►
From YouTube: 2019-02-18:: 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).
A
A
Yeah
thanks
to
that
being
set
up
in
the
agenda
for
today,
the
bashing.
You
have
a
couple
first
couple
items
there
around
error,
handling
gun
right.
There.
B
B
A
B
As
I
said,
I've
tried
to
incorporate
every
feedback,
I
got
but
would
be
nice
to
get
some
reviews
of
the
actual
final
implementation.
B
A
D
Excuse
me
yeah,
so
I
thought
it
was
mostly
done,
but
over
the
weekend
a
bunch
of
got
a
bunch
of
feedback
and
I
haven't
looked
at
it
specifically,
but
it
all
looked
pretty
straightforward.
D
It's
really
it's
not
available
from
Ubuntu,
but
let's
see
some
in
I
believe
or
well.
Okay,
I
can't
remember
exactly
who's
handling
this,
but
there's
a
PR
open
for
ansible,
the
ansible
playbooks
for
the
for
the
sepia
cluster
that'll
install
this
from
pip
and
I
think
we
need
just
to
make
sure
that
works
for
sent.
Os
2,
there's
an
updated
package,
that's
going
to
land
in
in
rpms,
but
it's
just
not
available
right
now.
So,
for
example,
right
now
the
test
works
on
Tinto
as
only
into
college
II.
D
D
Don't
know
how
that
affects
merging,
but
certainly
not
not
exactly
sure
how
to
proceed
on
that.
D
D
B
I've,
not
the
state,
Orchestrator
and
they're.
Really
lighted
works.
What
good
so
far,
I
couldn't
get
the
center.
At
seven
box
running
on
my
sister,
the
wife
thought
that
the
development
environment,
vagrants
grip
to
Ubuntu,
mm-hmm,
Bionic
and
I
was
able
to
deploy
all
these
on
Ubuntu
so
yeah
it.
Oh
that's.
D
A
I've
got
some
notes
on
rook
there
to
discuss
the
first
one
is
around
the
the
PR
for
head
and
remove
Mons
an
Orchestrator,
Jeff
I
think
we're
ready
to
merge
that
right.
We
just
had
some
questions
about
potential
things
to
do,
but
I
think
we
can
definitely
hold
off
on
that.
That
was
more
of
a
long-term.
E
If
we
want
to
add
adding
per
host
no
limits
or
whatever
you
know,
you
know
settings
or
whatever
eyes
early,
it
was
pretty
a
larger
scale.
Project
I'm,
not
sure
how
he
not
quite
sure,
I
envision
how
it
would
work.
You
know
you
know
aside
unless
we
did
have
a
way
to
do
the
label
perm
or
no.
But
that's
you
know
much
more
involved.
I
say
we
go.
Don't
do
that
now,
waiting
for
ready.
A
E
E
Something
where
you
you
know
add,
you
know,
add
as
the
label
with
the
node,
that
this
mod
should
be
assigned
to
to
the
mod
and
then
have
it
somehow
match
that
up
with
a
label
that
the
kubernetes
that
kubernetes
automatically
puts
on
the
node.
That
might
be,
you
know
doing
something
like
that
may
be
doable,
but
I'm,
not
sure.
If
you
can
even
express
that
with
the
way
the
you
know,
the
node
affinity,
stuff
works,
I'm,
not
quite
sure
how
that
would
work.
A
E
A
Next
I
was
so
last
week.
We
realized
there
was
an
issue
with
were
connecting
to
Nautilus.
It
was
really
independent
of
the
orchestrated
modules,
but
just
to
follow
up
on
that.
It
was
a
regression
on
the
book,
aside
from
setting
up
the
or
changing
how
the
mods
are
configured
moving
more
towards
a
mimic
style
at
a
configuration
and
not
depending
on
Sep
Kampf
anyway,
so
that
that
was
the
cause
of
the
break
there
and
with
the
work
Sebastian's
doing
massive
messenger
to
it'll.
F
G
Guess
it
just
everything
seems
to
be
working
now
addressed
all
the
comments
passing
see,
I
so
guess
one
last
review,
and
hopefully
we
can
get
this
merged
yeah,
it's
actually
taking
care
of
give
grades
and
everything.
So
we
should
don't
have
any
backward
compatibility
issues
and
also
not
having
any
problems
with
already
running
clusters
or
a
pretty
upgrade.
A
G
A
A
That's
so
after
the
messenger
2pr
goes
in
then
then
I'll
rerun
the
integration
tests
on
Nautilus,
so
we
can
get
that
I'll
have
some
reassert
reassurance
that
not
else
is
working
going
forward
catch
things
like
the
configuration
differences,
that'll
be
good
and
then
a
lot
of
the
thing
where,
let's
see
Jeff
you
early
now,
you
probably
know
this
best
as
far
as
where
we
are
with
getting
the
rook
module.
Orchestrator
module
loading.
E
Thing
it
looks
like
it
should
make
it
into
stable
tomorrow,
so
here's
the
the
thing
for
it.
So
it
looks
like
it's
one
more
day
to
stable
so
about
tomorrow.
We
should
have
that
resolved.
Then
we
only
need
to
do
is
put
the
Python
kubernetes
dependency
back
in
this
text
file
and
in
yeah
not
sure
what
we
do
for
debian
naturally,
but
we'll
need
to
do
that
for
and
then
you
know,
presumably
once
we
have
a
have
the
ACEF
container
images
pulling
in
the
manager
rook,
then
we
should
be
good
to
go.
A
A
B
A
F
H
Yes,
I
am
working
with
with
my
integration
test
platform.
Okay,
so
basically
I
am
playing
with
a
with
the
pieces.
I
can
explain
that.
Well,
basically,
what
I
am
doing
is
to
use
bootstrap
cluster
that
is
made
of
two
containers,
one
of
the
one
for
the
manager,
one
for
the
monitor
and
then
build
to
our
machines,
set
up
with
a
grant
in
order
to
test
the
unseeable
or
cos
theta.
H
A
part
of
that
I
have
another
container
with
the
ANSI
Voltron
a
service
a
running,
and
this
is
what
what
I
am
doing
now,
trying
to
set
up
environment
where
we
can
make
integrative
integration
test
and
to
end
okay,
so,
basically,
in
with
without
they
all
the
other
pieces,
I'm,
probably
they
were
the
test.
I
have
told
that
it
could
be
nice
to
have
these
tests,
don't
with
behave
I,
don't
know.
H
If
what
probably
you
know,
that
is
one
kind
of
test
that
use
a
specific
language
in
order
to
make
easy
to
specify
or
to
do
the
test,
and
this
is
what
what
I
am
doing
in
the
in
this
comment.
A
part
of
what
investigating
playing
with
a
permit
is
basically
in
order
to
to
be
prepare
for
right
next,
but
this
town
downstream.
C
I
yeah
I
played
a
bit
last
week.
I
was
able
to
make
it
work
on
on
a
note,
openSUSE
vagrant
box
I
even
took
care
of
the
of
packaging.
The
remote
Oh
Python
library
into
openSUSE
and
yeah,
but
I
haven't
I,
didn't
have
the
time
to
to
look
at
the
code,
but
from
from
the
things
that
I
did
it,
it
looks,
looks
great.