►
From YouTube: Ceph Orchestrator Meeting 2021-09-14
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
A
We're
gonna
start
with
that.
First,
I
don't
know
that's
what
I
wanna
hear
about.
A
Yeah
I
mean.
Were
you
able
to
reach
out
to
us.
B
Yeah
briefly,
at
least
the
I
mean
the
discussion
we
had
so
far.
Was
it
inviting
basically
inviting
them
to
be
a
storage
provider
of
rook?
Does
that
make
sense
for
them?
Are
they
interested?
So
they?
You
know
they
thought
about
it
for
a
few
days
like?
Oh,
that's,
an
interesting
idea,
you
know,
let's
think
about
it.
Then
they
came
back
and
said.
Well,
they
don't
really
see
that
as
something
they're
interested
in
right
now
and
it's
like
well,
okay,
I
mean
they
want
open
governance.
B
So
I
don't
think
it's
a
that
really
changes
our
perspective
on
wanting
to
use
topology.
It
was
just
it
would
have
made
things,
maybe
a
bit
more
streamlined
and
not
having
to
worry
about
cncf
yeah
but
yeah.
Currently,
the
plan
is
still
that
they
would
stay
separate,
but
the
yeah.
The
topic
I
still
need
to
bring
up
with
them
is
that
they
is
whether
we
could
get
to
dynamic
provisioning
for
raw
devices
instead
of
just
with
lbs
through
topo
lbm.
We
haven't
discussed
that
yet.
B
Yeah
we're
a
couple
of
people
on
the
team
ron
joseph
and
someone
else
are
diving
more
into
prototyping
with
top
lvm
and
trying
to
see
what
I
guess
wouldn't
have
a
better
understanding
of
how
things
all
fit
together
before
we
go
start
that
level
of
discussion.
Okay,
okay,
cool.
A
A
Yeah
I
mean
I
think
these
are
pretty.
These
are
easy
steps
to
take
to
disable
the
balancer.
I
think,
most
of
the
time
it
doesn't
matter
during
upgrades,
except
that,
if
the
upgrade
stalls
for
some
reason
like
if
you
hit
an
error,
one
of
the
demons
might
be
down,
and
so
it
might
make
sense,
then,
but
it's
easy
to
add
that.
A
I
don't
know,
maybe
we
can
make
it
a
configurable
option.
I'm
not
really
sure.
I
think
the
main
case
where
I
would
be
worried
about
is,
if
we're
doing
an
upgrade
and
then
we
hit
an
error,
and
so
the
upgrade
just
pauses
and
you
might
have
down
devices
and
then
then
you
hit
the
five
minute.
Otherwise
we
normally
restart
within
the
five
minute
issue.
A
C
Okay,
I
I
miss,
I
typically
the
other
ways
like.
If
a
user
has
a
full
lsd,
they
could
not
do
the
upgrade.
A
Yeah,
I
don't
know
I'm
ambivalent
on
it.
It's
easy
to
add.
It
would
be
easy
to
just
make
it
do
a
blanket
no
out
when
the
upgrade
starts
and
then
ops,
but
it's
also
easy
not
to
not
do
that.
So
I
think
I
don't
know
I
haven't
heard
any
users
complain.
I'd,
be
curious.
If
any
have
any
new
users.
A
A
Yeah
anyway,
okay,
all
mine's
lost.
Can
we
create
a
health
warning
if.
A
C
A
Yeah
I
mean
if
they're
like,
really
really
lost
like
you
lost
the
data.
Then
there's
this
like
really
painful
process
for
it.
Rebuilds
the
monitor
from
the
osds
that
you
can
take
releases
the
process
and
if
you
have
the
data
somewhere
but
like
you
can't
start
the
team
and
then
you
can
just
literally
copy
the
direct
introductory
somewhere
else
and
resurrect
monitor
without
too
much
trouble.
E
Hey
daniel
is,
is
you
the
the
one
that
that
interested
in
this
map,
who
is
the
the
one.
E
C
C
Yeah
I've
been,
I
figured
just
trying
to
I
don't
know.
I
haven't
personally
used
the
maintenance
mode
recently.
I
don't
know
if
maybe
it's
just
broken
but
and
they're
playing
in
case.
You
are
something
here,
but
probably.
E
D
A
All
right
all
right
recover
binding
to
all
interfaces
for
prometheus.
E
E
So
if
you
make
a
change
in
the
configuration
using
public
and
debate
networks,
that
is
not
going
to
be
there'll
be
address
that
the
manager
is
going
to
use
okay
as
far,
but
I
think
that
we
decide
to
change
the
liveness
pro
in
order
to
use
a
command
instead
of
an
http
request.
Okay,
in
order
to
continue
continuing
using
listening
in
all
interfaces.
Okay,
because
I
think
that
the
program
this
is
caused
because
a
problem
in
the
in
the
docker
implementation
is
the
true.
A
Yeah
I
mean
it
makes
sense.
I
think
the
thing
is
that
so
there's
there's
the
binding
behavior,
whether
you
bind
to
the
wild
card
or
the
manager's
current
ip.
But
even
if
you
bind
to
the
wildcard
address,
you
need
to
advertise
a
url
right.
E
B
I
think
also
a
better
possibility
for
rook
might
be
well
what's
the
right,
liveness
probe
for
the
manager.
Maybe
it's
not
the
end
point
where
we
need
to
figure
out
the
ip
I
mean
if
we
can't
just
bind
to
you,
know
0.0.0.0
or
you
know
basically
whatever
interface,
if
it's
a
specific
interface,
that's
where
we
have
a
problem,
but
if
there's
some
other
command,
we
can
run
for
the
lab
lightness
probe,
which
is
you
know?
B
E
Yeah
just
probably,
but
to
check
what
is
the
status
of
the
stator
okay,
we
have
a
response,
then,
what
the
manager
is
working.
A
A
But
the
it
doesn't,
let
you
configure
the
port.
E
A
B
C
B
C
That's
the
old,
then,
whatever
it
made
this
piano.
A
A
What
are
they
should
we
go
through
these
other
work
issues
travis
since
we're
on.
Oh.
B
A
B
A
A
Correct,
okay
sounds
good
yeah,
and
then
I
guess
in
the
in
the
documentation.
I
put
that
like
snippet
of
batch
in
there
like
there's.
Nothing
like
it
cursed
me
also
that
it
might
also
be
possible
just
to
rename
the
pool
like
if
there's
a
there's,
a
seo
that
lets
you
rename
radio
spools.
That
might
also
work.
A
A
Cool
okay
and
then
what's
this
last
tracker,
this
was.
E
E
That's
it
so,
in
order
to
configure
that
you
need
to
go
to
directly
to
the
host
where
the
the
container
is
running
and
go
into
the
the
container
and
make
the
configuration
that
path,
so
one
possibility
is
what
make
possible
to
to
access
this
configuration
directly
from
the
third
set.
I
think
that
this
is
easy.
A
Yeah
so
having
I
mean
we
could
just
what,
if
we
just
did
like
stuff
iscsi,
but
that
aptitude
cli
inside
you
know.
But
if
we
just
created
a
new
set
of
the
cli.
A
A
E
Taking
a
look
at
this
a
few
months
ago,
but
my
memory
is
a
disaster.
Okay,
we
can
ask
for
paul.