►
From YouTube: 2020-03-16 :: Ceph Orchestration Meeting
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).
C
A
D
E
B
D
D
B
A
A
A
A
B
Yeah,
well
is
just
because
I'm
meeting
that
we
have
is
this
morning:
okay.
Well,
it
seems
that
with
both
of
k-staters,
the
Westie
in
the
nation
OST
remove
is
implemented.
What
not
nothing
wrong,
but
is
something
that
probably
we
should
well
return
to
to
life.
I
think
that
when
I
think
that
it's
good
to
have
the
possibility
okay,
but
there
is
a
couple
of
use
cases
for
OST
remover.
One
of
these
is
the
replacement
and
to
you
to
use
to
really
in
order
to
reduce
the
acetate,
Heidi
and
I.
B
A
So
as
I
understand
it,
if
we
can
reuse
the
OSD
ID
on
the
same
node,
when
we
replace
a
disc
that
there
will
be
less
data
movement
in
you
know
due
to
crush
and
and
so
it
would
optimize
data
movement
and
do
you
last
native
now
how
critical
that
is
of
a
scenario
it's
a
hard
thing
to
solve,
at
least
on
the
Brookside.
It's
not
clear
to
me
how
we
can
replace
that
OSD
on
the
you
know
and
make
sure
it's
on
the
same
note
and
things
use
the
same
ID.
A
E
B
If
the
disk
are
in
different
computers
in
different
notes,
that
is
not
going
to
be
possible,
but
the
operation
is
going
to
be
executed.
So
it's
something
that
what
the
the
user
can
try
to
do,
but
it's
something
that
what
is
internal
is
who
is
going
to
be
detected
if
it
is
possible
or
not
it's
clear.
Thank
you.
Thank
you
guys.
B
A
B
This
me,
so
these
are
because
well
well,
I
think
that
what
I
have
been
taking
the
project
Orchestrator
project
in
mind,
inter
in
the
theft,
racket,
okay
and
I-
think
that
it's
very
difficult
to
find
out
what
functionality.
What
features
are
implemented
in
what
moment?
Okay
and,
for
example,
to
know
why,
traditionally,
we
have
divided
the
functionality
between
tape
0,
they
want
they
to
okay,
in
order
to
the
things
that
the
user
can
can
do.
B
Ok,
this
is
something
that
is
difficult
to
well,
you.
You
need
to
go
feature
by
feature
trading
and
and
understanding
what
what
what
is
in
each
feature
in
order
to
sort
the
different
items,
so
I
think
that
it
would
be
it
would
be
nice
to
to
have
tax,
for
this
is
just
to
come
in
another
thing
that
I
think
that
is
important.
This
is
this.
A
B
What
we
need
to
do
improve
a
lot
of
things
here
in
the
project
area,
because
what
for
me,
I
have
trying
to
have
had
a
general
idea
about
aquatic.
What
is
the
current
state
and
it's
difficult
to
to
understand
what
things
are
done
and
what
what
percentage
of
of
the
of
the
feature
of
the
complete
functionality
is
done?
So
I
think
that
we
can.
But
it's
just
an
idea
and
I
will
try
to
implement
I.
Think
that
to
know.
B
And
a
matter
of
that,
it
would
be
nice
to
have
a
demo
software
functionality
I
think
that
well,
we
can
start
to
not
for
all
the
things
that
we
do,
but
I
think
that
is
important
if
we
finish
one
feature
when
it's
good
to
have
a
small
demo
of
the
of
the
feature
in
order
to
show
all
the
people.
Oh,
it
works,
ok
and
to
allow
people
to
to
make
suggestions,
and
this
is
sort
from
a
back
lot
of
things.
B
A
C
If
you
want
to
like
manually
intervene
if
you
want
to
move
and
monitor
from
one
node
to
the
other,
something
manually,
if
you
do
that,
while
it's
already
in
the
state
of
moving
manage
monitors,
amounts
and
they're
just
trying
to
move
it
right
back
so
a
way
to
like
pause,
the
cluster,
so
you
can
manually
move
things
around
I,
don't
know
if
you
want
to
get
comments.
Another
okay.
A
C
C
Not
sure
about
the
next
one
to
be
honest
at
armor,
abstractions
on
sue,
say
and
then
the
last
one,
the
stuff
ATM
shell,
the
works.
So
currently
we
need
place
F.
Am
it
just
pulls
the
staff
image,
whatever
the
latest
one
is?
You
can
manually
like
specify
a
specific
image
you
want,
but
then,
when
you
run
the
Ceph
ATM
shell
command
to
open
up
the
shell,
so
you
interact
with
stuff,
still
pulled
the
same
image
in
the
master
image,
a
stone
they
wanted
to
be
able.
C
C
We're
good
for
now
my
only
other
was
that
I
agree
with
long
ago
if
we
could
have
some
sort
of
bellow
or
JIRA
or
even
if
we
can
do
this
in
the
south
tracker
to
like
Joe
day,
one
zero
one,
two
three
and
like
separate
them
who's
working
on
what
like
what's
done,
what's
been
completed,
the
demos
thing,
all
that
would
be
super
helpful,
I
mean
I,
know
you
that's
not
really
like
I
know.
This
is
all
Savior.
B
A
F
B
F
Team
is
starting
to
test
OCS
on
bare
metal
machines.
Finally,
we've
got
through
a
lot
of
our
problems
with
those
CP
installs
and
so
forth,
and
starting
to
get
some
significant
scale
and
some
of
the
features
that
we
used
to
have
with
traditional
stuff
I'm,
not
seeing
yet
how
you
do
those
things
with
OCS,
they're,
no
and
so
I
want
to
get
your
opinion
on
it.
I've
got
a
couple
of
examples
of
that,
so,
for
example,
with
traditional
little
stuff,
you
typically
could
use
multiple
NIC
ports.
F
A
Good
question:
it's
eva's:
let's
try
it
out
mulpus,
but
the
but
right
now
I
work
in
OCS,
don't
sort
of
configure
it
natively
or
work
with
a
natively,
so
you'd
need
to.
You
know,
specify
a
set
comm
through
rook
to
pick
up
whatever
settings
you
want
for
that
private
or
public
network.
So
you
can
put
them
on
different
subnets
yeah.
F
D
F
D
Yeah
and
the
thing
is
that,
because
it's
not
implemented
in
this,
yes,
then,
even
if
you
change,
because
we
use
a
config
map
or
extra
set
config
options,
we
want
to
add,
but
because
OCS
will
reconcile
events
on
on
that
config
map.
If
you
updated
to
change
the
set
conflict
and
it
will
just
be
deleted
and
reset
to
the
previous
state,
so
it's
users.
Yes,
then
you
won't
be
able
to
enable
what
that
option
to
Network
segregation
will
host
it
working.
What.
F
F
A
F
F
A
A
D
C
D
F
My
understanding
is
that
and
correct
me
if
I'm
wrong,
that
when
they
paid
me
to
switch
in
OCS,
to
use
the
local
storage
operator
to
acquire
Hart,
basically
storage
hardware
and
there's
good
reasons
for
doing
that,
but
it
sort
of
gives
up
a
level
of
control
that
we
used
to
have
with
Brooke.
In
terms
of
you
could
set
up
the
underlying
devices,
you
know
with
certain
volumes
or
whatever,
and
it
would
all
be
there
and
you
could
just
plug
into
that.
F
A
You
configure
the
TV's.
We
basically
just
have
documentation
right
now
with
how
to
do
that,
set
it
up
on
the
devices
you
want
to
expose
and
then
tell
Brooke
how
many
of
them
you
want
to
consume
or
tell
us
yes,
and
then
it
passes
it
on
to
remember.
But,
yes,
you
lose
a
little
of
that,
the
specific
nature
of
being
able
to
define
exactly
which
devices
it's
setting
up
the
PV
is
in
advance.
That's
the
new
thing
right.
F
So
that's
one
area:
that's
another
area
where
I'm
gonna
try
to
see
if
we
can
kind
of
get
that
back,
because
I
think
both
of
the
things
I've
talked
about
so
far
I
think
are
really
critical
for
bare-metal
performance.
They
may
not
be
as
important
in
the
public
cloud,
but
if
you
want
to
have
a
variable
right
and
then
the
last
one
is
for
HDD,
we
always
used
to
use
SSD
devices
to
store
the
right
ahead.
F
D
D
You
will
have
to
pass
a
world
device
like
when
I
say
wall
device
you,
you
will
have
to
consume
the
entire
device
that
is
being
passed,
though,
because
that's
the
way
the
implementation
works
today.
Unless
you
have
a
really
small
device,
you
won't
be
able
to
to
get
just
a
small
amount,
just
a
piece
of
that
storage
and
use
it
as
a
backing
device
for
for
neo
SDS.
But
that's
that's
the
the
downside
at
this
point.
D
It's
like
we're
starting
everything
from
scratch
again,
so
before
we
get
there,
we
will
have
to
wait
for
a
month
and
I
know
it's
frustrating
because,
like
we,
we
are
just
back
to
a
simple
setup
like
ok
onedrive
for
is
D,
which
is
not
so
appealing
when
you
do
permit
old
and
when
you
bought
super
expensive
hardware
to
get
the
best
out
of
it.
So
right
now
it's
it
is
well
that's
how
things
are
I,
guess
well,.
D
They
are
definitely
definitely
it
is
really
something
we
want
to
to
fix
at
least
this
year,
I
believe
so
yeah.
We
are
not
letting
you
down
on
this
one
and
we
it
is
like
phase
2.
The
support
is,
there
is
just
that
the
implementation
has
a
bit
smaller,
I
guess
mm-hmm
yeah,
but
it's
we're
moving
toward
that
goal.
F
A
A
A
B
E
B
F
A
F
E
Maybe
the
developer
summit
on
next
week,
actually
so
think
you
have
seen
the
pipette.
We
need
to
schedule
a
session
for
the
orchestrator
plans
for
Pacific,
including
rook
and
self
ADM,
that's
going
to
be
on
March
25th
and
we
need
to
get
you
a
jingle.
A
blue
jeans
session
for
it
and
I
need
to
coordinate
it
with
lens,
because
we
have
two.
We
have
two
meetings,
one
for
the
despot
which
is
going
to
be
Emir
only
probably
and
one
for
the
August
fighters,
which
is
going
to
be
Amir
plus
us.
D
A
B
B
E
B
It
doesn't
matter
if
it
is
tax
or
another
method.
Okay,
the
important
thing
is
that
I
think
that
it
will
be
nice
to
have
the
different
features
group
it
in
some
way
enough
in
functionality,
groups,
for
example.
When
we
talk
about
orchestrators,
we
have
divided
the
functionality
traditionally
in
the
zero
this
one,
the
okay.
B
We
would
have
a
lot
of
features
in
this
moment
in
David,
mind,
depraved
the
theft
rocket
project,
okay,
but
it's
very
difficult
to
know
what
degree
of
completion
of
completion
we
have
in
each
of
these
groups.
Okay,
for
example,
yes,
I-
want
to
know
what
is
the
current
state
of
the
rtw
functionality.
I
need
to
go
feature
by
feature
button
buyback
in
order
to
to
have
a
complete
vision
of
the
current
state
for
this
facility.
Okay,.
E
E
Little
groovin,
but
actually
at
least
it's
the
first
way
to
actually
have
a
look
at
the
issues.
To
be
honest,
we
have
I
think
we
have
about
90
issues
at
overall
80
issues.
So
if
you
just
have
a
look
at
all
issues,
then
that's
that's
all
we
have
tracked
right
now,
so
I've
tried
to
incorporate
everything
there
poor
mentioned
to
Sage.
Last
week,
I've
tried
to
incorporate
everything
that
Patrick
mentioned
to
me.
Last
week,
I've
tried
to
incorporate
everything
that
I
just
covered,
that
that
we
have
discussed
in
the
in
eastern
apps.
E
B
E
B
B
B
Okay,
state,
an
orca
state
or
what?
What
is
the
functionality
that
we
need
in
all
these
things,
because,
for
example,
with
on
this,
is
part
of
the
of
the
other
things
that
I
think
that
we
need
in
the
interpret
aura
at
least
I
haven't
got
this
information
of
it.
For
example,
the
question
that
I
have
have
tries
if
I
want
to
know
what
is
the
things
that
when
we
need
to
implement
in
in
octopus
or
rtw,
there
is
any
way
to
know
this.
This.
E
B
Okay,
this
is
the
kind
of
things
that
I
think
that
can,
in
this
moment,
is
it's
very
fuzzy.
Okay,
we
have
functionality
and
lot
of
features
lot
of
packs.
But
what
is
the
things
that
we
really
need
for
octopus
in
this
area
of
features,
for
example,
RTW,
for
example,
file
system,
for
example
out?
Yet,
do
you
understand
what
what
I
mean
so
early?
We
should
define
this.
Is
this
target
okay
and
then
group
all
the
features
and
all
the
books
around
these
targets.
E
So
you're
right,
you're
right
the
one
one
problem
is,
of
course,
that
the
the
feature
set
of
the
orchestrator
interface
is
kind
of
defined
by
self
ATM.
At
this
point,
and
as
a
development,
this
is
really
on
going
it's
kind
of
hard
to
actually
define
what
needs
to
be
there
like
I,
can
go
over
the
interface
and
tell
you
what
is
the
feature
set
of
the
orchestrator
in
the
face.
I.
B
Is
nobody
with
a
clear
idea
of
what
we
need
to
do
product
to
boost
in
this
area
or
functionality?
This
is
what
I
mean.
Okay,
so
I
think
that
it
would
be
nice
well
to
have
a
meeting
even
a
set
of
meetings,
in
order
to
define
this
very
clearly
and
and
then
try
to
organize
the
project
around
this
areas
of
functionality
and
the
target.
B
I
think
that
they
should
be
that
if
I
want
to
know
what
is
the
current
status
in
for
RDW,
we
have,
we
should
have
a
percentage
of
50
percent
or
twenty
bucks
pending
and
three
four
features
depending
okay
is
this
kind
of
things
there
is
actually.
E
E
This
is,
of
course,
just
a
check
mark
right,
so
you
you
can.
What
what
we
need
to
do
is
actually
make
it
much
more
detailed
like
creating
our
JW's
updating
our
JW's
supporting
multiple
zones,
multi-zone
deployments,
though
so
each
feature
like
mons
and
stuff
like
that
needs
to
be
spread
out
into
the
whole
set
of
functionality
that
is
exposed
by
the
orchestrator
interface,
but
you're
right
right
now
we
don't
have
it
yes,.
B
Yes,
okay,
let's,
let's
start
to
do
to
do
that?
Okay,
it's
not
it's
just
to
see
how
we
can
improve
this,
and
if
it
did
it,
if
this
has
sense,
okay
is
something
that
I
have
found
when
I
have
started
to
try
to
to
see
the
project.
The
whole
project,
okay,
I,
think
that
it
could
be
good
for
everybody.
Okay,
no
I,
I
I
can
start
to
propose
things
and
to
suggesting
things,
and,
of
course,
if
everybody's
interested
in
that
we
can
participate
of
the
work,
but.
B
This
should
be
completely
link
it
with
the
project.
Okay,
because
is
something
that
I
think
that
we
need
in
order
to
to
to
realize
in
what
point
of
compression
we
are
in
each
moment.
Okay,
well,
I,
think
that
is
important
to
know
if
we
are
far
away
or
very
near
of
having
one
fusion
additive
working
and
tested.