►
From YouTube: Ceph Orchestrator Meeting 2020-06-15
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
Okay,
I'm
hello,
everyone
to
today's
pockets
with
a
we
click
on
the
agenda
for
today
is
quite
small.
Actually,
so
we
have,
we
had
a
very
slow
week
and
really
not
much
to
report,
though
we
have
a
new
backpack
or
request
for
Sofia
Morgan.
We
have
a
new
way
to
to
easily
add
new
tautology
tests
to
save
idiom.
I
hope
that's
going
to
be
used
in
the
future,
thanks
to
Russia
or
a
neighbor
that
we
have
I'm
working
on
the
ETCs.
A
If
that
selfish
left
at
conf,
which
is
seems
to
be
an
important
feature
to
main
manage.
Let's
pile
on
the
horse,
though
safe
IDM
is
really
in
the
rook
in
that
regard.
Panzer
and
that's
all
I
have
for
safe
IDM,
and
we
don't
really
have
my
dad
much
much
of
a
on
the
list.
Regarding
the
topics
except
drive
groups
breath
group
specifications,
I.
B
B
So
if
a
drive
group
comes
with
information
about
what
hosts
the
DRI
groups
apply
to,
and
then
also
of
course,
what
devices
on
each
node
that
they
that
need
to
be
configured
with
those
Dedes,
so
the
so
the
so
rook
only
has
drive
groups
at
the
post
level
and
then
select
as
far
as
which
nodes
are
selected.
We
want
to
use
the
kubernetes
placement
mechanisms
like
node
affinity
for
driving
that
so
any
any
node
selection
will
be
done
with
the
koreans
placement.
B
That
will
allow
work
to
do
the
node
selection
and
configure
that
OS
DS
and
then
will
rook
will
basically
pass
on
the
drive
your
spec
for
devices
in
to
set
volume.
So
that's
that
volume
can
create
those
teams
based
on
the
DRI
groups,
specs
for
devices
and
absolutely
put
a
link
there
to
the
PR
that
you
have.
As
far
as
that,
translation
to
placement
and
I
know
any
questions
or
its
we're
all
on
the
same
page
about
that.
B
A
That
kind
of
accumulated
eye
placement
specification
with
required
during
scheduling,
ignored
during
execution
and
stuff
like
that
and
I
just
didn't
know
what
do
you
wanna
have?
But
what
do
you,
which
kind
of
placements
back,
do
you
actually
want?
Do
you
want
some
kind
of
part
of
Finity
or
I
was
a
bit
unsure
sure.
B
Well,
if
you
don't
specify
any
of
any
placement,
we
would
just
apply
the
you
know
that
drive
your
spec
to
all
nodes,
so
all
nodes
would
get
those
devices
configured
if
they're
found,
and
then
the
placement,
you
add,
would
just
cause
rook
and
terminate
to
add
a
filter
on
what
nodes
those
devices
are
configured
on.
So
if
you,
you
know,
you
want
to
add
a
list
of
nodes.
You
know
there's
one
type
of
placement
for
that.
A
A
Okay,
davell
environment,
for
safe
with
safe
ATM,
so
we
have
three
options:
the
way
to
go
now
right
we
have
using.
We
have
the
option
to
use
restart,
f,
ATM
that
will
spawn
a
local
cluster,
that's
not
deployed
using
containers
and
then
the
ability
to
spawn
new
containers
on
any
horse,
and
we
have
safe
area
in
bootstrap.
That's
one
that
jamika
added
a
few
days
ago.
That
is
a
bit
more
related
to
a
real
cluster,
but
it
mounds.
A
C
But
I
think
they
all
serve
different
purposes,
which
is
bad,
but
though
I
did
this
this
patch,
because
I
use
we
start
mainly
for
cafe
diem
development,
and
that's
that's
completely
fine.
So
the
workflow
for
this
works
pretty
good,
so
I,
the
Napoli,
so
truster
and
then
I,
add
so
I
remove
my
localhost
from
being
a
Sofia,
yeah
member
and
then
I.
Add
my
two
VMs
VMs
that
I
spawn
the
the
containers
that
I
spawn
our
spot
inside
IBM
fine
can
be
cleaned
up,
it
can
be.
C
Can
you
snap
shows
is
that
everything
and
stuff,
and
if
I
do
clone
changes,
the
manager
will
only
run
on
my
local
machine,
so
it
will
immediately
pick
up
the
code
changes
that
I
do
in
my
local
machine,
which
is
also
fine
that,
however,
it
doesn't
work
very
good
if
you
want
to
change
binaries
or
containers
that
are
running
into
binaries
in
in
your
VM.
If
you
want,
for
example,
changed
SFO
is
decoder
or
something
for
that
we
would
need
to
rebuild
the
containers.
Oh
I
think
there
is
no
end-all,
be-all
solution.
C
So
the
patch
that
that
you
linked
in
the
ether
pad
is
basically
just
told
I
always
did
two
commands
when
starting
the
V
star
cluster
and
then
well,
basically
removing
my
local
host
from
the
list
of
posts,
and
that's
just
you
know
possible
into
in
one
command
and
a
bit
more.
Generic
and
I
think
this
is
the
workflow
that
other
people
also
do.
I
assume.
A
B
D
D
Here,
I
am
basically
what
is
a
tool
that
allows
you
to
create
a
clusters
or
visual
machines
in
different
environments,
one
of
the
environments
is
cavium,
okay
and
it's
very
easy
to
set
up
trusted
of
of
beautiful
machines
using
your
preferences.
Okay,
in
fact,
I
have
repeated
a
CLA
plan
that
allows
you
to
create
a
cluster
with
M
machines
repeated
to
to
be.
D
We
use
it
with
Fifi
DM
today,
this
plus
the
modification
that
I
did
in
the
in
the
Buddhist
are
common
in
order
to
say,
for
this
allows
you
to
create
the
cluster
and
to
what
new
new
diamonds
in
any
of
the
machines.
You
seen
the
code
that
you
have
him
the
indicator,
and
this
is
tall.
All
everything
is
done
in
just
one
common,
okay,
I
can
say.
E
D
Ten
easy
to
use,
yes,
basically,
is
to
donate
the
container
that
is
running
ke
CLI
and
then
with
Gaza
one
command.
You
can
use
the
one
of
the
plants
that
are
studied
in
in
that
not
folder.
In
the
repository
and
with
that,
you
configure
the
cluster
label
tourism
cluster
and
with
a
bootstrap
cluster
script,
you,
basically,
you
configure
each
of
the
real
built
on
machines
in
order
to
use
safe
idiom
and
with
a
bootstrap
come
on.
You
can
serve
the
the
wealth
or
the
orchestrator
for
this
in
all
the
diamonds.
D
Wanted
to
to
write
something
more
to
explain
all
the
or
main
development
environment,
but
not
too
much
time
in
the
last
two
weeks
with
the
modification
that
I'm
doing
doing
now.
Okay,
but
I,
think
that
what
is
very
easy
to
to
make
modifications
with
this
I
can
make
a
demo
if
you,
if
you
want,
if
people
is
interested.
D
A
E
Yeah,
it
looks
very
promising,
I
mean
I,
think
the
thing
I
liked
about
the
c-store
ever
change.
Your
approach
with
the
dashboard
was
that
there
was
a
way
to
allow
you
to
change
the
other
processor
self
as
well,
yeah
and
yet
without
having
to
rebuild
the
container
image.
I
think
that
could
be
easily
added
to
of
this
approach,
though,
with
vm's
and
get
and
both
a
user
development
and
as
close
as
you
can
as
to
a
production
dam
set
up
as
you
can,
which
sounds
like
the
best
of
both
worlds.
To
me.
Yes,.
D
Yes,
I,
in
fact,
what
I
use
it
part
of
the
mapping
set
up
folders
that
you
use
in
the
siesta,
I
think
so
well.
I
think
that
we
have
room
for
improvement,
but
I
think
that
is
promising,
because
basically
it
allows
you
you
to
have
a
complete
environment
system
working
a
working
system
in
just
three
four
minute
minutes:
okay,
it's
just
the
part
that
takes
more
time
is
to
download
the
safer
Emacs
container,
and
this
is
all.
A
E
Know
I
think
I
would
go
that's
what
good
directions
to
go,
though,
and
I
think.
Once
once
we
have
this,
we
have
written
that
more.
We
can
maybe
try
to
get
yf-12
ready
to
start
crying
to
use
it
as
well.
I
think
that's
many
where
that
see
start
came
from
a
perspective
of
fellows,
who
are
usually
developing
on
a
single
host
using
the
B
start
script,
which
is
the
bit
relatively
faster.
That's
up,
but
not
really,
representative,
of
a
real
environment.
A
Kristina's
Casey
and
I
really
takes
time
to
set
up
a
cluster
and
in
setting
up
virtual
machine
and
with
some
machines
and
stuff
like
that,
I
don't
see
Casey
Li
a
replacement
for
for
visa.
A
A
D
For
example,
the
only
thing
that
you
need
to
do
is
just
to
disable
and
re-enable
the
manager
model,
and
this
is
you
don't
need
to
rebuild
any
any
vomiting.
You
don't
need
to
walk
again,
any
definitely
much
much
definite
container.
The
only
thing
that
you
are
doing
is
using
that
theft
code,
that
is
in
the
in
the
safe
container
limit
and
the
code
that
you
have
in
your
own
folder.
B
D
No
comforta
Natalie
I
haven't
time
to
do
things
within
the
in
the
room,
part
okay,
so
yes,
the
next
step
is
just
to
create
a
new
pipeline
in
order
to
run
the
Deaf
manager
intubation
test
in
a
daily
Natalie
basis,
okay,
and
to
avoid
to
to
run
this
test
in
in
every
pull
request
but
well,
I
have
started
highly
put
one
Jenkins
working
in
my
in
my
machine.
Okay
and
I
started
with
testing
things,
but
not
timed
in
order
to
what
to
have
a
pool
to
to
provide
auto
generator.