►
From YouTube: 2019-02-25:: 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
B
I
I
think
the
my
first
question
is
just
you
know,
they're
a
bunch
of
things
in
here
generally,
we
do
want
to
have
the
state
time
spread
priority
sub
system
message,
with
one
exception,
where
there's
at
the
very
beginning
or
adding
a
prefix,
that's
odd
type,
basically
I'm,
not
a
beep,
debug,
debug,
logs
or
Montag.
Oh.
A
B
Everything
you
have
to
pass
in
a
couple
extra
arguments
to
the
set
demons
and
I'm,
depending
on
what
version
of
rook
you
were
running.
It
may
or
may
not
have
been
passing
that
in
yet
I
can't
remember
if
all
the
demons
have
been
updated
and
those
requests
been
immersed
yet,
but
the
second
acts
an
extra
set
of
flags.
Basically,
there
are
there
per
team
and
debug
logs,
which
is
mostly
what
you're
seeing
here
that.
C
B
Are
going
to
be
prefixed
with
debug,
and
then
you
also
have
the
cluster
logs,
where
you
have
channels
for
its
defaults
audit,
dashboard,
I
think
this
might
be
I,
don't
know
there
are
couple
different
channels
and
then
we'll
put
that
in
there
also
else,
but
the
idea
they
I
haven't
actually
looked
at
this
to
verify.
So
it's
we
should
do
this,
but
the
idea
is
that
we'll
have
a
standard
format,
so
all
sort
of
uniform
and
we
can
we
can
tweak
if
they're
not,
then
we
can
just
tweak
the
syntax.
D
D
D
B
Sage,
so
there's
there's:
there
are
a
couple
more
changes.
We
have
keyed
up
for
the
messenger
to
I
think,
generally
speaking,
everything
else's
future
complete,
so
we
just
need
to
get
all
the
QA
Suites
cleaned
up,
Oh,
probably
name
frnd
on
our
second-
are
see
like
maybe
at
the
end
of
this
week
and
then
hopefully
a
final
release
week
after
that
that'd
be
the
sort
of
aggressive,
aggressive
timeline.
B
F
B
So
we
should
just
just
asking
Alfredo
in
the
channel
actually
channel
and
see
if
he
look
that's
out
on
purpose
or
just
because
or
at
a
time
or
whatever
I
think
we
should
find
Adam.
The
other
sort
of
question
around
the
builds
was
that
and
we
had
a
couple
people
asking
about:
Debian
builds:
they
dropped
them
for
mimic
because
of
GCC,
but
Buster
has
the
right
GCC.
So
in
theory
we
should
be
able
to
add
the
MEK
in
okay.
F
G
F
G
B
D
E
E
E
As
soon
as
it's
really
branched-
and
we
have
to
make
sure
that
we
are
making
use
of
the
tracker
by
everything
that
needs
to
be
back
potted,
which
is
hopefully
everything
for
now
I've
added
an
Orchestrator
category
to
the
manager
project
in
the
tracker,
so
that
we
can
categorize
properly
into
the
orchestrator
tag,
makes
it
a
bit
easier
to
find.
Although
there's
only
one
issue
right
now
in
the
tracker
for
us.
E
E
H
Miss
urgent
was
merged
last
week,
then,
on
the
fly,
is
switching
to
beasts
front
end
for
nautilus,
then
doing
memory
auto-tuning
for
OSD
and
and
ESS
make
sure
that
we
don't
run
out
of
memory.
Well,
basically,
for
for
HCI
environments,
we
will
be
capping
memory
in
CPU
use
it.
So
we
don't
really
want
to
end
up
into
a
situation
where
OS,
DS
and
M
DSS
go
above
their
own
cache
caches.
So
the
idea
is
to
simply
set
those
flags.
D
And
one
note
on
essence:
we
got
the
patch
merged,
so
Nautilus
is
working
now
with
rook,
with
the
one
have
yet
that
you
have
to
change
the
base
image
of
the
rook
operator
to
be
Nautilus,
though,
as
as
soon
as
we
have
the
the
Nautilus
image
pushed
up
to
Steph
Steph,
then
upon
making
that
change
to
make
Nautilus
the
base
image.
And
then
everything
should
work
out
of
the
box
or
Nautilus
yeah.
F
Was
just
gonna
say
I
could
I
could
add
in
a
line
that
would
end
up
just
being
temporary
for
this
week
to
another
poll
requests
that
would
essentially
use
the
master
branch's
version
of
NFS
for
the
nautilus
build
for
this
week.
If
we
really
want
to
get
started
with
that
and
rook
instead
of
waiting
for
a
few
days
or
at
the
end
of
the
week
or
next
week
or
however
long
it
would
take,
yeah
go
for
it.
Okay,
great.
E
E
Yeah
lens:
that's
some
problem
with
a
face-to-face
meeting.
Yeah
he's
going
to
send
out
a
new
rule
for
that.
So
it's
not
going
to
be
week,
twelve,
but
afterwards,
yeah
he's
he's
going
to
send
out
another
tool
and
I
will
just
forward
it
likely.
E
E
Next
item
is
that
we
had
some
discussion
about
the
drive
group
specification
and
one
question
is:
do
we
want
to
support
file
store
in
the
orchestra?
At
all
sure
the
the
external
Orchestrator
might
want
to
support
file
store?
But
do
we
want
to
have
that
in
the
orchestra
CLI
or
do
just
go
for
what
was
blue
store
and
just
forget
about
five
store,
I.
B
E
H
C
E
B
G
G
When
it's
like,
oh
yeah,
you
know
a
selector,
basically
in
there,
and
so
you
know,
I
want
to
rip
out
a
bunch
of
these
other
fields
because
they're
not
used
and
it's
misleading,
and
you
just
add
them
in
as
we
need
them
and
and
I
I'm
also
proposing.
Maybe
we
should
go
ahead
and
just
rip
out
the
file
store,
boost
or
boolean,
because
you
know
why
you
know
product
is
supported
in
the
India
orchestrators
know.
We
don't
really.
E
G
D
G
E
B
G
D
G
G
We
need
to
have
something:
that's
this
thing
and
also
well
yeah.
Maybe
it
shouldn't
be
right
access
both,
but
guys
people
without
privileges,
but
you
never
know
someone
can
screw
up
and
go
to
safety
features
in
here.
But
that's
one
thing,
but
also
do
we
need,
you
know,
do
I
need
to
add
other
fields.
So
we
have
this
drag
groups
back
and
then
we
have
these
drive,
selection,
things
or
something
or
just
selection,
I
feel
that
is
another
object.
We
have
lists
of
those
objects
inside
the
draw
group.
Spec
object.
G
D
B
E
E
B
B
E
B
So
like,
if
I
long
term
broad
strokes,
my
hope
is
that
we
can
do
a
couple
things.
One
eventually
have
a
a
sort
of
a
trivial
bootstrap
script
that
will
just
start
up
a
monitor
and
a
manager
on
the
localhost
and
turn
on
the
SSH
Orchestrator.
But
you
can
then
de
to
provision
the
rest,
your
cluster,
every
one
thing,
and
then
the
other
is
that
we
have
the
documentation.
Right
now
is
like
there's
a
couple
sections
on
install
different
ways
to
install
that
are
like
super
now.
B
One
of
them
is
using
set
deploy
and
it's
like
super
complicated
and
the
other
one,
just
like
a
link
to
sensible
and
there's
no,
the
one
about
manual
deployment,
that's
like
totally
useless,
except
for
developers
and
then
everywhere
else
in
the
documentation.
They're.
Just
random
references
to
like
septa
boarding
like
the
upgrade
steps,
for
example
or
like
I,
don't
know
whatever.
B
That
would
be
nice
to
eventually
make
a
full
pass
across
the
docs
and
rewrite
all
of
these
in
terms
of
using
the
orchestration
API
still
I
for
everything,
so
that
the
the
overall
safe
Doc's
are
sort
of
agnostic
about
this
stuff.
But
the
big
project,
of
course,
but
I'll
play
that's
where
we
can
get
you
eventually.
B
Yeah
yeah
generally
after
the
release,
we
want
to
do
like
a
series
of
blogs
on
all
the
new
stuff,
and
so
this
would
be
a
great
one
to
do.
Yep.