
say
is
if
you
do
put
that
video
together
and
do
a
write-up
about
you
know
deploying
the
operator
on
openshift.
We
have
a
recipe
section
of
our
okd.io
site
and
that's
where
I
would
embed
the
video
into
there
and
link
to
whatever
write-up
you
guys
did.
So
if
you
want
to
record
something
and
make
a
readme
file,
you
know
specific
to
okd
deployments.
K
K
K
K
G
K
K
Because
because
I
knew
I
knew
charo
or
or
vadim
or
someone
would
show
up
and
have
my
back
so
anyways
so
next
time
in
two
weeks
time
will
be
the
date.
So
we
are
still
working
into
the
holiday
season.
So
that
would
be
the
eighth
of
of
december
and
I
actually
may
ask
charo
to
actually
host
this
or
vedim
or
that,
because
I
I'm
in
an
all-day
meeting
on
the
8th,
9th
and
10th
and
on
the
10th,
which
is
actually
10.
K
Oh
there's
two
and
there's
two
at
the
same
time,
so
I'm
in
that's
even
worse,
two
of
them
said
there's
a
staff
meeting
for
three
days
and
then
there's
a
pragmatic
marketing
product
management,
training
thing
for
at
the
same
time
and
on
the
10th
of
december,
but
in
japanese
time,
which
is
actually
the
night
of
the
ninth
of
december
or
the
night,
something
like
that.
Some
weird
time
we're
we're
hosting
an
open
shift
commons
gathering
in
all
japanese.
K
K
K
I'm
going
to
be
cranky
on
that
day
and
he's
going
to
do
something
on
january
4th
and
they
asked
to
come
in
because
I
think
it's
something
that
I
want
to
see
if
it's
anything
of
interest
for
the
okt
group,
and
that
would
be
another
thing
that
would
be
a
nice
recipe
to
have
as
well.
So
there
you
go
so
all
right.
Everybody
we've
run
over
seven
minutes,
which
seems
like
the
shortest
meeting
I've
ever
had
with
you
guys.