►
From YouTube: 2017-03-07 17.01.37 SIG-cluster-lifecycle 166836624
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
But
I'm
happy
to
like
you
know,
let's,
let's
try
and
figure
out
who
we
actually
deal
with.
I
think
the
release
coordinator
from
gosh
names
escaping
me
right
now.
It's
probably
the
right
guy
to
to
talk
to
about
figuring
out
who's
handling.
That
arm
we
need
to.
Let's
see,
we
need
to
finish
up
other
documentation
I'm
on
the
hook.
For
some
of
the
token
stuff
I
can
try
and
clean
up
some
other
documentation
is
necessary.
A
A
On
the
the
deb
packages
for
1.6
turn
on
the
secure
flags
for
the
cube
list,
the
RPM
has
not
been
done
yet,
so
we
need
to
make
sure
that
we
get
the
flag
changed
in
the
RPM
and
I
have
at
least
one
customer.
That's
been
poking
me,
that's
kind
of
like
scared
that
we
remove
old
version
depositories
and
so
there's
a
bug
that
we
should.
A
B
A
So
this
is
why
I
mean
this.
Is
why
so
and
then
the
other
thing
that
we
should
do
is
on
the
on
the,
and
this
is
just
a
general
thing-
is
I:
have
instructions
for
how
to
install
the
latest
112
of
docker,
because
we're
going
to
start
failing
prereqs
for
the
latest
version
of
docker
and
so
like
we
should
eat.
You
know
like
those
aren't
officially
supported,
so
we
should.
We
should
probably
just
notify
users
of
what's
going
on
there
and
give
them
instructions
on
the
easy
way
to
get
the
old
versions
of
docker.
B
B
A
B
B
A
Okay,
okay
sounds
great,
I'm
going
to
be
at
GCT
next
later
this
week,
so
it's
going
to
be
a
busy
couple
of
days
coming
up
one
of
the
other
things,
and
this
is
longer
term.
It's
not
going
to
make
it
for
the
16
really
is
all
the
release
for
the
scripts
and
architecture
for
building
the
debs
pushing
using
these
things.
A
All
that
stuff
was
only
written
to
be
run
by
Googlers,
and
so
what
it
means
is
that
the
best
we
can
do
for
testing
sort
of
the
latest
cube
admin
is
to
sort
of
like
kabul,
some
stuff,
together
with
the
hypercube
image
and
manual,
pushing
and
stuff
like
that.
So
it's
really
not
a
proxy
for
sort
of
the
full
release,
artifacts
and
so
I
think,
probably
in
the
17
cycle.
A
E
Do
so
I
have
a
batch
of
PRS
that
should
be
ready
this
afternoon
that
fixed
our
current
continuous
integration,
testing
and
n
4q
badman.
So
if
the
debs
it
brings
up
a
cluster,
you
mean
the
newest
version
of
cube
admin,
I'm,
also
hoping
to
make
that
a
PR
blocking
job
later
this
week,
if
it's
pretty
stable
and
all
that
infrastructure
is
there,
it
just
keeps
getting
broken.
A
So,
like
I
still
think
that
this
is
issue
like
and
I
have
some
PR
that
I
submitted
to
the
to
the
release,
depot
and
I
think
they'll
probably
get
held
up
in
the
1.6
doctors,
everybody
busy,
but
there's
you
know
we
should
still
be
able
to
do
that
whole
thing
when
you're
not
running
under
jacobs
when
you're
doing
it
manually,
because
it's
like
right
now,
like
all
that
stuff,
just
assumes
that
you're
at
Google
and
assumes
that
you
have
you
know:
user
local,
google,
that's
writable!
You
know
I'm
impressed
like
that.
A
F
A
That
is,
you
know,
a
good
balance
between
being
real
in
terms
of
multi-node
and
also
being
fast
and
I'm
hopeful
that
the
doctor
and
doctor
stuff
can
actually
play
a
part
of
that
I
haven't
had
the
time
to
really
dig
in
and
in
figure
out
sort
of
how
close?
That
is
a
reality
and
whether
that's
going
to
be
useful
for
people
testing
stuff
out
so.
B
F
Were
you
did
you
see
Luke
for
the
doctor
and
doctor
stuff
about
supporting
a
few
proxy
flag
variable
for
something
that
happens
with
few
proxy
when
it's
inside,
because
it
expects
to
be
on
the
the
local
network
or
something
there's
a
flag
you
have
to
pass
to
the
Cuba?
Have
you
seen
this
issues?
I
haven't
gone.
F
A
Well,
I
still
think
we
want
to
be
I
would
love
to
be
able
to
take
my
local
bills.
That
I
think
is
good
and
build
enough
artifacts
that
I
could
launch
like
a
GC
or
an
Amazon
cluster
off
of
those
things,
and
have
it
come
up
where
it's
essentially,
you
know
the
exact
same
flow
as
if
it
were
a
release
you
know
would
like
maybe
a
single
flag
cake.
A
H
A
I
There
anyone
on
the
call
from
Canonical's,
because
Red
Hat
has
the
atomic
install
which
allows
you
to
install
containers
to
which
you
can
bypass
the
repo
entirely
so
long
as
you
have
the
containers
and
assistant
dean
unit
file,
which
basically
uses
kicker,
then
you
can
bypass
the
repos
entirely.
As
your
is
your
baseline.
This
trip
well.
A
B
A
I
H
For
the
depth,
it's
all
really
possible
to
do
local
bill
just
get
clowned
police
repo
CDW
and
go
from
the
old
Galu
I
think
so
that
will
do
it
and
produce
flock
of
them
files.
But
still,
as
you
say,
it's
not
pushed
anywhere
or
something
so
you
then
have
to
install
and
manually
but
yeah.
Well
we're
looking
looking
at
it,
but
I
guess
thats
related
issue.
H
H
A
B
J
Yeah
so
I've
deployed
a
bunch
of
q,
brunette
who's
clusters.
Now
and
you
go
on
to
the
web
and
you
you
curl
something
from
like
we
for
a
plug-in
and
there's
hard
coded
IP
addresses
in
it
that
don't
tend
to
fit
well
and
the
networks
that
I've
got
so
I
have
to
set
across
them.
And
it
would
be
really
nice
to
have
like
a
templating
engine
for
some
of
the
pieces
that
we
have.
J
J
J
J
K
J
So
helm
has
values
for
its
templating
engine,
so
the
user
can
override
them
very
easily
on
the
command
line,
so
they
could
do
a
helm.
Install
we've
dash,
dash
set
IP
range,
equals
whatever,
and
then
it
would
fill
in
that
value
provided
by
the
user.
If
they
don't
choose
the
default
into
the
template
as
it
instantiates
it.
K
H
A
I
think
it's
important.
You
mentioned
two
things
that
we
would
get
out
of
helm
here.
One
of
them
is
a
templating
engine
and
the
other
one
is
the
ability
to
essentially
hit
a
dashboard
to
do
upgrade.
I
think
it's
important
to
separate
out
those
one
of
the
concerns
that
I
have
about
how
masasi
the
templating
and
the
dashboard
and
that
stuff
is
all
combined
I.
A
Like
you
know,
the
upgrade
procedure
for
your
cluster,
including
these
low-level
things,
can't
all
be
done
from
the
helm.
Dashboard
right
at
home
is
not
going
to
drive
the
upgrade
of
the
cluster
itself,
so
I
just
don't
think
that
they
usability
for
that
particular
part
of
the
system,
really,
in
my
mind,
justify
some
of
the
way
to
running
another
server
site
com.
So.
J
H
H
So
I
think
we
should
still
cube.
Admin
should
still
deploy
the
DNS
thing
I
right
now.
It's
order
detects
cluster
IP.
The
service
subnet
and
check
season
sets
the
right
value,
but
I
think
it
would
be
very
useful
I'm
on,
like
the
document
that
after
now,
you
got
your
posture
funny
and
it
has
some
command.
You
can
come
from
from
this
app
repository
as
well,
the
same
for
example,
and
then
you
we
have
some
starter
file
for
you
here.
H
J
It's
it's
a
Clark's
issue,
and
it's
also,
you
know
all
the
upstream
plugins
and
stuff
right
now
assume
that
you're
just
going
to
cubed
CTL,
create
F
URL
and
they
have
to
split
them
up
into
like,
like
canal
has
like
three
or
four
of
them
now,
because
they
have
different
options
that
you
may
want,
and
they
can't
have
no
templating
engine.
So
they
can't
like
write
just
one
with
a
couple
options
that
you
can
tweak.
D
H
J
So
having
a
template
of
engines,
I
think
is
really
important
to
to
this
and
having
some
way
of
easily
performing
upgrades
on
things
and
right
now
those
are
kind
of
lacking
I.
You
may
be
right
on
all
might
be
overkill,
but
they're
they're
kind
of
mission
seems
to
align
pretty
well
with
just
having
note
some
upstream
packages
that
you
you
can
upgrade
or
you
go
through
the
regular
life
cycle
of
the
kubernetes
application
and
like
the
dashboard
and
some
other
things
really
are
just
applications.
J
H
I
think
we
should
distinguish
between
cubed,
em
and
Combinator
school,
and
that's
the
case.
If
you
run
on
it,
I
think
the
only
thing
that's
kind
of
needed
here
is
a
document
that,
like
after
you
have
your
cube,
ADM
cluster,
you
should
maybe
install
helm
the
fun
helm
in
it,
and
then
you
have
a
lot
of
options
to
install
the
dashboard
you
might
canal
may
may
be.
It
may
have
some
helm
reports
for
you
as
well
and
make
it
easy
for
you.
H
J
There
may
be
some
kind
of
support,
at
least
for
handing
off
nicely.
How
should
I
said
it
so
like,
for
example,
if
if
Cuba
idiom
ends
up
the
playing
cube
proxy
and
the
user
might
want
to
use
a
home
package
to
manage
that,
after
its
after
cube,
ATM
is
done.
It's
its
job
to
have
enough
information
that
a
helm
upgrade
cube
proxy
would
work.
H
Yeah
it
Justin
maybe
has
something
that
just
a
quick
note.
We're
gonna
well
work
in
progress
already
to
split
up
cubed
and
interfaces.
I
have
a
duck
signed
document
for
it,
and
I
expect
it
to
be
ready.
Total
ready
in
17.
We
have
two
of
like
ten
commands,
may
be
implemented,
so
in
17
I
expect
you
to
be
able
to
run
it
run
all
the
faces
individually
and
you
can
omit
box
and
in
at
the
future,
if
you
like,
just
as
it
would.
J
K
I
mean
I
think
so
I
think
there's
a
general
move
to
try
to
make
sure
we
don't
have
to
use
templating
so
to
replace
templating
with
component
convict
and
so
another
word
you
would
have
a
config
map
and
I
think
then,
and
the
content
map
would
be
correct
by
the
salt
and
the
ideas.
The
user
would
change
the
config
map
if
they
wanted
to
change
something
and
the
controller
or
whatever
it
was,
would
automatically
reload
I.
K
J
Map
but
may
need
to
be
comforted
like
like
image
tags.
You
know,
I
may
want
to
go
from
one
for
172
to
173
right.
K
I
think
the
Greek
window-
and
it
like
tell
me
to
do
great
experience
in
terms
of
upgrading
and
I,
can
tell
you
what
cop
stuff,
which
is,
what
cops
has
a
small
wrapper
around
I
again
I
owe
a
I
owe
a
doc
describing
it
and
like
trying
unify
the
various
proposals
out
there.
It
has
a
small
wrapper
around
poop
cuddle,
which
basically
so
we
add
another
manifest
file,
which
is
like
a
list
of
all
the
versions
of
cute
dashboard
that
are
available.
You
can
go
into
the
cops,
Rico
and
might
get
some
add-ons.
K
You
can
see
some
examples,
and
so
you
can
do
it
called
channels
for
now,
because
I
don't
know
why
it's
just
me,
but
anyway,
channel
is
like
because
I
channels
ad
on
ad
ad
on
dashboard
or
something
and
it
will
fill
the
dashboard
you
can
be
like
it
will.
It
keeps
some
metadata
around,
so
you
can
upgrade
easily,
but
it
is
just
a
very
thin
wrapper
around
cuoco
to
apply
and
yeah.
K
The
problem
is,
it
doesn't
do
templating
yet,
and
there
was
I
think
I
think
a
lot
of
templating
solutions
out
there
and
there
was
a
proposal
in
core
and
it
seems
to
be
non-contingent
stop
moving
forward.
So
maybe
we
should
take
that
on
look
at
the
helm
in
sick,
caster
lifecycle.
Look
at
the
helm,
template
other
ones
that
are
out
there
decide
whether
whether
we
want
to
like
take
on
the
complexity
and
weight
of
helm.
We
might
not
even
be
at
a
heavy
right
and
then
yeah.
J
C
J
G
A
J
J
B
Think
it
would
be
good
to
see
if
he'll
interested
in
going
down
that
route.
I
also
think,
let's
go
see.
The
inside
of
your
car
Joe
I
also
think
that
it
may
well
be
worth
breaking
this
conversation
off
into
a
little
working
group
meeting
where
people
who
are
interested
in
this
topic
and
expend
a
whole
hour
flashing
it
out,
because
I
think
that
you're.
G
B
Of
finding
a
way
of
getting
some
of
the
attributes
that
helm
offers,
but
I'm
pretty
sure
that
people
can
have
lots
of
opinions
about
the
instrumentation
sure
like
can
I
suggest
that
we
that
we
have
a
separate
meeting
that
people
can
show
up
to
if
they're
interested
in
this
topic,
yeah,
okay,
let's
compare
something,
then
we
can
get
to
the
rescue
gender
who
would
volunteer
to
send
an
email
to
the
list
proposing
that
meeting,
Kevin's
hat.
B
K
I
can
try
the
first
one
might
be
one
that
is
more
kicked
off.
Another
sub
working
group
I
suspect
because
it
is
sort
of
a
big
in
dorney
topic,
but
if
I
gave
a
sort
of
introduction
to
what
it
is,
and
hopefully
we
can
people
that
can
decide
whether
they're
interested,
which
is
the
idea
of
in
we,
have
these
different
pieces
that
installed
equipment
a
cluster
like
cloud
formation,
terraform
q,
ADM,
coughs
and
Cooper
Nettie's
itself
manages
some
resources
and
the
challenge.
K
A
I
I
want
to
say,
I
think
this
sounds
like
a
really
interesting
topic.
I
think
one
of
the
things
that
we've
taken
a
little
bit
of
a
different
tack
with
cube
admin
than
other
tools
is
that
we
have
been
willing
to
push
some
of
the
stuff
necessary
to
make
things
streamline
into
court
right
the
bootstrap
token.
It
is
one
of
those
things
the
TLS
stuff
that
Mike
did
right
is
another
one
of
those
things
that
fundamentally
simplify
how
easy
it
is
to
bring
up
clusters
and
I.
Think
that
extending
that
into
like,
as.
G
A
Look
at
these
right
now,
the
the
cloud
provider
is
a
big
bag
of
crap
right.
It's
in
terms
of
like
everything's
gone
together
in
terms
of
all
the
different
roles:
right,
not
a
bag.
It's
a
kitchen
sink
right
in
terms
of
like
everything
gets
sort
of
put
in
there.
If
we
can
find
ways
to
sort
of
faster
that
kitchen
sink
out
and
then
make
some
of
it,
be
activatable
x,
cube
admin
and
sort
of
move
those
lines
a
little
bit.
A
I
think
that
seems
really
interesting
to
me
and
I
could
imagine
this
plays
in
some
of
those
to
the
last
discussion
that
that
may
be
like
you
use
your
cloud.
Formation
is
type
of
thing
to
bring
up
a
master
node
that
ends
up
being
a
zygote,
and
then
it
manages,
and
then
you
install
some
sort
of
extension,
that
is
a
sort
of
uber
cloud
provider
that
then
you
know
scales,
notes
at
that,
and
so
you
know
that
won't
work
in
all
situations,
but
I
think
that
seems
totally
totally
interesting
and
reasonable
week.
Yeah.
J
H
K
H
But
anyway,
the
cloud
control
cloud
stuff
in
general:
it
is
on
its
way
out
and
we
should
think
with
Tim
Hawking
and
Sid
has
money
from
venture
about
that.
It's
working,
progress
and
I
think
you
should
already
be
able
to
use
the
cloud
controller
manager.
It's
binary
in
the
release.
Stop
that
should
be
exactly
like
the
controller
loops
in
controller
manager
right
now,
so
I
think
it
should
work
out
of
the
box
with
AWS
and
GT,
and
everyone
else
and.
H
The
the
hard
part
is
volume
and
then
removing
things
from
the
queue
blood.
That's
the
non
straightforward
like
to
talk,
but
anyway
it.
It
would
be
unrealistic,
discussion
this.
So
please
like
I'll
a
working
meeting
prototype
or
something
and.
G
I
will
do
that,
give
you
an
example.
I
mean
I'll
just
mention
that
we
found
it
rather
quite
hard
to
have
good
morning.
These
managing
and
they'll
be
on
the
earth
managing
a
dns
right,
so
I
know,
there's
words
of
the
dns
and
the
blv.
That
stuff
would
improve,
but
we
found
it.
Lolly
eh
actually
manage
that
external
it
both
of
those
things.
G
Essentially
we
perform
for
or
setting
up
the
NS
MLB,
and
when,
when
we
move
from
one
floor
to
another,
essentially
then
we
do
do
not
grade
settle,
for
example,
or
or
just
just
otherwise
having
learning
the
students
to
be
really
what
will
be
holding
price
useful
for
us.
So
we
actually
ended
up
settling
all
cross
forever
into
action.
G
G
K
G
B
I
was
going
to
say
something
similar,
which
is
that
I
I
mean
I.
Imagine
that
many
users
are
going
to
have
opinions
felt
this
right.
I
can
imagine
many
users
feeling
quite
uncomfortable
about
their
cuban
se,
plastered
being
able
to
provision
more
BMS
on
their
infrastructure,
for
example,
depending
on
like
very
security
and
compliance
requirements
that
they
may
have
so
I
think
we're
just
making
a
bit
careful.
I
agree.
K
J
B
K
Yeah
they're,
probably
these
people
bringing
me
better
so
on
this
one
phrase
also
going
to
be
composing
other
working
groups,
but
this
is
I
am
policies.
Gte
now
has
a
son
form
of
I
am
AWS,
obviously
has
I.
Am
that
is
perfectly
complicated,
so
no
one
understands
it,
and
this
is
about
locking
down
the
I,
am
role
so
that
pod
don't
get
access
to
the
node.
I
am
permission
and
also
trying
to
give
cards.
Particular
I
am
permissions
and
ideally
letting
a
cube
api
server
who
controller
manager.
K
Those
are
like,
if
you
could
have
everything,
there's
that
would
be
great,
there's
just
a
pull
request
that
went
in
like
it
sickened
me
and
16.
It's
just
up
a
rule
which
will
isolate
pod.
So
that's
that's
a
hook.
It
doesn't
you.
Let
us
find
additional
roles,
so
I,
don't
know
that
maybe
I'll
start
with
more
of
a
spreading
a
paper
or
like
a
feature,
adding
more
duck
into
a
feature
and
then
propose
a
working
group
in
a
couple
of
weeks
or
something
if
there's
anyone,
that's
interested,
but
the.
A
A
We
should
still
so
here's
the
deal
so
like
I
think
like
the
overarching
goal
here,
it's
like
we
should
make
it
easier
for
people
to
turn
on
the
cloud
provider
or
two
when
using
cube
admin
right,
whether
that's
documentation,
whether
that's
tooling,
whether
that's
extensions
and
I,
think
you
know,
documenting
and
and
and
getting
crisper
about
sort
of
how
you
interact
with
the
I,
am
systems
and
sort
the
implications
of
that
I
think
is
falls
under
that.
Does
that
sound,
reasonable
Justin
there?
It.
K
Sounds
good
I
had
this
situation
when
the
cute
I
am
would
be
or
the
IM
thing
would
be
separate
and
lips,
but
then
they
put
it
in
core,
but
there's
definitely
a
piece
or
there
has
to
get
fire
wall
piece
that
has
to
be
established
and
we
just
haven't
figured
out
actually
how
to
do
the
firewall
piece
correctly
yet-
and
this
is
certainly
at
least
half
of
the
puzzle
that
they
put
into
core,
but
I
don't
think
it's
the
other
half
which
is.
I
need
you
here,
I'm
not
insecure.
A
K
Ahead,
the
next
one
age,
I
think
everyone
on
the
list
should
have
received.
It
should
have
been
invited
to
cops
office
hours,
so
this
is
basically
about
trying
to
keep
successful
life
cycle
for
the
cross-cutting
concerns.
Now
that
there
are
so
many
tools
and
I
guess,
blue
cube
as
well
should
be
in
there.
Although
I
don't
know
whether
maybe
be
tube
is
more
like
actually
working
on
integrating
with
what
is
the
correct
approach
long
term,
but
I
think.
C
K
It
may
be
booked
you
into
exception,
but
I.
Certainly
you
know,
like
cops,
has
a
lot
of
discussion
about
things
that
are
specific
to
cops
and
are
not
a
general
interest
to
like
how
should
we
run
cluster
in
general
and
I'm
sure
like
to
BTM
has
the
same
and
Argo
has
the
same
and
other
I'm
not
trying
to
excuse
me
once
if
I
forgot
name
just
put
them
in,
but
we
added
the
cop's
office
hours,
which
are
sort
of
a
developer
touch
space.
K
Ask
us
anything,
discuss
your
PR
is
how
should
I
do
this
typing,
like
which
we
do
every
other
week?
The
intention
of
doing
that
is
that
that
way,
we
don't
have
to
have
the
discussion
in
this
forum
and
I
think
if
everyone,
if
the
other
projects
did
that
I
think
that
would
be
awesome
also
because
we're
only
like
running
out
of
time
right
when
you
say
we
want
to
keep
the
more
time,
we
can
keep
further
bigger
issues,
better
I
think
so.
I'm.
B
B
What
we
could
try
doing
is
lifting
this
meeting
in
half
and
say
that
I
don't
know
which
order
which
way
round
would
be
better.
But
if
we're
going
to
talk
about
cube,
ADF
stuff
and
we're
going
to
talk
about
big
picture
stuff
in
the
meeting,
then
we
can
just
say:
well:
let's
let
there
be
a
convention
that
we
will
try
and
do
is
like
in
a
certain
order
sort
of
people
to
show
up
to
the
first
or
second
part
or
both
depending
on
their
preference,
nitrogen.
K
B
E
G
H
G
H
H
B
Yeah
I'm
I
would
like
to
start
with.
I
would
like
to
start
an
experiment
where
we
are
on
the
side
of
like
less
meeting
time,
just
because
I,
like
I'm
grams
and
I,
know
a
lot
of
other
people
on
this
call
around
but
I
my
preference
would
be
for.
Like
can
we
start
with
the
cup
of
meeting
and
half
experiment,
and
if
we
find
that
we
run
out
of
time
like
repeatedly
in
one
of
those
halves,
then
we
can
like
extend,
expand
it
and
have.
F
Also
get
say
how
many
times
of
our
current
leading
gone
and
we
haven't
covered
everything,
so
I
think
trying
to
cram
now
two
things
into
the
space
of
one
thing
that
you
said,
but
always
almost
go
over.
If
there's
I,
don't
know
how
that's
going
to
work
right,
we
typically
don't
get
through
the
entire
agenda.
In
one
hour,
like
last
week,
we
didn't
the
week
before
we
didn't
yeah.
B
We
typically
only
have
some
temple
sent
over
overflow
or
sober.
What
I
would
say
is
like
we're
talking
about
mostly
discussing
same
content
that
we
already
discussed
in
the
same
amount
of
time,
just
being
a
bit
more
organized
felt
like
different
categories,
so
Joe's
doing
something.
Where
do
the
standards,
which
I
think
means
move
on?
Maybe
I'm
going
to
make
a
call
or
nice.
A
C
A
B
Going
to
I'm
going
to
make
a
call
on
this
and
say
that
we
try
it
by
splitting
the
meeting,
then
the
only
question
remaining
is:
do
we
try
to
you
too
baby
and
soccer?
So
do
we
try
a
new
big
picture,
stuff
of
which
order
would
be
more
useful
people?
I
think
it
doesn't
really
matter
so
I'm
going
to
propose
that
we
start
with
a
big
picture
supper
any
objection.
B
K
B
H
H
H
Cool
cool,
sweet
and
so
yeah
I,
don't
just
awesome,
yeah,
cool
and
yeah.
I
also
have
I
think
with
David
EADS
on
an
issue
with
the
where
we're
still
like
the
dummy
pod,
we're
deploying
is
hanging
around,
has
been
doing
that
since
14
I
think
I
found
an
issue
thanks
to
today,
which
help
and
I'm
going
to
submit
a
bug
feel
for
that
as
well
yeah
Luke,
most
beautiful.
H
H
Okay,
yeah.
We
have
that
one
integrate
twitter
bootstrap
discount,
let's
discover
with
Combinator
school.
If
Joe
could
give
a
short
status
update
in
the
dock
as
off
by
ego,
it
would
be
great
and
we
have
to
do
the
same
with
the
other
one.
H
Let's
see
if
I
find
the
link
to
that
one
and
yeah,
so
we
should
update,
give
us
a
short
description
on
on
the
features
issue
issues,
so
the
pn
groups
there
zappy
include
knows
what
we're
after
and
yet
please
test
cube
area
mat
head
when
1
16
b
tattoo
is
released.
I
expect
it
to
be
like
both
feature
complete
and
like
nearly
bug
free,
so
I
think
we
should
be
in
good
good
shape,
but
that's
requires
testing
as
well
so
like
if
there
have
been
a
lot
of
refactor,
a
lot
of
new
issues.
H
F
H
Let's
discuss
that
on
slack
later
and
I
have
a
solution
to
the
deletion
of
damage,
the
dhammapada
I
think,
and
the
last
issue
is
cubed
in
research
should
not
trained
and
remove
the
node.
We
added
this
in
15,
but
due
to
the
are
back
things
that
got
added
into
16,
we
have
to
remove
the
functionality
is
the
class
administrator
that
should
drain
and
remove
pods
node
from
the
cluster.
H
So
that's
a
small
thing
as
well,
but
that
was
my
point
then,
on
death
and
Oh
Marco
Paul
sent
well
I,
don't
know
if
it
was
the
Cabana
this
damn,
but
Brian
grant
forwarded
it.
The
announcement
incubation
thing
to
seek
love
life
cycle,
the
mailings
and
I
just
think.
We
should
briefly
talk
about
it
here
and
see
what
we
think
feel
free
to
to
present
your
work.
Yes,.
N
Hello
I
hope
you
can
hear
me
yeah,
so
basically,
you
can
see
you
can
look
at
cube
now
as
a
set
of
our
wrapper
for
cool
batman.
The
motivation
why
we
did
that
because,
because
we
needed
to
speed
up
the
deployment
time
of
a
cluster,
so
weak
algo
x
quite
a
lot
of
time,
and
basically,
if
we
use
uber
native
to
the
data
processing
of
biology
stuff
and
we
have
our
workflow
systems,
let's
send
basically
containers
through
jobs,
coo
coo
burnett
is
in
order
to
process
data.
N
N
Cloud
provider,
so
we
needed
to
be
able
to
spawn
cluster,
to
amazon,
to
google
and
to
oven,
stark
installation.
So
this
was
a
very
brief
summary
of
why
we
came
up
with
a
with
a
new
deployment
to
and
right
now
we
have
quite
a
bit
of
stuff
into
it,
but
it
feels
not
so
many
line
of
code.
We
and
we
want
to
the
couple
like
the
graphs
lfs
parse
from
the
deployment
of
of
the
cross,
the
kubernetes
cluster.
N
H
Cool,
thank
you.
I.
Think
that
sounds
great
and
I'm
I'm
glad
that
you
did
it
on
top
of
queue.
Barium,
that's
really
cool
I
think
it
has
quite
a
lot
of
similarities
with
cops
and
with
cargo
itself,
and
since
it's
in
cops
and
fog
of
road
back
to
integrate
and
build
on
top
of
cube,
am
that's
all
along
quantum
goal
to
have
cube
amss
like
startled
at
the
bottom?
That's
not
the
heavy
stuff,
the
boring
things
and
then
do
the
fun
stuff.
B
I
think
this
looks
really
great
I
mean
as
a
way
of
like
capturing
I
mean,
and
if
it's
useful
to
people,
then
that's
that's
awesome
it
if
it's
a
way
of
sort
of
encoding,
a
set
of
things
that
people
might
want
a
way
of
that
people
may
want
to
configure
their
clusters
and
and
people
find
it
useful
than
it's
great.
Is
there
a
specific
ask
here,
though,
is
the
question
like?
Should
this
go
into
incubation
or
or
what
or
is
it
just
like?
Hey
we
built
this
thing.
This
is
cool
this
yeah,
so.
N
Basically,
we
are
part
of
a
European
projects
for
ya,
metabolomics
stuff,
and
basically
one
criticism
that
we
got
is
that
we
didn't
push
push
this
upstream.
So
I
was
looking
to
understand
if
we
can
make
it
an
incubator
project
or
if
we
can
make
it
part
of
an
existing
projects,
but
I
think
we
will
be
more
key
and
so
have
it
as
a
separated
incubator.
Roger
like
it
will
be
less
work
for
for
us
to
integrate
it
in
in
cargo
and
I.
N
Think
I'll
go
is
also
a
little
bit
different
scope
like
they
want
to
install
on
bare
metal
on
a
lot
of
different
distributions.
While
we
are
gonna
we're
going
many
will.
Maybe
you
are
going
to
put
a
support
centos
at
a
certain
time,
but
it's
not
really
number
one
priority
for
us
like
we
just
need
to
have
a
cluster
after
running,
doesn't
really
matter
for
the
distribution
that
we,
which
we
have
so
yeah
a
long
story
short.
We
are
interested
in
employ
it
as
an
incubator
like
that
would
be
great
and
I
have
one
question.
N
So
we
have
a
I've
been
reading
a
bit
bit
incubation
process.
We
have
a
couple
of
champion
and
I
think
this
European
project
is
a
very
good
champion.
We
already
have
a
beta
release
of
this
European
project,
metabolomics,
processing
tool
that
is
based
on
on
cube
now
and
the
question
is:
how
should
we
look
for
response
or
because
this
is
the
second
requirement?
So
how
does
this
work.
H
H
H
I
think
it's
great
that
you
have
this
project
and
but
I
still
wonder
if
less
an
easy
way
to
integrate
with
cargo
like
to
make
cargo-
maybe
modern,
more
modular
and
have
it
as
an
as
options
in
cargo
to
do
these
things
like
like,
we
have
a
an
option:
file
or
a
config
configuration
management
thing,
or
something
that
you
could
enable
this
cube
now
mode
whole
thing.
Would
that
make
sense,
I
think.
I
To
rephrase
that
you
might
want
to
have
a
specialized
playbook
with
potential
specialized
configuration
for
how
you
do
forking,
because
right
now
the
default
Forks,
usually
for
instable
or
not
said
hi
I,
know
that
we
wanted
to
blast
clusters
fast.
We
had
to
you
know,
have
a
fast
that
machine
with
a
lot
of
memory.
Is
that
works
for
like
200,
and
that
allowed
us
to
build
out
a
cluster
in
like
20
minutes
of
several
hundred
nodes
and
there's
there's
ways
to
optimize
it
instead
of
custom
configuration
playbook.
K
O
O
Or
modular,
to
get
so
many
things
worse
as
much
as
went
to
put
one
release
made
on
appearing
to
every
loop
because
well
because
every
variable
is
reevaluate
on
each
generation.
So
it's
like
a
perfect
storm
right
now
we
have.
We
need
to
really
zoom
down
on
playbook
suppose
that
we
only
call
certainly
in
combination
with
certainty
issues
but
really
you're
spending,
half
the
time
doing
evaluating
protecting
our
money.
So
yes
is
the
problem
and
we're
working
on
it,
but
I
just
feel
a
stretch
Ricardo
in
the
negative
mother
cell
yeah.
O
O
B
B
H
H
I
think
it
might
be
a
little
bit
hot,
so
I
I
think
it
would
be
great,
like
work
with
cargo
and
see
if
you
could
like
join
forces
and
try
to
incorporate
what
you've
done
in
the
speed
mode
or
work
with
Justin
and
see
like
I
know.
Justin
is
working
on
getting
GC
supported,
cops,
bare-metal,
eventually
and
I
mean
I.