►
From YouTube: 2017-08-29 17.03.03 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
Cool
hello,
everyone:
this
is
Sig
first
life
cycle
on
the
29th
of
August
in
217
and
three
days
away
from
code
freeze.
Is
that
right?
Yes,
correct,
exciting,
okay,
so,
probably
crunch
time
yeah.
If
people
have
agenda
items
they
want
to
bring,
then
please
add
them
to
the
list.
Currently,
there's
just
a
few
things
that
I've
added
so
the
main
item.
I
guess
is
the
second
one
that
I
added
the
agenda,
which
was
I,
thought
it
probably
makes
sense
to
try
and
go
through
all
the
pull
requests.
A
That
cell
posted
upgrades
depends
on
because
that's
the
one
of
the
features
at
least
the
main
feature
that
I'm
interested
in
making
sure
that
we
get
into
one
day
and
before
that,
the
question
that
I
had
was
so.
We
talked
about
this
a
little
bit
in
one
of
the
more
specific
meetings
where
we
were
doing
the
sickest
lifecycle
goals
meeting
and
we
talked
about
whether
or
not
to
turn
on
self
hosting
by
default
in
1.8,
and
our
tentative
conclusion
was
no.
We
should
not
turn
on
self-esteem
by
default
in
1.8
the
key
Badman.
A
The
reasoning
for
that,
as
written
in
the
agenda
is,
we
think
that
it's
self
hosting
that
we're
happy
where
there's
going
to
land
at
the
last
minute
and
last
time
we
did
a
release.
Cube
admin
broke
kind
of
badly,
and
so
we
want
to
make
sure
that
we
don't
do
that
to
releases
in
a
row.
Otherwise,
we'll
probably
start
having
people
think
that
we're
shipping
unreliable
software-
and
we
don't
want
people
to
think
that,
and
we
actually
don't
want
to
ship
unreliable
software.
A
B
If
I
recall
correctly,
the
reason
that
we
really
wanted
self
hosting
was
to
make
upgrades
work
right.
The
most
important
thing
was
for
us
to
get
upgrades
implemented,
which
we
we
thought
would
be
sort
of
easier
to
implement
or
better
done
with
self
hosting
I
think
it
turns
out.
In
retrospect
that
was
not
the
case,
and
we
do
have
upgrades
landing
that
that
will
be
in
1.8.
B
A
Yeah
and
just
to
clarify
the
cube,
admin
upgrade
commands
that
Lucas
and
other
and
others
implemented,
works
with
all
hosted
and
non
self
hosted
clusters
in
the
non
self
hosted
case.
It
will
just
upgrade
your
static
manifests,
which
works
in
single
master
scenarios,
but
won't
work
in
the
future
in
a
che.
All
won't
work
as
well,
but
yeah
in
a
self
hosted
case.
It
will
do
a
self
ising
upgrade
and
so
so
yeah
we
kind
of
get
the
best
of
both
worlds.
B
And
we
were
planning
on
doing
that
at
some
point
in
the
future
anyway,
during
one
of
the
upgrade
cycles,
and
so
I
think
we
were
gonna,
have
a
split
mode
of
self
hosted
non
self
hosted
for
a
while
and
tried
to
sort
of
merge
those
together
into
also
posted
some,
so
I
think
that's
actually
that
much
different
than
what
we
thought
we're
going
to
do
it
just
when
we
set
the
default
and
sort
of
get
more
people
over
on
the
self
hosted
side
will
be
a
little
bit
later.
Cool.
A
B
I
mean
when
we
did
our
sig
update
in
the
wider
community
meeting
when
I
gave
that
a
couple
of
weeks
ago,
I
announced
that
we
were
gonna,
do
an
1/8,
so
I
think
we
should
just
make
it
clear
that
slip
to
1/9,
but
I
think
people
will
sort
of
assume
or
I
think
some
people
at
least
realize
that
that's
coming.
We
should
also
make
sure
that
we
update,
sig
p.m.
which
I
guess
is,
is
next
Tuesday
morning
I'll.
A
A
A
Okay,
thank
you.
Ok,
so
I
made
a
probably
incomplete
list
of
PRS
that
need
to
land
in
time
for
1/8
to
actually
get
self
hosting
working
reasonably
well,
even
though
it's
not
going
to
be
on
by
default,
please
if
people
notice
things
like
mess
or
think
this
the
wrong
links,
please
shout
otherwise,
should
we
just
go
through
them
one
at
a
time.
A
C
Alright
deal,
can
you
hear
me?
Yes,
Hey
great
yeah,
so
you
know
it's
written.
We
thought
we
kind
of
had
discussed
it
enough
and
didn't
need
a
design
doc,
but
a
lot
of
pushback
so
went
back
and
wrote
a
design,
doc
and
that's
been
discussed,
and
so
the
plus
side
is
that
it
there's
nothing
really
big.
That's
gonna
be
different
about
the
implementation.
The
downside
is
that
the
design
dog
is
still
not
yet.
C
You
know
fully
approved
by
someone
on
cig
acts,
so
we
need
to
complete
that
before
we
can
actually
continue
getting
the
PR
approved
so
yeah.
If
someone
has
a
swathe
the
gaps,
the
the
current
blockers
just
to
get
the
design
doc,
you
know
rubber-stamped
and
then
we
can
move
on
to
getting
the
PR
reviewed
and
merged.
So.
B
C
A
B
A
That
would
be
great,
I
mean
it's.
If
we
don't
get,
if
this
doesn't
land
before
Thursday,
then
like
we
probably
don't
even
want
to
well,
we
probably
want
to
heavily
caveat
self-hosting
in
our
Doc's,
because
my
understanding
is
that
it
means
that
some
flesh
requesters
won't
like
reboots
and
that's
not
something
we
want
to
give
anyone.
A
C
We
want
to
say
please
review
this
quickly,
yeah
the
way
you're,
viewed,
I
think
there's
just
a
couple
open
questions
that
I'm
waiting
for
someone
to
say:
okay,
yes,
option
A
or
option
B
D
and
then
I
can
very
quickly
fix
up
my
PR
and
just
you
know,
I,
don't
think
the
reason
why
we
didn't
do
design
doc
is
because
I
don't
think
the
PR
is
all
that
fancy.
So
you
know
again,
maybe
famous
last
words,
but
it
could
be
smooth
if
we
just
get
unblocked,
decided.
B
To
me,
look
like
the
the
blocker
was
just
getting
sig
ABS
review
bandwidth,
both
on
the
design,
dark
and
the
code,
and
that
the
code
was
largely
implementing
what
we'd
sort
of
verbally
agreed
to
before
that
and
just
sort
of
in
the
in
the
details.
We
need
to
get
agreement
to
get
it
merged.
That's
right.
Okay,.
A
A
A
B
A
Cool
that
sounds
good
fingers
crossed.
Are
there
any
dudn't?
Do
you
happen
to
know
which
thing
it
was
that
might
need
to
be
rebased
just
so
that
I've
got
a
link?
Let
me
see
if
I
can
find
it-
okay,
great,
maybe
just
pop
it
in
the
document
under
the
section
others
to
do
link
to
them.
If
you
could
thank
you,
then
the
other
PR
is
Doc's.
I,
think
Doc's,
pr's,
technically
don't
have
to
get
in
before
the
feature.
Freeze.
D
D
Than
Lucas
so
like
he
knows,
if
someone
else,
maybe
someone
who
is
not
that
for
me,
stress,
addressed
his
issue
has
a
second
look
over
it
kind
of,
like
you
know,
idly
someone
who
hasn't
used
that
at
all,
sometimes
total
real
alpha
beta
tester,
who
you
know,
could
really
give
deeper
insights
about
what
is
missing.
What
is
like
something
that
we
just
implicitly
assume
people
know
and
might
trip
people
or
whatever
so
once
I've
gone
that
beginning
being
next
week,
it's
having
someone
else,
volunteering
to
give
it
a
try,
would
be
awesome.
A
Cool
yeah
that
sounds
good
I'm
happy
to
do
a
pass
over
it.
I
know
I'm
a
bit
close,
although
I
haven't
actually
been
doing
much
implementation
work,
so
maybe
I
qualify.
A
Maybe
ping
me
on
Thursday
or
I'll,
make
a
note
for
myself
to
take
a
look
at
the
or.
A
B
That
was
a
Tim,
also
mention
in
slack
that
he's
out
of
the
office
at
the
moment,
and
it
looks
like
the
pier
has
not
merged
the
one
I
linked
to
I,
don't
know
when
Tim
gets
back
or
if
he's
playing,
I'm
being
able
to
push
that
over
the
finish
line.
But
that
was
one
of
the
other
things
we'd
written
down
as
sort
of
a
blocker
for
herself
herself
hosting
was
checkpointing
in
the
queue
block.
Yeah.
A
I,
actually
that's
what
I
thought
we
were
talking
about
the
beginning,
but
obviously
wasn't,
but
that
my
comment
about
like
not
surviving
a
reboot.
This
is
the
PR.
That
means
that
self
posted
clusters
can
survive
a
reboot
as
far
as
I
understand
it
right.
There
Damon
said
surges
so
that
we
can
survive
an.
A
Yeah,
so
does
anyone
on
the
call
no
status
on
this
Tim
said
four
days
ago,
thanks
for
the
feedback,
color
update,
ASAP.
B
Okay
and
and
Tim
ste-zel
office
this
week,
which
doesn't
sound
good
for
getting
it
and
by
code
freeze,
but
it
kind
of
looks
now
like
the
the
thing
that
Diego
is
working
on.
We
might
be
able
to
push
over
the
finish
line,
but
if
the
think
Tim
is
working
on,
if
there's
no
way
that's
going
to
get
in
then
toughest
is
gonna,
be
pretty
sort
of
severely
compromised
in
the
one,
a
release.
A
A
A
A
A
B
A
Yeah
absolutely-
and
we
also
landed
a
feature
to
make
it
easier
to
install
pre-release
clusters
right,
something
like
that:
yo
easier
instructions
for
users
on
how
to
try
the
functionality,
and
we
would
therefore
still
be
able
to
invite
feedback
by
showing
people
how
to
get
it
but
yeah.
It
would
be
a
shame,
though,
if
we
cannot
she
week,
I
mean
we.
A
B
A
B
B
B
A
Yeah,
okay,
cool!
Well,
that's
everything
on
my
points.
Does
anyone
know
of
any
issues
or
well
so,
okay,
I,
haven't
really
been
tracking.
Anything
apart
from
self-hosted
upgrades,
but
I
do
know
that
there's
a
bunch
of
other
stuff
that
I
think
there's
a
bunch
of
other
stuff,
that's
landing!
Four
one
eight!
Is
that
right?
Oh
no,
not
much!.
B
B
G
B
So
we
have
eleven
eleven
feature:
issues
a
1.8
checkpoint
in
the
queue
latest
ones.
We
talked
about
that,
the
famous
that
surge
update.
We
talked
about
that.
There
are
one
two
three
four
cops,
so
we
haven't
talked
about.
I'll
come
back
to
those.
We
have
extensible
configuration
of
cube
admin
which
is
also
supposed
to
hit.
Oh.
B
Let's
say
the
cube
admin
phase
command,
we've
got
upgrades
of
self-hosting,
which
may
not
work,
but
we
have
upgrades
without
self
estimate
if
doesn't
and
better
multi-platform
compatibility
which
was
I
just
mentioned.
So
this
is
Lucas
is
a
or
description.
This
is
be
able
to
create
a
cluster
where
you
have
heterogeneous
nodes
where
some
are
running
PowerPC.
Some
we're
running
armed
and
some
are
running
AMD
64.
It's
not
clear
to
me.
That's
quite
done,
but
so
we'll
talk
about
what
it's
basically
a
labeling
thing.
B
Do
we
call
it
beta
or
do
we
keep
calling
alpha
better
bootstrapping
and
discovery
and
I
think
we
did?
We
did
have
some
of
the
discovery
stuff
land
out
of
process,
controller
managers
for
different
clouds.
That
is
I'm
not
sure,
if
that's
exactly
how
close
that
is,
but
it's
getting
it's
definitely
getting
closer
and
then
add-on
management
which
we
just
in
you
can
comment
on.
We
had
targeted
that
for
alpha
and
I'm,
not
sure
we
quite
day
it's
alpha.
F
A
B
On
the
agenda,
one
that
I
should
mention
if
any
of
those
issues
are
assigned
to
you,
if
you
could
go
put
an
update
on
them
before
next
Tuesday
look
slice,
I,
assume,
I
haven't
been
to
a
sig
p.m.
meeting
it,
but
I
assumed
the
sig
p.m.
meetings.
We
will
sort
of
cover
that
list,
especially
at
the
next
one,
which
is
right
after
the
the
code
freeze
and
sort
of
talk
about,
what's
actually
landing
in
1x8.
So
anybody
who
owns
those
issues
updates
would
be
appreciated.
Otherwise,
I'll
just
be.
B
E
Okay,
Jacob,
you
have
the
next
item:
hey
so
I've
been
hosting
these
knowledge
sharing
sessions
for
our
end-to-end
test
infrastructure.
Yesterday,
I
just
gave
the
second
session
and
I
think
based
on
the
content
left,
we
probably
have
one
more
session
to
do
so.
I'm,
just
advertising
that
it'll
be
in
two
weeks.
Approximately
there's
an
oh
I
forgot
to
link
to
the
issue
up
on
the
issue
in
the
notes
where
I've
been
coordinating
availability.
E
E
A
Yes
and
as
you
said
on
chat,
it
would
be
great
if
we
could
find
a
way
of
granting
that
permission
without
sharing
a
shared
password.
More
of
a
like
infrastructure
of
the
kubernetes
project.
Question
like
can
we
figure
out
how
to
do
groups
in
zoom
anyway,
I
just
remembered
that
as
Robert
was
going
through
the
list
of
features
in
the
features
repoed
that
we
skipped
over
the
costs,
features
and
didn't
come
back
to
them?
So
I
wanted
to
ask
it
Justin
you
want
to
go
through
those.
Yes,.
F
So
we
are
obviously
not
bound
by
the
feature,
freeze
date
being
a
like
federated
project.
I
think
that
we
will
probably
get
two
out
of
the
three
but
I'm
not
going
to
commit
to
which
two
but
I
think
yeah.
We're
looking
good
to
continue.
Batman
continually
bound
on
pull
request
and
with
more
than
anything
else,
that's
very
difficult
to
actually
forecast
these
things,
but
we
have
most
of
the
pieces
there.
So
I
was.
B
F
When
I
asked
this,
the
answer
is
that
we
are
the
like
the
front-runner
or
the
canary
that,
yes,
we
we
have
to
be
in
the
features
repo
if
we
want
to
be
considered,
for
you
know,
documentation
or
like
the
release,
notes
or
anything
around
a
release,
and
so
yes,
it
is
a
shortcoming
in
the
process
that
yeah
that
we're
so
tracked
in
the
same
system,
but
are
not
following
the
same
rules.
Okay,.
B
A
F
F
Gce
sport,
the
GCE
support
from
alpha
to
beta
the
the
big
blocker
on
that
is
just
figuring
out
making
sure
the
API
is
ready.
The
GCE
cloud
objects
that
we
create
are
stable
and
we
know
that
one
of
them
is
not
what
we
want
in
terms
of
the
networking
configuration
there's
a
newer,
better
networking
configuration
where
you're
using
the
old
like
traditional
networking
configuration.
Then
we
want
to
fix
that
before
we
declare
it
as
sort
of
stable.
Everything.
F
Sort
of
you
know
we
can
fix
this
week
out,
but
changing
like
the
network
settings
is
a
pretty
disruptive
event.
So
that's
a
that's
the
think
blocker
for
going
to
beta
on
GC
support,
another
one
of
which
is
cups
server,
which
is
a
kubernetes
api
server
using
the
kubernetes
api
machinery.
We
have
early
versions
of
that.
We
are
continuing
to
put
more
objects
into
the
API
like
our
cluster
secrets
and
things
like
that,
and
so
that
is
going
well.
F
We
are
able
we
have
a
branch
which
can
basically
run
run
with
cops
and
actually
create
a
cluster
from
a
server
not
using
a
controller,
but
so
using
the
cop
CLI.
So
that's
progressing,
it's
not
going
to
be
complete,
but
it's
progressing,
and
the
third
feature
is
support
for
bare
metal
and
again
we
have
a
branch
in
a
more
experimental
state,
but
we're
getting
there
and
likely
we'll
tie
that
to
the
server
and
yeah
it's
more
progress
on
it.
I
think
is
like
slow
and
steady
progress
on
on
this
server
and
bare
metal.
A
I've
just
been
chatting
with
Tim
by
the
way,
and
he
says
regarding
that
checkpoint
in
PR-
he's
not
sure
who
would
be
able
to
pick
it
up
and
he
thinks
that
effort
would
be
better
spent
on
focusing
on
end-to-end
tests
and
upgrade
tests,
I
suppose
his
summary
was.
We
should
really
our
efforts
on
testing
in
his
opinion,
seems
fair
enough.
We
don't
have
anyone
putting
their
hands
up
to
say
that
they're
gonna
try
and
take
that
PR
over
the
line.
Then
you
know
what
else
to
be
ready.
B
A
A
A
A
B
It's
also
up
to
say
gap,
saying
I,
guess
the
question
is
like:
how
much
do
we
want
to
push
them
because,
presumably
they
have
other
stuff
they're
trying
to
land
before
the
release
freeze
also-
and
if
this
isn't
sort
of
the
last
thing
to
get
us
to
stop
posting,
do
we
want
to
sort
of
sort
of
burn
some
of
our
our
friendly
capital
on
getting
it
end
or
not?
I.
A
A
B
B
A
H
H
H
E
H
So
yeah
you'll
have
to
actually
remove
those
cuz.
We
won't
know-
and
these
these
are
shoes
just
to
be
clear-
are
things
like
bugs
and
other
things
that
we're
looking
at
for
the
quality
of
the
released?
So
if
there's
something
that
comes
up
as
a
bug
that
you
think
should
hold
the
release
off,
that's
kind
of
a
work
we're
trying
to
have
in
the
one
milestone,
so
essentially
by
the
time
we
give
her
neck
cuts
release
there
shouldn't
be
any
anything
in
the
milestone.
That
is,
that
is
not
really
squawking.
If
that
makes
sense,.
A
B
Can
narrow
it
down
a
little
bit
yeah
cool?
Do
you
want
to
leave
that
sure
Thanks
I
just
removed
one,
so
the
one
that
sent
was
assigned
to
Tim
I
bumped,
that
to
1.9
so
they're
11
left
I'll
just
go
down
and
whatever
order
they
appeared
on
my
screen.
Hopefully
it's
the
same
for
everyone
else.
After
following
Jason's
link,
the
first
one
is
enabled
a
bootstrap
signer
and
token
cleaner
controllers
by
default.
B
Lucas
says
it's
needed
to
get
something
to
beta
I
thought
that
we
did
that,
although
that
looks
like
there
is
an
open
PR
from
29
days
ago,
that
has
not
been
merged.
So
looks
like
that
is
maybe
not
done,
and
Lucas
is
out.
So
it's
not
clear.
If
that's
I
mean
it's
been
a
month,
it's
not
clear.
If
that's
been
being
pushed
forward,
his
Luke.
Do
you
know
if
Lucas
is
still
chattable
well
he's
outdoor?
Is
he
completely
off
the
grid?
There
he's.
A
B
So
second
one
is
to
is
to
update
some
of
the
GCE
shell
scripts
to
remove
some
deprecated
flags
in
G
cloud.
It
looks
like
at
least
one
p
are
merged
for
that
in
the
last
day.
So
I
think
we
should
keep
that
one
in
the
one
eyed
milestone
it
looks
like
Jeff
Crafton
has
been
pushing
that
forward
and
and
he's
not,
he
hasn't
not
assigned
himself,
but
he
has
sending
PRS
for
it.
So
hopefully
he
continues
to
do
so.
B
E
So
we
got
punted
to
a
cluster
lifecycle,
because
someone
detected
an
anomaly
after
upgrading
the
cluster
that
it
looked
like
it
wasn't
set
up
in
the
same
way
that
a
fresh
cluster
would
be
set
up
so
I
route
caused
that
problem
and
that's
fixed
now.
So
now
it's
kind
of
back
on
them.
We
should
unassign
our
cig
because
it's
really
their
failure
at
this
point
is
my
understanding
of
it.
B
B
We
bumped
it
to
1.8
3
months
ago
and
I
don't
see
a
whole
lot
of
progress
since
the
end
of
May.
So
I
think
this
one.
It's
a
sign
to
our
sig,
because
there
isn't
really
a
better
sig
to
own
the
separation
of
the
cloud
controller
manager,
but
Siddhartha's
is
owning
that
one
and
well
I
think
he's
still
in
hooks.
B
We
should
ping
him
to
see
if
we
should
pump
it
the
next
one's
about
our
back
and
able
an
enablement
CJ,
actually
just
sent
a
pull
request
for
this
I
think
yesterday,
the
day
before
so
we'll
leave
that
there
for
now,
until
we
decide
if
that
PR
is
going
to
make
the
code
freeze.
Otherwise
we
hunt
it
came
out
of
add-on,
update,
revamp.
B
Less
complicated,
it's
a
move,
cube
proxy
and
to
attainment
set.
So
this
is
something
that
Zhi
Hong
is
working
on.
I
think
he's
at
the
point
where
he
believes
we
can
add
a
flag
to
do
this
optionally
in
the
cube
of
scripts,
but
not
turn
on
by
default,
which
to
me
is
a
little
bit
concerning,
because
the
reason
I
wanted
him
to
do.
B
This
was
because
we're
already
doing
this
in
cube
admin
and
some
people
on
sig,
node
and
sig
network
think
that
it's
not
safe
to
do
on
gke,
which
makes
me
concerned
and
I've
been
trying
to
get
them
to
fix
inner
lying
issues
to
make
it
safe
for
them
to
consider
giving
a
production.
So
we
can
do
it
everywhere,
made
consistent.
B
The
next
one
is
cubelet
checkpointing
or
something
so
this
is
very
similar
to
the
other
one
that
we
just
bumped
out
of
1/8,
because
Tim's
PR
is
not
going
to
get
in
and
we'll
bump
that
to
1/9
and
the
last
one
is
support
high
availability
clusters,
and
that
was
a
p2
for
1/8,
which
we
didn't
quite
get
to.
We
believe
we
have
some
designs
and
we'll
get
to
it,
1/9,
so
I'm
going
to
push
that
to
one
mine
also.
I
B
A
G
A
A
E
So
Jess
isn't
here,
but
she
is
taking
on
the
upgrade
test,
at
least
according
to
the
old
upgrade
procedure
where
you
do
cube
admin
in
it
again
with
the
same
flags
and
the
new
version,
and
he
did
mention
rallying
the
efforts
around
testing
upgrades
I'm,
not
sure
if
he
mentioned
anything
else,
but
we've
pretty
good
coverage
now.
The
problem
is
that
we
still
need
a
human
to
look
at
the
test
grid
on
a
an
occasional
basis
and
notice
when
things
are
red
and
then
jump
in
to
help
root
cause.
What's
going
on.
E
This
has
mostly
been
a
few
individuals
and
we've
actually
found
real
regressions
that
aren't
you
madman
lately,
like
the
cubelets,
had
a
few
regressions
that
we've
had
to
push
back
on
them,
so
I've
been
hosting
the
knowledge
sharing
sessions
to
try
to
make
it
easier
to
farm
out
and
the
knowledge
and
responsibility
of
being
able
to
maintain
this
stuff.
But
if
there
any
other
specific
ideas
or
specific
asks,
that'd
be
great.
Another
great
improvement
that
we
had
pretty
recently
was
cube.
E
Admin
was
only
able
to
use
the
images
from
the
control
plane
from
the
last.
You
know.
Alpha
release
or
whatever,
and
now
we
actually
have
CI
images,
I
believe
fingers
crossed
I,
think
that
one.
So
that
means
that
we
actually
notice
when
things
go
red
right
when
they
go
red.
Instead
of
like
two
weeks
later,
when
the
next
yeah.
B
B
B
We
also
have
all
of
sort
of
the
legacy
cube
up
based
upgrade
tests
which
are
sort
of
much
more
thorough
and
testing
kubernetes
upgrade
functionality
between
versions,
and
they
take
a
really
long
time
to
run
and
I
historically
have
been
very
difficult
to
keep
running
healthy
and
it's
been
hard
for
us
to
get
a
signal
out
of
them
for
the
last
couple
of
releases
until
very
late
in
the
release
cycle.
So
people
are
thinking
about
helping
out
without
very
testing.
That
would
be
another
place
that
we
could
use
extra
hands.
A
A
Nice-
and
we
also
now
so
you
said
the
the
limitation
is
that
humans
have
to
look
at
it.
They
have
to
remember
to
look
at
it
are
we
do?
We
have
I,
remember
at
some
point
in
the
past.
I,
don't
know
how
up-to-date
my
knowledge
on
this
is
that
there
were
cube
admin
tests
that
were
failing
that
we're
not
noticed
when
the
release
was
cut.
Are
we
also
have
we
also
fixed
that
have
we
fixed
that
problem,
but
are
there
cube
admin
tests
that
block
the
actual
release
of
1.8
or
0.
E
I,
don't
know
if
we've
cut
the
release
1/8
branch,
yet
usually
we
wait
until
the
branches
cut,
because
the
way
the
tests
get
triggered
are
based
on
the
branch
name.
But
there
is
a
new
master
blocking
tab
on
test
grid
and
there's
an
issue
to
add
the
cube
admin
test
to
it
and
I
haven't
gotten
around
to
it.
Now
it's
kind
of
hoping
someone
else
would
be
able
to
you,
because
I
have
other
stuff
I've
been
trying
to
work
on,
but
it's
at
least
being
tracked
that
we
should
be
there.
E
B
Yeah
I
think
the
question
was
more
about
the
the
relief
blocking
tests
which
we
don't
have
set
up
yet
so
there's
the
tests
that
are
like
release
1
8,
that
the
test
sort
of
burned
down
group
looks
at
when
we're
cutting
release.
I
think
the
question
was:
can
we
get
our
test
and
that
test
grid?
Oh
right,.