►
Description
No description was provided for this meeting.
If this is YOUR meeting, an easy way to fix this is to add a description to your video, wherever mtngs.io found it (probably YouTube).
A
Okay,
so
today
is
january,
13
2022..
This
is
distribution,
team,
demo
and
the
subject
today
is
the
listing
operator
on
operator
hub
io.
A
So
there
are
quite
a
few
things
that
are
going
on
with
this,
and
I
wanted
to
kind
of
split
it
into
two
pieces.
One
is
talk
about
the
lay
of
the
land,
so
to
speak,
just
where
we
at
how
things
are
and
everything
else,
and
then
what
are
the
issues
that
are
related
to
that.
A
So,
to
begin
with,
let
me
start
sharing
my
screen,
because
this
is
gonna
be
much
easier
to
talk
about
things
there
we
go
sure,
okay,
so
here's
the
ticket.
So
up
until
now,
our
olm
experience
was
maybe
not
exactly
great,
but
it
was
okay
because
we
were
bundling.
A
A
So
this
is,
we
can
either
save
it
as
a
snippet
or
immortalize.
It
some
other
way
or
if
we
don't
want
to
immobilize
it
doesn't
matter,
but
what
it
does.
It's
a
wrapper
creating
the
kind
cluster
deploying
some
circ
manager
in
there
and
I
specifically
broke
out
every
step.
So
we
can
see
what
this
thing
is
doing,
so
the
olm
bundle
is
doing
several
things.
So
it's
doing
the
build
manifest
then
generates.
The
bundle
then
patches
the
bundle,
meaning
some
of
the
activities
need
to
be
done
after
the
bundle
has
been
auto-generated
by
operator
sdk.
A
So
patch
bundle
are
those
steps
that
are
happening
afterwards.
So
if
we
go
to
the
script
itself,
you
can
see
that,
for
example,
we
substituting
the
manager's
image
name
to
operator
image
operator
tag,
so
we
can
override
those
on
the
fly
and
the
other
thing
that
I've
added
as
part
of
trying
to
resolve
issues
right
now
is
injecting
the
ingress
class
into
the
bundle.
A
But
this
actually
introduces
its
own
issue.
So
what
we're
to
do
we're
going
to
just
take
that
one
out
for
now,
so
that's
the
patch
bundle
and
then
what
we
do
afterwards
is
validating
bundle,
compile
and
publish
bundle
that
needs
to
go
into
some
a
container
registry
of
some
sort
that
we
referencing
here.
A
A
Only
after
that
can
we
deploy
the
subscription,
which
will
instantiate
the
operator,
and
then
we
have
to
wait
again,
because
until
the
operator
is
there
we
cannot
just
dump
the
gitlab
resource
into
into
the
cluster.
So
once
we
get
the
operator,
that's
the
last
step,
we
just
deploy
the
gitlab
in
there.
A
A
Kind
get
clusters,
no
clusters,
good,
all
right,
so
this
script
is
like
seriously
maybe
somewhat
over
engineered
piece,
but
it
kind
of
hides
all
the
otherness.
That's
happening
under
the
hood,
so
we'll
just
do
because
I
was
doing
several
iterations
over
it
already
and
the
iteration
number
impacts
the
tag
that
I'm
using
for
publishing
the
catalog,
the
bundle,
the
name
of
the
cluster
and
everything.
So
I
just
extracted
it
out.
A
So
I'm
using
the
gitlab
chart
version
6561,
I'm
using
the
operator
tag,
the
one
where.
A
A
There
is
no
difference
between
the
two,
quite
honestly,
so
once
we
launch,
that
is
what
it's
going
to
do,
is
going
to
spin
up
the
cluster
populate
it
and
you
will
see
what's
going
to
happen
next.
So
what
we're
going
to
do
is
kind
of
go
with
the
current
implementation
and
that
implementation
worked
by
the
way.
Up
until
now,
it
was
working
just
fine.
A
So
there's
one
extra
step
that
we'll
have
to
do
before
deploying
the
git
lab,
we'll
need
to
like
using
the
operator
we'll
need
to
instantiate
the
ingress
class,
and
then
everything
works.
A
So
while
it's
doing
this,
I
wanted
to
kind
of
touch
a
little
bit
on
what's
inside
the
bundle,
because
that's
probably
an
important
thing
altogether,
so
is
the
csv.
The
cluster
service
version
manifest,
and
that
thing
is
a
really
really
long
one,
and
it
includes
the
core
of
everything
that
we
really
really
need.
There
is,
if
we
get
the
outline-
and
just
do
it
like
this,
so
inside
the
spec.
A
A
A
All
in
custom
resource
definition,
sorry
that
one
shouldn't
have
missed
that
one.
So
that's
what
bundle
includes
right
now
and
how
it
does
it?
We
feed
the
operator
sdk
the
full,
manifest
that
we
normally
like.
We
currently
give
our
users
say:
do
cube
ctl,
apply
and
there's
that
operator
yaml,
that's
the
manifest
that
we
give
to
the
user
and
that
one
includes
everything,
including
the
ingress
class
right
there.
A
So
one
operator
sdk
processes
that
it
decides
to
strip
the
ingress
class
from
the
output
period.
So
that's
what's
happening
right
now.
So
that's
why
I
have
this
patch
that
I
shown
earlier
this
guy
just
blatantly
copying
the
ingress
class
into
the
directory
and
modifying
the
metadata
name.
So
it
fits
with
the
rest
of
things,
but
that's
about
it.
But
there
is
nothing
else
to.
B
A
Good
question,
though,
because
I'll
show
where
it's
coming
from
it's
coming
from
here,
so
we
have
the
config
manifest
basis,
and
this
is
the
csv
the
default
csv
we
start
with.
So
it
takes
all
the
metadata
from
this
csv
and
then
it
generates
whatever
it
generates.
It
generates
on
top
overwriting
some
of
the
values
here.
So,
for
example,
this
one
you
will
notice
somewhere
at
the
top.
It
says
the
operator
sdk
version
yeah
like,
for
example,
it
says
operator
sdk,
v
1.1.0,
but
if
you
actually
go
into
the
csv
here.
A
And
if
you
scroll
up
that's
there.
A
B
A
A
A
B
A
C
A
D
The
definition
of
job
security,
yeah
yeah
just
keep
making
operating.
A
Yep,
so
now
that
it's
this
whole
thing
is
complete,
so
we
can
actually
take
a
look
at
package
manifest
yeah,
it's
right.
There,
it's
been
instantiated.
So
now
we
can
proceed
with
the
next
step
from
our
auto
deploy
script
and
the
next
step
is
just
deploy
subscription,
which
means
that
so
let
me
show
what's
happening
in
the
cluster.
A
A
B
So
you
subscribe
to
a
service
at
a
certain
version
right
or
you
can
subscribe
and
tell
it
to
automatically
update.
I
think
I
remember
that
from
that's
another
date.
A
Let's
refresh
my
memory
as
well,
because
I
don't
worry
more
subscription
yeah,
we
just
say
which
csv
we
subscribe
to,
but
it
says
starting
csv,
so
I'm
guessing
it
will
kind
of
take
in
the
future
updates.
B
B
That's
probably
down
the
road
I
know
because
we'll
have
to
consider.
I
would
be
curious
to
see
what
olm
lets
you
update
to
like.
If,
if
someone
forgets
to
click
that
upgrade
button
for
a
while
and
they
come
back
a
year
later,
would
it
let
you
go
straight
to
the
newest
version,
or
would
it
make
you
go
step
wise,
at
least
for
the
minor
versions,
which
is
what
we
usually
require.
A
Yeah
I'll
have
to
tinker
with
that.
So
for
the
moment,
we'll
wrap
the
scope
back
to
the
issue
at
hand,
and
that
is
not
dismissing,
which
you
just
said,
because
it's
important
so
we
already
done
the
subscription.
So
all
we
have
to
do
right
now
is
just
deploy
the
gitlab.
So
that's
what
that's
going
to
do
so
it
actually
spits
out
the
manifest
that
was
applied,
so
the
gitlab
resource
has
been
deployed
and
if
we
go
into
the
space.
B
A
So,
let's
just
delete
this
gitlab
deployment.
Now
this
is
fine.
So
now
we'll
do
the
step
three
and
we'll
just
use
a
different
chart
version
to
begin
with,
but
we're
not
going
to
do
it
because
the
ingress
class
is
missing
too
so
or
we
can
do
it
without
the
ingress
class.
Just
to
show
what's
gonna
happen.
A
So
let's
do
this.
E
A
Posted
the
output
and
the
ticket
right
there,
those
are
the
errors
from
the
olm.
D
A
B
A
This
create
the
upstream
issue
and
instruct
users
to
create
the
ingress
class
manually,
so
this
manual
instruction
can
actually
be
embedded
into
our
olm
bundle
and
that's
what
I
was
going
to
kind
of
take
a
look
at
and
if
that
is
the
way
we
want
to
go,
we
should
probably
so
here's
the
documentation
and
the
prerequisites.
A
A
So
if
we
just
copy
paste,
the
csv.
A
A
So
if
you
add
into
the
prerequisites
section
an
additional
blurb
about
getting
the
ingress
class,
that
would
at
least
notify
users
in
some
shape
or
form
that
what
needs
to
be
done,
it's
not
going
to
help
them
at
least
it's
going
to
notify
them,
and
that
was
one
of
the
solutions
proposed
all
right.
Okay,
so
we
almost
done,
but
the
migrations
are
still
running
so
I'm
guessing
the
web
service
is
gonna
spin
up
in
a
minute.
A
And
once
that
is
done,
then
we
can
yeah
right
there,
so
so
ingresses
all
there,
but
the
ingers
class.
B
So
the
two
of
the
two
upsides,
at
least
from
our
perspective,
are
a
you
already
have
to
apply
a
manifest
manually.
The
get
lab
cr
so
asking
for
one
more
10
line,
manifest
apply,
isn't
too
bad
and
b,
you
don't
technically
have
to
use
nginx
like
our
nginx.
You
know
it's
it's
mostly
if
you
want
support
for
get
over
ssh,
which
obviously
is
a
very
common
feature
request,
but
not
technically
a
hard
prerequisite.
So
obviously
we
want
to
push
to
get
this
working,
but
I
guess
not
necessarily
hard
blockers
at
the
moment.
A
Let
me
see
which
one
contained
nginx
stuff.
A
There
we
go
in
48,
so
the
48
bundle
manifests
it
loud,
nginx,
just
to
be
a
little
bit
more
visual.
A
A
B
B
B
I
don't
think
we
need
to
keep
that
one
or
do
we
need
to
kick
that
one
too.
I
don't
think
so.
Now
it's
just
the
default
back
end.
That's
what
I
was
thinking.
A
A
A
E
E
A
Class
that
won't
work
yes,
but
in
olm
world
things
are
very
fixed
in
a
lot
of
ways.
Okay,
it's
very
opinionated,
at
least
at
the
moment.
We
only
have
one
deployment
into
one
particular
namespace
gitlab
system,
so
right
all
the
prefixes
and
everything
else
are
fixed
as
well.
True,
the
reality
is
that
yeah
nothing.
E
B
Yeah
I
mean,
since
the
user
has
to
apply
that
themselves.
They
could
change
that
spec.controller
value
and
just
make
sure
that
they
pass
that
as
a
value
in
their
gitlab
cr
to
engine
x
when
it
spins
up
but
yeah
you're
right.
A
lot
of
things
are
fixed,
given
that
olm
is
installing,
with
certain
prefixes
etc.
B
Just
curious
if
you've
gotten
any
responses
yet
on
that
issue.
Oh,
let's.
A
Yeah
and
actually
of
note,
I
did
provide
the
full
blown
like
if
anybody
wants
to
tinker
with
operator,
bundles
and
stuff
like
that,
this
actually,
the
cli
version
of
what
those
scripts
are
doing
very
condensed
so
minimal
amount
of
tooling
is
required
to
kind
of
run
it
locally.
If
you
want
to
tinker
with
that
cool.
Otherwise,
our
scripting
is
better
because
I
wrote
it.