►
From YouTube: Ceph Orchestrator 2023-02-14
Description
Join us weekly for the Ceph Orchestrator meeting: https://ceph.io/en/community/meetups
Ceph website: https://ceph.io
Ceph blog: https://ceph.io/en/news/blog/
Contribute to Ceph: https://ceph.io/en/developers/contrib...
What is Ceph: https://ceph.io/en/discover/
A
I
was
looking
at
redo
stuff
yeah.
The
first
issue,
I
think,
has
already
been
fixed.
Redeploying
rgw
in
the
circus
fix
I
think
there
was
like
three
different
trackers
for
this
okay.
B
Performance
worth
what
I
had
originally
said
on
mute
is
I'm
more
interested
in
the
two
issues
under
the
second
bullet
point:
the
not
the
automatic
redeploy.
So
if
you
know,
if
you
don't
mind-
and
we
could
start
with
the
some
items
to
review,
slash,
discuss,
I'd
appreciate
it,
because
I
gotta
leave
at
the
half
hour
mark
yeah.
A
B
Yeah,
it
looks
like
it
hasn't,
been
updated
in
a
long
time
unless
that's
the
wrong
image.
So
the.
A
Arc
TS
won,
we
don't
use
the
one
on
dark,
Hub,
anymore,
I
still
say.
B
Right,
okay,
so
off
of
the
link
tracker,
it
links
to
equate.io
one
yes,
I'll,
link
it
in
this.
A
Channel,
probably
the
one
it
should
be
on
the
queio
Steph
yeah.
It
is,
and
we
have
there's
this
thing.
A
B
I
think
so
again,
I
don't
know
if
that's
the
one
we
use.
That's
the
one,
that's
linked
from
the
tracker
in
question,
my
goodness
I
can't
open
the
right
tab
to
save.
A
B
Yeah
so
I
think
the
issue
is
possibly
that
it
needs
to
be
run
more
regularly
either
through
GitHub
or
Kio
Auto
build
because
I
think
what
the
the
Dimitri
seven
I
was
complaining
about.
Is
that
it's
way
out
of
date
and
it's
got
the
vulnerabilities
and
stuff-
and
it's
that's
definitely
the
case
of
the
one
I
link
to
they
may
not
be
like
the
most
major
issues
in
the
world,
but
Quay
is
detecting
14
High
vulnerabilities,
the.
A
Varies
definitely
should
get
on
one
of
the
2.2.
It
was
I
think
there
was
a
pull
request
that
W7
we're
going
to
do
this,
for
the
guy
was
trying
to
update
it.
Actually,
if
you
can
find
that
real
quick.
B
Yeah,
just
I,
don't
know
how
this
repo
or
how
this
this
container
repo
got
populated
with
Gail
pushed
it
manually
or
whatever
it
might
be.
Worthwhile
setting
up
some
kind
of
like
monthly
Auto,
build
just
to
keep
the
base
image
up
to
date.
I
have
to
mess
with.
A
B
A
B
A
Sort
of
remember,
when
they
got
made
it
was,
we
were
having
trouble
with
the
rate
limiting
stuff
from
docker,
and
so
we
just
made
a
keep
alive,
V
image
and
put
it
on
there
and
I.
Guess:
we've
just
sort
of
used
that
one
for
a
long
time
and
we've
never
updated
any
of
them
I
guess
we
need
to
sort
of
track.
What
other
things
we're
pushing
that
are
into
the
Seth
image
a
bit
better
in.
B
B
B
B
Definitely
definitely
would
be
a
good
start,
even
if
it
just
another
manual
push.
But
my
thought
is
it's
just
we're
just
going
to
get
into
the
situation
again.
You
know
six
months
down
the
line
from
there.
It's
like,
oh
you
know,
this
thing
is
not
up
to
date.
Oh
well,
it's
because
it's
a
manual
process,
the
manual
process,
gonna.
A
Yeah,
so
we
have
a
few
of
them.
We
have
to
update
I,
guess
AJ
proxy
here
from
that
PR
people
I've
needs
to
I.
Guess
all
our
whole
Ingress
thing
needs
it.
I
can't
remember
if
there's
any
other,
that
we
push
ourselves
I,
don't
think
so.
I
think
we
just
use
the
Upstream
ones
for
like
grafana
and
Docker
I.
Don't
remember,
yeah.
A
I'll
have
to
talk
about
it
with
him
he's
not
here,
unfortunately,
so
he
can't
get
super
far
on
this,
but
yeah.
A
Yeah
I'm
looking
yeah,
we
use
other
people's
stuff
for
pretty
much
all
the
other
ones.
It's
like
it's
just
as
we
push
our
own
grafana
h,
a
proxy
and
keep
alive
but
I.
Don't
think
I
think
some
someone
else
I
think
the
dashboard
team
handles
the
grafana
one
I'm
sure
they
do
that.
A
Okay,
I
think
it's
really
mostly
just
the
these
Ingress
ones.
You
have
to
worry
about
right
now,.
A
I'll
bring
it
up,
I,
don't
know
how
much
time
he
has
right
now
to
dedicate
to
this,
but
we'll
see
about
it
or
maybe
you
can
just
tell
me
how
to
build
or
something
I
think
I.
Think
I
think
I
need
permissions
to
push
to
this.
This
repo.
Do
you
used
to
work
on
the
stuck
container
stuff
a
lot
so
I
think
he
has
a
bunch
of
permission
that
I
don't
have.
B
A
A
I,
don't
have
to
rebuild
it,
I
guess
every
once
in
a
while.
Even
if
it
was
super
uncommon
like
once
a
month
or
something.
A
B
That's
fine
I,
I
I'm,
no
worries
I.
Just
I
didn't
realize
you
when
you
said
IPv6
I
realized
you
but
I
didn't
realize.
B
B
A
I
had
a
while
back,
but
essentially
I
got
denied.
Basically
like
we
don't
we
can't
you
don't
have
the
resources
to
do
that
right
now,.
B
A
A
Galloway
once
back,
when
he
was
running
stuff,
he
wasn't
able
to
do
it.
At
least
at
that
time.
I
could
ask
again
no.
B
Yeah,
like
the
the
whole
Hardware,
gets
installed
in
the
lab,
so
I
I,
guess
my
silly
wonder,
and
possibly
because
I
haven't
used
enough
IPv6
Hands-On,
but
if
it
was
just
local
networking
like
we
weren't
asking
to
get
IPv6
routed,
addresses
I
wonder
if
he
could
start
with
something
simple.
Just
having
like
like
the
moms
and
the
manager
and
everything
talking
to
each
other,
I,
don't
know
it,
it
might
be
challenging.
I'm,
just
I
mean.
B
A
Progress
asking
about
stuff
video
in
particular
right
because
we
don't
test
any
of
that
stuff.
Obviously
we
care
a
lot
about
SSH
and
all
that,
no,
that
part
to
work
I
have
never
had
a
chance
to
test
it
anywhere.
Unfortunately,.
C
Yeah
I
remember
fixing
some
backs
last
year
related
to
that,
but
I
don't
remember
like
if
there's
any
kind
of
special
setup.
A
Yeah,
basically,
right
now,
our
yeah,
so
we
don't
have
anything
really
for
it.
I
think
we
do
try
to
support
it
like
if
people
bring
up
stuff
in
particular
is
broken,
we'll
try
to
fix
it,
but
it's
pretty
much
only
tested
through
unit
tests
right
now,
because
we
don't
have
any
lab
stuff
for
it.
C
B
Yeah
I,
don't
I,
don't
think
it
makes
sense
to
me
like
if
you're
gonna
set
it
up
on
one
see
you're
gonna
set
up
on
one
and
not
the
other,
but
I
believe
with
the
dual
stack
question
means
is
you
know
current,
you
know,
I
have
an
interface
I,
have
both
an
IPv6
address
on
an
ipv4
address
on
it.
What
happened.
A
A
We
just
I'm
trying
to
think
what
do
you
even
use
those
IPS
for
so
I
know
for
the
only
thing
that
we
always
sh
to
the
host,
and
then
we
like
set
up
like
monitoring
on
fake
stuff.
Usually
this
is
the
yeah.
The
use
cases
for
and
I
feel
like
for
the
context
stuff.
A
We
do
have
a
bunch
of
testing
or
not
proper,
like
unit
testing
around
that
those
get
set
up
properly,
but
we
don't
have
anything
to
test
like
Shing
the
nodes
yeah
with
IPv6
dresses
at
all,
which
that's
the
really
important
one,
because
if
that
doesn't
work,
nothing
works,
but
it's
kind
of
hard
because,
like
what
would
I
I
don't
know
without
again
without
Labs
distance
to
test
it
on
there's,
nothing
really
be
done,
like
I,
don't
think
a
unit
test
if
you
have
any
use
for
that
yeah
like
I.
B
C
So
the
first
question
in
the
truck
is
travel
by
providing
both
like
pv4
and
pv6.
The
trip
this
setup
makes
sense.
They
have
always
seemed
like
it's
only
using
one
version.
Let's
mix
all
of
them,
so
I'm,
not
sure
if
you
can
have
like
monitors
running
on
different
IPS
with
a
mix
of
ips
I'm,
not
sure,
if
that's
that
makes
sense.
A
I
know
there's
some
weird
conflict
options
about
this
I
want
to
say,
like
they
try
to
make
that
work,
but
I.
Remember
there
being
some
problems
with
it,
like
somebody
had
like
every
reason,
I
think
I've
heard
about
it
was
because
somebody
tried
doing
it
and
think
something
didn't
work
or
whatever
like
you're
printing
on
like
IPv6
disabled
ipv4.
It
wasn't
supposed
to
or
yeah
I
think
they're
attempting
to
make
that
work,
but
I
don't
know
if
it
actually
does.
A
C
A
Yeah
pretty
much
all
the
IPv6
bugs
we
see
come
from
Upstream
trackers
from
Upstream
users,
I,
don't
ever
see
like
Downstream
stuff
through
IPv6.
A
Oh
I,
don't
know
that's
hard.
Baseball
I
can
maybe
bring
it
up
again
if
the
network,
if
the
lab,
is
ever
stable
for
like
more
than
a
couple
weeks
at
a
time,
I'll
bring
it
up
again.
Yeah
we've
had
problems
the
ad,
probably
just
the
networking
stuff
General
last
week,
so
I'll
tell
you
some
networking
software
and
everything
broke
for
a
while.
That
was
not
trying
to
do
anything
new
like
that
was
just
like
updating
the
software.
A
C
Seems
like
somebody
from
qae
was
testing.
This
combination,
maybe
there's
some
request
for
for
that
support,
because
kind
of
combinations
is
trying
are.
B
A
Yeah
yeah,
we
can
kind
of
a
little
repo
on
there.
Just
have
our
own
little
OneNote
cluster,
that
that's
whatever
yeah
yeah,
but
even
then
I
mean
I,
can't
I,
don't
know,
I
wouldn't
build
a
document
that
we
can
support
it.
Yet
so
we
get
like
a
real,
proper
thing,
so
yeah
I
could
look
into
that.
But
I,
don't
know
what
the
I
guess
I.
If
there,
if
it
was
broken,
I
get
a
lot
out
of
it
because
then
then
we
know
it
won't
even
work
that
far
it
worked.
I
wouldn't
get
much.
C
Well,
especially
with
IPv6,
we
have
seen
couple
of
places
where
it's
not
supported
properly,
so.
A
C
That's
for
English
there
are
some
Upstream
issues
related
like
that's
IPv6,
Ingress
I
saw
some
issues
so.
A
We're
going
to
try
to
fix
those
at
least
those.
Maybe
we
can
get
the
Upstream
person
to
try
them
once
they're
they're
fixed
if
they
come
back,
but
you
know
I
could
also
maybe
manually
test
those
a
little
bit
might
be
some
way
to
do
that.
But
at
least
I
can't
have
like
the
automated
testing
that
we
do
like
the
proper
one
over
that
yet,
but.
B
A
At
least
for
now,
but
we
can
about
to
revisit
it
I
think
and
once
the
lab
has
been
more
stable,
ask
about
it,
but
I
think
what
I
remember
them
saying
before.
A
Well,
you
said
one
effect
there
guys
it
seem
like
they
were
saying
before
the
lab
was
like
really
full.
They
had
no
room
for
other
machines,
but
that
doesn't
do
with
networking
I
think.
B
That
makes
no
sense,
because
just
this
is
more
of
a
software
thing
than
yeah
say.
A
A
B
And,
like
I
said,
we
should
definitely
be
willing
to
volunteer
to
do
some
of
the
the
lifting,
because
it's
mostly
an
install
time
thing,
except
for,
like
the
router.
B
A
Anyway,
while
you
weren't
here,
we
talked
about
the
first
two
trackers.
C
Yeah,
the
other
one
they
just
put
it
because
since
we
are,
we
can
keep
Lefty
just
to
receive
the.
A
Earlier
keep
alive,
I'm
gonna:
ask
you
about
it
a
little
bit
because
I
think
you
need
permission
to
push
to
that
repo
or
we're
putting
that
image.
I,
don't
think
I
have
them.
So.
Thank
you,
because
you
work
with
stuff
container
before
I
think
you
can
do
it.
I'll
talk
to
a
little
bit,
but.
C
C
A
A
2.1.5,
it
was
push
a
little
over
a
year
ago,
February
of
last
year.
A
No,
it's
not
two
to
Oda
five,
like
the
tracker's
talking
about
we're
a
little
bit
more
update
than
that,
but
we
are
still
not
on
a
2.2.x
which
yeah
is
the
more
recent
ones.
C
Nine
links
in
the
chat
what
yeah
it
was
just
you
remember
somebody
else
supposed
to
asking
to
update
the
PHD
proxy.
A
C
Exactly
so,
maybe
if
we
will
be
upgrading
the
each
year
proxy,
then
maybe
we
can
just
update
to
keep
our
lives
yet
sometimes
so
we
do
the
testing
all
the
ones
for
both
images.
First.
A
We
need
images
pushed
on
there.
Then
we
can
run
a
test
that
actually
tries
updating
what
the
images
point
to
in
our
in
our
code.
So
the
first
thing,
I
guess,
is
just
getting
the
image
up.
Then
we
can
do
its
actual
testing
on
them.
After
that's
what
we
did
last
time,
we
first
put
these
up
there.
They
used
to
point
to
the
docker
images
and
then
we
move
the
quickly
push
the
Quay
ones
up,
and
then
you
had
like
a
pull
requests.
A
I
tried
changing
both
of
the
defaults
and
then
that
got
pushed
through
after
they
got
tested.
A
Yeah
yeah
I'll
have
to
talk
about
at
least
help
with
getting
the
image
up
itself,
and
then
we
can
the
power
quest
to
fix
the
set
the
image
after
a
super
trivial.
So
let
me
try
it
after,
but.
B
C
What
yeah
I
mean
for
each
epoxy?
At
least
they
have
this
long-term
supported
image,
not
sure
if
people
empty
to
have
something
similar.
C
A
Now
to
check,
we
look
at
that
one.
If
that
one
has
something
like
that
at
least
the
very
least
you
want
to
get
into
some
2.2.
something
image,
and
we
can
check
after
that.
If
there's
one
that's
properly
supported
yeah
was
that
also
that
first
tracker
I
think
is
already
fixed.
The
thing
getting
redeployed
service
changes,
I
was
gonna,
have
to
go,
find
the
performance
I
think
fix
it,
but
I
think
it
got
fixed,
at
least
in
Maine
a
while
back.
Oh
okay,
thank
you.
A
We
I
think
the
change
what
the
change
did
was.
If
you
change
the
spec,
you
reapply
a
spec
when
we
save
it,
we
we
trigger
off,
like
it's
got
like
a
service
action.
The
service
action
is
what
handle
this
stuff
and
so
internet
is
a
service.
Action
will
get
affected
by
the
change
in
this.
A
Not
every
not
everything.
Actually
we
do
is
a
service
action,
so
not
everything
but
like
say
like
the
setting
their
stuff
for
the
spec
or
putting
the
cert
into
the
config
key
store,
which
we
do
for
rgw.
That's
a
service
action,
so
that
will
get
redone,
but
that's
look
what
everything
like,
if
you
guys
think
in
particular
that
you
think
should
be
done.
A
C
A
C
This
big
option-
oh
I,
see
I,
see
okay,
yeah.
A
C
I
was
here
thinking
not
like,
not
in
other
services,
depending
on
that
spec,
but
like
for
the
main
Service,
the
owner
of
the
spec,
do
do
we
redeploy
the
service
whenever
any
changes
done
on
on
this
big
file?
Or
it's
not.
C
A
I
mean
we
saw
we've
done
on
a
case-by-case
basis,
because
there's
a
certain
cases
where
I
don't
want
to,
like
you,
have
some
huge
OSB
spec
that
like
has
made
like
a
gender
osds,
and
you
change
like
one
thing
to
also
like
say.
It
also
includes
this
other
disk
or
something
I
don't
want
to
redeplay
all
of
them
because
it
doesn't,
it
doesn't
do
anything
like
they're
going
to
be
the
same.
C
A
A
And
then
again
we
have
the
service
action
thing
which
that's
anything,
that's
not
like
deployment
team
in
itself
that
we
have
to
do
with
service
wise
service
wide.
That
depends
on
the
spec
we
we
can
put
stuff
in
there.
Those
are
also
a
thing,
so
each
like
service
type
has
a
service
action
and
you
pass
it
like
a
spec
file,
whatever
spec,
you're
doing
and
it'll
do
something
based
on
that
spec
for
RDW.
A
It
sets
the
that
sells
cert
into
the
conflict,
key
store,
and
so
the
pro
one
of
the
I
when
I
originally
had
fixed
but
I
think
is
pretty
much
the
same
issue
as
this
earlier.
It
was
just
somebody
had
changed
the
rgw
spec
and
then
the
service
action
wasn't
getting
re-triggered,
so
it
never
got
set
properly
in
the
content
key
store
on,
because
we
only
were
doing
service
actions
whenever
we
deployed
a
new
demon,
so
I
just
changed
it
in
that
pull
request
to
do
it.
A
Whenever
you
reapplied
the
spec,
so
yeah
is
there
anything
we
want
to
do.
That's
like
service
wise.
Instead
of
specific
to
an
individual
demon,
we
can
put
it
there
and
things
for
a
specific
demon
that
should
belong
as
dependencies
for
the
demon
type.
We
can
use
the
spec
as
part
of
setting
up
those
dependencies
like
we
do
that
for
iSCSI.
A
Does
that
as
a
trusted
IP
list,
if
that
gets
modified,
that's
considered
a
dependency
for
iSCSI
and
so
it'll
redeploy
the
iSCSI
demons
to
trusted
IP
list
make
sure
they
all
have
because
they
have
to
have
identical
trusted
IP
lists
in
order
to
work
properly,
so
it
really
deploys
all
the
iscosity
demons
for
that
service
effect
yeah.
We.
A
Want
to
find
a
requests
that
did
it
see
if
at
least
link
it
to
that
track.
The
way
there's
not
even
a
duplicate
tracker
linked
on
there,
that's
a
verified
actually
did
what
I
think
it
did
as
well.
Let
me
see
make
sure
it's
actually
closed
too
I.
A
A
All
right,
the
other
thing
I
had
in
here
for
today
was
I
linked,
there's
an
out
of
date,
compatibility
dock.
A
B
A
Last
text
into
one
and
some
guy
was
asking
that
on
the
mailing
list,
but
I
think
the
thing
is,
we
haven't
ever
updated
it
because
things
have
sort
of
been
okay.
Since.
A
We
got
to
I,
don't
know,
pass
past
those
three
I
guess
after
Batman
three
I,
don't
think,
there's
been
anything
that
I
remember
broke
anything
so
he
never
bothered
updating
the
dock,
like
16
2-1
onwards,
with
anything
above
3.0
has
seemed
to
have
worked.
Okay,
so
I'm
wondering
maybe
we
just
want
to
add
in
like
a
make.
They
changed
that
from
like
through
that
from
3.0
to
3.x
and
then
have
another
thing.
That's
like
four
dots,
two
or
something
and
see
like
that.
One
also
works
I!
A
Don't
know
you
mean
this
one
foreign,
yes,
I,
I
like
that
in
the
other
pad.
Okay,
oh
yeah,
it's
exactly
that
that
table
right.
There
is
what
I'm
talking
about.
So
we
used
to
have
problems
with
it,
as
you
see
from
all
like
the
true
false
stuff
stages,
but
I
think
since
3.0
in
the
later
Pacific
versions,
I,
don't
remember
there
being
anything,
that's
broken
I
think
there
was
only
there
was
that
host
of
containers,
internal
thing.
A
That
was
a
problem,
but
that
wasn't
breaking
everything
that
was
like
a
problem
like
specifically
ice
because
he
and
so
I,
don't
think
I
would
put
in
here.
Like
oh,
it's
incompatible.
A
A
Yeah,
so
that's
what
I,
some
guys,
that's
not
on
the
email
list
ends
up
responding
to
them.
Saying,
like
I,
think
you
should
be
fine
to
use
like
4.x
the
video
we
don't
know
of
any
problems.
So
I
was
trying
to
think
of
what's
a
good
way
because
I
don't
want
to
like
extend
the
table
infinitely
doing
like
oh
3.1
3.2
and
they
say
they're
all
just
say
like
true
or
whatever
release.
C
So,
in
fact,
with
any
Goodman
version,
higher
than
3.0
should
be
compatible
with
any
set
version
higher
than
16.2.1.
Is
that
true.
A
I
think
so
so
I
think
whatever
work
for
1621
should
also
work
with,
like
even
main
branch
right
now,
but
I'm
pretty
sure
it
is
just
anything
higher
than
3.0
works,
with
any
release.
Hiring
16
footer,
though,
like
that,
where
the
graph
ends
like
I
mean
it's
like
outside
of
like
Beyond
those
rows
and
columns,
and
that
that
charts
is
actually
working.
But
we
see
a
good
way
to
signify
that.
C
Maybe
we
can
like
modify
this
first
sentence
and
put
this
more
generic,
like
we
can
say,
like
any
Pokemon
version
higher
than
3.0
with
insta
version
higher
than
16.1
should
be
good
to
go.
Yeah.
C
Back
for
people,
especially
from
Upstream,
the
normally
use
latest
version
of
of
safe
and
for
and
of
Pokemon
yeah.
A
Like
I
know,
Downstream
create
I,
didn't
even
test
them
up
at
this
point,
users
like
4.2,
and
it's
been
testing
that
with
like
a
bunch
of
Quincy
versions
and
stuff,
even
Pacific.
They
have
any
problems
as
far
as
I
know,
but
I
might
modify
it
even
the
beginning,
part
to
say
like
in
the
early
likes
fpms
days
like
some.
There
was
a
lot
of
fantasy
issues
with
podman
like
nowadays,
but
in
this
release
onwards,
with
this
podman
onwards,
exactly
as
we
know
and
like
if
you
need
more
details.
A
A
Probably
okay
yeah,
while
I'm
on
this
paper,
if
you
look
at
the
bottom
of
the
page,
is
also
are
you
out
of
date,
I
think
I,
don't.
C
C
B
A
Ad
I,
don't
remember,
I,
think
because
I
don't
know
what
version
the
the
current
did
you
installed
you're
playing
but
I'm
pretty
sure
it's
a
Ford
or
something.
C
I
still
yeah
I
still
see
some
Upstream
trackers
that
fade
from
time
to
time,
and
they
seem
to
be
using
very
old
version
to
treat.
Some
notes
are
like
attacked
in
that
version
and
that
don't
get
upgrades
because
I.
A
The
bottom
of
that
page,
because
I
don't
think
that's
super
accurate
anymore
either
like
there's.
C
Like
a
60,
Amazon
or
development,
I
think
nowadays
it's
quite
stable.
So
maybe.
C
On
yeah
I
I
mean
yeah.
He
probably
did
this.
All
this
section
probably
doesn't
make
sense
anymore,
I
mean
with
stable
it's
being
used
in
production.
So
if
this
is
me
back,
you
can
just
put
some
link
to
the
safety
name
tracker.
Maybe
so
they
can
just
know
how
to
fill
the
bags
with
the
the
proper
fields
that
could
be
useful,
but
because
this
is
the
right
place
to
put
it.
A
Yeah,
like
I,
probably
want
to
leave
the
thing
at
the
bottom
about
pausing,
should
they
say
like
something:
that's
still
useful
information
if
something's
broken,
but
the
parts
about
what
types
are
incomplete
is
wrong.
That
needs
to
be
removed,
or
maybe
it's
a
general
sentence
like,
if
am,
is
relatively
stable
now
and
then
I
can
remove.
Most
of
these
things,
I
can
maybe
leave
like
this
fam
exporter
in
there,
because
that
one
is
very
new.
C
Yeah
I
think
you
can
just
like
remove
all
this
section,
because
Stephanie
is
or
just
put
like
stability,
and
it's
already,
you
know
been
used
in
production
and
emu
wishes.
A
That's
when
you've
we've
seen
user
production,
Downstream
and
then
I'm
sure
people
have
used
Quincy
a
little
bit
Upstream
at
this
point,
yeah.