►
From YouTube: 2019-06-18 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
B
So
we
have
the
1.2
release
last
Friday,
or
was
that
we
could
go
already.
Okay,
so
that
that's
out
it's
been
resolved.
There
definitely
a
couple
more
issues
that
we
want
to
get
patched
at
some
point:
I,
don't
know
how
urgent
they
are,
but
I
haven't
reviewed
the
board
this
morning
to
see
what's
needed
in
there,
but
there
have
been
a
couple
instances
of
those
like
high
CPU
OS
Dee's,
not
running
Your
Honor
I.
C
B
B
B
A
B
B
D
B
A
B
A
All
right,
so
we
can
move
on
to
the
community
topics
here.
The
first
one
that
we
had
on
the
list
here
is
that
new
change
approval
process
has
been
merged,
has
been
ratified
in
merge
two
master,
so
you
can
read
more
about
the
details
here
for
the
process
and
it's
linked
from
the
agenda
document
today.
A
But
basically,
this
this
model
is
modeled
after
the
kubernetes
upstream
change
approval
process-
and
this
enables
we've
been
talking
about
this
for
a
while,
but
this
enables
the
storage
riders
to
have
more
autonomy
and
independence
on
getting
their
changes
merged
into
master.
So
we
have
a
set
of
reviewers
and
approvers,
and
you
know
that
so
which
expands
the
list
of
folks
that
can
you
know,
review
pull
requests
and
help
facilitate
get
them
merged
into
master.
So
this
will
unblock
a
lot
of
storage
fighter
teams
from
waiting
for
a
pull
request.
Review
from
the
maintainer
team.
A
So
I
think
that
this
was
going
to
help
the
velocity
on
a
lot
of
the
storage
providers
work.
So
if
there's
book
will
start
implementing
this
process
today,
since
it's
now
been
ratified
and
merged
into
master
and
there's
some
automation
that
we
would
like
to
improve
in
the
future
that
I've
captured
in
33:20
around
potentially
using
the
kubernetes
prowl
system
to
to
have
bots
that
enforce
and
assign
pull
requests
to
users
and
automatically
merge
and
do
all
that
interesting
fancy
stuff.
But
right
now
it's
a
manual
process
that
is
completely
described
in
the
contributing
guide.
B
Yeah
I
really
see
the
bot
as
being
a
critical
piece
of
the
next
step
so
that
we
can
make
sure
storage
the
fighters
are
controlling
all
of
their
areas
and
Irbil
proper
goal
of
rupal
is
needed
for
common
areas
or
anyway,
it'll
just
be
a
good
thing
to
have
the
bots,
because
right
now
it
is
manual
and
it
will
require
people
respecting
that
their
areas
of
ownership.
You
know
nothing
technically
prevents
someone
from
clicking
the
merge
button
in
another
area
right
now.
It
just
needs
to
be
respected.
A
It
is,
it
is
a
manual
process
right
now,
so
there
is,
you
know
now
that
it's
written,
ratified
and
merged
master,
which
was
just
like
an
hour
or
so
ago.
The
next
step
here
is
to
create
the
teams
with
the
correct
permissions
and
add
members
to
those
teams,
as
defined
here
so
I
need
to
create
these
two
teams
right
after
this
meeting
here.
A
I
can
do
that
and
add
the
right
people,
as
defined
in
the
code
owners
file,
add
them
to
those
teams,
and
so
then
that
will
give
them
permission
to
you
know
to
merge
and
have
write
access
and
all
the
right
things
will
be
there.
So
they
have
that
captured
here.
It
has
a
follow-up
as
well
cool,
so
so
I'm
glad
that's
in
master
and
apologies
that
it
took
a
little
bit
longer
than
it
then
I
had
intended,
but
it
is
there-
and
this
will
hopefully
relieve
some
pain
for
everybody
now
mic
thingy.
E
It
won't
be
a
step
Lacan,
mostly
because
putting
on
a
cephalic
haunting
could
be
really
costly,
with
the
multiple
multiple
tracks
and
getting
everything
organized.
So
instead
we're
looking
at
just
doing.
We
typically,
you
know
we
more
smaller
events
are
set
days
which
are
a
one
track
full
day
event
with
content,
and
but
this
one
in
particular
we
would
be
pitching.
It
has
both
of
Rick
and
set
day,
so
it
won't
just
be
just
such,
but
and
people
can
expect
more
rep
content.
There,
imbalance
with
stuff.
E
We
would
be
open
to
that
as
well,
so
it
would
be
something
that
we
would
organize
from
the
foundation
standpoint
on
getting
sponsors
and
whatnot
to
to
get
together
for
putting
in
the
venue
and
the
catering
and
whatnot
together.
So
but
as
I
mentioned
before,
as
opposed
to
just
a
Saturday
I
mean
if
we,
if
we
pitched
it
both
as
a
rep
and
Saturday,
it
would
be
more
than
just
like
say
one
talk
on
work.
There
would
be
more
talks
on
Rick,
so.
B
E
That
can
kind
of
help
me
with
some
ideas
for
venues
or
checking
things
out
if
something
gets
picked
out,
but
otherwise
it'll
just
be
funded
through
this
foundation,
so
the
sponsors
could
go
pay
through
the
foundation
and
it'll
just
be
like
I
mentioned,
like
a
one-day
of
that
we
do
a
bunch
of
stuff
days
already
and
different
countries,
but
as
opposed
to
this
just
being
a
set
they'll,
be,
of
course,
a
reckons
s
day.
So
we'll
be
looking
for
more
content
on
trip.
A
Okay,
I
got
you,
so
it's
it's!
It's
so
like
with
the
Colo
official
co-located
events
like
there's
a
whole
process,
and
you
have
you
know
you
have
to
get
the
like
the
room
on
site
there
and
stuff,
but
this
can
be
like
a
more
casual
thing
like
at
another
venue
in
San,
Diego,
somewhere
near
downtown
and
near
the
Convention,
Center
and
stuff
like
that
and
just
kind
of
like
in
our
set
day
but
more
of
having
more
involved.
Yes,.
E
And
yeah
and
probably
will
look
at
maybe
doing
it
the
day
before
Cuba
in
San
Diego
and
then
we
also
have
John.
We
did
the
Brooke
and
Seth
tutorial
book
on
Barcelona
and
he
also
lives
in
San
Diego.
So
we
can
definitely
use
his
content
as
well
we're
putting
on
a
tutorial.
We
can
kind
of
talk
about
how
we
want
to
structure
that
and
with
a
one
day
of
that
or
if
we
want
to
do
like
tutorial
first
and
then
the
sessions
the
next
day.
E
A
That
sounds
pretty
cool
Mike.
Is
that
a
John
Johnston
hrus
that
you
yeah
I
just
had
lunch
with
him
last
week,
I
know
exactly.
He
runs
some
meetups
here
in
San
Diego
that
I've
spoken
at
recently
about
Brooke
he's
a
good
guy
and
yeah
that'd
be
cold.
Yeah
I
definitely
have
them
looped
in
that
that
sounds.
That
sounds
interesting.
Is
there
you
know
you've
done
you've
done
some
other
events
before
Mike
is
there?
A
E
I,
don't
really
have
any
secret
process
for
that
I,
just
I,
we
we've
put
on
events
in
different
locations
and
I
kind
of
have
a
seal
is
like
why
how
it
successful
those
events
were
based
on
you
know
just
previously
putting
them
on
in
those
locations.
I
will
say
that
one
thing
that
we
did
notice
from
the
tenants
for
cephalic
on
was
that
we
actually
just
benefitted
straight
off
of
stand-alone
registrations
and
as
opposed
to
registrations.
I
came
through
people
registering
through
cube
con
and
then
also
adding
on
instep
lacan.
E
So
there
were
more
stand-alone
registrations,
which
I
thought
was
interesting,
so
people
actually
went
out
of
their
way
from
different
countries
to
attend
Steph
Lacan.
So
I
will
say
that
we
kind
of
have
a
little
bit
of
traction.
There
I
feel
like
we'll
have
the
same
benefit.
We've
been
having
a
lot
of
people
asking
for
the
cephalic
on
to
be
put
on
in
the
US
I.
Don't
think
we're
quite
ready
for
that
yet
just
because
of
dealing
with
visas
and
things
like
that
in
the
u.s.
E
right
now,
but
a
set
day
co-located
with
cube
con
would
definitely
be
doable
through
the
budget
that
we
normally
have
for
these
events
and
typically
for
the
size.
For
these
events,
we
look
at
about
150
people
and
we're
pretty
good
about
filling
up
that
those
spots
for
these
one-day
events.
If
we
want
to
grow
that,
we
can
do
that
as
well.
We
just
have
to
kind
of
talk
about
the
budget
and
figure
that
out
in
terms
of
how
many
people
were
expecting
and
then
pitch
a
price
to
potential
sponsors,
cool.
A
E
A
In
terms
of
reach
scope
of
that
type
of
messaging,
that
slack
is
gonna,
be
a
way
better.
Take
for
you
I
think
the
the
mailing
list,
Google
forums
mailing
listing-
is
fairly
low
traffic,
so
I
you
can
feel
free
to
pitch
that
on
slack,
that's
perfectly
acceptable
and
saying
you
know,
get
some
in
see
who's
engaged,
or
you
know,
interest
and
stuff
like
that.
That's
great
cool.
C
B
What
really
is
the
book
charter
and
what
you
know,
what
does
it
mean
to
encourage
storage
providers
to
join,
and
you
know
how
what's
anyway,
what's
how
can
we
build
and
grow
that
community
of
again
story?
Writers
is
kind
of
ultimate
question
and
then
there's
other
questions
that
fall
out
of
it
like
how?
How
can
storage
providers
have
their
own
ship
schedule
or
give
aid
they
need
their
own
ship
schedule?
B
B
B
D
A
Alright,
let's
see
so
Jose
opened
up
a
poll
request
recently
about
the
device
device
sets
proposal,
design
and
Jose
I.
Think
you're
on
the
call
here
today
and
thank
you
very
much
for
incorporating
the
early
feedback
around
that
this
functionality
could
be
useful
for
in
a
general
sense
and
all
storage
writers.
So
I
really
appreciate
you
taking
that
feedback
in.
A
F
So
all
right,
based
on
a
comment
made
just
recently
by
Blaine,
which
I
agree
with
I,
think
it
would
be
helpful
to
describe
this
broad
more
accurately
as
a
storage
device
as
a
storage
class
device
set,
because
the
purpose
of
this
construct
is
to
consume
PVCs,
intended
for
underlying
storage
via
storage
classes.
F
That
is
its
ultimate
goal,
and
the
idea
here
is
to
slightly
abstract
away
the
notion
of
an
underlying
storage
device
such
that
the
main
thing
you
would
do
is
simply
specify
a
count
of
how
many
devices-
and
you
want
in
this,
set
the
capacity
that
each
device
should
have
in
that
set
and
the
storage
class
those
devices
are
provisioned
from
now.
Obviously,
this
is
mostly
targeted
for
cloud
environments,
where
kubernetes
allows
you
to
dynamically
provision
remotely
attachable
storage
like
AWS
EBS
volumes.
F
F
A
I'll
say:
have
you
heard
any
recent
updates
about?
If
the
you
know,
local
volume
provisioner
would
get
dynamic,
provisioning
enabled
I
thought
I've
always
thought.
That's
been
kind
of
a
weak
point
of
it.
So
far,
at
least
from
a
usability
perspective
that
it
makes
it
harder
to
use
local
volumes
is
that
you
have
to
statically
provision
them.
Is
there
been
any
update
that
you've
heard
on
on
dynamic
provisioning.
F
No
update
at
this
time
in
particular
just
because
it's
a
really
nasty
problem
yeah.
So
so
you
know,
as
I
mentioned,
the
core
use
case,
for
this
is
cloud
environments,
I'm,
simply
leaving
the
option
open
to
use
storage
classes
backed
by
local
PBS.
If
a
user
wants
to
simply
manage
their
devices
that
way
right.
A
F
In
particular,
AWS
has
a
notion
of
local
ephemeral
storage,
where
the
instance
is
provisioned,
with
storage
devices
already
attached,
but
of
course
that
would
only
show
up
as
local
devices
on
the
note,
so
you
could
specify
you
could
either
statically
provision
local
Peavey's
against
the
devices
you're
provisioning
or,
if
you're
wrapping
this
up
into
some
broader
app.
That
has
knowledge
of
the
instances
you're
creating
you
could
write
a
small
reconciler.
You
could
write
a
small
controller
to
automatically
create
those
local
PBS
and
configure
them
to
be
to
provision
against
a
storage
class.
F
See
your
comment:
I
initially
took
it
and
said
yes,
this
could
currently
make
sense
to
just
to
just
provision
it
or
to
abstract
it
away,
so
that
it
could
be
used
across
rook
Opera
across
multiple
rook
operators.
If
they
so
chose
to
implement
it,
then
I
had
a
thought
and
thinking
is
dangerous.
So
the
main
issue
here
is
that
for
rooks
F
in
particular,
I
was
naively
unaware
that
you
can
define
multiple
OS
DS
per
storage
device
and.
B
C
F
But
yeah
it's
discussions
of
its
discussion
of
these
kinds
of
features
and
ideas
that
make
me
think
it
might
not
be
useful
to
try
and
make
this
generic
for
other
rook
operators,
but
if
there
could
be
such
a
design
where
we
could
just
leave,
you
know
similar
to
what
the
SEF
cluster
CR
has
now.
If
we
could
just
leave
a
config
section
of
arbitrary
key
of
arbitrary
string
string
pairs
for
storing
key
value
configuration,
then
each
operator
could
individually
specify
or
read
certain
or
whatever
options
at
once
from
that
list
of
string
string,
pairs.
F
The
only
reason
I
wouldn't
want
to
do.
That
is
because
it
is
anathema
to
my
current
understanding
of
kubernetes
design
philosophies
which
I
don't
know
how
much
we,
how
much
is
desired
to
keep
that
in
rook
right,
because
rook
is
not
kubernetes,
it
does
not
need
to
adhere
to
those
design
principles.
All
I
know
is
that
kubernetes
would
rather
avoid
arbitrary,
arbitrary
fields
of
string
to
wherever
possible.
A
Yeah,
you
definitely
see
that
on
storage
classes,
which
can
make
things
kind
of
it's
nice
to
have
a
flexibility,
but
it
also
can
make
you
know
provisioning
a
bit
difficult.
If
you
have
you
know
if
you
want
to
express
more
complicated
types,
but
yeah
that
storage
classes
at
least
do
that.
You
know
arbitrary
string,
string,
key
value
pair
configuration
style,
yeah.
F
So
they
sound
really
keep
in
mind
a
dish
to
tackle
this
since
I.
Don't
want.
You
know,
since
this
could
eat
up
a
good
chunk
of
time.
There
are
current
there's
currently
a
thought
to
hold
a
meeting
tomorrow
to
further
deeply
discuss
the
implementation
details
of
this
design,
and
if
so,
there
is
a
current
SEF
community
meeting
called
the
container
huddle,
and
we
were
thinking
about
using
that
meeting
time
to
go
into
this.
B
To
open
it
up
to
you
know,
it's
not
just
a
set
meeting,
it's
with
the
stuff.
This
to
us
discussed
the
device
set
design,
so
others
are
certainly
welcome
to
join
and
we'll
we'll
get
the
meeting,
or
at
least
the
blue
jeans
link
out
there
for
others
who
want
to
join
so
be
a
half
hour
before
this
meeting
tomorrow
and.