►
From YouTube: OKD Community Development Meeting 07-25-2023
Description
The OKD Working Group's purpose is to discuss, give guidance to, and enable collaboration on current development efforts for OKD, Kubernetes, and related CNCF projects. The OKD Working Group includes the discussion of shared community goals for OKD 4 and beyond. Additionally, the Working Group produces supporting materials and best practices for end-users and provides guidance and coordination for CNCF projects working within the SIG's scope.
https://okd.io
A
B
C
All
right
you
go
because
I
think
we're
we're
on
videos.
Aren't
we
first.
B
Okay,
yeah
all
right.
Let's
let
me
share
my
screen
here
then
and
go
okay,
so
we
have
got
a
couple
different
agenda
items.
Please
put
your
name
in
the
attendees
section
here,
and
so
the
first
thing
is
the
video
series,
and
so
if
we
go
to
the
okd
foundations,
video
series
Brian
was
kind
enough
to
to
put
all
of
this
into
a
nice
format
here
and
add
some
notes
and
update
it
and
then
Timothy
added
some
stuff.
B
So
let's
sort
of
navigate
through
this
when
I
added
some
stuff,
so
let's
sort
of
navigate
and
see
if
we're
all
on
the
same
page
right
so
again,
the
goal
is
to
provide
a
series
of
videos
that
outline
the
foundations
of
okd
to
use
towards
general
education
and
helping
build
community
support
and
we're
targeting
September
for
completion.
B
So
Timothy
is
proposing
I'm
going
to
jump
down
to
what
he
has
here
and
this
sort
of
what
we
talked
about
at
the
working
group
meeting
last
week
because
he
was
there
and-
and
so
he
sort
of
suggested
this
last
week.
Does
this
look
about
right
to
folks
so
a
15
minute
on
ignition
and
butane
and
those
are
the
questions
that
he
would
answer
or
topics
he
would
cover
within
those
and
then
15
on
F
costs.
B
A
Well,
I
guess
my
question
is:
does
it
mean
that
your
home
lab
needs
to
be
able
to
support
multiple
versions
at
the
same
time,
because
it
says
he's
covering
f
costs
as
costs?
It
almost
says
that.
A
B
So
I
think
one
of
the
things-
and
this
is
this-
is
an
interesting
aspect
of
the
core
OS
system,
as
it's
used
in
okd.
Is
that
basically
there's
a
new?
You
start
out
with
a
backup.
You
start
out
with
a
base
version
of
F
costs
or
S
cause,
we'll
talk
about
F
costs
for
our
purposes
here
and
then
changes
are
made
and
packages
are
installed
to
make
it
not
a
version
of
f
costs
anymore.
It's
it's
like
a
hybrid
f-cos.
B
Basically
with
some
changes
to
it
and
I
mean
the
assumption
is
always
that
the
hardware
that
you
started
on
is
going
to
be
able
to
handle
whatever
this
mutated.
Hybrid
version
is
and
then,
when
you
update
your
cluster,
the
new
versions
of
f
cost
that
that
the
nodes
get
updated
to
that
you'd
be
able
to
handle
that.
So
there
is
an
underlying
assumption
there,
which
may
or
may
not
be
true.
Maybe
we
should
we
is
that
something
we
should
flesh
out,
and
this
is
that
what
you're
suggesting
like
maybe
sort
of
touch
on
that
aspect.
B
Do
you
have
access
to
the
to
the
planning
board
to
if
you
could
have
that
as
a
as
a
comment
yourself
that
way,
it's
in
your
own
words
and
and.
B
C
C
So
we're
looking
at.
How
can
you
debug
and
how
do
you
understand
how
the
nodes
get
the
version
of
the
operating
system
that
they
get
when
you
actually
run
an
install
or
an
update?
So
this
is
about
the
technology
that
lays
down
the
node
operating
system,
but
you
can
do
this
All
in
Virtual
machines.
If
you
want
to
play
around
with
this
technology,
you
don't
need
physical
Hardware.
A
Well,
I
think
also
I,
understand,
there's
a
normal
VM,
there's
containers
and
now
there's
a
containerized
VMS
are
out
there.
No.
C
C
This
is
about
how
the
node,
which
could
be
bare
metal
or
could
be
virtualized,
gets
its
underlying
operating
system,
but
okd
then
manages
going
forward.
So
this
isn't
a
manual
install.
This
is
when
you
run
the
openshift
install
command,
how
do
nodes
get
their
base
operating
system
and
how
is
that
been
managed
going
forward?
And,
more
importantly,
if
you
wanted
to
tweak
that
image,
how
do
all
the
mechanisms
work?
So
you
could
go
and
customize
that
image,
if
you
wanted
to
say,
add
an
additional
package
or
install
a
different
certificate
or
do
anything
like
that.
D
C
D
There
that's
a
good
point
because
that's
the
purpose
of
the
ignition
and
butane,
where
you
can
change
your
ignition.
So
your
your
underline,
select,
let's
say:
you're,
creating
a
VM
you'll
use
the
ifcos
iso
the
base
image.
You
install
that
you
get
that
up
and
running
and
then
at
boot
time,
when
you
boot
that
it's
going
to
ask
for
ignition
file,
I
mean
if
I'm
we're
talking
manually,
obviously
that
the
assistant
install
and
all
the
other
installers
apis
do
that
automatically,
but
a
manual
intervention.
D
Is
you
tell
it
where
the
ignition
file
is
and
there's
a
raw
targyz
file?
That's
has
the
very
specifics
of
all
the
fundamentals
of
core
OS
in
it.
So
it
points
to
that
rotar
gz
overwrites,
all
the
whatever
it
needs
to
do
in
the
kernel
and
then
uses
the
ignition
to
say:
oh
I'm,
a
master
or
I'm
a
worker
or
I'm
the
bootstrap,
and
and
that's
how
it
goes
about
doing
it.
C
D
Oh
go
ahead.
Sorry,
Jenny
I
was
going
to
say
with
our
new
release.
There
is
a
pipeline
build
with
layering
involved,
and
so
maybe
Christian
can
give
her
a
bit
of
a
brief
overview
when
he
does.
The
video
on
s-cos
I
know
I'm
laying
it
on
on
on
Christian,
but
you
know
he's
he's
really
cleared
up
on
the
layering.
B
Yeah
we
could
do,
we
could
have
Christian
or
we
could
have
one
of
the
other
core
OS
folks
I
can
think
of
a
couple
of
the
core
OS
folks
that
were
sort
of
part
of
the
initiative
to
get
layering
going
in
core
OS
in
general,
but
they
might
be
interested
in
in
speaking.
B
All
right,
I'll
reach
out
to
Christian
and
see
if,
if
he
wants
to
do
it
or
has
the
time
to
do
because
he's
already
already
he's
already
gonna
be
doing
video
stuff
for
the
other
aspects
and
if
not,
then,
or
if
we,
if
we
want
something
more
comprehensive,
we
can
reach
out
to
the
F
cost.
Folks,
okay,
so
Larry
do
you
want
to
add
a
comment
actually
to
that?
Just
so
that
we
don't
forget,
Luigi,
go
ahead
and
add
the
layering
to
that.
D
Yeah,
okay
I'll:
where
should
I
edit
in
into
the.
B
E
I
was
gonna,
ask
I
saw
that
there
was
one
of
the
videos
that
you
were
intending
to
create
is
about
the
builds.
E
A
B
I
think
doing
it.
If,
if
you
feel
that
the
this
is
my
thought,
if
you
feel
that
the
way
you're
doing
it
is
going
to
stay
somewhat
the
same
for
like
the
next
year
or
two
and
there's
not
gonna,
be
major
changes
in
that
process.
I
would
say
yes,
but
we
don't
want
to
put
something
out
in
video,
that
and
or
in
a
blog
post
or
anything,
that's
like
this
is
the
way
we're
doing
it
and
then
suddenly
you
know
it
changes.
B
B
That's
sort
of
an
overview
of
the
let's
see
where
is
it?
The
State
of
Affairs
blog
post
by
Jacob
Christian
had
Jacob
do
this,
and
so
it
talks
a
little
bit
about
it.
B
But
it's
very
you
know
it's
just
a
couple
of
paragraphs.
Basically,
you
know.
So
if
are
are
we
shreen?
Are
you?
Are
you
thinking
that
it's
going
to
stay
the
same?
You
know
that
the
process
is
going
to
stay
the
same
for
maybe
the
next
year
or
two
or.
E
E
Building
is
in
general,
I
think,
is
more
or
less
stable.
Now
that
we
transition
to
Moc
now
the
layering
part
is
still
something
that
we
are
working
on.
So
probably,
unless
we
have
something
that
is
perfectly
stable,
it's
not
really
then
necessary
to
demo
it
or
to
add
it.
E
And
like
what
I
was
wondering
about
this
series,
like
is
about,
do
we
want
to
show
what
we
are
doing
on
mlc
clusters,
which
is
the
like
the
official
way
of
building
the
okd
released
in
order
to
to
to
publish
it?
Or
do
we
want
to
focus
more
on
how
to
run
say,
to
build
pipelines
on.
B
E
B
I
think
more
of
the
former,
because
really
what
we're
trying
to
do
is
for
now,
anyway,
give
people
the
tools
to
understand
how
okd
becomes
okd
so
that
they
can
better
participate
in
the
community
in
terms
of
providing
providing
assistance
to
each
other
writing
documentation,
helping
you
know,
helping
contribute
to
the
project.
Etc,
so
I
would
say
the
former
actually
yeah,
okay
I've
been
doing
a
video
on
that
I
I!
Don't
want
to
speak
for
the
group.
What
does
the
group
think
I
think
it's
a
great
idea,
but.
C
E
I
we
did
something
with
Alessandro,
Christian
and
timote
at
the
defconf.
E
B
E
B
Okay,
if
you
could
drop
the
link
in
the
in
that
item,
so
okay,
the
foundations,
video
series,
number
38.,
excellent,
all
right
Brian,
you
were
going
to
say
something.
Sorry
I
didn't
mean
to
cut
you
off.
C
No,
no
no
I
I
was
just
agreeing
with
going
to
agree
with
Shireen
I
think
it
is
it's
another
aspect
that
if
we
can
enable
people
to
go
and
experiment
and
create
their
own
builds,
I
think
that's
also
a
very,
very
valuable
aspect
to
the
community.
Where
we
can
sort
of
say
people
can
experiment,
they
can
create
their
own
bills.
They
can
include
their
own.
C
Their
own
sort
of
modifications,
I
I,
think
that's
also
a
very
valuable
activity
because
it
will
help
people
sort
of
spark
sort
of
innovation
around
RTD,
because
I
think
majority
just
consume
and
I
think
we're
very
keen
as
well
to
to
allow
people
to
innovate
on
the
base
platform.
C
C
Be
yeah
and
having
the
pipeline
that
is
outside
of
pro
is
I.
Think
the
the
thing
that
unlocks
that,
because
up
to
now,
it
took
quite
a
lot
to
actually
be
able
to
figure
out
how
to
create
a
build
and
how
to
create
a
release.
Well,
I
think
now
that
we've
got
the
pipeline
everything's
in
the
public
domain
everything's
available,
and
we
probably
just
need
some
pointers
in
terms
of
how
to
put
it
all
together
to
create
your
own
instance
of
that
pipeline.
D
Yeah
good,
a
good
point:
I
think
that
was
the
whole
objective.
When
we
started
we
were,
we
were
looking
at
the
fcos
Jenkins
pipeline
when
we
looked
at.
Let's
see
how
we're
going
to
do
this
and
you
know
Alex
Alex
and
myself
in
a
previous
company,
we
worked
extensively
on
tecton,
so
we
said,
let's
see
if
we
can
get
teched
onto
work
and
better
still
how
about
getting
kubernetes
in
Docker.
D
You
know
kind
running
on
your
laptop
and
see
if
we
could
build
an
s-cos
there
and
that's
the
whole
That's
The
Power
of
this
and
I
think
the
more
we
show
people
how
you
don't
need
a
75,
000
cluster
server
to
Bob
us.
You
need
a
simple
setup
on
your
laptop,
follow
these
instructions
and
you're
good
to
go
and
I.
Think
that's!
That's
the
power
of
this
and
you
100
right
around
to
take
to
get
people
to
get
the
community
to
understand
that
we
don't
need
prayer
I!
D
B
Excellent,
very
cool,
oh
and
Christian
did
at
the
last
meeting
say
that
he
is
still
going
to
record
the
videos
for
us
because
he
was
giving
presentations
internally
in
red
hat
and
that
he
was
going
to
record
that
and
then
also
record
a
fresh
one
for
us
for
the
stuff
that
he
volunteered
for
for
this.
So
that's
the
okd
on
F
cost
to
build
modifications,
cool,
awesome.
Yeah.
B
Please
add
any
comments
to
this
item
where
I
want
to
do
30
minutes,
30
minutes,
so
that
we
we
have
time
to
talk
about
operators
and
stuff.
So
let's
look
at
so
it
sounds
like
we're
good
for
ignition
butane,
F
cost
and
S
cause
sounds
like
we've
got
that
covered
yeah.
Here's
the
video
I
did
comment
just
earlier
today.
Here's
the
video
that
Brendan
Shepard
did
that
happened.
B
He
happened
to
be
doing
an
MCO
based
video
anyway,
like
and
and
offered
for
us
to
either
use
this
or
he
would
record
a
new
one
if
we
had
suggestions.
B
So
if
folks
could
look
at
this
over
the
next
like
two
weeks-
and
you
know,
come
back
to
the
group
and
say:
do
you
think
this
covers
everything
we
need
for
MCO
or
do
we
want
to
ask
Brendan
to
report
anyone
his
his
videos
are
very
professional
and
he's
got
a
great
manner
of
speakings
and
he's
offering
to
do
this
and
that
would
be
I
think
fantastic
to
cover
the
MCO
stuff.
B
So
it's
a
test
for
all
of
us
to
watch
the
MCO
video
where
at
least
Breeze
through
it
a
little
bit
and
see
if
it
covers
what
we
need,
what
we
want
and
if
not,
then,
when
we
come
back
in
two
weeks,
let's
talk
about
what
we
would
want,
DMC
won't
go
one
to
cover.
Is
that
sound
legit?
Does
that
work.
B
Oh
and
so
there's
the
blog
post.
We
talked
about
that
a
little
bit.
A
couple
of
folks
did
put
merge,
requests
on
it
to
tweak
it
a
little
bit.
If
you
get
a
chance,
look
at
this
as
well,
it's
it's
already
been
posted,
but
that
doesn't
mean
we
can't
make
changes
if
there's
anything
there
that
doesn't
jive
or
that
can
be
stated
more
elegantly
or
more
effectively.
B
You
know
put
in
a
merge
request
on
it,
and
then
we
can
tweak
this
because
you
know
blog
posts.
People
really
rely
on
blog
posts
to
help
them
understand
stuff
right,
even
more
so
I
think
than
video.
When
someone
does
a
web
search
for
you
know,
okd
and
cicd,
or
something
like
that.
This
is
going
to
pop
up.
So.
B
Yeah,
okay,
anything
else
on
the
video
series
does
that
seem
pretty.
Does
that
cover
everything?
And
then
we
just
need
to
to
do
our
work.
Get
that
done
so
Bruce
and
El
Nico
and
myself
will
need
to
do
troubleshooting,
Essentials
video
sometime,
probably
in
the
next
couple
weeks
and
I-
think
that's
all
of
them
and
Andrea
is
doing
the
installation
options
for
the
different
types
of
installations.
B
All
right
moving
on
now
to
operators
so
Brian,
do
you
want
to
do
you
want
to
recap
what
we
talked
about
last
week
to
give
folks
the
set
the
stage
for
folks
that
weren't
there.
C
Yes,
certainly
so
I've
been
sort
of
digging
through
the
operators
trying
to
get
my
head
around
what
a
pipeline
needs
to
do
at
the
meet
at
the
at
the
work
group.
Last
week
we
had
a
conversation
around
whether
we
should
align
operators
in
the
okd
catalog
with
the
ocp
catalog,
and
the
challenge
with
that
is
that
much
of
the
information
needed
to
be
able
to
do
that
is
within
the
the
registration
sort
of
paywall
within
Red
Hat.
C
The
information
for
example.
They
don't
write,
they
don't
tag
within
the
public
githubs,
where,
when
a
release
is
and
what
versions
from
everything
so
to
actually
find
out
the
commit
boundaries
that
went
into
a
specific
operator
that
has
a
tag
of
a
version
number
in
the
in
the
operator
hook,
catalog
that
is
actually
within
the
entitled
content
of
openshift.
C
It's
not
within
the
open
source
domain.
So
there
is
a
problem
with
trying
to
get
that.
One-On-One
alignment
and
the
recommendation
is
that
we
don't.
We
don't
try
and
do
that,
because
there
doesn't
seem
to
be
illegal
way
that
we
could
do
that
without
accessing
these
entitled
content
of
ocp.
C
So
that
was
that
was
really
what
the
discussion
was
around
last
week
within
the
within
many
of
the
GitHub,
the
operator
GitHub
repos.
There
are
branches
per
version,
so
the
412
version,
the
413
version,
the
414
version.
So
we
should
just
be
able
to
pick
the
right
branch
in
the
repo
for
the
version
that
we're
targeting.
C
It
does
suggest
that
within
the
pipeline,
we
have
to
have
the
ability
to
build
multiple
versions,
so
we
build
the
for
example.
Currently
we
built
before
13
version
as
the
current
release,
but
we
probably
also
want
to
start
building
the
414
version
for
the
development
release
So
within
the
pipeline.
We
we
would
have
to
be
able
to
accommodate
that
functionality
as
well.
C
Since
then,
I've
been
doing
some
experiments,
I
haven't
written
anything
up
yet
because
I've
just
been
trying
to
work
it
out
and
the
more
I
look
at
it.
The
the
more
little
alleys
I
go
down.
There
seem
to
be
sort
of
like
issues.
So
if,
if
I
just
explain
where
my
head
is
at
the
minute
in
terms
of
what
this
pipeline
has
to
do
so
when
we
look
at
the
operators
that
are
there,
there
are
some
operators
where
what
they're
installing
is
in
the
public
domain.
C
So
if
you
look
at
something
like
the
pipelines
operator,
what
the
pipeline
operator
does
is
install
pipelines.
So
all
of
the
containers
that
make
up
pipelines
have
to
be
available
in
repo
for
the
operator
to
work.
So
that
means
we
in
in
many
cases
we
have
to
build
those
containers
as
well,
because
some
of
those
like
the
pipeline
containers
are
in
the
entitled
registry,
so
we
would
have
to
go
and
build
potentially
the
pipeline.
C
Other
operators,
like
the
the
dev
workspace
that
the
content
for
that
is
in
key.io,
so
it's
in
a
public
public
repo.
So
for
that
operator
we
wouldn't
have
to
build
the
actual
containers
for
the
functionality.
So
the
first
thing
we've
got
to
do
is
work
out
what
the
operator
is
going
to
install
and
all
those
containers
available
in
the
public
repo
or
we
have
to
build
them.
C
Then
we've
got
to
go
and
work
out
the
operator
that
will
actually
go
and
do
the
install
and
then
again
we've
got
a
problem
of
what
the
version
is,
because
the
version
that's
in
the
the
repo
could
point
to
a
registry,
that's
within
an
entitled.
So
how
do
I
map
the
source
code
that
created
the
version
that
the
operator
points
to
with
a
commit
within
the
functionality
of
the
of
the
container?
So,
for
example,
the
the
pipeline
operate,
the
pipeline
functionality
container?
C
D
Absolutely
this
is
this,
is
the
problem
with
we
have
Brian
is
the
dependency
management
you
could
have
you,
like.
You
said
like
the
githubs
operator
or
the
the
pipelines
operated
fantastic.
All
the
code?
Is
there
everything,
but
then
it
uses
some
weird
elasticsearch
logging
or
something
like
that
which
is
locked
behind
and.
E
C
We
want
to
control
repo
that
will
list
the
containers
that
need
to
be
built
and
somehow
we
need
to
find
the
commit
IDs
or
the
tags
that
we
need
to
build
that
then
maps
to
the
operator
and
again,
what's
the
commit
that
we
have
to
do
then,
and
then
we
have
to
create
the
operator
bundle
which
gives
you
the
upgrade
path
effectively
that
that
pre,
which
which
we
can
build,
which
then
links
in
so
I'm
getting
the
point
where
I
think
I
understand
how
all
these
things
work
the
challenges,
the
challenge
we're
going
to
have
is:
how
do
we
actually
work
out?
C
What
all
these
versions
and
commit
levels
are
going
to
be
I'm
trying
to
get
it.
So
everything
is
automatic.
I
don't
want
someone
to
have
to
go
in
every
week
and
spend
sort
of
three
hours
trying
to
test
different
combinations
of
git
git
commits,
and
things
like
that,
so
I
want
it
to
be
automatic.
So
then,
the
next
thing
I'm
looking
at
is
what
test
functionality
is
within
the
repos,
so
we
could
literally
go
and
walk
back
to
the
connectory
until
we
find
the
one
that
works.
B
D
Yeah
so
Brandon
you're
100.
This
is
this
is
the
the
big
issue
that
we
have.
Even
when
we
like
we've,
we've
created
a
couple
of
operators
in
our
team,
and
so
you,
you
have
to
have
a
really
well
defined,
make
file
to
be
able
to
build
it
vendor
the
because
usually
there's
vendoring
and
you
might
disable
vendoring
and
or
the
whole
pile
of
nonsense
there
once
you've
built
the
binary
then
like
we
have
to
then
build
the
bundles
and
then
there's
also
the
index.
D
But
what
is
great
is
that
most
of
us
most
of
the
repos
on
the
openshift
that
will
build
these
operators
are
have
branches
that
are
versions,
so
you
could
go
to
release
413,
and
you
know
that
you're
going
to
get
413
and-
and
that's
great
I,
think
where
the
problem
might
be
is
if
we
are
working
on
a
live
Branch
like
a
414,
that's
been
updated
daily,
there
could
be
breaks
there
and
that
could
sort
of
set
you
off.
But
if
we
go
back,
one
version
I
think
we've.
C
Thus
I
was
going
to
ask
so,
for
example,
if
someone
reports
a
bug
on
a
413,
is
the
413
Branch
always
guaranteed
to
work,
or
is
that
effectively
the
development
branches
Branch
for
fixes
up
for
413?
So
there
may
be
certain
commits
with
a
partial
fix
on
or
I
might
guarantee
that
every
commit
on
the
413
branch
is
a
tested,
known
working
version
for
the
413
platform.
Yes,.
D
If
it's
G8
so
in
other
words,
so
you'll
hit,
if
the
413
you
will
be
guaranteed
for
13,
GA
will
work,
and
so
we
have
the
z,
z,
screen,
Z
or
Z
stream
releases.
And
so,
if
we
go
4,
13
1,
4,
13,
2,
4,
13
3,
those
those
have
to
work
and
they
have
to
they
have
to
be
fully
tested
before
they
are
released
in
a
z
stream.
So
we
we
have
that
confidence
that
they
are.
C
D
D
so
we'll
take
that
414
fix
this
and
then
back
Port
it
to
413.,
and
that
backcourt
makes
sure
that
it's
almost
like
a
cherry
pick,
cherry
picker
put
it
into
413,
and
then
that
gets
built
on
our
the
automatic
release
pipelines
and
it
will
then
say:
okay,
this
is
working
everything's
good,
it's
been
obviously
all
the
QA,
all
the
testing
everything's
been
verified
and
then
it
gets
back
ported
and
then,
when
that
merges,
it
will
come
out
in.
D
C
C
C
B
D
You're
100
right
because,
for
example,
we
have
something
like
a
node
observability
operator,
so
the
Operator
Operator
is
really
the
bundle.
It'll
have
a
controller
inside
there
and
obviously
the
control,
there's
the
reconcile
lip
and
it
might
have
dependencies.
It'll
say:
okay,
for
example,
I've
got
to
watch
a
demon
set
because
we're
going
to
be
deploying
something
on
each
note
and
it
calls
an
operand
and
the
operand
could
be
an
ebpf
gatherer.
D
It
could
be
a
CPU
memory,
Network,
metrics,
gatherer
and
then
temperature
file,
so
the
operator
might
install
and
instantiate
an
operand
which
is
another
Reaper.
D
In
the
in
the
in
the
the
main,
the
sort
of
the
main
operator
will
say:
oh
I'm
I'm,
going
to
be
using
operand
ABC.
C
C
I
want
to
have
a
generic
pipeline
that
will
build
all
operators,
but
have
the
control
that's.
This
is
where
I'm
thinking
about
it.
I
have
the
control
in
a
version
controlled,
git
repo
that
says
this
operator.
These
are
the
containers
that
need
to
be
built.
This
is
the
operator
that
needs
to
be
built.
This
is
the
bundle
that
needs
to
be
built.
C
This
is
how
the
index
is
built
and-
and
things
like
that,
so
for
each
operator
that
we
want
in
a
catalog
we're
going
to
have
a
control
file
somewhere,
which
effectively
tells
the
pipeline
what
it
has
to
do.
That's
my
current
thinking
of
a
design
and
that's
what
I'm
playing
around
with.
So.
If
anyone
that's
the
wrong
approach,
no
it's
it's.
D
100,
that's
I,
think
you
100
on
the
right
track
and
that's
what
we
try
to
do.
I,
don't
know
if
you
know
if
you
looked
at
the
project
at
all
okd
operator
Pipeline
and
that's
what
we
try
to
do
there.
In
other
words,
we
we
made
it
as
generic
as
possible
to
build.
We,
we
tested
on
a
very,
very
few
I.
Think
three
or
four
operators
made
sure
that
we
could
build
it.
Obviously
it
relies
on
the
make
file.
So
the
the
the
weakness
really
is
the
make
up.
Yeah.
D
C
Looked
at
that,
but
that
assumes
that
the
the
target
containers
are
already
available,
so
it
doesn't
look
down
to
say
what
are
the
containers
of
this
operator
is
going
to
install
not
the
operator
containers
but
the
actual
end
result.
Operators
like
the
database
containers
or
the
pipeline
containers
or
the
other
Argo
CD
containers.
Yes,.
D
The
on
a
an
agent
that
needs
to
be
deployed
and
that
you
can,
as
Shireen
mentioned
in
the
chat
there
is
there's
the
cluster
service
version
and
inside
there
you'll
have
links
and
generally
in
the
in
the
crd,
they
could
have
like
an
agent
name
of
the
agent
file
that
you
could
pick
up
there
and
you
can
say:
oh
I
I
have
a
dependency
on
this
project
so
before
I
build
this
I
need
to
build
that
project.
So
you
could
get
that
information
from
there.
C
I
think
the
openshift
one
calls
the
community
one
then
apply
some
patches
and
then
does
some
other
things
and
then
you
look
at
it
and
there's
about
12
containers
it
points
to
that
are
with
inside
the
red
hat
registry
and
I,
just
like
whoa
yeah,
sorry
Jane.
Why
are
we
doing
this
Can?
You
capture
the
chat
because
there's
some
useful
links
in
the
chat
that
we
probably
want
to
put
in
hack,
MD.
B
Yeah
I'm,
actually
just
about
to
do
that
as
we
speak,
I'm
actually
gonna
well
I'm
doing
two
well
yeah.
Why
don't
let
me
copy
and
paste
these
and
then
I
was
actually
going
to
screen
share
it
for
for
purposes
of
our
of
our
folks
that
are
oh
hold
on
one.
Second,
here
sorry.
B
E
Sure,
okay,
let
me
see
so
this
kind
of
describes
the
operator
right.
So
so
you
get
things
like
a
description.
You
get
base64
for
an
icon
for
the
operator,
you
get
the
r
back
that
the
operator
will
use
here.
What
you're
seeing
is
examples
of
the
custom,
the
custom
resources
that
the
operator
knows
how
to
how
to
deal
with
later.
You
will
see
here.
This
is,
for
example,
the
description.
So
this
is
what
will
pop
up
in
the
console
when
you
look
at
it.
E
What
you're
seeing
here
is
a
bunch
of
labels
that
will
be
later
used
by
olm
in
order
to
say,
can
I
can
I
not
install
this
operator?
How
do
I
upgrade
it
some
stuff,
like
that?
What
you
see
afterwards
in
this
pack
I
think
at
some
point
you
will
be
able
to
see
the
deployment
of
the
operator,
so
you
will
be
able
to
see
okay.
So
what
we
see
here
this
is
the
custom
resource
definition
like
how
what
is
it?
E
What
is
an
elasticsearch
cluster
instance,
but
let's
scroll
down
a
little
bit
further
at
some
points,
still
more,
let
me
see,
I
was
I.
Was
there
a
while
ago
and
I
stopped
the
the
page
you
will
see.
So
this
is
still
the
the
custom
resource.
E
If
you,
if
you
find
you
control,
find
deployments,
we
will
be
able
to
scroll
to
the
part
where
it
it
kind
of
describes.
What
is
this
operator
and
how?
What
do
I
need
to
deploy
to
the
cluster
in
order
for
it
to
work.
E
Yes,
one
more,
no
still
more,
two
more
here,
I
think
that's
the
one
that
we
want
yeah.
So
you
see
the
section
that
is
just
above
line:
five,
six
nine!
This
is
a
deployment
of
of
the
Pod.
So
this
says,
if
you
scroll
a
little
bit
up,
we'll
see
the
name
of
the
container.
So
here
we
have
the
elasticsearch
operator
latest
that
is
being
deployed
and
we'll
see
all
of
the
images
that
are
needed.
E
Okay,
so
that's
how
olm
knows
what
needs
to
be
deployed
in
the
elasticsearch
operator
namespace,
so
that
the
operator
can
do
its
job
right.
B
E
There
it's
it's:
just
a
concatenation
of
a
bunch
of
kubernetes
manifests
you'll
even
find
the
older.
C
At
the
images
you'll
see
that
they're
all
in
key.io,
which
means
they're
public.
Yes,
operators,
that's
in
the
red
hat
registry,
which
is
not
public,
which
means
that
we
would
have
to
then
build
the
container
and
the
challenge
that
we
have
is.
If
you
look
at
the
one,
that's
highlighted,
that's
version
6.8.1,
the
repos
aren't
often
tagged
to
say
what
is
the
commit
in
the
repo
that
built
that
container,
which
in
itself
can
be
a
challenge
to
work
out
where
the
repo
is
the
build
battle
container
and
where
to
commit
the
equates
to
version
6.8.1.
E
C
E
I
agree
with
you
100:
it's
not
it's
not
at
all
easy,
so
the
the
section
that
probably
is
the
most
relevant
to
what
what
you're
trying
to
do
is
the
section
here
that
will
be
called
related
images.
So
here
what
we're
seeing
is
the
deployment,
but
you
have
another
section
there.
That's
called
related
images
and
the
related
images
section
just
says
this
is
the
list
of
all
the
images
that
I
need
in
order
for
this
operator
to
work.
E
D
E
We
say
GVK
just
for
you
guys
to
know
in
some
cases,
let
let's
say
elastic,
church
or
could
be
maybe
an
example
say:
elasticsearch
operator
absolutely
needs
actually
elasticsearch.
No,
but
maybe
if
we
take
Prometheus
and
and
and
grafana
and
stuff
say
that
Prometheus
needs
to
have
a
grafana.
E
It's
it's
not
a
real
case,
but
just
imagine
for
a
second
that,
in
order
for
Prometheus
operator
to
work
properly
on
on
a
cluster,
it
needs
a
graphana
operator
to
be
installed
and
we
need
that
graph
funnel
operator
to
be
respecting
a
certain
version
range
that
information
also
being
included
somewhere
in
one
of
those
I.
Think
it's
also
in
the
cluster
service
version.
Now,
as
as
Luigi
is
explaining.
E
Usually
operators
are
Standalone
and
we
don't
have
that
type
of
situation,
but
it
can
happen.
So
it
can
happen
that
an
operator
explicitly
says
in
its
cluster
service
version.
I
need
that
other
operator
to
be
installed
on
the
cluster
or
otherwise
I
need
this
cluster
to
understand,
custom
resource
so,
and
so
so,
for
example,
say
we're
we're
talking
about
I,
don't
know,
pipelines
and
pipeline
runs
and
tecton
events
and
stuff
like
that.
E
Like
say
we
you
you're
installing
an
operator
and
you
you
for
for
pipelines,
operators
and
you
and
that
operator
might
say,
but
I
need
another
operator
to
provide
me
with
tecton
triggers.
D
E
B
D
D
You
know
the
red
hat
infrastructure
and
you
won't
be
because
like,
for
example,
just
trying
to
think
off
the
top
of
my
head,
the
three
scale
operator
is
a
good
one.
Three
scale
operator
you
won't
be
able
to
get,
and
we
have
access
to
that
in
that
information.
We've
got
the
git,
the
git
hash,
that
the
commit
the
boudoirs,
because
we
can
access
that
red
hat
operator
index
image,
but
it
it
doesn't
help
us
from
a
community
perspective
and
I.
D
Think
that's
where
the
disconnect
is
and
I'm
also
trying
to
think
how
we're
going
to
be
able
to
do
something
like
that.
If
something
somebody
wanted
to
use
the
three
scale
operator
in
in
okd
land,
how
would
they
do
it?
It's
going
to
be
really
really
difficult.
Yeah.
B
Is
it
can
we
start
really
small
and
just
say
when
it
and
basically
have
a
tool
that
scans
looking
for
something?
That's
not
in
Quay
or
key,
depending
on
your
pronunciation?
Can
we
have
just
something
that
an
initial
pipeline
that
runs
that
says?
Okay,
this
is
something
that's
not
in
a
publicly
accessible
like
I
could
try
pulling
the
image
and
if
it
can't
pull
the
image
flag
that
and
then
at
least
this
way,
we
have
a
quick
way
of
checking.
If
this
is
something
that
needs
some
some
hand
holding
to
to
build.
A
A
You
know
for
always
checking
on
something,
but
they
have
something
that
is
kind
of
like
ai-ish
and
it's
and
it's
doing
the
sketch
scheduler's
job
and
it's
kind
of
like
making
it's
kind
of
like
giving
if
a
scheduler,
it's
kind
of
like
doing
the
schedules
job,
but
it's
trying
to
be
more
efficient
than
the
scheduler.
D
Yeah
I'm,
sorry,
sorry,
Duane
I,
don't
know
if
this
is
an
operator
or
not.
I
must
be
honest.
It's
the
first.
C
Unfortunately,
the
techton
is
one
of
them.
Pipelines
is
one
of
them.
D
For
me,
the
two
big
ones
are,
the
August,
CD
gets
Ops
and
then
the
pipelines
and
they're
really
difficult
I.
C
D
C
But
yeah
I
mean
they're,
not
easy,
but
but
so
if
anybody
wants
to
work
with
me
on
any
of
this,
please
shout
up
because
but
what
I'm
doing
is
I'm
just
doing
some
skunk
work
little
projects
at
home
just
trying
to
work
through
some
of
these
issues
and
working
out.
But
the
idea
is
that
for
each
operator,
We'll
add
them
to
the
catalog
one
by
one,
but
we
have
to
work
out
all
of
these
issues
and
come
up
with
a
config
file
that
we
can
then
just
create
that
will
capture
all
of
this.
C
What
I
want
so
that's
the
plan,
but
what
I
wanted
to
do
was
at
least
have
enough
of
the
infrastructure
understood.
So
what
we're
going
to
have
to
do
is
throw
everything
away
and
redesign
it
when
we
get
to
something.
That's
not
simple.
So,
as
I
said,
I
created
a
repo
and
did
the
simple
one,
which
was
the
the
the
the
dev
workspaces
for
car
ready
containers.
C
So
I've
got
that
working
in
it
just
as
a
manual
step,
but
then
what
you're
coming
against
is
all
of
these
complex
ones
and
I
just
want
to
have
at
least
understand
what
the
pipelines
sort
of
got
going
to
have
to
do
so.
I
don't
get
to
the
position
where
we
release
something
and
then
it's
all
going
to
be
thrown
away
and
re-implemented
because
it
can't
handle
so
yeah.
That's
where
I
am
at
the
minute.
B
D
D
In
other
words
that
are
totally
there
are
subscriptions
you,
you
can't
get
them
and
give
you
a
list
and
see
which
ones
we
could
maybe
like
from
a
low
hanging
fruit
type
of
thing
like
amq
streams,
for
example,
oh
yeah,
I,.
A
C
E
D
C
Can
look
at
them,
but
I'm
just
very,
very
aware
that
I
don't
want
to
put
anything
that's
behind
the
registration
into
the
public
domain
by
sort
of
pinching
it
from
there.
So
I'm
I'm
trying
to
be
very,
very
clean,
but
I'm
not
going
to
pull
anything
from
an
ocp
registration
and
embed
it
within
a
cluster
because
a
it's
going
to
get
out
of
date
very
quickly
but
B.
It's!
We
need
this
to
be
an
automated
continual
update,
type
process
that
can't
rely
on
that.
So
yeah.
B
B
Do
folks
want
to
take
a
a
week
off
so
that,
basically
this
could
end
up
being
a
month
from
now
that
we
would
meet
or
are
folks
comfortable,
just
working
through
August.
D
Yeah
I
I
would
I
would
lift
my
hand
up
and
say
please
could
we
could
we
have
a
break
because
I've
got
to
be
I'm
going
to
be
helping
family
move
in
August,
and
so
it's
back
and
forth
from
one
country
to
the
next
country.
So
oh
wow.
E
I
I
had
my
vacation
already
I'm
off
also
during
the
month
of
August
for
a
couple
of
days,
but
otherwise
I'm
I'm
here
so
I'm
I'm
happy
to
to
to
do
as
as
you
guys,
prefer
and
and
to
take
that
time
off.
Whenever
is
best
for
everybody
else,.
C
Lecturine
I've
had
my
my
a
week
after
I
was
planning
in
summer
and
if
we
actually
ever
get
a
heatwave
here
in
the
UK,
the
rest
of
Europe
is
cooking
yeah.
We
are
in
one
of
our
dumper
cooler
Summers
so
far.
If
we
ever
get
some
sunshine
I
might
take
some
days
off
and
go
sit
on
a
beach
for
a
day.
But
if
people
want
to
sort
of
identify
a
fortnight
or
two
weeks
as
a
sort
of
a
break,
I'm
happy
for
that,
otherwise
I'm
going
to
be
here
and
work
through.
B
So
could
we
take,
that
would
mean
skipping
the
eighth
and
then
we
would
come
back
on
the
22nd.
B
D
Yes,
I'll
I'll
be
so
it's
basically
the
last
three
weeks
in
in
August.
That's
going
to
be
a
bit
up.
E
B
I
need
to
take
a
break
and
I
actually
taking
a
vacation
with
my
son
at
some
point
towards
the
end
of
August.
So
how
about
we
skip
the
22nd?
Does
that
work
for
everyone?
So
in
this
way
we
can
meet
on
the
eighth,
keep
the
momentum
that
we're
going,
that
we
have
going
and
then
skip
the
20
seconds
and
give
everyone
a
breather
before
we
head
into
September.