►
From YouTube: Operator Framework SIG October 22 2019 Full Meeting
Description
Operator Framework SIG October 22 2019 Full Meeting
co-chairs: Diane Mueller and Rob Szumski
join the SIG here: https://groups.google.com/forum/#!forum/operator-framework
find more details about Operator Framework here: https://github.com/operator-framework/community
A
Hello,
everybody
we're
just
getting
started
here.
Let's
see
if
we
have
I
can
share
this
pop
over
to
the
the
agenda,
which
is
in
a
new
arena
or
area
this
year.
This
week
we're
experimenting
our
and
moving
our
agenda
building
process
over
to
projects
in
the
operator
framework
under
the
community
repo,
and
you
should
be
able
to
see
my
screen
now.
A
Okay,
so
today
the
agenda
and
I'm
just
gonna
quickly
go
through
here
is
that
I've
set
up
is
the
update
on
the
operator
framework.
There's
been,
you
know
a
number
of
discussions
on
the
mailing
list,
so
I'm
gonna,
let
Rob
handle
that
and
then
I
think
Daniel
Messier
is
joining
us
and
he
can
give
us
an
update
on
the
operator
hubs
or
operator
hub
do
and
those
that
are
in
the
queue.
A
B
B
Some
of
our
release
cycle
in
terms
of
timing
is
typically
when
food
builders
gonna
do
a
release
and
that
a
lot
of
the
members
of
this
state
are
working
on
upstream
that'll
get
consumed
in
the
SDK,
along
with
other
kind
of
the
user.
Experience
related
features
that
we
have
so
the
seal
is
gonna,
remain
compatible
kind
of
behind
the
scenes,
and
so
that
will
drive
a
lot
of
the
timing.
So
if
you
see
a
new
big
cube
builder
to
myself,
I'll
actually
show
my
screen.
B
If
you
see
a
cube
builder
release,
then
that
is
what's
going
on
so
here's
the
version
0.1,
you
can
see
I'm,
not
gonna,
read
through
all
this.
Some
new
features
and
things
that
were
added
one
thing
that
you'll
see
is
we're
doing
some
regular
maintenance
on
the
helm,
SDK
an
update
on
that
in
terms
of
where
we're
going
with
that
is.
We
know
that
three
is
right
around
the
corner,
and
so
we
are
planning
to
pull
that
in
when
it
gets
pretty
stabilized
and
update
the
SDK.
B
So
you
can
build
helm
three
based
operators,
a
nice
thing
about
that
is
it
doesn't
require
a
chiller.
A
bunch
of
other
cool
things
as
well.
Coming
with
that,
as
always,
there
are
some
breaking
changes
that
happen.
This
is
part
of
their
transition
to
coop
builder.
You
know
you
have
to
make
some
changes
from
time
to
time.
Those
are
called
out
and
the
release
notes.
So
the
team
is
done,
a
really
good
job
with
this.
So
if
you
have
any
questions
about
this,
please
open
issues
send
emails
to
the
mailing
list.
B
That's
about
the
update
on
the
SDK.
We
can
jump
over
to
the
lifecycle
manager
and
take
a
quick
look
at
their
releases,
so
very
similar
inversion
scheme.
So
0.12
came
out
22
days
ago
a
quick
look
at
the
change
log,
a
number
of
just
kind
of
everyday
bug
fixes
and
things
like
that.
One
thing
you'll
see
is
mentioned
for
private
registries.
This
is
a
big
focus
of
ours.
B
As
you
know,
we've
started
on
the
easiest
use
case,
which
is
these
online
clusters,
where
everybody
can
pull
container
images
from
all
over
the
Phillies
from
docker
Hagen
from
Quay
and
etc.
But
we
also
want
to
make
Olin
work
just
as
well
for
private
registries,
and
so,
if
you
want
a
mirror
down
content
of
an
operator
and
to
put
it
in
your
private
registry,
that's
behind
your
firewall.
We
want
Olin
to
work
for
that.
So
that's
a
lot
of
the
work
that's
going
on
now
and
we'll
go
into
the
future.
B
That
will
involve
some
changes
to
the
CSV.
Just
you
know,
kind
of
declaring
which
images
that
you
support,
so
that
you
can
then
go
sync
them
into
your
cluster
I.
Think
that
is
about
the
update.
I,
don't
want
to
take
too
much
time
away
from
the
already
packed
agenda
that
we
have
make
sure
we
have
time
to
talk
about
the
the
local
storage
operator.
But
if
there's
questions
really
quick
on
some
of
the
things
that
are
going
on.
B
Wanted
to
mention
that
one
of
the
things
about
Kugler
that
is
really
great
is
we're
not
duplicating
that
work
like
I
said
so
we're
working
with
kubernetes
community,
and
so
what
this
frees
us
up
to
do
is
spend
more
time
on
a
better
user
experience,
better
troubleshooting
fixing,
bugs
etc
also,
possibly
opening
up
more
time
to
consider
new
SDKs.
I
know:
we've
talked
about
this
in
the
past
before,
but
once
again,
if
you
have
interest
in
like
a
job
at
SDK
or
a
Python
SDK
we'd
love
to
hear
your
use
cases
on
the
mailing
list.
B
I
know:
we've
discussed
I,
think
Java
a
little
bit
in
the
past,
so
continuing
to
look
at
that.
It's
a
big
investment.
So
we
want
to
make
sure
that
it's
the
right
thing
to
do.
But
if
you
have
interest
in
that,
please
let
us
know
or
later
on
in
the
state
call
or
features
things
and
Dan
I
don't
know.
Are
you
going
to
talk
about
any
of
the
CN
CF
donation
process
yeah?
So.
A
So
so,
while
I
read
look
while
we're
waiting
for
Daniel
to
do
that,
if
like
put
a
link
for
hacker
MD
to
use
for
public
notes,
so
if
you
could
add
your
name
as
attendees
into
that,
that
would
be
great
and
I
will
endeavour
to
keep
track
of
what
we
talked
about
today
and
I.
Think
it
was
a
key
I
was
using
before
too
so.
C
C
So,
in
order
to
give
context
to
that,
we
created
that
a
charter
which
is,
if
you
look
at
it
very
similar
to
the
Charter
that
openshift
or
okd
as
a
community
version
of
overshift
has-
and
this
is
really
about
some
standard
stuff
like
what's
the
mission
statement,
what
is
considered
in
scope?
What
are
we
doing
as
part
of
the
framework
and
what
are
some
of
the
more
logistical
aspects
of
that?
So
one
example,
for
that
is
our
operations
model,
so
there
will
be
a
group
of
chairs
that
meet
this
community.
C
This
is
a
draft
by
the
way,
this
is
up
for
discussion
and
it's
available
as
a
pro-v
quest
in
the
community
repository
and
operator
framework.
So
this
is
obviously
up
for
public
discussion.
The
structure
is
heavily
inspired
by
how
open
shift
and
OPD
does,
but
in
the
end,
there's
going
to
be
a
group
of
people
who
are
running
the
community
as
chairs
and
Dan
q,
obviously
being
one
of
them,
but
also
core
maintenance
of
the
various
projects,
as
well
as
product
management.
C
So
I
don't
want
to
make
this
too
long,
but
this
is
a
fairly
standard
implementation
of
an
open
source
of
string
governance
model
that
you
will
find
in
many
other
projects
like
helm,
for
instance
as
well,
and
that
is
I
think
all
I
wanted
to
say
here
so
go
check
that
out
and
see
if
and
how
you
want
to
contribute
to
that.
Another
part
that
will
come
with
formal
governance
is
also
more
community
liaison
in
terms
of
more
public
meetings
like
this
one,
for
instance,
but
focus
on
particular
topics.
C
So
in
the
SDK
community,
for
instance,
we
are
discussing
of
having
a
community
call
like
a
standard
from
office
hours,
I
think
all
the
regular
cadence
to
discuss
all
topics,
that
of
element
for
the
SDK,
and
that
would
also
be
a
forum
to
be
a
little
bit
more
transparent
about
the
roadmap
about
upcoming
releases,
their
time
frame
and
what
they
actually
include.
This
is
not
formalized
yet,
but
I
think
we
will
let
the
individual
sub
projects,
mainly
the
SDK
and
all
them,
run
that
either
as
part
of
this
sick
or
and
then
separate
Linux.
A
The
main,
the
main
point
that
make
right
now
we
are
a
special
interest
group
underneath
and
we're
meeting-
is
underneath
the
open
ship
commons
umbrella,
and
this
would
probably
when
we
do
the
CNC
F.
Well,
we
move
this
to
a
working
group
under
the
CN,
CF
or
instantiate
that
in
some
other
way-
and
we
probably
still
have
an
operator
sig
hanging
around
under
open
ship
Commons,
because
it's
of
interest
to
the
open
ship
community
specifically-
but
this
was
really
a
move
to
make
this
more
of
an
openly
governing
project.
C
As
it
seems
now,
the
sick,
app
delivery
group
is
at
least
interested
in
having
us.
They
are
very
new
as
well
so
we'll
see
if
that.
That
is
how
it
turns
out,
but
we
have
actually
presented
the
offer
a
framework
for
inclusion
in
the
Cinzia
and
that
particular
special
interest
group,
which
is
also
tied
in
to
seek
apps
in
general,
but
I,
think
sick
of
delivery
from
a
charter
and
agenda
perspective
fits
very
well
and
what
we
want
to
do
with
operators,
in
particular
with
operator
lifecycle,
management.
A
A
We've
done
this
when
we
get
to
a
point
where
we
feel
like
we've
got
a
good
comment
and
feedback
on
it,
we'll
put
it
out
for
actually,
we
probably
should
put
out
an
official
call
for
comment
on
this,
and
so
before
we
move
on
to
the
operator
hub
update
I.
Just
maybe
it
would
be
a
more
appropriate
this
time
to
talk
about
the
status
of
the
the
TOC
stuff.
A
The
and
all
I'm
going
to
really
say
about
that
is
if
you
on,
we
are
set
to
be
on
the
agenda
for
December
5th,
CN,
CF,
toc
meeting
so
post
kook
on
North
America
and
the
pool
Rick.
The
official
pull
request
from
where
you
can
comment
and
see
comments
is
here
and
we
would
love
it
if
people
put
in.
You
know
supportive
comments
of
this
effort
in
here
or
ask
their
questions.
So,
if
you
can,
I
hope
I'm
sharing
my
screen.
It
doesn't
look
like
nope
I'm,
just
wandering
here.
A
So
if
you
take
a
look
here,
you
will
see
the
requests
and
we
are
requesting
incubation
status.
We
think
we
have
enough
people,
obviously
on
this
call,
there's
30
folks
and
another
another
group
of
people
that
aren't
here
who
are
already
have
operators
and
operator
hub
and
a
lot
of
people
using
operators
in-house
at
their
enterprises,
and
we
are
heavily
committed
to
it
under
the
hood
at
open
shifts.
A
C
There
are
four
things
that
I
wanted
to
cover,
like
just
turn
in
trouble.
Finding
my
presentation
right
now,
but
I
think
it's
it's
fine.
If
you
go
through
it
like
this
morphine's
who
operate
after
day
or
what
the
first
thing
is
actually,
hopefully
something
some
of
you
have
already
noticed.
We
are
now
trying
to
streamline
the
process
of
creating
a
packaged
version
of
your
operator
suitable
for
deployment
with
the
lifecycle
manager,
and
for
that
we
have
created
a
you
I.
C
This
will
flow
in
order
to
create
your
cluster
service
version
5,
which
is
really
the
heart
of
that
packaging
process
right.
So,
if
you
have
never
packaged
an
operator
before
for
operator
or
operator,
lifecycle
manager,
it's
mainly
about
creating
a
yama
manifest
that
describes
your
operator,
both
under
more
metadata
Heaviside,
like
the
display
name,
a
logo,
a
description
all
the
stuff
that
you
want
to
see
in
a
catalog
like
operate
after
that.
Also,
all
the
content
on
offer,
after
all,
is
coming
out
of
this
a
metadata
format
and
UNAIDS.
C
On
the
other
hand,
there's
a
lot
of
data
in
it
that
are
is
needed
on
cluster.
In
order
to
run
your
operator
now,
it
can
be
a
little
cumbersome
because
it's
a
lot
of
data
and
a
very
large
manifest
to
write
this
all
yourself
or
get
it
right
even
using
an
example.
So
what
we
have
done
is
we
have
created
a
package
or
bundle
editor
which
allows
you
to
do
that
in
a
graphical
fashion,
and
there
are
two
ways
you
could
go
about
this
first.
C
If
you
already
have
a
constant
service
version,
llamó
and
your
custom
resource
definition
manifest.
You
can
just
upload
those
here
and
this
UI,
wizard
or
editor
will
pre-populate
as
much
as
possible
of
that
and
allow
you
to
edit
this
or
if
you
are
starting
from
scratch,
and
you
have
no
CSV
at
all.
You
just
have
your
standard.
Two
Vanilla's
manifests
like,
for
instance,
your
deployment
or
also
cluster
roles,
role
bindings,
all
this
stuff
which
you
need
for
deploying
you
operate
a
manually.
Don't
you?
C
E
C
Of
your
operator,
let's
say
cockroach
to
be
2,
1,
12
or
13.
You
could
just
go
in
here
and
make
edits
and
in
a
UI,
with
validation
with
guidance,
the
proper
errors,
let's
say:
1
2
2,
for
instance,
increase
your
capability
model.
You
don't
need
to
write
it
in
yellow
file.
You
can
use
the
UI
for
that.
So
this
this
really
helps
you
do
that
in
a
more
human
friendly
way
with
less
opportunity
for
errors,
and
it
also
works
off
the
standard
cube
manifest.
So
you
don't
have
to
have
the
class
the
service
version.
C
You
don't
have
that
an
existing
package
already,
if
you
have
just
this
done
a
few
minutes
this
operator,
this
just
fun
where
the
toggle
things
it
will
be
able
to
pre-populate
as
much
of
that
as
possible,
including
some
of
the
more
complicated
stuff
like
see.
Are
these
oh
well,
this
one
didn't
quite
work,
I
guess,
but,
for
instance,
the
deployment
can
be
edited
here.
So
if
you
need
to
change
that
somehow
you're
doing
that
with
a
proper
editor,
there's
no
issues
anymore
with
event
and
syntax
errors
on
something
like
that.
C
So
this
is
in
beta
and
we'd
love
to
have
your
feedback
on
this,
because
we
obviously
want
you
to
create
as
easy
as
possible.
These
bundles
and
update
I
just
have
mentioned
Stauffer
and
after
I/o
or
on
your
cluster,
and
this
is
hopefully
an
easier
way
to
do
that,
rather
than
you
know,
dealing
with
all
together
yourself.
C
What
you
get
as
a
result
is
a
download
it's
a
zip
file
which
contains
at
least
your
CSV
or
your
CRTs,
and
a
package
channel,
which
is
exactly
what's
needed
in
order
to
make
a
contribution
to
operator
after
I/o,
and
you
can
always
also
switch
back
to
full
llamo
edit
mode.
If
you
need
that
or
you
want
to
see
what
a
particular
setting
did
and
get
that
in
an
editor
here
as
well
that
even
has
syntax
highlighting
and
colecovision,
you
can
also
preview
that.
C
So
you
see
how
that
would
work
and
how
that
would
look
like
on
right
after
I/o,
so
yeah
I
would
love
to
get
your
feedback
on
that.
If
you
have
feedback,
there
is
and
repository
in
the
operator,
framework
ork
for
the
site
itself
operate
a
hospital
and
we
would
love
to
have
your
feedback
there
and
you
just
create.
C
Well,
you
create
an
issue
here
in
the
indicator.
Click
on
start
for
that,
and
that's
that.
The
other
thing
that
I
wanted
to
make
you
aware
of
is
that
we
have
updated
our
contribution
guidelines.
We
are
now
including
something
that's
called
TCO,
and
it
should
be
very
well
known
to
people
who
have
at
once
at
least
committed
a
hunch
art
to
the
public
hundred
repository
developer.
Certificate
of
origin
is,
is
basically
a
very
loosely
format,
formal
or
informal
process
to
sign
off
contribution.
C
So
we
get
a
sense
of
authenticity
and
an
ownership
there
again
very
loosely
defined
just
appended
it
on
your
PR,
especially
if
you
modify
an
existing
operator.
The
only
criteria
here
for
that
is
that
your
name
on
github
must
match
whatever
you
put
in
here,
and
this
should
also
match
the
original
submitter
of
the
PR.
C
In
addition
to
that,
we
have
also
changed
the
file
layout
of
the
bundle
format.
This
is
the
metadata
that
drives
operator
hafta
Rio,
and
this
is
the
Casa
service
version
stuff
that
I
was
talking
about
before
so
before
we
had
everything
in
one
big
directory,
and
you
were
supposed
to
you-
know
just
drop
in
your
additional
CSVs
in
your
new
or
updated
crts.
C
That's
that
and
the
other
thing
I
wanted
to
draw
your
attention
to
is
that
we
have
added
an
automated
way
to
test
your
operators
with
a
local
cluster.
So
obviously,
whenever
you
add
an
operator
to
operate
after
all,
by
adding
or
by
creating
a
pull
request
on
the
community
or
for
this
vehicle,
there's.
F
C
Of
tests-
that's
that
is
run
I
mean
in
order
to
check
the
operator
for
completeness
syntactic
correctness
and
if
it
deploys
with
all
M
and
if
it
behaves
correctly
on
cluster
now
it
may
be
a
little
bit
cumbersome
to
you
know
see
that
only
coming
in
via
PRS
and
github
checks
that
are
created
as
part
of
GPRS.
You
may
want
to
test
it
before
you
actually
make
a
contribution,
and
you
can
do
so
already
with
manual
instructions.
So
we
have
that
here
as
well.
C
So
we
walk
you
through
this
entire
process
of
testing
your
operator
department
with
all
M
on
either
kubernetes
or
openshift
step
by
step,
but
we've
also
automated
this
now.
So
there
is
in
the
top-level
directory
of
the
community
operators
report,
a
Meg
file
which
you
can
just
call
on
a
particular
operator
on
your
local
disk,
and
this
make
file
has
several
sub
comments
that
represent
different
test
modes
and
what
it
will
do.
C
It
will
basically
run
the
same
amount
of
linting
very
and
on
Cluster
test
as
all
the
tests
that
are
done
by
the
pipeline
behind
operator.
Half
the
diode.
This
gives
you
a
chance
to
catch
arrows
and
syntax
or
packaging
quite
earlier,
so
you
don't
need
to
wait
on
PR
feedback
and
intervention
from
some
of
the
maintenance
of
operate
out
the
rail.
So
what
you
need
for
this
is
either
a
Linux
or
Mac
machine
it'sthe
doctor
installed
make
obviously
as
well
and
either
an
existing
kubernetes
cluster
or
if
you
have
it
installed,
mini
cube.
C
The
tool
will
basically
try
to
sense.
If
you
have
a
running
cluster
already
by
inspecting
you
a
few
convict,
if
you
don't,
it
will
offer
you
to
create
a
mini
cube
cluster
just
for
the
test
run,
and
it
will
create
that
for
you,
it
will
basically
install
all
em
in
that
for
you.
It
will
package
your
operator
into
the
catalog,
so
it
can
be
installed
by
all
and
it
will
trigger
the
installation,
and
then
it
will
run
a
couple
of
behavioral
tests
on
your
operator
to
see
if
it's
basically
adheres
to
non
best
practices.
C
These
best
practices
are
the
ones
identified
by
the
scorecard
from
utility,
which
is
part
of
the
Opera
SDK
long
story
short.
If
you
have
a
Linux
or
Meccans
roof
doctor
and
make
install,
you
can
run
this
particular
command,
make
operator
tests,
and
then
you
give
it
the
path
to
your
bundles
on
disk,
and
that
will
also
already
work
with
the
new,
a
nested
format,
and
then
it
runs
all
these
tests
for
you.
So
at
the
very
end
of
this,
you
should
have
a
clear
preview
of
how
the
pattern
will
treat
your
submission.
C
E
C
Right
now
that
is
unfortunately
in
no
way
in
the
scorecard
to
state
which
test
should
be
run.
We
are
working
on
an
improved
version
of
the
scorecard
utility
to
do
just
that
actually,
but
the
current
version
does
not
do
that,
so
the
conversion
role,
you
know
blindly
run
those
standard
tests
that
are
that
are
shipped
with
the
with
the
utility.
C
The
default
test
of
the
clock
are
not
still
variable,
so
if
they
tell
you
something
is
it's
not
right,
then
there's
probably
a
good
reason
for
it.
Okay,
we
know
it
can
be
partly
sometimes,
but
usually
it
it
catches
all
deal.
You
know
missing
metadata
stuff
and
by
missing
examples
and
stuff
like
that.
A
Anyone
else
have
questions
for
Daniel
on
this
Joan
docker
looks
like
you're
trying
to
say
something
or
ask
something:
no,
we're
gonna
move
on
and
I'm
gonna
take
over
your
screen
again
and
I
just
wanted
to
give
a
couple
of
updates
on
Kuk
on
North
America
and
the
open
ship
Commons
top
just
a
reminder
to
everybody
that
as
well
that
the
Red
Hat
summit
call
for
papers
ends
today
the
22nd
when
I
wrote
this.
It
was
tomorrow.
A
So
if
you
have
a
set
that
you'd
like
to
submit
give
it
get
it
in
today,
if
you
have
a
talk
that
you
are
giving
a
kook
on
North
America,
that
is
about
your
operator
or
tangentially
about
operators.
Please
let
us
know
so
one
of
the
things
I'm
going
to
try
and
do
for
cuckoo
on
North
America,
I,
sort
of
write,
a
a
guide
to
operators
and
all
things,
operators
at
Keuka,
North
America,
for
people
to
have
all
the
talks.
A
A
I
just
was
informed
that
they
rented
a
room
just
for
me
to
hold
meetings
in
at
Keuka
in
North
America,
which
kind
of
like
is
making
more
work
for
Diane,
but
that's
okay,
because
what
I
was
thinking
was
that
we
would
do
our
next
virtual
sig
meeting
the
week
before
Kuk,
on
which
the
data,
which
is
would
be
I,
think
the
11th
or
the
12th.
Actually
sorry,
the
12th
of.
A
So
you
do
one
virtually
prior,
because
it's
craziness
at
at
coop
con
to
try
and
really
have
an
official
sig
meeting,
but
if
people
were
open
to
it,
the
room
once
I
know
what
the
size
of
it
is
that
they've.
Given
me,
it's
probably
12
to
20
people
in
the
room.
We
could
have
a
face-to-face
session
and
have
everybody
get
together
and
chat
at
Keuka.
A
So
once
I
have
the
room
number
and
the
size
I
will
send
out
a
notice
to
the
group
to
see
if
people
are
up
for
that
so
and
I'm
trying
coordinate
that
with
everybody's
presentations
at
tubecon,
so
that
everyone
who's
coming
to
could
come
and
meet
up
as
well.
So
that
was
kind
of
what
we
had
on
the
schedule
today
for
visual
business
and
as
we
are
want
to
do.
I
have
Chris
I,
have
a
short
ansible
update,
I
think
from
Chris
all
right.
A
D
A
D
No
permission
required
all
right,
so
I
will
drop
a
link
to
my
slides
somewhere
in,
but
I
can
speak
through
them.
So
ants
will
operator
Earth,
Day
I'm,
one
of
the
right
update
on
an
school
is
built
in
the
operator
SDK.
We
have
been
doing
a
lot
of
work
around
it
and
trying
to
popularize
it
and
get
more
people
thinking
about
moving
workloads
from
wherever
they
are
now
automated
with
their.
You
know,
existing
ansible
automation
to
kubernetes
with
you
know
at
least
exam.
D
So
we
feel
like
that.
You
know
they
offered
a
framework
kubernetes
ansible
or
match
made
in
heaven.
It's
all,
yellow
all
the
way
down,
so
that
makes
it
super
easy
to
limp
at
the
very
least
picking
the
hard
one,
existing
automation
for
complex
or
clothes
into
Kate's,
really
kind
of
opens
eyes
for
people
and
then
in
general.
You
know
this
is
based
off
polling
rooms
and
meetups
and
conferences.
I've
spoken
at
more
people,
I,
followed
I,
have
familiarity
with
ansible
than
go
by
about
four
to
one.
D
So
plugging
people
in
with
ansible
into
operators
I
think
makes
a
ton
of
sense.
We
actually
hosted
a
like
official
workshop
at
ansible
fests
building,
kubernetes
operators,
ansible
hands-on
workshop.
It
sold
out
100
registrations
like
almost
immediately.
There
were
a
total
of
183
seats
in
the
room.
I
did
not
count,
but
the
room
was
absolutely
packed.
I
can
share
that
content.
There
I
want
to
have
a
huge
thank
you
to
do.
D
Michael,
rib,
Matt,
Sean
Melvin,
that
Hillsman
Jeff
Garlin,
Tim,
Atmel
and
Matt
that
have
all
done
tremendous
work
to
make
that
workshop
possible,
as
well
as
in
school
operators
in
general.
We
also
have
a
new,
like
the
demo,
is
called
a
Mik
router
operator
and
the
Crowder's
a
small
project
that
routes
requests
between
multiple
memcache
instances.
So
it's
pretty
simple.
You
know
which
is
kind
of
you
know.
D
This
is
the
declared
state
I
want
to
make
router
instance
or
MacGruder
process,
and
you
know
to
memcache
DS
services
running
and
then
routing
between
that
the
operator
builds
that
and
a
lot
of
work
has
been
done
by
Jeff,
Melvin
and
Matt
on
that
as
well.
So
that's
actually
what
we
demonstrated
at
ansible
fest,
and
it
went
very,
very
well
and
a
lot
of
feedback.
A
lot
of
good
feedback
on
that
where
to
next,
but
we
need
more
use
case
ideas
that
can
be
demonstrated
in
less
than
an
hour.
D
The
hardest
part
I
think
we
found
with
ansible
operators
is
like
great
I
can
do
massive
database
failover
and
you
know
horrendous.
You
know
operations
that
would
normally
take
1820
hours
before
and
you
know
one
you
know
on
ansible
playbook
and
you
know,
convert
it
to
an
operator
and
off
I
go
what
like
going
from
hey
these
people
vaguely
know
kubernetes
to
full-blown
like
operator
like
that,
and
a
small
amount
of
time
is
very
hard.
Also,
you
have
to
tackle
the.
Where
do
we
have
real
data
here?
Can
we
verify
that?
D
You
know
all
the
things
that
we're
doing
in
the
workshop
actually
works?
That
takes
time
when
you're
dealing
with
large
data
sets,
like
you
know,
page
a
pairs.
So
we
need
more
ideas
like
that.
We
have
a
brand
new
project
under
the
orders
operator,
framer
projects
and
github,
and
we
will
be
adding
more
fun.
Things
like
we
need
to
eventually
get
figure
out,
since
ansible
has
decided
to
go
to
the
culture
out
where
the
kubernetes
module
or
ansible
might
need
to
live
long
term,
if
that's
gonna
main
be
maintained
by
Red,
Hatters
or
the
community.
D
A
A
Further
ado,
I
wanted
to
let
and
unmute
herself
and
give
her
talk,
which
I've
cued
up
as
how
to
use
local
storage
operator
or
or
how
her
group
did
it.
But
she's
done
a
lot
of
work
around
rook
and
SEF
and
container
storage
and
things
so
I'll.
Let
her
talk
and
introduce
herself
and
take
over
the
screen.
Yeah.
A
F
So
first
thing
I'm
finding
out
is
that
it
appears
we're
no
longer
using
this
local
storage
operator
or
the
open
shift
container
deployment.
So
I
can
put
this
in
here,
but
yeah.
We
are
not
using
it
anymore.
We
are
using
block
mode
and
file
mode
PVCs.
The
difference
is,
if
you
look
down
here.
F
F
So
that's
news
to
me.
I
think
when
the
RAI
next
project
went
basically
got
reconfigured
to
be
a
taco
project
that
was
dropped
because
there's
no
need
to
support
bare
metal,
so
what
I
thought
I'd
do?
Just
since
you
know,
the
local
stores
operator
is
something
that
is
deployed
again
during
the
OCS
4.2
deployment,
but
it
is
not
currently
used,
but
what
I
have
I
just
deployed
up
until
4.2
GA
I
also
deployed
these
operators.
This
is
done
right
now,
through
a
manifest,
because
open
chef,
container
storage
is
not
GA.
F
Yet,
in
addition
to
the
the
operators
we
we
also
have
two
new
dashboards,
and
these
are
specific
for
open
chef,
container
storage.
If
you
you
know,
if
you
chose
to
install
port
works
or
storage,
OS
or
Rob
and
IO
or
s,
you
would
not
have
these
dashboards
right
now,
it
says
open
chef
container
stores
is
not
available.
F
If
we
look
at
the
the
CI
here
and
then
it's
a
project
that
got
created,
open,
chef
storage,
so
all
of
the
resources
for
open
chef
container
storage
are
going
to
be
in
this
project.
You
see
that
we
do
create
a
local
storage
operator.
We
also
create
the
Nuba
operator.
The
OCS
operator
is
the
meta
operator
that
sort
of
orchestrates
the
other
operators
and
then,
of
course,
the
Rick's
F
operator,
which
is
the
upstream
project,
that's
been
around
for
a
bit.
F
So
this
these
dis
currently
consists
of
what
you
get
with
OCS
4.2,
like
I,
said
just
today,
finding
out
that
we're
no
longer
using
a
local
storage
operator
to
create
block
and
and
file
mode,
PVCs
I
haven't
deployed
the
cluster.
Yet
I,
just
booted
up
the
operators
at
the
dashboards
and
I've
got
the
I've
got
the
OCS
operator.
F
So
right
now,
if
we
look
at
what
we
have,
we've
basically
just
done
the
OCS
in
it,
but
we
haven't
deployed
the
cluster,
so
I
can
go
to
here
and
I
create
the
cluster
when
it
that
comes
up.
It
tells
me
which
workers
I
have
available
I've
played
on
AWS
and
I.
Did
some
I
put
some
things
in
the
install
config
yeah
mo?
This?
F
Is
the
minimum
size
on
AWS,
basically
16
CPUs,
64
gigabyte
mem
that
we
are
recommending
for
us
has
to
be
deployed
on,
so
that
would
be
true
for
VMware
and
eventually
I
assume
for
bare
metal
next
release.
So
if
we
go
back
here,
you'll
see
that
we
have
information
for
at
least
AWS
on
the
zone
is
the
not
for
VMware
yet,
but
I
just
have
to
basically
pick
three
workers
that
are
not
in
the
same.
F
A
Z
Oh
pick
this
one
and
because
they're
all
the
same
Bricker
type
I,
don't
really
have
to
to
worry
about
which
worker
I
pick,
but
the
Masters
are
not
scheduled.
Bowl
I
think
that's
a
future
enhancement
to
not
show
them
because
they,
you
cannot
schedule
on
on
the
masters.
I
believe
I
pick
three
workers.
It
is
required
that
you
pick
a
minimum
of
three
because
of
the
the
need
for
availability
and
the
fact
that
stuff
is
going
to
be
a
replica.
Three
I
create.
F
So
then
I
can
do
a
few
things.
Go
back
to
dashboards
I'll
start
to
see
a
lot
of
events
on
the
right
side,
not
up
yet
so,
let's
back
to
our
CLI,
your
watch-
and
this
is
where
local
storage
operator
was
used.
F
F
And
they
are
basically
an
EBS
volume
that
is
in
creating
a
PVC,
a
foul
mode
PVC
for
the
monitors
for
the
sub
monitors.
So
we'll
see,
here's
the
first
monitor
coming
up
right
now
and
it's
going
to
go
ahead
and
mount
it
turns
out.
The
monitors
do
need
some
persistent
storage
in
addition
to
the
u.s.
T's,
the
contrary
to
say
when
you
know
and
rook
now,
ruck
up
stirring
has
this
feature
as
well,
but
currently
with
OCS
for
downstream.
F
We
only
support
using
a
storage
class,
either
in
VMware
or
and
AWS
2,
to
create
the
PVCs.
If
you
you're
not
able
to
directly
use
local
storage
that
that
will
change
in
OCS
4.3
in
January.
So
if
you
have
local
storage
connected
via,
especially
on
something
like
VMware,
where
you
want
to
consume
local
storage
or
or
bare
metal,
then
it
would
be
similar.
F
Also,
if
you
haven't
seen
it,
we
have
all
these
CSI
sidecar
pods,
most
of
the
pods,
have
more
than
one
container
you'll
see
a
SFF
fest
bus
sidecar
odd
for
each
one
of
the
each
one
of
the
workers
and
you'll
also
see
these
provisioners
and
you'll
see
the
RVD
plugin
for
each
one
of
the
workers
as
well.
So
this
is
all
to
support
CSI,
which
for
CSI
for
this
version
we
support
deleting
and
creating
volumes.
F
We
currently
don't
support
cloning
or
snapshotting
volumes,
that
that'll
be
tech
preview
and
in
4.3,
so
we're
now
seeing
the
storage
come
up
for
the
OS
sees
these
terabyte
PVCs
that
have
been
created.
These
are
going
to
be
block
mode
compared
to
file
mode
for
the
the
Mons
and
Anna
sort
order.
Here
we
see
some
prepare
pods
here.
Preparing
the
these
prepare
pods
are
used
to
create
the
OSD.
So
that's
sort
of
the
last
step
here.
F
You
also
notice-
and
this
may
or
may
not
go
out
of
pending,
there's
a
busy
right
now
for
this,
but
you
notice
that
nuba
needs
persistent
storage
as
well,
so
we
have
the
Nuba.
This
is
the
server
for
nuba
it's
currently
pending,
as
is
its
persistent
volume,
and
the
reason
it
is
pending
is
because
it's
trying
to
claim
from
a
storage
class
that
is
not
backed
by
stuff.
F
Yet
so
as
an
outcome
of
this
deployment,
we're
going
to
have
three
if
it's
successful,
we'd
have
three
stores
classes,
one
for
stuff
already
or
block
one
force,
ffs
file
and
one
to
do
bucket
claims.
So
bagger
claims
are
brand
new
to
OCS
and
probably,
as
far
as
I
know,
I
don't
know
any
other
storage
capability
for
openshift
that
supports
bucket
claims.
F
If
someone
knows,
let
me
know,
but
I,
don't
think
port
works
or
open
abs
or
storage,
OS
or
Robin
IO
I
support
my
current
claims,
so
these
are
bucket.
Claims,
be
really
similar
to
really
similar
to
like
a
PVC
they'll
have
OBC
and
you'll
have
OB
that'll
back
the
OBC
and
the
OBC
will
be
namespace
dependent
and
then
the
OB
just
like
Peavey's
will
be.
You
know
globally,
globally,
viewable.
Of
course,
if
you
want
to
delete
the
OBC,
you
would
delete
at
OBC
level,
not
at
the
of
evil.
F
F
There
is
a
problem
right
now
with
the
crust
rule
and
stuff,
and
you
also
see
the
object.
Service
is
not
populated
yet
because
noob
is
not
up
so
at
the
time
of
GA
in
about
a
month
from
now,
this
stuff
will
all
be
fixed.
The
GA
for
this
is
November
19th
for
downstream
OpenShift
OCS.
So
I
can
answer
other
questions
or
we
could
go
in
and
look
at
some
more
things
here,
but
any
questions
or
I
mean
I
I'm.
Sorry,
it
wasn't
totally
about
local
storage
operator,
but
that's.
A
Quite
okay,
so
that's
actually
good
to
know
and
thanks
thanks
for
doing
this,
does
anyone
have
any
questions
for
Annette
at
the
moment?
And
you
know
we
can
always
end
a
few
minutes
later.
I
see
that
I'm
just
going
to
pause
for
a
minute,
Jeff
Garlin
is
on
the
call
and
I
just
want
to
kind
of
maybe
cue
him
up
either
to
do
a
talk
at
the
next
operator
framework,
sig
or
if
he
wants
to
give
a
quick
update
on
his
work.
A
Just
you
know
the
status
check,
but
for
and
that
I
think
maybe
where
can
you
also
ran
a
very
nice
video
yesterday
from
me
when
we
were
talking
about
it
yesterday
we
can
add
the
link
to
that
into
the
the
notes
here,
because
I
think
that
would
be
a
good
thing
for
people
to
be
able
to
watch
as
well.
The
overall
video
that
you
shared
yeah.
F
A
All
right,
so
we
only
have
five
minutes
left
so
I
think.
Maybe
what
we'll
do
is
we'll
pause
and
Jeff
is
chatting
in
the
chat
here.
I
can
see
him
saying
next
time
so
we'll
if
we
could
file
out
an
agenda
item
for
the
next
meeting,
because
Jeff's
been
doing
some
interesting
work
around
the
ansible
stuff
as
well
and
we'll
move
that
forward.
A
That
use
case
would
be
wonderful
and
if
you
need
any
help
offering
or
writing
them,
qing
mian,
slack
or
somewhere.
Today
and
I'll
try
and
give
you
a
hand
if
you'd
like
that
and
we'll
be
talking
again,
I'm
sure
very
shortly
on
in
in
chat
or
in
on
the
mailing
list.
So
here
comes
the
video
deployment
that
I
was
talking
about.
That's
a
mojo
Red
Hat
one,
so
that
would
not
be
publicly
accessible.
A
A
So
I
love!
That's
why
we
do
these
things
so
that
we
can
get
get
feedback
from
our
colleague
on
what
we're
doing
wrong
so
anyone's
interest
in
this.
Please
reach
out
to
Annette
I,
think
in
that
you're
on
the
the
Google
group
for
the
operator
framework,
and
it's
not
big,
if
not,
will
make
you
get
there
today.
F
A
One
more
mailing
list
you
should
be
on
and
we'll
talk
to
everybody
again
on
the
11th
I'm
not
feel
like
you
say
we
limit
the
12th
of
November
I'll,
send
out
a
notice
and
again
please,
let's,
if
you
have
a
topic
like
Jeff,
that
we
want
to
talk
about
at
the
next
meeting
or
if
there's
a
topic
that
you
think
we
should
be
talking
about
or
an
agenda
item.
Please
add
it
to
the
project's
page
on
the
the
operator
framework
community
project
and
please
take
a
look
at
the
governance
document
over
the
next
interim.
A
As
my
guests
apologize
for
those
of
you
already
paid
and
are
coming,
but
we're
going
to
have
a
number
of
operator
that
are
used
in
machine
learning,
workloads
being
presented
and
the
panel
on
that
within
the
video
and
celadon
and
the
prophets
tour
and
a
whole
bunch
of
people
who
have
done
that.
So
that's
that'll
be
interesting
and
we'd
love
to
have
you
in
the
room
for
that
as
well.
So
that's
again,
that's
in
the
Bay
Area
San
Francisco
on
October
28th
and
said
they
saw
San
Francisco
Marriott
just
by
the
airport.