►
From YouTube: 2018-11-06 Rook Community 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
Then
that
sounds
good.
We'll
talk
about
that
agenda
item
when
we
get
there,
then
yep.
Let's
move
on
to
0.9,
you
know.
Once
again,
we
have
a
fair
amount
of
items
in
the
milestone
here
in
the
hopes
that
contributors
can
pick
those
up
and
we
will
approach
them
as
we
get
closer
to
the
milestone
targeting
a
release.
Are
there
any
big
issues
that
people
want
to
bring
up
in
this
forum
here,
potentially,
maybe
an
update
on
si
si.
C
Si
I
don't
see
a
woman
today
in
the
meeting,
but
I
know
that
he's
he
has
been
working
on
CSI
fairly
actively.
There's
I
think
before
we,
you
know
really
declare
done.
There's
some
integration
to
be
done.
You
know
si
si.
Some
instructions
were
written
up,
I,
don't
think
that
PR
emerged
yet
or
get
it
for
documentation
on
how
to
use
it,
so
that
is
available,
but
it's
really
a
separate
process
from
the
normal
installation
for
work.
So
we
need
integration
steps
there
that
and
we
had
that
hasn't
been
worked
through.
A
A
Cool,
it's
a
my
a
testing
on
just
you
both
no
upgrade
scenario.
Just
a
clean
slate
everything's
been
working
just
fine
for
me.
I've
had
any
problems
with
it
at
all
great
you
tried
out
the
dashboard
on
mimic.
I,
don't
need
a
dashboard.
I've
got
cube,
control
yeah,
you
know,
like
you,
Ord
I,
don't
need
to
click
on
anything.
I've
got
a
keyboard.
D
E
Have
a
question
I'm
CSI
plugin
in
this
poor
across
which
we
just
sent,
we
actually
do
have
CSI
integration,
but
I
was
wondering:
do
we
need
to
do
anything
like
generic
way,
because
what
we've
got
right
now
in
a
Jefferson?
You
guys
welcome
to
review
the
pull
request.
It
is
a
separate
CSI
installation,
but
we
saw
that
it's
important
to
have
a
documentation
point
to
that
one.
So,
as
far
as
these
polar
course
is
concerned,
it
has
a
GFS
CSI
MD
document,
but
other
than
that
we
need
to
do
anything
else.
A
I
think
Demetri,
that
you
know
we're
gonna,
be
kind
of
have
a
better
understanding
once
we
have
the
ability
or
I
get
to
take
the
time
to
look
at
that
per
request.
But
the
original
thinking
I
believe
so
far
has
largely
been
that
CSI
plugins
are
going
to
be
storage
providers
specific
and
they're
spread,
no,
probably
not
going
to
be
much
overlap
in
terms
of
you
know,
experience
there.
If
there's
some
reuse
or
commonality,
they
can
work
that
you
know
it
can
be
in
a
common
place
in
the
book.
A
E
C
E
B
A
A
A
It's
only
with
1.11
about
12
kubernetes,
and
I
have
no
idea
how
to
troubleshoot
it
either
because
everything
just
kind
of
you
know
I'd
like
can't
even
run
top
or
if
you
have
to
look
at
processes
or
the
message
or
I
can't
do
it,
we
can't
figure
anything
out
to
even
poke
at
no
pokes
so
I
don't
know,
but
yeah
we're
never
giving
demos
in
Shanghai
next
week,
I'm
just
gonna
use
one
button,
I'm
not
gonna,
take
any
chances
or
I.
Don't
really
have
enough
time
right
now
to
keep
looking
into
notice.
A
B
E
A
A
E
E
A
Yeah
to
be
it'll,
be
really
nice
to
you
know
for
other
folks
to
give
it
a
test.
Spin,
you
know
see
what
the
experiences
are
and
kind
of
dig
into
it
a
little
bit
more
because
it
is
a
whole
lot
of
new
code,
and
you
know
it's
also
a
an
integration
we've
been
excited
about
for
a
while
too
so
I'm.
Definitely
looking
forward
to
it.
Yeah
thanks
cool,
okay,
one
last
call
for
any
others.
You're
done
any
issues
that
people
want
to
bring
up.
A
A
A
E
E
It's
okay
I
mean
if
you
can
just
walk
through
the
changes,
I
guess
a
little
bit
yeah.
So
what
we've
done-
and
this
is
probably
enough-
first
picture
load.
E
If
so,
what
we've
done
so
far
is
we
grab
existing
chef
yeah?
There
we
go
and
we
essentially
added.
We
try
to
reuse
as
much
code
as
possible.
What
Francis
reusing
discovery
demon
sets
information
about
the
discs,
and
then
we
added
some
a
different
features.
As
you
see
in
our
series,
we
tried
to
keep
existing
parameters,
as
is
in
our
shared
ease
like,
for
instance,
data
dear
host
pass
has
the
same
names
in
idea.
We
added
more
like
France.
E
We
need
11
and
higher,
but
is
that
we
give
ability
to
manage
edge
of
s
is
just
it
was
set
yet
with
preparing,
obviously
through
crg,
all
the
necessary
pieces,
like
data
volume,
for
instance,
if
provided
it
can
be
an
energy
of
which
makes
sense
for
the
data
set
to
be
attached
in
case
of
state
all
set.
We
then
also
added
features
like
like
sub
devices
and
react.
Those
are
interesting
features
like,
for
instance,
if
you
deleted
this
ERG
and
then
we
want,
for
some
reason
bring
up
the
cluster
again.
E
But
what
keep
the
device
configuration
you
can
just
satisfy
your
reactant
automatically
rediscover
the
network,
but
will
reuse
existing
settings
from
data
volume
or
from
dated
year
we
also
added
Zeb.
Do
is
zap
devices
which
will,
if
you
do
this
reactor,
you
can
basically
result
the
cluster
for
easy
of
use.
Then,
additionally,
if
you
scroll
down
specifically
to
storage
configuration.
E
Yeah
like
in
this
this
one,
it
changed
a
lot
comparing
to
chef,
obviously,
because
we
do
things
a
little
bit
different,
but
essentially
what
you
can
see
here.
He
added
dynamic
mechanism
of
detecting
SSDs
and
edging
is
installed
like,
for
instance,
if
you
have
a
proportion
of
four
HDDs
and
say
once
as
detail
that
much
detected
and
the
pair
on
a
target
will
be
automatically
built
in
addition
to
row
disks
similar
self-build
store.
We
also
provide
a
capability.
E
Just
only
be
specific
at
the
cluster
level,
they
cannot
be,
you
know,
be
overridden
and
the
not
level
new
use
of
the
stateful
sets.
This
is
not
the
case
for
the
row
disks
and
yeah.
The
placement
is
pretty
much.
The
same
did
not
change
too
much,
obviously,
because
we
using
staples
in
terms
of
for
us
to
properly
select
which
not
needs
to
be
selected
or
not,
we
will
be
using
labels
essentially,
so
the
our
curator
automatically
sets
up
the
labels
on
the
nodes
of
selection
of
the
users.
E
Choice
and
those
labels,
then
is
correlated
with
when
the
staple
set
is
starting
up,
that
works
beautifully
actually,
and
once
we
delete
in
this
year,
D,
we
then
prior
to
final
finalizar,
gets
removed.
We
delete
in
the
labels
that
works
well
as
well,
so
like
this,
for
instance,
he
restarts
mg
R,
which
is
the
manager
it
provides.
Gr,
PC
and
integration
is
the
CSI
addition.
It
provides
parameters
for
the
dartboard
and
in
the
future
we
can
add
some
support
for
the
GUI,
which
we
have
so
that's
pretty
much
it.
Then.
E
If
you
go
back
there
is,
there
are
other
crg,
so
what
they
I
can
talk
about
those
a
little
bit
and
then
I
can
answer
the
question.
So
if
you
scroll
up
a
little
bit
yeah
there,
so
what
we
currently
have
from
this
pull
request
is
just
an
affair.
So
there
is
a
document
called
edge
FS
and
if
I
see
Arnie,
if
you
go
to
that,
one
I
can
go
here.
Show
you
what
it
is.
E
E
It's
very
simple
one,
but
it's
it
has
awareness
of
the
namespaces
which
refers
provides
in
the
first
is
couple
of
planning
across
multiple
data
centers
and
connecting
different
buckets
together
in
transparent
way
and
also
its
multi-tenant.
So
you
can
basically
specify
the
service
name.
It
just
needs
to
be
the
same
as
the
CD
name.
This
NFS
zero
one
has
to
match
the
name
of
the
service
definition
in
the
cluster.
So
if
you
scroll
down
a
little
bit,
there
is
a
little
documentation
how
to
do
it.
E
You
know
hard-coded
static
volumes
here,
because
in
case
of
CSI
you
can
then
define
how
do
you
wanna
provision
those
experts?
It
can
be,
for
instance,
dynamic,
in
which
case
the
buckets
will
be
dynamically
created
and
destroyed.
It
depends
on
the
availability,
so
we
decided
basically
keep
it
simple
over
time.
E
We're
gonna
add
more
parameters,
but
at
the
moment,
just
in
the
face
we're
going
to
be
adding
more
charges,
do
does
does
it
to
share
this,
which
cannot
be
definitely
gonna
have
is
s3
and
I
is
GW,
which
is
s3
is
s3
and
as
you
double
in
the
segment
gateway
to
connect,
multiple
datacenters
together.
This
will
come
in
couple
of
weeks
and
hopefully
gonna
be
in
a
pull
request
for
0.9,
so
that's
basically
eat.
If
you
guys
have
questions,
I
can.
D
One
thing
that
would
help
the
review
of
this:
this
is
all
great
for
you
guys
spent
quite
a
bit
of
time
on
this.
It's
potentially
breaking
up
into
multiple
commits
the
optimizing
for
readability,
maybe
put
all
the
dogs
and
I
command
all
the
generated
code
in
the
separate
commits
the
actual
code
for
edge
FS
operator
in
another
commit
so
that,
because
it's
a
big
change,
it's
likely
gonna
take
much
longer
to
review
this.
D
E
A
Dimitri
I
just
sent
to
the
chat
a
link
to
the
cockroach
DB
initial
pull
request,
I'm
gonna
upload
over
here
to
the
screen
too,
and
this
might
be
a
good
example
of
a
Bassam
was
talking
about
of
how
the
commits
are
broken
up
into
functional
areas
like
the
integration
tests
and
the
documentation
and
the
operator
controller,
and
the
API
is
the
CR
DS.
So
this
may
be
a
good
model
to
follow
that
helps
the
readability
of
the
pull
requests
by
separating
out
each
commit
so
easy.
E
A
Awesome
yeah
and
if
you
lose
that
link
just
ping
me
on
slacker,
whatever
now
I'll
be
happy
to
send
that,
to
you
sure
thanks
already
does
Dimitri
does
that
yeah
and
by
the
way
too,
I
want
to
make
a
note
too
that
you
know
this.
This
effort
here
looks
like
it's
very
thorough
and
it
looks
like
there's
a
lot
of
attention
to
the
detail
there
and
kind
of
fleshing
things
out
pretty
far
so
I'm
definitely
impressed
with
it
and
I
think
it's
looks
like
a
really
solid
effort
from
from
your
team,
so
yeah.
E
A
A
E
I,
my
understanding
is
that
the
other
page
gets
automatically
generated,
but
if
you,
if
you
go
to
documentation
now,
if
you
want
to
get
over
there,
then
what
we
thought
it'd
make
sense
is
to
restructure
this
page,
because
this
page
now
kind
of
reading
about
this,
you
can
I,
don't
understand.
What
is
this
like?
How
can
I
do
block
well?
I
can
do
block
within
a
different
ways
hub,
like
not
just
chef
right.
How
can
I
object,
not
just
chef
can
be
mean
a
you
say.
D
B
Well,
restructuring
the
whole
thing.
There
is
one
of
the
things
I
have
planned
when
I'm
finally
done
with
kind
of
well
getting
done
this
stuff
first
year
at
nine
as
I'm
Allison,
not
only
looking
into
restructuring
the
docks
and
being
able
to,
for
example,
this
site
are
in
the
documentation
having
that
a
multi,
multi
level
stuff
like
that,
which
will
also
grant
us
more
possibilities
in
structuring.
You
know
the
whole
documentation
there,
yeah.
B
Least
well,
if
I
would
use
as
a
list
right
now
have
in
mind,
I,
don't
think,
there's
an
issue
to
put
it
like
that,
but
I
don't
think
I
would
be
able
to
get
it
into
your
serial
line,
because
we
could.
It
may
sound
a
bit
weird
but
I
think
if
we
could
get
a
dummy
people
comment
into
the
release
branch
of
serial
of
mine.
We
could
read
trigger
the
generation
of
the
documentation
for
serial
of
mine
if
it's
in
master
or
well.
B
I
have
some
plans
for
that
I
haven't
been
able
to
write
them
down
or
something
it's
I
have
it
on
my
mind,
I
want
to
have
the
page
looking
much
better,
much
more
inviting
and
especially
on
the
focus
with
the
new
upcoming
search
providers.
Much
more
focused
on
kind
of
like
a
bit
going
into
the
cuvette
news,
documentation
there,
where
it's
like
hey
kind
of.
What
do
you
want
like
do
oneself?
Or
do
you
want
this
or
you
know
it's
really
showing
you
the
possibilities
in
the
end,
what
you?
C
A
B
Have
been
slight
changes
for
me
if
I
I
just
want
to
know
to
do
what
to
put
them
out
there,
because,
for
example,
if
you
click
on,
could
you
go
to
Luke
that
I
oh
go
to
the
main
page?
If
you
click,
if
you
click
on
Doc's,
it
just
is
one
of
the
changes
we
added
already.
It's
kind
of
that
you
get
to
the
latest,
detect
version
documentation
and
not
as
it
was
previously
tomas
delicious
wish
to
master
contribute.
B
B
To
put
it
kind
of
like
this
in
the
end,
if
the
user
had
issues
with
that
even
sounds
will
be
bad.
You
could
always
kind
of
go
like
you
saw
the
red
egg
box,
which
said
to
well
not
use
math,
unless
you
know
like
you're
testing
or
something
I
had
also
kind
of
from
a
broad
message.
To
put
it
like
that,
we
should
we're
like
oh
yeah.
B
B
If
you
go
to
know
the
documentation
version,
you
will
also
have
a
box
which
is
like
hey
John
and
older
version.
Justice
you're,
not
on
the
latest
version
and
also
displayed
like
they're,
like
hey,
latest
releases
co2
late.
So
if
we
clean
our
serial
at
nine,
it
will
automatically
be
0.99
as
far
as
I
know
and
I
will
probably,
if
I
find
the
time
go
ahead
and
kind
of
do
a
bit
of
a
mock
up
with
a
friend
of
mine
together
offered
loose
the
home
page
to
get
that
straightened
order.
A
Cool
yeah
I,
like
that
add
the
longer
term
vision,
and
you
know
the
the
intermediate
steps
that
we
can
do
and
0.9
or
more
more
soon.
You
know
mix,
mix,
tons
of
sense
to
me.
Dimitri
I,
wouldn't
even
be
opposed
to
you
know,
kind
of
as
part
of
the
next
inta
initial
pull
request
here,
potentially
me
know
making
some
updates
to
these
links,
and
maybe
you
know
making
a
little
bit
more
friendly
as
part
of
that.
Proquest
is
not
entirely
out
of
the
question
for
me
honestly.
I
can.
A
I
have
no
problem
with
that
whatsoever.
Thank
you
for
for
bringing
up
that
to
our
attention
that
I
definitely
appreciate
that.
Okay.
Does
that,
and
so
that
covers
your
two
issues
here,
Dimitri
all
right,
awesome,
that's
I,
really
like
what
I'm
seeing
here,
I'm
really
looking
forward
to
going
forward
with
all
that
Alexander.
You
are
highlighting
the
next
next
thing
there
I
see,
I,
don't
think
Yanis
is
on
the
call
today,
but
I
know
I
saw
that
he
brought
up
this
issue
about
our
Jenkins,
build
for
his
Cassandra
pull
request,
that's
outstanding!
A
B
As
far
as
I
mustered
from
the
discussion
between
Rohan
and
Youngs,
instead,
some
kind
of
script
I
think
it
was
yeah
that
promise
message.
Is
there
any
deaf
channel
they
go
out
how
to
how
to
make
this,
probably
something
with
also
generation
four
go
things
and
it's
it
needs
material
well,
and
we
don't
have
that
right
now.
I
think.
The
problem
kind
of
here
is
that
everybody
which
uses
Rukh
would
then
required
finished
material
installed
on
their
machines.
B
A
B
D
D
A
B
B
There's
still
an
issue
of
0.7
or
well
we're
serial
date
with
lung
host
names
when
you
want
to
upgrade
it
can
cause
year.
Well,
the
existing
our
seems
to
still
function
but
unknowns
to
new
OS.
These
create
to
be
created
for
I.
Don't
know,
I
have
no
clue
why
right
now,
but
like
acetate
panel,
a
kind
of
process,
if
you
have
a
short
host
name
for
the
service,
for
example,
node
one,
it
works
fine
perfectly.
You
have
sir
seven
with
metadata
device
and
everything,
even
you
know
like
this.
B
At
the
end
setup
you
update
you'll,
see
rotate,
it
works.
It's
working,
fine,
the
new
or
assisted
created,
an
S
expected
they're
going
into
crash
loop
it
off
because
he
all
those
these
are
still
using
the
distal
resources.
And
then,
if
you
delete
them,
everything
is
fine.
So
a
quick
process
works.
But
if
you
have
a
long,
hostname,
okay
I
think
I'm
not
sure
how
many
chances
it
was,
but
like
30
or
50,
or
something
the
upgrade
will
fail.
C
C
B
Student
through
the
code,
a
bit
and
well
just
search
for
a
note,
name
and
I,
came
across
a
few
sections
where
well
for
config
maps,
where
it
doesn't
matter
too
much
of
how
long
the
name
is
that
there
was
the
truncate
node
named
function,
not
used
so
I
kind
of
oversaw
some
sections
below
it
was
like
SL,
it's
a
config,
not
H,
it
doesn't
matter
for
them.
They
I
think
they
are
names
for
a
conflict
or
are
not
in
s
related
things.
A
B
Has
been
no
issue
yet
because
well
yeah!
Well,
I!
We
looked
into
the
issue
two
weeks
and
while
just
I
think
yesterday
sitting
there
again
for
two
hours,
they're
trying
to
debug
it
I
was
like
well.
What
is
the
big
difference
between
mini
given
to
your
cluster
right
now
or
well?
My
multi
neared
environment
are
well
the
node
name,
so
it
just
when
I
had
changed
in
here
and
then
I
basically
broke
completely
so
or
well.
You
know,
went
haywire
our
sister,
so
that
was
the
first
time.
A
D
C
Yeah
we've
had
a
couple
of
people
with
with
larger
data
sets
that
hit
this
issue
where,
when
the
mountain
happens
after
the
mount
with
the
Flex
driver
on
the
kubernetes
decided
that
mount
it
does
a
recursive
traversal
of
all
the
files
and
and
sets
the
FS
group
and
basically
the
ownership
and
the
selinux
relabeling.
And
it
takes
a
long
time
to
attach
so
the.
Although.
C
So
yeah
yeah
for
large
datasets
it'll,
certainly
slow
it
down.
So
it
seems
like
from
everybody's
comments.
You
know,
I
was
I,
wasn't
sure
like.
If
we
disable
it,
there
is
a
way
to
disable
it
with
the
Flex
drivers
and
it's
kind
of
a
global
setting
for
the
Flex
driver,
so
it
effect
walk
and
file.
It
seems
like
the
right
thing
to
just
go
ahead
and
disable
it
and
we
just
set
it
at
the
top
level
using
the
same
code
that
kubernetes
has
to
do
it,
but
not
recursively.
A
A
B
C
D
A
D
B
Then
I
may
have
also
experienced
this
with
my
data
sense
on
file
system,
but
I
was
kind
of
like
well
as
I'm
running
desktop
and
basically
I
was
like
well,
my
probably
so
my
kind
of
Aboriginal
person
right
now
was.
Are
we
have
enough
intention
on
the
original
or
we're
not
the
original
of
the
kubernetes
kubernetes
issue,
d69
six,
nine,
nine
good.
B
C
That's
what
this
issue
is
about,
it's
about
changing
that,
so
they
don't
do
that,
but
anyway
they're
they
haven't
well
I!
Guess
this
issues
only
26
days
old
I
thought
there
was
an
older
issue
as
well
that
just
as
is
still
open,
but
nobody
was
making
progress
on
so
since
we
have
a
way
to
work
around
it
and
it's
a
pretty
big
issue
for
some
work
users,
yeah
I,
propose
we
go
ahead
and
make
a
change
on
our
side.
C
B
I
think
we
need
to
change
it
just
for
some
reading
also-
and
you
could
be
near
this
issue,
that
seems
like
a
bigger
change
in
point
of
there
seems
to
be
something
available,
but
that's
the
continued
runtime
interface
or
something
like
that
does
not
apply
the
selinux
tough,
but
you
container
runtime
does,
which
kind
of
all
sit
down.
Well,
the
FS
cool
thing
being
well
being
probably
a
thing
we
can
fix
bullying.
We
would
need
to
look
into
that.
If
that
also
is
fixed
by
ropes
doing
it.
B
D
B
A
All
right
Travis
does
that
cover.
Then
the
discussion
you
wanted
to
have
for
these
items
yeah
all
right
excellent.
Maybe
we
can
quickly
get
into
this
or
we
could
take
it
offline,
I'm.
Okay,
with
that
as
well,
I
think
there
were
two
outstanding
so
from
my
effort
recently
to
walk
through
our
demos
and
make
sure
all
the
storage
shredders
are
working.
A
B
A
B
A
B
B
Library
doesn't
really
have
support
for
search
parameter
in
the
resource
machine
or
they
they
said
they
added
it,
but,
like
a
good
amount
of
users
are
still
reporting
that
issue
and
to
make
it
some
to
live
for
I,
think
Alpine,
3.5,
3.6
or
something
around
it,
or
even
a
bit
valia,
at
least
for
my
images
that
are
used
to
them.
I
had
to
switch
to
something
like
deviant,
which
will
uses
chillip,
see,
I,
think
there
was
even
Alpine,
which
should
appear
something,
but
it's
and
be
nice
to
put
luck.
B
A
B
Quickly,
a
point
for
that,
as
I
wrote
in
my
comment,
I,
don't
think
there
is
really
a
kubernetes
api
for
that
as
well.
You
know
you
normally
have
exactly
those
search
parameters
that
you
don't
need,
and
it's
I
think
it's
even
that
bad
from
practice
which
is
considered
bad
to
use
because
well,
if
you
go
to
another
class
and
it
has
something
different
than
clustered
local
you're,
basically
screwed
when
actually
changed
everywhere,
whether
you
have
it
so
I,
don't.
A
B
A
So
maybe
we
could
take
this
offline,
but
one
last
question
I
think
is
so
what
do
you
think
would
be
ideal
for
fixing
this
soon?
If,
if
Mineo
does
have
a
problem
with
you
know,
dns
resolution
or
you're
using
you
know,
search
parameters
and
such
in
resolved
Kampf?
What
do
you
think
would
be
the
right
thing
to
do.
B
A
D
A
D
B
Yes,
kind
of
more
or
less
just
as
an
update.
B
B
You
know
I've,
basically
edited
when
we
delete
the
file
system
during
the
integration
test,
a
test,
it's
actually
deleting
the
file
system
and
you
can
see
I
had
some
time,
testing
it
or
playing
around
with
and
yeah
I'm,
not
sure
why
we,
why
does
an
issue
for
the
condition
than
Indian
words
like
check
if
any
MDS
pots
are
still
there
or
yeah
I?
Think
I
would
need
to
get
with
Travis
together
to
see
if
he
knows
what
is
going
on,
because
right
now
it
it
doesn't
really
make
sense
what
the
problem
there
is.
A
C
Just
take
quick,
I
guess
time
checked
for
the
the
release,
so
we're
we're
talking
about
basically
the
end
of
November
having
the
or
dot
nine
releases,
our
target
to
give
only
a
few
days,
wiggle
room
before
coop
gun
right,
so
just
to
be
clear.
That's
where
we
still
are
I
think
we're
coming
down
to
maybe
at
the
next
meeting.
We
need
to
really
clean
up
the
nine
board
for
what's
realistic
or
I
mean.