►
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
All
right,
so
we
will
is
that
still
visible
for
folks.
C
Let's
see
so
so
I
made
I,
I
worked
on
uninstall
feature,
so
I
just
made
it
a
little
bit
more.
I
mean
more
robust
because
you
would
like
I
talked
about
this
last
week,
but
last
week
I
had
made
a
pr
that
had
a
bunch
of
different
types
of
changes
together
and
they
were
smart
to
to
ask
me
to
separate
things.
C
So
I
separated
just
the
uninstall.
That's
that's!
That's
that
pr.
Basically,
it
wasn't
uninstalling
everything.
The
first
era
was
just
return
which
would
potentially
leave
stuff
behind.
C
So
I
fixed
that-
and
I
may
have
better
warnings
for
the
user
and
the
reason
why
I
even
got
to
work
on
that
and
if
I
figure
I
noticed
that,
because
I
have
been
running
into
a
problem
with
aws
and
it's
really
bizarre.
C
So
basically
I
most
of
the
time
like
ninety
percent
of
the
time
I
can't
install
valero
on
aws
and
I
can't
run
the
e2e
tests
because
they
also
require
installing
blur.
So
nobody
else
is
getting
this
problem.
C
I
don't
know
I.
I
do
have
a
an
issue
that
I
described
what
it
is,
and
I
have
some
logging
errors
there.
The
logging,
the
errors
that
I'm
getting
is
basically
I'm
getting
a
timeout
when
when
so,
the
letter
goes
through
this
to
tries
to
install
the
crds
and
it's
always
in
the
same
order.
So
it
starts
the
backup,
the
backup,
storage
location
and
then
when
it
gets
to
the
the
delete
request,
it
gets
a
timeout.
It's
always
like
this.
It's
always
on
the
delete
request
times
out,
arrows
out,
stops
and
stop
installing.
C
Yeah,
it
works
if
I
use
quilt
control
to
install
it
by
hand
it
works.
I
have
so
no
so
anyway.
This
is
happy
with
your
happening
with
the
rc
as
well.
So
it's
not
an
issue
with
my
building.
The
binary
on
my
local
machine,
but
dave's
theory
is
that
it's
a
network
issue.
E
C
C
Did
I
I
tried
it
with
what
yeah
I
I
I
added
that
data
points
to
the
tickets,
because
that's
a
very
good
test
at
any
rate,
so
last
week
I
spent
a
ton
of
time
on
this
because
at
first
I
thought
that
the
uninstall
leaving
things
behind
and
there's
also.
I
also
that
I
was
just
looking
for
what
could
have
been
causing
this
in
the
code,
so
I
noticed
the
uninstall
pro.
C
I
noticed
that
our
e2e
tests
are
leaving
things
behind,
but
that's
not
a
big
issue
as
long
as
the
test
passes
it's
fine,
but
if
there
is
a
problem
in
the
middle
of
the
test,
then
the
tap,
then
the
tests
are
not
cleaning
up
artifacts
that
it
creates.
So
I
have
an
issue
for
that.
I'll
address
it
later,
so
I
just
thought
that
those
were
the
things
and
I
just
started
adding
code
to
fix
it,
but
then,
and
then
so
this
is
intermittent,
and
this
is
why
it
makes
it
even
worse.
C
Sometimes
it
works.
So
I
made
all
these
changes
and
it
worked,
and
I
thought
okay,
this
this
is
one
of
these
changes
fixed.
The
problem.
Wait
let's
move
on
and
then
it
started
happening
again
and
then
so
I
pulled
the
branch
with
my
changes
ran
in
it
was
happening
again.
So
none
of
those
issues
that
I
thought
were
the
issues
causing
this
were
had
any
relations.
So
so
I
just
I'm
saying
I
spent
a
lot
of
time
chasing
this
last
week.
G
C
This
is
comp
so
frankie.
Thank
you
for
off,
retire
dave.
This
is
completely
unrelated.
Oh
okay,
it's
completely
unrelated,
because
I've
run
I've.
I've
run
into
this
this
that
this
problem
in
all
kinds
of
ways
and
the
tests
have
nothing
to
do
with
it.
C
So
so
to
answer
your
question,
though,
it
really
doesn't
matter
how
many
things
are
there.
We
are
going
to
have
to
have
a
cleanup
function
that
we
put
into
all
of
the
all
of
the
returns,
so
it
cleans
up
before
returning.
So
if
there's
like
ten
or
one
doesn't
matter,
we're
gonna
have
to
do
that.
So
it's
your.
C
H
So
I
think
we
so
I
I
think
it's
showing
up
enough
and
the
weird
thing
is
it's
always
showing
up
in
the
same
place.
Like
you
say:
you've
switched
binaries,
so
I
think
we
need
to
start
basically
bisecting
things,
so
we
can
narrow
down
what's
causing
this
and
my
network
theory
I'd
forgotten
that
it
always
happens
at
the
same
crd.
H
E
C
H
H
But
well
yeah,
but
let's
make
a
simple
script
that
just
uses
like
the
the
stock
install
and
install
run
that
you
know
it
shows
the
problem
after
a
half
now
we'll
take
that
same
script,
we'll
run
it
from
like
my
either
mac
or
my
linux
machine
against
the
same
cluster
and
see
if
we
can
report.
H
Places
easily,
then
we
can
start
nailing
things
down,
and
I
want
to
try
things
like,
for
example,
if
we
do
all
the
crd
installs
with
cube
cuddle
versus
the
installer,
does
it
work?
Does
it
not
work?
So
we
start
to
bisect
and
and
start
to
figure
out.
What
is
the
problem
here?
Is
it
the
cluster?
I
think
the
cluster
we've
seen
that
the
cluster
works
for
other
people,
like
I've,
been
able
to
use
that
same
cluster.
Do
things
immediately
after
you're
having
the
problem,
but
maybe
it
isn't
a
cluster.
C
Yeah
so
another
data
point
for
people
who,
who
don't
know
is
we
use
two
different
clusters
to
run
this
on.
I
got
a
problem.
Both
clusters
dave
didn't
get
the
problem.
Well
dave.
I
used
dave's
cluster,
so
I
got
up
the
problem
on
his
cluster.
He
didn't
try
my
cluster,
but
we
should
do
that
from
different
machines
against
the
same
cluster.
That's
a
good
test.
H
C
Like
we
have
two
more
so
after
the
script,
we
do
have
to
do
same
for
drc
and
that's
what
I'm
going
to
be
working
on.
That's
it.
H
Okay,
dave
yeah,
so
mainly
released
up.
Excuse
me
last
week,
so
ed
tests
are
with
the
volume
snapshots.
Those
are
running
versus
vsphere
aws
and
I
got
azure
running
last
night,
mainly
config
things
and
no
real
actual
problems,
so
yeah,
so
we're
doing
good
there
and
that's
that's
pretty
much
it
for
me.
A
E
Hi
everyone
so
main
announcement
for
me
is
that
yesterday
we
got,
I
went
through
the
release
process
and
we
now
have
an
rc
candidate
available
for
1.6
yeah.
My
week
has
mostly
just
been
spent
going
through
the
last
steps
and
and
getting
that
out.
E
So
that's
what
we're
going
to
be
using
now
as
a
team
for
the
testing
before
we
get
the
official
release
site,
I
don't
really
have
a
huge
amount
to
add
other
than
that,
but
if
you
want
to
try
it
out,
please
feel
free
and
give
us
feedback.
It
would
be
super
helpful,
but
we'll
also
be
testing
it.
A
Yep
thanks
for
getting
that
out
and
yeah,
I
like
us
as
we
run
tests,
we
all
use
the
same
one.
We
always
use
the
same
thing
same
artifact,
and
that
way
we
can
actually
trace
down
issues
so
yeah
thanks
for
getting
this
out
any
questions
for
bridget.
C
C
I
would
test
the
image,
the
new
images
with
the
rc
and
then
cut
the
releases.
If
this
is
not
right,
somebody
let
me
know,
but
that's
what
I
was
going
with.
So,
if
someone
once
so,
I
don't
know
what
to
do
with
this
aws
issue.
As
far
as
cutting
the
release
for
the
plugins
well,.
A
Let's,
let's
put
that
on
hold
and
figure
it
out
before
we
cut
a
release
for
the
aws
plugin
yeah.
I
think-
and
we
can
address
that
separately.
Yeah.
A
Yeah
those
will
need
to
be
updated
for
bsl
changes
that
came
with
the
secrets.
But
yes,
let's,
let's
make
sure
it's
not
in
core
valero
the
issue
you're
having
before
we
cut
anything.
A
For
my
update,
I
have
been
out
for
several
weeks,
so
I'm
catching
up,
I'm
gonna
grab
doing
some
manual
testing
there.
I
don't
have
the
issue
here,
but
I'll
link,
it
there's
an
issue
with
the
manual
steps
that
we
should
all
be
running.
I'm
gonna
run
that
against
jc
gcp,
verify
all
those
work
and
otherwise
been
catching
up
on
emails
and
things.
A
One
thing
I
will
mention
here
carlesia
and
I
will
be
at
kubecon
eu,
with
dylan
murray,
from
red
hat
talking
about
data
protection
in
a
in
a
birds
of
a
feather
session.
We
have
a
date
for
that,
but
it's
escaping
me
at
the
moment.
May
8th
or
something
but
I'll,
we'll
get
those
into
the
weekly
updates
and
let
people
know
so
not
a
whole
lot
from
me.
D
For
me,
I've
just
been
running
tests
on
the
rc
to
make
sure
stuff
is
working
or
mainly
focusing
on
rustic
stuff,
because
I
did
the
upgrade
yeah
the
so
far
things
things
have
been
good.
I
want
to
run
the
same
test
that
I've
been
running
on
like
aws
to
make
sure
it
works.
A
A
B
Hello,
I
think
I've
met
nearly
ever
on
the
call,
but
I'll
just
say,
I'm
eleanor,
I'm
the
new
product
manager
for
valero
I'm
from
vmware.
This
is
the
beginning
of
my
third
week
on
the
project,
so
I'm
still
ramping
up
a
lot.
My
two
main
focuses-
and
I
just
want
to
give
a
very
quick
product
update
just
so.
You
know
that
my
two
focuses
are
the
valero
vision
and
roadmap.
B
Right
now
for
the
power
of
the
past
two
and
a
half
weeks,
I've
been
focusing
on
the
vision
I've
been
working,
especially
with
dave,
but
others
also
familiar
with
valero,
to
define
major
problems
facing
valero,
possible
solutions
or
sorry
major
problems
facing
valero
users.
Pardon
me
not
there
possible
solutions
to
those
and
then
evaluating
the
benefits
of
those
possible
solutions
for
the
valero
community.
B
My
next
step
is
to
circulate
this
product
grid
that
I've
been
using
to
contain
these
thoughts
with
valeria
community
members,
and
I'm
also
hoping
to
do
user
interviews
to
really
validate
a
bunch
of
assumptions
in
this
product
grid.
From
that
we'll
form
the
vision
and
then
once
we
have
the
vision,
then
we
can
define
initiatives,
epics
prioritize
features
and
build
that
longer
term
roadmap,
so
hopefully
in
the
coming
community
meetings
I'll
be
able
to
share
learnings
as
I
go.
So
that's
it
for
me.
A
All
right,
moving
on
to
the
discussion
topics
not
being
here
last
week,
were
there
any
discussion
topics
last
week
that
we
need
to
circle
back
on
I'm
gonna
scroll
down
here
I
don't
see
any.
A
Yeah,
it
sounds
like
a
lot
of
people
have
been
heads
down
on
the
release
and
really
focused
on
that.
So
some
other
stuff
has
slowed
down.
G
All
right,
so
this
discussion
topic,
I
was
wondering-
and
this
is
kind
of
a
hard
question
to
ask-
because
it's
kind
of
asking
how
hard
is
something-
and
it's
like
I
wanted
to
say-
is
it
possible
to
do
something
but
of
course
everything's
possible?
So
now
I'm
asking
I'm
going
to
change
it
to
how
hard
will
it
be,
but
you
know
how
we
have
to
install
in
order
to
do
a
backup
and
restore
with
valero.
You
have
to
install
valero
on
the
source
cluster.
G
You
have
to
install
valero
on
the
destination
cluster
and
we
were
thinking.
Would
it
how
how
much
work
would
it
be
to
maybe
use
valero
as
as
a
library,
so
basically
pull
in
packages
import
them
just
like
a
library
from
a
binary
that
works
remotely?
So
maybe
you
install
valero
on
cluster
a
service
cluster.
G
That's
not
the
source
cluster,
not
the
destination
cluster,
so
a
service
cluster,
and
then
you
start
pulling
items
from
the
source,
cluster,
saving
it
into
a
bsl,
backup,
storage,
location
and
and
then
push
those
items
to
the
target
cluster
or
destination
cluster
from
that
bsl.
G
I'm
not
sure
if
this
is
something
anyone
if
the
maintainers
can
answer
off
the
top
of
their
heads.
But
what
are
your
initial
thoughts
or
questions
as
to
maybe
why
the
heck
would
you
want
to
do
this?
You
could
ask
that
too.
A
Yeah,
so
for
the
service
cluster,
I
have
two
points
there.
So
the
first
is
you
actually
could
do
that
with
kubernetes
objects
by
passing
the
right
cube
configs,
however,
valero
makes
a
lot
of
assumptions
that
is
living
in
one
cluster
and
not
doing
you
know
multiple,
so
you'd
have
to
have
a
cube.
Config
pointed
at
the
original
the
source
and
then
you'd
have
to
swap
it
to
cube.
Config
pointed
at
the
destination,
that's
doable,
it's
just
kind
of
clunky
right
now.
The
second
thing
is
volume.
A
Snapshots
and
rustic
would
not
work
that
way
right
now
today,
because
the
volume
snapshots
would
assume
you're
in
it.
It
assumes
it's
in
the
same
cluster,
especially
rustic,
although
I
think
the
volume
snapshotters
could
work
if
it
was
like
a
public
cloud,
because
they
would
send
the
calls
to
the
cloud
api
for
that
region
in
the
nodes,
but
rustic
would
wouldn't
work
because
it
requires
local
node
access.
A
A
Going
to
the
other
question
for
backup
importing
backup
and
restore
packages
right
now,
you
could
import
them,
but
I
think
back
up
in
as
a
backup
restore
one
of
the
two
is
heavily
interwound
with
or
interwoven
with
the
controller.
I
think
it's
restore
so
it's
doable,
but
what
we'd
like
to
do
is
move
basically
clean
up
that
that
tech
that
to
move
that
that
logic,
that's
in
the
controller
into
the
restore
library
and
get
closer
to
this
yeah.
So
there's
an
issue
out
there.
A
It's
pretty
old,
mostly
we've,
we've
put
it
on
the
back
burner
because
stuff
works
today,
but
yeah.
That's
it
it's
doable
and
you
have
a
clean
up.
Yeah.
I
We
will
be
happy
to
take
a
look
there,
just
you
to
bring
everybody
the
the
rationale
for
this
problem,
slash
opportunity,
whatever
you
want
to
call
it.
I
If
you
want
to
migrate
from
legacy
clusters
to
a
new
cluster
it,
it
is
getting
more
and
more
difficult
to
install
valero
on
those
legacy
clusters
and
if
we
add
this
capability
to
have
the
letter
running
on
a
service
cluster
and
make
the
connection
to
the
source
cluster
over
a
client
go-
and
I
know
about
discs-
let's
say
for
this:
for
the
sake
of
this,
it's
always
stateless.
I
J
But
of
course
that
puts
us
in
other
areas
of
the
code
which
are
you
know
they
currently
work
for
us,
but
we
know
they're
unsupported
on
those
clusters,
so
that
this
would
help
us
get
away
from
that,
because
we're
trying
you
know
because
we
were
helping
customers
migrate
from
you
know,
clusters
from
kubernetes
to
say
you
know
one
seven
through
111
up
to
newer
clusters.
A
J
So
right
now
our
basic
migration
approach
is
that
you
know
we.
We
create
a
migration
cr
with
our
an
mtc
product
and
then
basically
we
do
a
valera
backup
and
we
and
we
are
doing
two-phase
backups
so
that
you
know
we
do
kind
of
the
our
open
shift
image
streams,
plus
the
pv
and
pvc
data
and
one
backup.
And
then
we
do
a
separate
backup
of
everything
else,
just
so
that
we
can
do
incremental
image
and
pv
backup
without
shutting
down
applications
for
those
that
can
do
that.
J
I
Yeah,
okay,
so
what
let
us
know
the
ticket
we
can
talk
offline
or
over
the
kubernetes
slack,
but
I'll
be
more
than
happy
to
have
me
and
frankie
looking
at
this
and
and
talk
to
eleanor
and
everybody
here
to
see
if
it
makes
sense
to
to
put
on
the
roadmap.
I
H
Can
I
get
an
award
yeah
yeah,
so
I've
done
some
of
this,
so
I
was
actually
able
to
the
poc
where
we
used
the
backup
or
as
a
library
I
hadn't
actually
gotten
around
to
restore
yet
so
I
hadn't
run
into
that
problem.
I
think
we
do
want
to
take
this
as
a
case
for
our
general
overall
use
cases,
so
we've
been
looking
at
you
know.
We've
been
looking
at
splitting
out
a
lot
of
faces
into
the
infra
layer,
and
I
think
this
is
another
driver.
H
Is
the
migration
use
case
and
being
able
to
build
eventually
like
migration
utilities,
which
are
specifically
for
upgrades
or
clustered?
You
know,
specifically
these
upgrades
or
import
situations
and
being
able
to
use
the
infra
layer
independently
of
valero.
The
application,
I
think,
is
going
to
be
important
here
and
yeah.
If
we
can
include,
is
a
library
sometimes
because
a
lot
of
these
things
minus
the
volumes,
there's
really
no
reason
why
we
need
to
be
running
in
the
cluster
yeah
and
even
for
things
like
you
know,
like
aws
or
even
vsphere.
H
If
you
have
the
credentials
and
you
can
look
into
the
cluster,
you
can
talk
directly
to
the
storage
system
rustic.
We
need
a
pod
running
in
there
and
some
of
the
other
stuff
we're
going
to
pod
right
away,
but
yeah.
G
H
In
the
architecture
there's
a
branch
in
astrolabe,
so
we
can
take
a
look
at
it
if
you
like
so
catch
up
with
me,
we
need
to
start
merging
that
stuff
back
into
the
the
mainstream
of
asteroid,
but
we
have
a
bunch
of
other
stuff
going
on.
A
I
have
one
more
question
about
the
security
aspect.
So
what
what
are
the
securities
concerns
that
folks
are
having
rafael
and
frankie?
Oh.
I
Okay,
so
the
for
this
gonna
solve
one
security
concern
which
is
connecting
a
legacy
cluster
to
an
object,
storage
that
not
necessarily
going
to
be
on-prem
or
off-prem.
So
that
means
that
you're
open
up
a
legacy
clustering
in
production
to
connect
to
a
source
using
a
credential
that
that
that's
a
security
problem
we
are
solving
now.
I
Of
course,
you
need
to
make
the
trust
between
a
service
cluster
to
this
legacy
cluster
over
the
cube,
adm
acuba
admin,
but
if
we
make
that
a
secret,
if
we
put
some
mitigation
for
that
trust,
one
one
mitigation
is
you
know
passwords
or
even?
I
This
is
another
thing
that
we
like
to
solve
in
this
problem.
If
we
have
an
object,
storage,
only
that
that
service
cluster
only
talks
to
that
object,
storage,
nobody
else,
not
even
the
source,
a
destination
cluster,
because
that's
one
of
the
problems
I
had
back
at
citigroup
using
valero
the
moment
that
you
take
a
backup
of
the
letter.
We
have
your
secrets
on.
H
I
Disk
right,
then,
you.
H
I
H
H
Well,
if
we
could
get
the
restore
working,
we
could
at
least
do
a
poc,
but
we'll
probably
have
to
work
through
the
restore
issues,
and
maybe
that's
something
we
put
on
the
run
one
seven
road
map,
or
at
least
we
get
a
poc
working
where
we
un-tease
that
it's
probably
not,
that
it
can't
be
that
bad
right.
Okay,.
D
But
if
you
keep
the
backup
on
local
storage,
then
maybe
you
should
that
local
storage
can
be
something
that
is
backed
by
min
io
and
then
you
can
initiate
the
restore
that
way.
H
D
Yeah
I
get
that,
but
for
restore
to
work,
you
need
to
have
a
backup
which
has
to
in
current
current
situation,
the
backup
has
to
be
in
a
backup,
storage,
location,
yep
right.
I
But
we
want
to
psl,
but
if
is
this
bsl?
It's
a
local
disk
on
my
service
cluster,
nobody
is
not
exposed
to
anywhere
I'm
solving.
The
security
problem
was
mentioned
before,
which
is
to
do
a
copy
paste,
I'm
still
kind
of
like
putting
I'm
persisting
this
data
somewhere
that
potentially
can
be
readable
at
rest
by
somebody
else.
D
Yeah,
I
think
what
what
we
are,
what
we
will
need
is
like
restore
from
not
just
a
backup
but
a
carbon.
H
Well,
I
so
so.
The
idea
here
is
to
take
this
whole
thing
outside
of
valero,
where
it's
no
longer
a
valero
back
up
and
to
restore
that
we
really
have
a
migration
utility.
That's
using
like
our
kubernetes
serializer
to
serialize
and
deserialize,
and
there's
really
even
no
need
for
local
storage.
They
should
be
piped
together
if
you're
doing
it
right
in
the
moment.
K
A
It
effectively
comes
down
to
a
bunch
of
cube.
Ctl
gets
piped
into
exactly
cto
apply
with
maybe
fancy
manipulation
in
between
in
the
pipeline,
but.
I
H
Yes,
I
think
that
would
be
a
good
poc
for
us
to
work
on.
Maybe
we
can
collaborate
a
bit
on
getting
that
restore
part
working
in
the
astrolabe
code.
G
G
I'm
definitely
willing
to
put
in
some
time
into
that.
G
A
Okay,
anything
more
on
that
topic.
A
C
Carlesia
yeah,
so
I
did
this
checklist
because
I
was
trying
to
keep
keep
track
of
things
myself
for
me
like
what
do
I
need
to
do?
What
have
I
done?
What
do
I
used
to
have
leftover
to
do
then?
I
figured
maybe
we
should
be
doing
this.
I
don't
think
this
list
is
probably
it's
probably
I'm.
It's
definitely
not
accurate.
For
example,
I
started
having
questions.
C
C
No,
no
so
so
there
so
there's
drc,
and
then
there
is
the
release
and
then
there
there
will
be
betas
in
between
what
are
the
sets
of
tests
that
we
want
to
run.
I
think
I
think
if
we
list
it
out,
okay,
because
what
we
do
is:
okay,
let's
run
this
manual
test,
but
I
don't.
I
have
a
hard
time
keeping
track
of
things
unless
I
have
a
checklist
for
myself
and
I'm
wondering
if
it's
the
same
thing
same
case
for
for
for
all
of
us.
C
I
don't
know,
but
the
the
my
question
is
so
when
we
did
the
rc,
you
specified
a
list
of
like
these
are
the
tests
there
are
before
the
rc
is
cut,
so
when
they
are
all
passed,
then
we
cut
the
rc.
So
now
we
do
this
other
set
of
tests
for
after
drc.
So
that
part
is
I
now
because
I
basically
just
copied
what
you
instructed.
C
So
I'm
having
the
same
questions
for
this
question
for
develop
for
the
actual
release.
Do
we
want,
because
I
know
what
the
whole
entire
like
the
full
set
of
tests
is.
My
question
is:
do
we
run
that,
after
the
release
and
before
the
release
before
and
what
do
we
do
for
the
beta
releases?
Do
you
know
what
I
mean
is
like?
Do
we
run
up?
I
I'm
just
having
these
questions
that
I
don't
know
the
answer
to
okay.
H
So
the
so
the
reason
I
had
like
a
subset
of
tests
for
a
release
candidate
is
because
these
are
tests
that
we
can
run
pretty
much
mechanically
and
if
we're
iterating,
which
is
what
we
should
be
doing
until
we
get
to
the
release
candidate,
we
can
quickly
run
these
tests
again.
So
I
didn't
want
us
to,
for
example,
run
all
the
manual
tests
on
the
release
candidate,
because
that
basically
consumes
the
team
for
a
week
yeah.
So
that's
that's
impractical.
H
So
what
I
wanted
to
do
was,
and
so
what
we
should
be
doing
is
the
release
candidate
should
be
as
close
to
final
as
possible,
and
I'm
not
sure
where
actually,
where
beta's
fit
in
this
betas
actually
should
be
before
the
release
candidate.
So
I
don't
think
we're
doing
betas
this
go
around,
so
betas
are
a
chance
for
people
in
the
field.
H
So,
by
the
time
we
get
to
release
to
the
rc,
we
should
have
a
lot
of
confidence
that
it's
just
going
to
pass
and
move
forward
and
by
the
time
we
cut
the
release
every
test
that
we
can
think
of
to
do
that.
We
think
we
should
do
should
have
been
performed
should
have
passed
or
if
we
catch
a
bug.
We,
you
know.
G
H
Make
a
decision
on
the
bug
is
this
a
release?
Block
or
not?
You
know
if
there's
sometimes
there's
things
you
say:
yeah,
there's
a
typo
in
the
docs.
Okay,
we're
gonna
live
with
that,
so
so,
where
we're
at
right
now
is
we
we're
doing
all
we're?
The
release
candidate
has
a
limited
set
of
tests
just
because
those
are
practical
to
run
frequently.
C
No
so
so
that
makes
complete,
like,
I
think,
that's
very
sensible.
So
let's
say
we
we
we,
we
cut
the
release,
we're
still
testing
the
the
release
candidates.
So
let's
say
where
we,
we
know
what
set
of
tests
to
do
for
the
for
this
point.
So,
let's
say:
okay,
two
weeks
from
now
we're
going
to
prepare
to
do
the
actual
release.
C
H
H
If
we
find
bugs
in
rc1
that
we
need
to
fix,
we
cut
a
new
build
and
we
go
back
to
the
beginning
and
we
redo
our
test
cycle.
So
by
the
time
we
get
to
the
end
of
the
test
cycle,
we've
done
all
of
our
tests
that
we
can
think
of
against
the
release
candidate
bits.
There
are
no
changes
in
those
bits
and
then
they
get
marked
as
released
make
sense.
H
C
H
C
Yeah,
so
that's
the
that's
one
of
the
questions.
The
other
question
is,
I
mean
the
main
question
is:
is
it
useful
for
us
to
have
this
checklist.
H
The
the
checklist
is
good.
I
think
we
need
to
expand
a
little
bit
in
terms
of
what
exactly
we're
testing,
but
I
don't
want
to
get
too
crazy
in
that.
I
was
asking
that
we
record
our
sessions,
so
I
started
a
page,
it's
internal.
At
the
moment,
I
started
a
page
with
like
test
logs
that
show
this
is
the
test.
We
run
this
the
output
of
the
test,
and
so
as
we
do
our
manual
testing
I'd
like
to
ask
the
people,
you
know
record
their
sessions
and
just
put
that
in
as
their.
E
H
H
H
A
lot
of
time
right
now
writing
in
detail
each
step
we
should
take,
but
I
think
if
we
record
what
we
did,
that's
probably
good
enough
and
then
we
can.
You
know
we
can
take
that
and
either
make
it
into
a
detailed
document
for
the
next
release,
or
ideally
we
make
it
into
another
test.
A
Okay,
so
there's
no
contributor
shout
outs
for
prs:
we've
kind
of
slowed
down
the
merge
rate,
basically
frozen
it
if
we
find
bugs-
or
if
you
find
a
bug,
please
submit
it
with
for
the
rc
and
we
will
evaluate
it
and
see
if
we
need
to
cut
a
new
one
or
what
we
need
to
do
for
the
contributor
shout
outs
on
the
q,
a
carlesia,
it
looks
like
we
don't
have
any.
A
Okay,
just
reminding
folks
the
discussions
page
is
super
useful
and
we
ask
that
people
ask
q
a
type
questions
in
the
community
support
q,
a
get
a
question,
github
discussion,
section
all
right,
any
final
subjects
for
the
meeting.