►
From YouTube: 2019-03-12 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
Okay,
the
recording
is
started,
and
this
is
the
March
12
2019
Brook
community
meeting,
so
we
will
go
ahead
and
start
with
looking
at
the
0.9
pax
releases.
We
just
did
one
recently
so
I.
Don't
think
I
need
to
get
this
thing
off
my
screen,
it's
in
the
way
there
we
go
I.
Don't
think
that
we
have
much
on
here
to
worry
about
right
now
that
I
remember
at
least
yeah.
B
A
A
B
A
A
Kicking
out
or
punting
some
of
the
issues
we
know
we
don't
want
to
do
in
1.0
and
then
also
start
pinging.
Some
of
the
other
storage
provider
owners
like
pinging
Dmitri
about
sorry
not
to
be
true,
yeah
us
about
some
of
the
Cassandra
stuff
and
then
pinging
Dmitri
for
at
Jeff
s,
etc.
I
think
I
think
that
there's
a
fair
amount
of
these
that
are
just
not
gonna,
be
in
the
1.0
milestone,
Dimitri.
A
A
A
A
B
A
I
think
that
it's
a
little
hard
to
have
that
conversation
to
when
we
don't
have
an
accurate
picture
of
the
full
scope
that
we
want
to
include.
Like
you
know,
I,
don't
think
we've
identified
what
the
critical
issues
that
you
know
that
we
don't
want
a
ship
1.0
without
so
I
think
that's
kind
of
a
first
step
that
we
would
need
to
be
have
some
more
clarity
most
likely.
What
we're
going
to
want
to
do,
though,
is
you
know,
I'm
not
married
to
the
exact
last
week
of
March
I.
B
Yes,
and
before
we
can
even
commit
to
that,
we
need
to
commit
well.
What
are
well
are
we?
What
are
we
shooting
for
in
their
lease?
And
if
this
really
is
a
time-based
release,
we
have
to
get
more
strict
about.
You
know,
making
it
a
time-based
release
and
not
a
feature
based
release
type
of
thing.
I
think
that's
the
challenge.
We've
we've
been
in
a
feature,
release
mentality.
You
know,
since
the
beginning
of
the
project
which,
which
is
worked,
fine,
but
are
we
a
time-based
release,
in
which
case
you
know
how
do
we?
A
Well,
I
mean
yeah
and
I
think
that
that
also
requires
having
a
well
informed
understanding
of
what
features
are
required.
You
know
like
if
there's
something
that
would
block
the
release
that
then
we
would
not
ship.
Without
that,
then
you
know,
being
a
strictly
time-based
release
becomes
a
little
bit
more
malleable.
I
would
say
you
know
so.
D
D
Preventing
different
rook
backends
from
releasing
independently,
like
it
seems,
I
think
by
my
logic.
It
seems
like
all
the
back
ends
like
they
have
their
own
containers
and
could
theoretically
like
have
their
own
release.
Cadence's
and
then
rooks.
Release
process
could
be
more
around
like
how,
like
it's
evolution
in
the
shared
utilities
to
all
of
the
backends.
A
Yeah
I
think
that
we
don't
have.
We
just
don't
have
the
infrastructure
for
that
right
now,
blade.
You
know
with
our
you
know
automated
release
process
it
doesn't
allow
it
doesn't,
have
the
ability
to
filter-
or
you
know,
specify
which
specific
providers
you
would
want
to
include
in
a
release
and
I
think
gotcha
yeah,
and
we
don't
really
have
you
know
facilities
right
now,
either
for
kind
of
managing
versions
across
providers
when
they
diverge
that
hasn't
really
been
defined.
A
Yet
of
how
that
you
know
is
handled
exactly
so,
we
do
need
to
do
some
thinking
through
it.
I
think
that's
where
we
want
to
go.
We
probably
even
eventually,
you
don't
want
to
go
to
potentially
to
a
model
where
you
know
each
storage
provider
has
their
own
repo.
Even
you
know,
source
code
levels,
so
there's
definitely
work
to
do
to
break
apart
kind
of
the
you
know,
single
rook,
monolithic,
I,
guess
for
lack
of
a
better
term
approach,
and
we
can
start
having
those
conversations,
but
we
I
don't
think
yeah.
D
B
Between,
what's
upstream
and
what's
you
know,
people
have
gonna
have
their
own
separate,
downstream,
builds
potentially.
So,
if,
if
we
think
about
that,
maybe
it
would
stay
together
long-term
and
you
know
but
yeah
it
would
be
nice
to
separate
for
sure
I
guess
it
is
just
a
resource
question
right
time
issue
a
refactoring.
A
A
B
D
A
Yeah
this
is
I
mean
this
is
kind
of
capturing
the
beginning
of
that
idea,
but
it
was
kind
of
scoped
to
promotions
and
channels.
You
know,
like
specific
storage
providers,
having
you
know
specific
or
differing,
you
know
versioning
graduation
type
of
processes.
So
I
don't
think
this
captures
that
completely
yeah
yeah.
So,
let's,
let's,
let's
get
an
issue
on
that.
D
B
C
A
C
A
C
B
D
B
A
A
Okay,
let's
say:
let's
go
ahead
and
move
on
then
yeah
I
am
so
I
am
so
I
mean
I'm
at
the
cloud
Expo
you're
up
here
in
London
and
I'm,
giving
a
couple
of
demos
for
Brooke
both
in
the
talk
I'm
giving
tomorrow
morning
and
then
also
at
an
open
shift
like
user
group.
That's
here
in
London
tonight,
I'll
have
to
leave
for
it
like
45
minutes
or
so,
and
so
I
was
just
trying
to
get
the
no
usual
normal.
A
My
sequel,
WordPress,
it's
you
know,
Seth
demo,
working
with
mini
cube,
then
I
ran
into
a
couple
of
issues
with
it.
That
I
was
surprised
that
I
ran
into
there's
a
couple
that
I
don't
know
why
they're
happening
and
I'm
just
kind
of
working
around
them,
and
so
I
wanted
to
bring
some
of
them
up
here
to
you
know
a
see,
if
there's
anybody
has
you
know
good
ideas
about
them
or
any
hints
or
leads,
because
that
would
be
useful
for
it's,
not
gonna.
A
I'm
not
gonna
have
enough
time
before
it
night
at
the
OpenShift
meetup,
but
before
I
present
tomorrow
at
the
conference,
I
might
have
time
to
do
something
about
them.
So
this
has
anybody
seen
this
with
mini
cube
where
you
use
mini
cubed,
an
SH
up,
and
then
that
ends
up
failing
with
when
let's
try
to
create
a
symbolic
link.
Has
anybody
seen
this
issue
before.
A
A
B
A
B
A
A
C
B
A
Oh,
that's
what
you're
doing
I
see
yeah
I
had
to
create
a
new
snapshot
here,
that
you
know
that
I'm
gonna
be
using
while
I'm
in
London,
but
your
workaround
is
to
continue
using
what
what
what
does
work?
Yes,
that
it
is
okay,
the
next
one
I
want
to
talk
about
this.
This
is
still
still
happening
here
where
the
mini
qvm
just
goes
kaput
and
you
know
everything
hangs.
You
can't
access
anything
good
ole.
Mr.
mr.
A
Watkins
ran
into
this
as
well,
and
so
his
workaround
of
updating
the
PBC
requested
storage
size
from
22
to
did
seem
to
work.
For
me
as
well,
I
was
I,
was
hey
I
hit
it
multiple
times
this
morning
and
then
I
made
that
change,
and
it
works
the
first
time
after
that,
I
don't
know
if
that
means
that
that
hundred-percent
fixes
it
or
not,
but
it
did
help
the
first
time.
I
tried
it
at
least,
but
I,
don't
think
anybody
knows
what's
going
on
here.
Still
it's
quite
perplexing.
A
B
B
A
Yeah,
the
weird
thing,
though,
is
that
you
know
if
it's
backed
by
a
RBD
image-
and
those
are
you
know
like
thinly
thinly
provisioned-
that
you
know
the
size
I
would
think
shouldn't
matter.
I'm
surprised
that
that
would
have
an
impact.
I,
don't
know
what
exactly
is
going
on
or
under
the
covers,
but
you
know,
and
sometimes
it
does
work
too.
So
it's
like
a
race
condition,
type
of
thing,
yeah.
B
A
C
A
I
know
Noah,
try
it.
He
gave
me
eight
gigs
to
lose
to
and
still
still
died,
so
it
does
not,
and
whatever
every
time
and
every
time
I've
looked
at
this
so
far
to
it
yeah
it
doesn't
seem
to
be
a
memory
pressure
issue.
There's
like
not
the
load
average
is
there's
nothing
like
its
it.
You
know,
there's
not
we're
not
running
into
using
any
swap
or
anything.
It's
doesn't
seem
like
a
memory
issue
on
the
VM.
A
C
A
C
A
This
is
this
seemed
to
be
an
easy
one.
It
looks
like
we
missed
a
when
we're
updating
to
use
apps
to
be
one
deployment
everywhere.
It
looks
like
we
missed
the
I
think
the
my
sequel
one
and
the
WordPress
one.
We
didn't
update
to
use
a
required
selector,
which
is
apparently
part
of
the
required
validated
field
in
the
deployment
specs.
So
this
should
be
easy
to
update,
but
it
is
it's.
It's
it's
occurring
with
our
the
manifests
that
are
committed
to
master
right
now,
that's
a
little!
That's
an
easy
one.
So
do.
A
I
in
the
interest
of
time
today,
when
I
was
running
into
a
lot
of
things,
not
working
I
did
the
most
simple
things
possible
in
most
of
the
time.
Instead
of
fixing
things
I
put
in
workarounds
that
perverted
her
back
to
the
way
that
they
used
to
work
without
any
issues,
so
I
fix
year
was
moving
it
back
to
apps
v1
beta
one
that
doesn't
have
that
required.
Selector
I
was
doing
anything.
I
could
get
the
demo
working
today.
C
A
B
Yeah
so
it
you
know
a
month
or
two
ago,
when
I
tried,
upgrading
our
client
to
used
1.13
as
well,
and
I
tried
to
update
some
other
things
like
the
external,
provisional
provisionary
and
yeah
externa.
Probationary
code
was
still
using
G
log,
so
I
just
updating
edge
FS
didn't
work
for
me
to
be
able
to
pull
in
1.13,
so
the
Winnie
up,
so
that
sort
of
external
provisioner
needs
that
change
to
use
K
log
as
well.
A
Well,
maybe
when
I
so
I
went
through
the
exercise
of
updating
some
window.
Thirteen,
because
in
my
demo-
and
you
did
what
about
thirteen
and
so
I
I
did
a
number
of
things
to
make
that
work
here
so
like
I,
hoped
I
opened
an
issue
to
capture
that
and
including
everything
I
did
so
that
we
want
to
update
to
one
about
thirteen.
We
have
this
information
and
we
don't
have
to
stumble
through
all
these.
A
You
know
you
know
exactly
how
diff
don't
think
of
me
to
update
the
kubernetes
references
Travis,
because
you
did
it
at
112.
It's
pain
yeah,
but
as
part
of
that
I
updated,
six
storage,
Lib
external,
the
external
provisioner
to
be
2.0,
and
maybe
they
got
rid
of
G
login
that
and
that
from
you
know,
from
102
to
dotto,
and
you
started
using
K
log,
maybe
because
I
didn't
run
into
any
G
log
issue
at
all
inside
of
the
external
provisioner.
Well,.
A
Might
not
have
been
oh
God,
it's
it's
come
it's
better
than
it
was,
but
the
kubernetes
version
dependency
management
of
go
packages
is
still
kind
of
a
pain
in
the
butt
like
you
know,
you
update
things
to
1.13
and
then
stumble
like
cube.
Open
api
doesn't
have
any
releases
or
tags
or
anything.
So
you
have
to
you
know,
be
it's
very
specific
about
the
version
you're
using
or
otherwise
it'll
go
out
of
date,
and
you
know:
okay,
it
go.
The
matching
one
version
for
1.13
is:
is
v10?
Okay,
yeah
got
it.
A
It's
such
a
such
a
headache
and
it's
a
the
iteration
cycle
to
you
know,
run
run,
make
vendor
and
have
depth
build
out
its
entire.
You
know
graph
and
try
to
solve
all
the
dependencies
and
then
something's
missing,
but
this
so
I've
captured
all
that
work
here,
and
you
know
obviously
like
that
change
is
not
gonna
go
in
today
or
may
prime,
maybe
not
name
of
one
that
oh
I
don't
know
because
that
will
require
you
know
a
fairly
comprehensive.
A
B
D
A
Don't
know
all
I
know
is
that
it
was
panicking
and
then
then
it
stopped
once
I
changed
the
edge
fs1,
so
I
didn't
look
into
it
further
than
that
all
right.
So
that's
I,
think
those
are
all
the
issues
and
things
that
I
ran
into
while
getting
this
demo
up
and
we
have
everything's
being
tracked
so
when
I
have
workarounds
for
everything
right
now,
so
we
should
be
okay
for
now
and
then
we'll
follow
up
on
these
issues
to
get
the
long
term
fixes
in
and
we
can
move
along,
I
think
Travis.
B
B
C
I
want
to
know
if
they
assign
either
I
think
it
would
be
good
if
you
can
look
into
TGIF
or
help
with
what
that
we,
like
with
the
kubernetes
project.
When
you
have
like
comments,
letters
like
slash
a
sign
or
something
like
that
that
we
have
this
automatically
so
really
basically
still
have
to
society
group,
but
that
it
will
happen
automatically
as
right
now
at
least,
we
still
need
to
do
a
little
signing
as
far
as
I
know,
yeah.
C
C
A
I
think
it'll
be
it'll,
be
good
to
have
a
clear
understanding
of
what
scenarios
we
want
to
solve.
You
know
like
what
the
workflows
are
like
if
we,
if
there
are
other
things
that
need
to
be
set
up
like
organizationally
or
process-wise
it
to
be
able
to
effectively
consume
those
BOTS
like
there
may
be
some
things
that
we
don't
have
set
up,
that
the
bot
itself,
you
know,
won't
solve
on
its
own.
So
thinking
through
that
a
little
bit
and
understand
me
what
bots
are
available
to
us,
like
all.
A
A
A
D
C
Yeah,
maybe
a
bit
less
of
our
work
there,
but
maybe
just
asking
if
they
have
something
I'd
like
that
on
the
roadmap.
I
can
see
that
they
maybe
look
into
something
like
exactly
for
a
while
imported,
like
that,
where
we
just
wanted
people
to
have
issues,
but
not
what?
If
I
like
a
projecting
monster,
yeah.
A
C
B
Yeah
just
talk
to
Sebastian
on
this
morning
and
yeah,
it's
kind
of
a
question
to
overall
about
how
many
users
do
we
have
of
of
arm
builds
I
just
glanced
through
our
docker
hub
repos
and
there's
like
on
the
order
of
hundreds
of
dr.
pols
for
the
arm
repos.
Where
there's
you
know
millions
for
the
for
the
amd64,
so.
A
B
C
C
D
C
For
maybe
you
already
have
something
going
there,
but
yeah
so
yeah
yeah
yeah
thankful.
So
if
you
would
need
to
see
because
I
from
some
people
I'm
more
or
less
her
here,
yeah
here,
like
yeah,
if
my
old,
raspberry
and
I
want
to
try
it
out
kind
of
like
them,
but
the
problem,
we
were
very
first
various
that
at
least
till
the
latest
one
is
Raspberry
Pi
2.
C
C
B
B
B
A
It's
a
good
question:
I
am
and
I've
always
been
a
little
caught
like
curious
about
the
poll
numbers
on
docker
hub
and
with
the
usage
of
the
manifest
tool
and
all
that
and
be
direction
and
stuff
I.
It
wouldn't
surprise
me
if
you
know
some
of
those
arm
numbers
aren't
getting
captured
correctly,
so
it
could
be
that
there's
more
to
me
and
then
what
is
indicated
right
now,
I'd
be
cautious
about
that
yeah.
C
A
Was
just
saying
that
arm
compared
to
a
to
x86
is
like
a
hundred
verses
and
you
know
million.
So
it
seems
like
it
was
a
massive
difference,
but
I
I've
always
been
a
little
skeptical
of
dr.
numbers
work
when
you,
because
of
the
manifest
tool,
is
supposed
to
specifically
get
the
right
version
at
runtime
and
I.
Don't
I've
always
been
a
little
little
suspicious
of
the
numbers.
Honestly
I
see.
D
C
A
The
yes,
the
I,
don't
know,
did
anybody
get
talk
accepted
they
sent
out
the
notifications
for
accepted
and
rejected
cube
con
talks
for
Barcelona
last
night
and
I
I
know
I
got
one
accepted,
and
then
we
have
the
intro
a
deep
dive
sessions
as
well
accepted
for
rook.
Did
anybody
else
get
me
any
good
news
about
a
talk
being
accepted,
not
I.
B
C
D
D
B
C
A
Where
you're
you
know
mirroring
to
another
zone
around
the
globe,
or
something
like
that,
because
I
when
I
wrote
that
abstract
that
wasn't
entirely
sure
what
we
had
to
show
and
it
got
accepted
so
now,
I
have
to
figure
out
what
the
that's.
The
big
content
of
the
talk
is
gonna,
be
some
thinking's.
Its
f
and
edgy
FS
will
be,
will
be
able
to
show
off
something
something
cool
from
from
those
that
has
to
do
with
multi-site
multi
cloud
type
of
you
know:
data
storage.
D
D
On
a
similar
note,
Jared
as
far
as
like,
like
I
guess,
integrations
with
rook
I'm
I've,
been
working
with
one
of
my
colleagues
here
to
try
to
figure
out
like
what.
What
methods
can
we
use
to
connect
rook
to
OpenStack
clusters
and
like
or
is
there
any
work
in
rook
or
in
OpenStack?
That
will
allow
it
to
go
to
go
better,
but
my
presentation
should
have
a
little
bit
of
that,
but
it'll
mostly
be
a
rook
introduction
at
the
end
of
April.
A
A
That
was
a
long
time
ago
and
I
think
that
he
did
something
with
you
know.
Sender,
basically
and
I
have
no
idea
what
it
was,
but
that
might
be
a
talk
that
you,
maybe
you
can
look
up
or
we
can
find
for
you
blame
that
shows
what
how
Steve
integrated
Brooke
with
with
OpenStack
cinder,
to
give
you
ideas
if
you
need
them,
yeah
that.