►
From YouTube: Ceph Orchestrator Meeting 2023-08-29
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/contribute
What is Ceph: https://ceph.io/en/discover/
A
A
Yeah,
so
we
want
to
talk
about
this.
Let
me
see
what
this
looks
like
all
right.
Some
actual
comments,
I
do
want
to
just
go
through
these
comments
and
stuff,
or
do
you
want
to
have
a
more
General?
C
I
mean
I
think
for
the
most
part,
these
are
just
kind
of
knits
they're,
not
really
too
important.
That
I
think
the
outstanding
question
I
had
is
like
just
the
packaging
structure
that
kind
of
discussion.
C
B
Present,
which
is
technically
not
the
case
inside
the
Zip
app,
but
when
we're
importing
it
for
the
tests,
I
I
I
get
your
point.
I.
Think
it's
a
good
point.
I
I
actually
would
prefer
to
take
the
shorter
context
than
stuff
idiom
context,
but
I
couldn't
come
up
with
anything
better.
That
was
very
unambiguous
about
the
new
package
name.
B
A
B
C
B
Have
things
used
as
normal
python
practices
as
I
can
and
be
unambiguous,
which
kind
of
ends
up
us
kind
of
having
to
be
very
explicit
and
verbose
in
places
right,
yep,
but
I
would
happily
take
the
other
suggestion
of
shortening
cephadium
context
to
context
that's
kind
of
the
way
I
originally
had
it
in
my
prototype
and
then
I
think.
While
I
was
talking
to
Adam,
we
we
made
it
longer,
but
I,
don't
know
what
are
your
thoughts
on
that
aspect?
Adam,
the
easier
part.
A
Like
just
this
part,
yeah
I
don't
really
care.
If
you
call
it
contact,
Vlogs
are
not
gonna,
be
any
sort
of
conflict
with
anything
else.
A
B
A
Like
we
have
a
big
object,
called
Step
ADM
context
that
we
have
to
cast.
So
let
me
do
that
same
the
class,
but
I
think
I'd
be
okay
with
the
shorter
name
for
that.
If
you
want
and
I
guess
for
this
other
part,
I
mean
I'm,
not
super
picky
on
what
neighborhood
it
sounded
like.
Maybe
maybe
it's
fbm
would
cause
us
to
like
move
things
around
or
whatever,
just
because
we
sell
this
fpm.fi.
C
A
B
Kind
of
it's
so
we
need
we
need
a
main
at
the
root
and
if
we
keep
the
dirt
we
end
up.
We
are
going
to
have
to
end
up
duplicating
it
or.
B
I
think
it's
doable,
but
I
think
it's
more
convoluted
if,
if
I
may
say
that
it's
a
good
idea
but
I,
don't
know
I
I
kind
of
hesitate
to
make
that
trade-off.
A
But
I
was
to
get
prepared,
I
just
mentioned
stuff
volume.
Randomly
is
I,
think
what
they
do
theirs.
They
have
a
dash
volume
here
and
then
they
have
pretty
much
all
the
actual
code
under
set
underscore
volume.
This
just
has
like
a
setup,
dot,
pi
and
some
talk
stuff,
and
then
they
have
a
subdirectory
where
they
have
everything
they
even
have
the
remained
up
high
in
here
it
looks
like.
B
We
can
certainly
move
to
that
model
again.
I'm
trying
to
take
baby
steps
tips
with
the
changes
to
Seth
ADM
py,
and
there
is
one
other
thing
I
just
thought
of
as
I
was
talking
moving
stuff.
80Mpy
will
further
make
back
porting
a
little
difficult
if
we
have
to
back
Port
things
fixes
I
mean
it's
not
a
big
deal
because
we're
already
making
a
lot
of
pain
in
this
area.
But
it's
it's.
You
know
another
droplet
of
churn.
There
I.
C
B
I,
don't
know
I,
I,
I,
I,
guess
I,
just
simply
don't
agree.
It's
it's
a
private
Library
I
mean
I.
I
did
say
something
to
Adam
like
we
should
probably
update
the
developer,
docs
or
add
some
sort
of
readme
to
explain
some
of
the
stuff
that
we're
doing
here,
maybe
not
in
this
initial
PR.
But
you
know
maybe
right
afterwards
have
a
separate
PR
for
the
docs,
explaining
the
new
the
new
structure
and
how
to
hack
on
it
in
the
future.
B
We
could
probably
just
put
a
note
there
that
you
know
this.
Is
the
library
force
f80m,
not
a
library,
to
use
f80m
externally.
A
B
A
random
thought
I
have
no
idea
if
this
will
work,
but
you
know:
there's
the
python
Convention
of
prefixing
things
with
an
underscore
I
mean
this
might
be
a
totally
goofy
looking,
but
what
if
the
directory
was
named
underscore
stuff?
Adm
I
mean
I
guess
the
other
issue
I
have
is
lib,
is
only
three
characters.
B
A
Yeah
I'm
personally
kind
of
a
bit
indifferent
to
it,
I
think
I
kind
of
wanted
the
files
to
be
more
organized
and
easier
to
sort
of
go
through
and
sort
of
the
naming
things
I
mean
I've,
always
been
sort
of
side
stuff,
so
I
was
going
to
just
say:
do
whatever
it's
easier
here,
I
feel
like,
even
if
we
did
all
this
moving
and
then
at
the
very
end
we
really
decided.
We
were
wrong
about
the
naming.
I
think
it
would
be
that
hard
to
then
change.
C
I
think
it
would
mostly
just
be
moving
everything
in
this
lib
directory
and
then
modifying
all
the
Imports
I
just
think
kind
of
movement
I
mean
I
kind
of
had
an
initial
thought
too,
like
oh,
this
is
just
a
lib
directory,
let's
call
it
that
which
obviously
doesn't
quite
work
out
with
the
python
package,
but
then
after
I
started
this
longer
I
realized
pretty
much
all
the
python
files
are
in
this
lib
directory,
except
for
the
main
function
file,
and
then
that
kind
of
felt
a
little
weird
it's
like
well,
why
aren't
they?
B
I
think
that's
the
way
to
move
eventually,
but
the
way
I
kind
of
see
it
is
a
lot
of
python
projects.
These
days
are
like
at
the
root
of
the
repo
we
there's
like
a
SRC
directory
or
the
name
or
let
me
put
it
in
chat
because
I'm
having
trouble
spitting
it
out.
B
Or
some
just
do
PKG
right
in
the
root
of
the
repo,
but
but
I
I
see
both
conventions
almost
as
equally
I.
Think
SRC
is
a
little
bit
more
common
these
days,
and
so
what
I'm
kind
of
envisioning
is
a
sef's
source.
Cephadium
is
equivalent
to
Source
in
this
model.
B
And,
yes,
we
have
this
Legacy
safadm.py,
but
eventually
it's
going
to
shrink
and
shrink
and
Shrink
until
we
can
probably,
instead
of
doing
a
moving
in
action,
they
generate
the
under
under
main
py,
with
like
a
stub
that
just
simply
points
to
something
inside
of
the
package
but
again
I'm
trying
to
keep
things
smaller
and
fewer
changes
in
the
in
the
initial
work.
I
see
that
as
kind
of
the
end
goal
as
a
parent,
as
opposed
to
something
I
want
to
do
up
front.
B
A
B
A
C
B
C
B
A
C
A
B
C
To
me,
that
makes
a
little
bit
more
sense
and
it
kind
of
fits
with
the
theme
of
this
PR
that
just
kind
of
take
care
of
that
structure.
All
in
one
go
instead
of
multiple
different
PRS
later.
C
B
C
C
B
And
that's
why,
like
again,
I
agree
with
you
I'm,
okay,
again,
I
I
picked
it,
so
you
can
kind
of
assume
that
I'm
I'm,
okay
with
keeping
Seth
80m,
live.
Even
if
it's
not
perfect,
it's
it's
three
more
characters.
On
top
of
the
name
of
the
thing
yeah
again,
if
it
was
starting
from
Day,
Zero
I,
totally
100
want
to
just
call
this
f80m
I.
Just
think
this
is
the
least
worst
option
at
the
moment.
A
C
And
I
don't
think
the
movement
of
the
main
movement
to
a
main
dot.
Pi
is
really
that
much
work
seems
like
that's
pretty
quick
to
do.
B
Yeah,
it's
not
I,
don't
think
it's
a
huge
amount
of
work.
I
agree
again,
it's
a
little
bit
more
churned,
but
it's
not
a
significant,
well,
more
churn.
So
again,
I
I
think
we're
I,
think
Mike
and
I
have
kind
of
like
staked
out
our
positions.
This
is
this
is
why
we
have
we
Quorum
right.
We
kind
of
need
a
tie
breaker
here.
It
would
be
better
if
he
had
more
people
to
be
like
I
have
an
opinion.
A
Yeah
and
I
probably
should
decide,
but
I
don't
know,
because
I've
kind
of
just
this
is
more
of
a
discussion
of
how
conforming
to
python
Norms.
C
A
B
Yeah
I
mean
if
you
want
I,
I,
wouldn't
re-base
everything
if
you
want
I
can
try
to
implement
the
changes
and
if
it
either
and
I'll
I'm,
definitely
going
to
give
you
a
fair
Shake.
If
I
do
it,
I'm
gonna
do
it
for
real
and
then
say
whether
you
know
I
can
I
can
I
can
pull
another
Branch
off
of
this
Branch
just
to
mess
around
with
that
that
aspect
of
the
naming,
I
wouldn't
repay
I,
wouldn't
have
a
clean
series
of
patches.
B
A
I
said
I'm
sort
of
different
to
it,
so
I
would
sort
of
lean
towards
you
could
leave
it
like
this
and
if
you
really
want
to
rename
it
at
the
end,
like
I
I,
just
really
care
about
getting
the
the
files
sort
of
split
up
and
organized
for
our
sake
and
then
what
it
looks
like
externally.
If
it's
like
a
library
or
package
or
whatever
to
me
that's
side
thing
so
I
would
leave
you,
because
it's
already
like
this,
we
could
do
it
like
this.
A
B
A
A
Then
we
can
do
this
one
like
this
context.
One
I
think
nobody's
gonna
have
an
issue
with
right.
I
haven't
related
to
the
other
two
comments
in
here:
yeah.
B
B
B
Yeah,
so
the
constants
that
it
was
looking
for
got
moved
to
a
different
file.
So
it's
a
combination
of
me
being
lazy.
So
I'm
glad
you
pointed
that
out,
because
it
definitely
needs
to
be
called
out,
but
I
also
kind
of
don't
think
it's
necessary,
so
I'm
kind
of
inclined
to
delete
the
line
in
its
own
commit,
of
course,
but
the
comment
was
meant
to
be
temporary,
but
I
forgot
about
it.
B
We
can.
We
can
talk
about
that
aspect.
I
think
the
idea
was
to
prevent
CEF
images
that
refer
to
Docker
IO,
but
I
think
it's
kind
of
a
misguided
thing,
because
we
can't
control
where
all
of
our
images
are
coming
from
I
mean
we
can
if
we
want
to
like
rebuild
everything,
but
if
someone
is
like
I
need
this
image
and
it's
on
Docker
IO
and
the
count
goes
up
we're
just
kind
of
fiddling
with
it's
just
make.
It's
like
make
work.
Every
time
you
make
a
change.
B
A
B
B
C
A
A
Yeah
I
guess
maybe
back
I
remember.
There
was
a
whole
thing
about
the
fully
qualified
names
and
in
the
short
names.
Maybe
this
was
around.
When
we
were
moving
from
Docker
to
Quay.
We
were
having
issues.
B
A
B
B
Yeah
and
for
what
it's
worth
podman
certain
versions,
you
might
be
able
to
configure
this
off,
try
to
push
you
away
from
using
unqualified
it's
a
security
thing,
because
if
you,
if,
if
some
say
a
popular
project,
starts
on
Quake
right
and
it's
like
Quay
IO
fubar,
someone
notices
this,
then
they
register
Docker,
IO,
fubar,
anything
that
checks
Docker
I
owe
first,
which
is
probably
most
things.
You've.
A
B
B
I
think
we
just
need
to
scrape
the
file
well.
The
interesting
thing
is
now
that
we
have
more
than
one
file
if
we
only
scrape
the
what
I
was
about
to
say,
which
is
scrape
constants
py,
but
then
people
can
start.
You
know,
you
could
say,
add
a
default
string
argument
to
some
random
function
that
refers
to
Docker.
I
o,
and
it
wouldn't
find
it.
But
I
don't
know
if
that's
what
you.
A
So
I
found
this
other
comment.
I
had
looks
like
I
actually
asked
about
this
yeah.
B
A
B
C
A
B
A
A
B
B
I,
this
is
probably
the
wrong
wrong
thing
to
say,
but
what,
if
we
converted
this
to
a
dedicated
Pi
test
test
like
if
we
had
a
file
under
like
the
tester?
That
said
something
like
test
underscore.
You
know
code
contents
and
in
there
was
a
test
that
actually
just
we'll
open
the
files
and
rent
them.
Instead
of
having
this
inside
of
a.
A
B
B
Yeah,
it's
at
the
very
least.
If
I
do
fix
it,
I'm
gonna
make
it
its
own
Rule
and
not
like
this
extra
thing
under
flake
yeah.
C
B
Give
that
a
whack
just
to
see
how
difficult
it
is
because
it
just
strikes
me
as
you
know,
you
have
to
do
the
whole
loudest
externals
thing
and
you
know
basically
say:
oh
I'm,
permitting
you
to
run
these
special
binaries
and
you
know,
and
then
I'm
thinking
about
bringing
find
and
X
args
into
the
mix.
It's
just
yeah.
A
B
In
this,
in
in
Source,
Pi
bind
it's
doing
it
with
a
bash
subshell,
so
I'm
assuming
bash
is
in
the
is
in
the
list
of
external
binaries,
which
is
basically
like.
Why
not
allow
everything?
Why
not
just
have
stars
I,
don't
even
know
if
it
have
supported,
but
yeah.
B
A
B
C
A
B
Yeah
I'll,
just
I'll,
just
literally
copy
and
paste
and
then
adapt
what's
going
on
inside
of
the
source,
Pi
bind
one
I
will
come
I
will
maybe
in
in
my
anger,
I'll
come
back
another
day
and
do
a
separate
PR
to
clean
them
both
up
and
make
them
less
wonky.
What
that
will
probably
require
touching
the
cmake
files
too.
B
Now
I'm,
like
motivated
to
change
I'll,
have
to
learn
the
C,
make
macro
language
in
order
to
bake
it.
So
I
never
have
to
do
this
again,
so
I,
it
can
actually
just
run
the
talks
coming
in
for
list
ends
and
just
run
whatever
the
default
ends
are
instead
of
having
to
explicitly
list
them
all
out.
Redundantly
in
the
CME
texts,.
A
A
Yeah
cause
we're
already
doing
a
big
cleaning
project
yeah.
All
right.
Do
you
want
to
talk
about
this
random
other
knit
as
well
over
here.
B
A
All
right
so
I
guess
we'll
you'll
directly
address
these
first
to
I.
Guess
part
of
this
one
and
we
might
hold
off
on
luciferium
limbs
f80m.
B
A
A
B
A
A
Yeah,
so
this
wouldn't
be
unprecedented.
I
think
this
you're
doing
this
to
allow
us
to
move
forward,
and
so
we
can
decide
the
naming
once
that's
the
last
problem
to
get
over
again.
A
B
A
Okay,
is
there
anything
else
you
wanted
to
go
over
with
us?
I
think
this
was
really
the
whole
thing
we
want
to
talk
about
today.
C
B
That's
right,
I
mean
I'm,
the
one
who
tried
is
trying
to
impose
it
on
the
team.
So,
yes,
I
try
to
use
it.
Whenever
I
can
I
I
think
what
happened
is
my
foray
into
the
world
of
go
where
there
is
a
language
level
canonical
tool
for
formatting
you
just
get
used
to
like
okay,
I,
don't
like
tabs
to
intent,
but
that's
what
go?
B
Format
does
and
I'm
just
like
fine,
whatever
go
format
run,
it
run
it
on
your
editor
at
save
time
or
whatever,
and
you
just
make
peace
with
the
fact
that
the
language
designers
had
a
particular
style
in
mind
and
and
black.
Well,
it's
not
at
the
level
of
go
format.
It's
not
like
a
python
core
tool.
It's
pretty
widely
accepted
in
the
community
at
this
point,
so.
A
B
A
B
A
C
B
Working
fine
there
there
is
one
subtle
thing
to
be:
I
think
is
worth
being
aware
of.
It
is
in
their
documentation,
but
there
are
places
in
our
code
where,
like
the
curly
braces,
are
expanded
onto
multiple
lines
and
the
black
documentation,
and
if
you
run
black
naively
it'll
compress
them
all
and
you
might
be
like.
Oh
that's
more
ugly.
This
is
one
thing
where
they
actually
say
that
we
we're
trying
to
give
you
the
choice,
but
you
implement
it
by
putting
training
commas
in
places.
B
So
if
you
put
trailing
commas
after
your
curly
braces
or
Square
braces
black
you're,
basically
telling
black
hey,
even
even
if
you
could
compress
this
all
onto
shorter
lines,
don't
again
I'll
I
can
try
and
find
the
docs
for
this.
But
it's
it's
good
to
be
aware
of,
because
if
you're
like,
oh
it's,
it's
kind
of
made
an
uglier
Choice.
It
was
better
before
and
it's
all
about,
like
an
expanded
dictionary,
just
sprinkle
a
cute
few
trailing
commas,
and
then
black
will
kind
of
do
what
you
want
to
do.
B
A
B
B
A
B
I
think
we
haven't
a
lot
of
the
code
that
it's
going
to
do.
A
real
number
on
it
tends
to
will
be
the
stuff.
That's
a
little
older
and
a
little
more
heavily
indented,
so
like
watch
out
when
it
comes
time
to
touch
some
of
the
like
bigger
functions
like
around
adoption
or
the
list
function,
but
I
personally
think
most
of
its
changes
are
sensible
and
the
ones
I
don't
like
I
have
just
made
my
piece
with
it's.
Just
it's
worth
it
to
get
the
standardized
formatting.
B
A
All
right
everything
else
want
to
go
on
about
this
or
is
just
need
to
address
the
things
we've
talked
about.
A
Sounds
like
we're
good
I
was
also
looking
at
the
right
button.
Let
me
stop
sharing.
A
I'm
looking
at
this
is
sort
of
watching
where
we
are
now
because
we're
gonna
feel
are
there
in
so
I
guess
we're
sort
of
at
the
very
beginning
of
this
phase.
Now
so
I
saw
that
okay.
A
It's
good
for
sharing
stuff
quickly
with
people
yeah
yeah.
It's
not
like
a
lots
of
actual
text,
editing
stuff
in
here,
yeah
I-
guess
we're
sort
of
here
now.
Yeah
like
we
have
these
other
tasks
we'll
get
to
afterwards,
but
I
think
we're
yeah.
B
So
this
is
basically
everything:
I
did
either
before
Adam
and
I
discussed
it
or
during
so
some
of
them
are
marked
as
like.
We
looked
at
them
and
decided.
We
couldn't
move
them
yet
stuff
like
the
firewall
and
the
assist
control.
A
B
Simple
moves
and
if
we
find
them
specifically,
not
agent
where's,
the
other
one,
the
ones
with
like
fetch,
underscore
I
get
a
lot
of
screen
corruption
when
you
scroll
so
I
I
think
it's.
B
They
yeah,
so
we
kind
of
assumed
it
would
be
harder,
but
when
I
skimmed
the
code
the
other
day
that
actually
might
be
easy
to
do,
but
again,
we've
kind
of
I.
Think
Adam
and
I
were
like
this.
This
initial
PR
is
good
enough
for
now,
because
it
gets
the
pattern
going.
Yeah.
B
A
B
It's
really
great
for
new
stuff,
anything
that
we're
adding
new
we
will
be
like
just
put
like.
Definitely,
the
rule
should
be
if
it's,
if
it's
not
clearly
under
one
of
the
existing
py
files,
just
make
a
new
file.
It's
it's
not
like
you
know,
sylvanians
pretty
big,
but
we're
not
going
to
have
thousands
of
files
we
might
have.
We
might
have
like
30
or
something
at
the
end
and
worst
case
scenario.
If
we
do
find
some
area
needs
like
a
subdirectory,
we
can
also
do
that
too.
B
Maybe
yeah
that'll
definitely
be
a
weekly
topic
for
there'll,
be
a
weekly
topic
for
what
am
I
trying
to
say
for
the
future
weekly
topic
for
the
future.
What
to
do
about
the
various
demon
name
classes
because
I
again,
I
discussed
this
briefly
with
Adam
in
the
office.
B
I
was
scribbling
on
a
whiteboard,
but
we
want
more
feedback
and
thoughts,
because
I
think
there's
like
two
directions
where
we
go
where
we
can
just
kind
of
leave
them
as
they
are
and
move
move
things
around
or
we
could
try
to
have
them
be
a
little
bit
more
uniform
because
right
now,
I,
don't
really
I
I
like
protocols
from
the
typing
protocols
Library
we
could.
We
could
look
at
that.
We
could
look
at
at
ABC's,
abstract
base
classes,
but
I
think
it
would
be
good
to
have
something
more
uniform.
B
So
when
the
methods
on
those
classes
are
used,
we
don't
have
to
do
this
if
ladder
thing
where
it's
like,
if
demon
type
equals
blah,
you
know
call
this
static
method
on
this
class.
If
you
should
be
doing
it,
like
you
know,
be
past
instance
call
specific
method.
That's
the
that's!
The
model
I
would
like
to
see
the
code
base
move
towards,
but
there's
different
roads
to
get
there.
B
A
A
All
right
for
this
week
is
there
anything
else
you
want
to
say
about
this.
I
stopped
sharing
again.
B
I
I
feel
the
need
I
feel
a
little
guilty
for
in
I'm
doing
air
quotes.
You
can't
see
me
winning
so
I'm,
not
not
changing
the
naming
at
the
moment,
but
I'll.
Just
repeat
what
I
said
earlier:
I
really
really
do
appreciate
the
feedback,
even
even
if
I
disagree.
B
B
I
kind
of
want
to
start
a
website
which
is
a
programmer's
thesaurus
kind
of
like
a
Wikipedia
type
thing
where
people
can
contribute
their
own
stuff,
but
it
would
be
specifically
for
programming
because,
like
I'll
hit
like
thesaurus.com
or
Merriam-Webster
or
whatever,
it
is
to
look
up
synonyms
and
it's
always
like
it's
like
normal
English
and
I'm,
like
normal
English,
isn't
programmer
English,
but
I
need
like
a
you
know.
B
I
I
want
to
call
something
a
container,
but
I
don't
want
to
see
container
because
in
the
context
we're
talking
about
that's
like
a
oci
or
Docker
container
I,
don't
want
to
call
it
a
bag
or
a
list
or
like
I
need
another
word
for
you
know.
So
this
is
an
idea.
I've
kind
of
jokingly
brought
up
on
and
off
for
years.
At
this
point,.
B
A
A
All
right
anyway,
I
think
we're
sort
of
talking
about
this.
Was
there
anything
else
you
wanted
to
briefly
touch
on?
We
only
have
about
10
minutes
anyway,.