►
From YouTube: 2021-08-12 Crossplane 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
All
right
recording
has
started-
and
this
is
the
august
12
2021
cross
playing
community
meeting-
feel
free
to
add
yourself
to
the
attendees
list
here.
If
you
want
to
be
included
there,
I
will
drop
a
link
in
the
zoom
chat
right
now
for
the
agenda
document.
So
you
have
easy
access
to
it
there
too,
and
you
can
hop
into
the
to
the
party
here
now
with
nick
with
editing
access,
though
he
may,
he
may
disable
or
remove
any
of
your
comments.
A
You
may
add
so
we'll
see
we'll
see
how
he
behaves,
but
let's
jump
on
into
the
agenda
here.
I
think
we've
got
a
fairly
long
agenda.
I
don't
know
how
many
of
these
are
small
items
or
or
bigger
discussions,
but
let's,
let's
start
making
progress
so
upcoming
next
release
is
the
1.4
release
with
our
typical
two-month
cadence.
That
would
be
putting
the
release
date
to
august
31st,
which
has
been
the
entire
time
here.
So
no
changes
there
we're
coming
up
on
that.
A
We're
like
two
and
a
half
weeks
out
or
so,
and
so
with
the
normal
typical
flow
we'll
be
entering
into
future
freeze
next
week.
A
A
So
if
there's
you
know
outstanding
features
or
you
know
bigger
ticket
items
that
we
want
to
get
into,
then
you
feel
free
to
go
ahead
and
work
with
folks
to
kind
of
get
those
reviewed
get
those
in
so
that
we
can
kind
of
try
to
slide
in
all
the
riskier
churny
or
sort
of
things
before
before
feature
free
starts.
A
And
then
looks
like
they
dropped
into
the
chat
here,
another
opportunity
for
folks
to
introduce
themselves
if
they
haven't
already
so
I'll
pause
for
a
quick
second
in
case
anybody
does
want
to
introduce
themselves
still.
A
Alrighty,
so
we
can
jump
into
the
1.4
board
here
I
think
most
yeah.
We
got
a
good
turnout
on
the
call
here
today.
So
let's
go
ahead
and
we
don't
have
to
visit
every
issue
on
the
board
here.
A
That's
not
necessarily
the
agenda,
but
the
the
purpose
of
right
now
is
to
time
talk
about
the
big
items
that
are
in
play
for
1.4,
as
well
as
what
may
be
needed
to
wrap
them
up
and
get
them
in
as
well-
and
you
know
maybe
some
good
stopping
points
and
then
also
other
items
that
may
not
make
it
or
that
are
of
importance
and
then
especially
things
that
folks
in
the
community
either
are
working
on,
or,
I
think,
are
our
important
items
that
they
would
like
to
voice
a
a
vote
for
so
I'll
go
ahead
and
kind
of
start
it
off
with
nick
since
he's
at
the
top
right
here
you
know
some
of
the
work
that
you've
been
doing
there
nick
you
kind
of
catch
us
up
on
that
real
quick.
B
Yeah
sure
the
the
two
key
things
that
I
have
in
flight
in
implementation
at
the
moment
that
are
sort
of
in
review
stage,
one
is
a
new
feature
of
crossplane
which
we're
calling
composition
revisions.
The
general
idea
with
this
is
up
until
this
point
in
crossplay
1.3.
B
B
You
might
want
to
like
slowly
roll
out
changes
through
your
organization
and
test
it
on
one
composite
resource
or
something
before
you
sort
of
roll
it
out
to
all
the
others.
So
there
is
a
pull
request,
open
and
designed
off
merged
for
for
adding
composite
resource.
Sorry,
composite
revision,
composition,
revision,
support
to
cross
plane,
I'm
hoping
that
it
will
lay
anti-cross
by
1.4
I'll
need
to
nudge
the
other
maintainers.
To
do
another
review
pass.
B
I
think
so
that
we
can
get
it
in
before
the
feature
freeze
next
week,
but
but
I'm
definitely
hoping
to
have
that
in
for
1.4
it
will
be
behind
a
feature
flag.
One
thing
that
we're
looking
at
doing
going
forward
is
all
features,
definitely
a
costly
core,
potentially
in
providers
that
are
alpha,
including
resources
that
are
alpha.
B
I
would
like
to
get
behind
a
feature
flag
and
have
them
off
by
default,
just
because
we
have
definitely
found
that,
even
though
we
have
api
types
and
code
that
is
marked
alpha,
it's
easy
to
miss
that
and
not
understand
the
expectations,
but
crosstalk
basically
doesn't
have
a
contract
for
for
alpha
resources.
B
So
yeah,
that's
one
thing:
the
other
thing
is
there
is
apr
that
does
a
sort
of
this
is
less
less
shiny,
exciting
features,
more
reliability
and
quality.
We
discovered
a
couple
of
races
and
a
security
issue
with
the
crossland
claim
reconciler
and
sadly,
when
I
say
discovered
a
security
issue
I
mean
I
was
in
there
refactoring
it,
and
I
saw
a
two
dude.
That
said:
hey,
maybe
there's
a
security
issue
from
me
about
six
months
ago
I
was
like
oh
yeah.
B
I
forgot
about
that,
and
it
turns
out
there
was
a
potential
security
issue
there.
So
there
is
a
problem
in
cross
playing
at
the
moment
in
1.3,
where
a
claim
which
is
a
namespace
resource
can
explicitly
target
a
composite
resource
that
does
not
belong
to
it.
Up
at
the
cluster
level
could
just
say:
hey
you
know
I
own
this
composite
resource
and
crossplane
will
just
honor
that,
unfortunately,
so
in
there
there
is
code
that
checks
this,
but
the
code
was
refactored
at
some
point
and
was
moved
out
of
order.
B
So,
basically,
currently
a
claim
if,
if
someone
has
r
back
access
in
one
name
space
but
not
in
another
namespace,
you
can
make
a
claim
that
points
to
an
xr
that
was
started
from
another
namespace
and
just
delete
it
or
change
it.
Things
like
that.
So
that's
not
great.
So
there's
a
fix
open
for
that
at
the
moment.
Again,.
C
This
is
a
small
question
there.
So
is
this
because
we
don't
check
the
claim
ref
on
the
xr
and
just
like
you
know,
let
it
work
with
that.
B
Yeah,
unfortunately,
so
we
do
check
the
claim
ref,
but
we,
but
we
check
it
too
late.
We
check
it
so
the
the
flow
at
the
moment
is
we.
If
it's
being
deleted,
we
don't
check
at
all
and
if
it's
not
being
deleted,
we
actually
currently
render
out
the
xr,
and
then
we
go
into
the
bind
stage
and
in
the
bind
stage
we
check
whether
they're
allowed
to
bind.
B
But
at
that
point
it's
too
late
right,
we've
already
overwritten
the
xr,
so
so
that's
all
being
refactored
and
reordered,
which
will
be
functionally
no
change.
There
was
also
a
potential
race
in
there
as
well,
which
is
kind
of
getting
integrated
details,
but
there's
basically
a
lot
of
hardening
of
the
claim
reconciler
logic.
B
So
those
are
the
two
big
things
I
have
personally
in
flight
at
the
moment.
D
Nick
on
on
the
the
fix
you
were
just
talking
about,
do
you
plan
on
backporting
that
to
all
active
branches
and
then
doing
immediate
patches
on
this.
B
I
hope
to
is
a
fairly
large
change,
so
it's
possible
that
we
could
get
into
some
on
one
hand.
I
don't
think
we've
touched
the
claim
reconciler
for
a
while
and
it
is.
It
is
constrained
to
the
claim
reconciler.
So
so,
hopefully
those
back
ports
will
be
simple,
but
yes,
it
does
seem
to
be
it's
a
security
fix.
So
it
does
seem
to
be
something
that
people
want
back
for.
D
Gotcha
also
just
a
heads
up
to
everyone
with
the
1.4
release
coming
out
that
by
default
means
that
the
1.1
release
branch
will
be
going
out
of
upstream
service
so
plan
accordingly,.
B
Right,
so
just
so,
we'll
have
1.4,
1.3
and
1.2
will
be
the
supported
sort
of
versions
that
we
patch
things
for.
A
Awesome
nick
yeah,
thanks
for
all
those
updates
there
and
some
of
the
things
you're
digging
into
right
now
and
then
you
know
the
effort
to
get
those
in
for
the
1.4
release
and
then
backboard
as
well
too.
Do
you
want
to
give
us
an
update
on
some
of
the
things
that
you've
been
focusing
on
as
well
too.
C
Yeah,
so
I
think
there
are
a
few
things
here
that
that
are
assigned
to
me,
but
my
focus
has
been
another
thing
for
a
while
now,
which
is
so.
We
are
working
on
a
new
code
generation
tool
called
terajet
that
will
be
generating
crds
and
controllers
for
arbitrary
terraform
resources.
C
C
Candidates
are
just
as
usual
like
aws,
azure
and
gcp,
but
I
believe
once
like
you
know,
we
have
all
those
like
you
know
three.
We
will
also
cover
like
a
lot
of
exceptional
cases
and
then,
like
you
know,
community
can
contribute
more
telephone
pro
generated
providers
from
terraform.
C
So
that's,
I
think,
the
the
biggest
thing
that
I'm
that
me,
hassan
and
alpha
are
working
on
and
hoping
to
demo
it
on
the
on
the
next
community
meeting
at
least
some
part
of
it
get
ready.
C
A
Yeah,
that's
definitely
really
exciting
stuff
to
vastly
expand
coverage
of
supportive
resources
in
the
near
term
and
often
did
you.
I
know
there
was
more
activity
on
the
design
document
for
cross
resource
references.
Is
that
nearest
states
to
be
ratified
or
what's
the
big
things
left
on
it.
C
Yeah
it
is,
it
is
very
close
to
merging.
It
is
just
I
think
it
needs
some
reverting
about,
like
you
know
its
goals
like
what
it
wants
to
achieve.
Initially
we
thought
it
would
be
like
you
know,
I
mean
we
still
think
like
some
of
us
don't
think
like
you
know,
that
is
the
case,
but
like
we're
still
kind
of
deciding
whether
we
want
this
to
replace
existing
cross
resource
references.
E
C
Just
be
an
additional
layer
where
you
can
refer
an
information
from
any
resource
to
any
other
resource,
and
in
fact
today
we
were
brainstorming
with
the
sun
and
opera,
and
I
think
the
having
this
would
help
with
cross
provider.
References
as
well,
which
is
just
like
not
possible
to
do
in
development
time,
as
opposed
to
our
like.
You
know,
current
cross
resource
references
right
and
usually.
E
C
We
were
thinking
was
that
you
would
use
native
provider
aws
or
native
provider
azure
for
all
most
of
your
resources
and
for
the
ones
that
are
not
implemented.
Yet
you
would
use
like
telephone
based
one
and
refer
to
them
like
now
refer
to
each
other
using
this
new
patching
mechanism,
let's
say
so.
There
are
a
few
things
like
no
conceptual
things
that
we
need
to
identify,
but
yeah
the
the
technical
stuff
like
know.
D
A
That
sounds
good
is
hassan
on
the
car.
Yes
hassan's
on
the
call
as
well
too.
I
know
you've
been
focusing
on
like
pluggable
secret
stores,
which
is
a
highly
requested
feature
from
from
the
community.
Do
you
want
to
give
us
a
quick
update
on
where
that
is.
E
Yeah
yeah
sure
yeah.
Currently
I
have
like
I'm
about
to
open
a
design
document
for
that,
basically
like
there
are
a
couple
of
options
that
that
we
can
follow
and
I'm
planning
to
open
a
design
document
with
like
listing
all
different
options
and
to
to
gather
some
feedback
from
the
community,
basically
like
for
for
people
who
are
not
familiar
with
the
issue.
E
This
would
plan
to
support
different
secret
store
or
secret
backhands
other
than
kubernetes
secrets
to
write
connection
details
so
that,
hopefully
we
can,
let's
say,
store
our
connection
details
in
a
bold
instance,
for
example.
So
I
I
I
know
that,
like
there
are
a
lot
of
you
know,
people
who
are
interested
in
the
in
the
issue,
but
if
there
are
any
other
feedback
until
I
I
could
open
the
design
doc.
E
Anyone
can
could
comment
on
the
tickets
with
with
any
suggestion,
otherwise
stay
tuned
for
the
design
dock,
and
we
will.
We
will
be
able
to
make
an
open
discussion
on
the
design,
though.
A
Perfect
yeah
that'll
be
a
great
opportunity
for
folks
from
the
community
to
weigh
in
on
some
of
the
options
there
and
where
we're
going
with
that
design,
especially
since
it
is
a
highly
demanded
feature
from
the
community.
That'll,
be
really
good
to
get
some
feedback
and
make
sure
that
we're
taking
this
in
the
direction
that's
going
to
solve
people's
problems.
A
So
thanks
for
that,
hassan
and
alpera,
do
you
do
you
have
something
you
want
to
go
ahead
and
update
the
community
on
as
well
too
for
for
1.4
and
some
of
the
work
you've
been
doing.
E
Yeah
regarding
this
support
merging
greater
than
replacing
objects,
it
said
a
companion
pull
request
in
cross
plane,
runtime,
which
got
merged,
and
I
have
also
received
comments
from
nick
for
the
crossplane
counterpart,
which
actually
exposes
the
api
as
part
of
the
patches
I
haven't
been
able
to.
You
know,
address
those
reviews,
but
you
know
I
definitely
would
like
to
if
possible,
get
this
for
1.4.
A
I
think
I
think
that
status
is
perfect
out
there
of
you
know
of
where
it
is
now.
There's
the
runtime
part
is
merged,
and
then
you
know
the
the
core
aspect
of
the
experience
around
it.
We're
hoping
to
get
in
for
1.4
incorporate
that
feedback.
So
that
sounds
perfect,
awesome,
helper,
okay
and
then
dan
did
you
have
some
things
that
you
were
focusing
on
to
get
in
for
1.4
as
well
too.
D
There's
a
couple,
I
guess
in
the
in
progress
column
the
exposing
pole
interval
on
external
client,
that's
implemented
for
aws
and
gcp.
I
think
I
still
need
to
add
that
to
azure,
but
that
just
basically
gives
you
the
ability
to
set
how
often
we
check
manage
resources
at
the
kind
of
global
scale
on
providers,
and
then
we
can
spread
that
around
to
provider
conf
or
crosslink
and
trip
as
well
and
yeah.
D
The
the
only
other
thing
is
the
too
many
update
calls
for
crds,
which
I
added
to
the
further
discussion.
At
the
end
of
this
call,
we
can
chat
about
it.
There.
A
Cool
sounds
good
all
right.
So
then
also
an
important
part
of
this
part
here.
Well,
two
things,
one
anything
else
from
contributors
on
the
call
here
that
they
want
to
update
us
on
or
things
that
they're
working
on
or
things
they
want
that
they're
planning
to
include
in
1.4.
A
All
right,
and
so,
if
that's
most
of
the
work,
that's
ongoing
there
I'd
also
open
the
floor
to
the
general
community
here.
Everyone
on
the
call,
if
there
are
things
that
are
on
your
mind
or
things
that
you
know,
issues
that
you
are
hoping
to
get
addressed
or
want
to
put
a
plus
one
on.
Then
you
know
that's
a
perfect
time
to
go
ahead
and
voice
your
support
for
some
issues.
Right
now,
too,.
B
Just
before
we
jump
into
that
one
thing:
I'd
consider
that
we
might
want
to
doing
the
spin
off
the
cuff,
but
one
thing
that
may
be
worth
doing
is
sort
of
giving
a
giving
a
bit
of
a
general.
B
Update
from
the
various
companies
that
work
on
crossplane
as
to
like
what
those
what
those
people
are
focused
on
when
they
are
contributing-
and
I
think
at
the
moment-
that's
upbound
is
one
of
the
you
know
the
primary
owners
of
crossblading
contributors-
and
I
you
know,
I'm
looking
at
this
board
and
I'm
seeing
nothing's
moved
into
the
done
column
yet
for
the
for
the
1.4
release,
and
I'm
worried
that
people
might
think
we're
not
doing
very
much
so
just
to
set
context.
B
A
lot
of
that
is
because
the
terror
jet
stuff
that
we've
asked
mentioned
like
we
have
we're
actually
doing
a
lot
on
the
provider
from
the
moment
rather
than
this
release
so
so
for
context.
It's
kind
of
mostly
me
working
on.
D
A
Yeah,
I'm
hoping
to
close
out,
you
know
a
fair
amount
of
these
as
well
too
over
the
next.
You
know
weekend
and
your
feature
freeze
and
code
freeze
as
well
too
yeah,
it
seems
yeah
seems
like
there's
a
lot
of
progress,
a
lot,
a
lot
of
things
that
people
have
been
focusing
on
as
a.
D
A
Okay,
cool
yeah
and
we're
always
always
available
for
and
we're
happy
to
receive
feedback
on.
You
know
issues
that
are
of
importance
to
the
community,
so
this
is
not
the
only
venue
to
do
that
feel
free
to
always
put
a
thumbs
up
on
issues
that
are
important
to
you
and
talk
about
them
on
slack
is
as
well
too
so
we
can
kind
of
get
some
visibility
onto
any
issues
that
may
be
important
for
folks.
A
Okay,
so
I
think
then
we
can
go
ahead
and
move
on
to
the
community
topic
section
of
the
agenda
here
so
up
front
real
quick.
We
have
a
couple
of
links
to
various
contents
like
live
streams
and
webinars
and
videos
and
all
sorts
of
stuff
like
that,
since
the
last
community
meeting
there
is,
I
think,
victor
just
released
a
new
video
today.
I
think
this
morning
about
how
to
use
cross
plane
and
flux
together
to
manage
your
infrastructure.
A
So
that
was
a
good
video.
I
got
a
chance
to
see
that
one
already
that's
the
top
link
here.
Aaron
did
a
live
stream
last
week
about
some
of
the
support
he
added
recently
for
a
gke
auth
provider,
so
that
provider
helm
works
very
well
going
forward
with
newer
versions
of
gke.
A
I
think
dan
and
steven
did
a
deep
dive
recently
on
crossplane
with
the
sibo
folks.
So
there's
a
oh
cool
nice
little,
oh
whoa,
a
place
too!
Oh
wow!
That's
cool
a
little
preview
there
for
you
so
another
another
good
kind
of
thing
to
start
diving
into
some
more
of
the
specifics
and
internals
of
crossplane
and
then
the
kubernetes
atlanta
meetup
that
we
had
talked
about
last
time.
The
recording
is
up
here,
so
I
shared
that
link
here
in
the
dock
as
well
too.
A
So
a
number
of
great
video
resources
to
check
out
and
stay
on
stay
on
top
of
various
use
cases
for
cross
claims
and
new
stuff
we've
been
doing
so
you
can
check
those
out
there.
I
I
we've
had
this
documentation
survey
here
for
the
crossplan.aodocs
for
a
couple
a
couple
weeks
now,
so
I
probably
won't
include
it
for
the
next
agenda,
but
I
think
that
we're
still
happy
to
receive
feedback
around
your
experience
with
the
documentation
how
things
are
presented.
A
A
quick
update
for
the
incubation
advancement
with
the
cloud
native
computing
foundation,
so
the
toc
vote
is
underway.
I
think
we're
up
to
like
four
binding
votes.
Now
and
we
need
a
total
of
either
eight
or
nine.
I
can't
remember
now
so
yeah
those
votes
are
coming
in
as
the
process
is
unfolding,
so
hopefully
by
the
next
community
meeting
we'll
have
an
update
where
we
have
the
binding
votes
and
we
can
go
ahead
and
make
the
announcement
that
that
the
incubation
proposal
has
been
accepted
and
ratified.
A
A
quick
announcement
about
new
maintainers,
so
a
number
of
folks
that
have
been
contributing
to
crossplane
as
a
project
for
a
while
now
and
doing
some
great
work
around
reviewing
pull
requests,
and
you
know
enforcing
the
quote:
quality
standards
and
all
that
sort
of
stuff
have
been
made.
Maintainers
on
a
couple
of
various
providers
and
projects,
so
hassan,
al
pere
and
aaron
are
all
newer,
maintainers
and
they're
happy
to
jump
in
and
start
or
continue.
Actually,
it's
kind
of
a
continuation
of
their
efforts.
A
But
you
know
continuing
to
review,
pull
requests
and
kind
of
keep
things
merging
and
helping
the
community
out.
So
thank
you
very
much
hassan
and
au
pair
and
aaron
for
all
the
work
you've
been
doing
and
the
commitment
to
the
project
as
well
too,
to
keep
growing
the
maintainer
team
and
scaling
the
community
as
a
whole.
So
that's
fantastic.
A
A
So
that's
something
that
they've
been
working
on
for
a
while
and
it
seems
fairly
robust.
So
far
I
didn't
get
a
chance
to
look
too
deep
in
it.
Does
anybody
want
to
maybe
dan
if
you
have
a
little
bit
more
context
about
some
of
the
scenarios?
That's
opening
up
for
folks
or
this
new
provider.
D
Yeah,
I
don't,
I
honestly
haven't
used
it
at
all
or
anything
like
that,
but
I
do
know
that
I
believe
it's
used
in
a
production
setting
today,
and
so
I
have
reasonable
confidence
in
its
quality.
We're
definitely
right
now,
it
doesn't
have.
You
know,
cross
plane,
ci
setup
with
it,
so
we're
working
on
getting
that
set
up
and
then
we'll
have
official
crossplane
images
being
published
for
folks
to
be
able
to
consume.
D
So
if
that's
something
you're
interested
in,
we
can
definitely
probably
be
a
cool
thing
to
get
some
of
the
folks
who
worked
on
that
on
a
live
stream
and
walk
through
it
and
kind
of
give
descriptions
for
the
resources
it
supports.
And
then
you
can
also
see
those
on
doc.crds.gov
also.
A
Yeah
really
really
excited
to
see
new
providers
coming
out,
and
especially
ones
that
are
have
a
good
maturity
to
them
as
well
too.
It's
kind
of
fleshed
out
a
little
bit.
A
Okay,
cool-
let's
see
next
topic
here,
is
around
kubecon
north
america.
I
think
it's
like
october
11th,
the
13th
or
something
along
those
lines.
The
we
received
an
email
this
morning
to
all
project
maintainers
in
the
cncf
in
general.
That's
about
getting
in
our
our
desire,
I
suppose,
for
how
we
want
to
engage
at
kubecon
the
project
pavilion
like
having
a
little
booth
and
stuff
like
that.
So
at
this
time,
at
just
up
this
morning,
they
are
still
pushing
forward
with
an
in-person
component.
A
So
we
could
have
a
booth
and
be
you
know
interacting
with
folks
in
in
person,
in
los
angeles
this
this
october.
So
I
wanted
to
just
kind
of
let
people
know
about
that
that
that
appears
to
still
be
on
the
table.
There
is
possibility
for
still
for
being
in
person,
and
you
know
attending
kubecon
and
staffing
and
being
part
of
the
crossplane
booth.
They've
already
got
some
interest
from
dan
and
grant
maybe
victor
as
well
too.
A
So
if
folks
are
interested
in
that,
let
me
know,
but
I
think
we've
got
a
few
folks
there.
So
we'll
probably
push
forward
and
have
you
know,
have
the
booth
there
it'd
be
have
some
in-person
component
of
as
well
too
there's
also
it's
a
hybrid
event,
so
there's
virtual
aspects
of
it
as
well
too.
So
I
think
the
office
hours
component
is
virtual,
so
there
will
be
multiple
opportunities
for
to
folks
to
folks
to
engage,
present
and
and
be
part
of
this
experience
as
well
too.
A
You
don't
have
to
be
in
los
angeles
for
that
to
happen,
so
you
know
any
interest
of
you
know
being
at
the
booth
or
hosting
or
participating
in
the
office
hours
or
things
like
that.
You
know
just
let
me
know
and
happy
to
include
as
many
people
as
is
want
to
get
involved.
Basically,.
A
All
right,
cool
and
then
nick
do
you
want
to
talk
about
some
of
the
we
talked
a
little
bit
about
in
the
status,
but
just
like
as
a
general
initiative
that
you're
undertaking
this
quarter.
B
Yeah
sure
similar,
but
basically
what
it
says
on
the
agenda,
what
it
says
on
the
10.
My
one
of
my
big
focuses:
this
quarter
is
going
to
be
quality
performance
and
security.
Investments
in
in
crossplane,
particularly
scalability.
We
increasingly
are
seeing
folks
come
to
us
and
say:
hey
you
know.
Cosplay
works
are
okay
with
you
know
thousands
of
composite
resources,
but
it
gets
kind
of
slow,
and
this
is
something
we
absolutely
invest
expected.
B
You
know
the
team
that's
been
working
on
cosplay
for
a
while
has
definitely
taken
a
you
know.
We
don't
want
to
prematurely
optimize
approach.
We
designed
for
sort
of
code,
readability
and
extensibility,
and
you
know
now
that
we're
finding
the
people
operating
scale
we
have
data,
we
can.
We
can
start.
You
know
investigate
some
things.
I
am
also
looking
at.
B
You
know,
as
I
mentioned
some
race
conditions,
security
things
like
that
one
one
sort
of
note
here
is
that,
unfortunately,
or
fortunately,
depending
on
how
you
look
at
it,
I
am
going
to
be
looking
at
this
from
a
areas
that
have
the
biggest
impact
possible.
So,
if
there's
updates,
we
can
make
to
our
overall
reconciliation
strategy.
B
Overall,
sort
of
controller
libraries,
crosslink
core
things
like
that,
I'm
probably
not
going
to
be
your
person
to
like
dig
into
individual
resources
that
are,
you
know,
you
know
one
management
source
or
the
other.
That's
having
quality
issues,
something
I've
got
to
focus
on.
You
know
if
something
affects
10
different,
managed
resources
or
all
cosplay
resources,
or
something
like
that.
That's
that's
what
I'm
going
to
try
and
focus
on
this
quarter.
A
You
know,
I
think
that
that
effort
there
nick
is
going
to
have
big
dividends
that
pay
off
as
well
too
man.
So
I'm
really
excited
to
see
that
you
know
that
you've
already
dove
into
it
and
starting
to
find
some
good
issues
and
investments
to
make
here.
So
thanks
a
lot
for
doing
that,.
A
B
Anyone
has
an
issue
that
feels
I
haven't
really
got
too
much
of
a
system
going
on.
The
one
thing
I
have
done
is
anything
that's
scalability
or
performance.
Limiting
we
do
have
a
label
that
is
the
performance,
a
label
in
some
of
our
repos.
If
anything
immediately
springs
to
your
mind
that
you
haven't
seen
me
pop
up
on
the
issue
in
the
last
couple
of
days.
B
D
B
You've
got
any
sort
of
quality
or
scalability
things
or
quality
or
security
things
that
I
should
be
looking
into.
Please
do
bring
them
to
my
attention.
I'm
trying
to
just
sort
of
get
a
picture
of
all
the
outstanding
issues
that
are
there
at
the
moment
to
sort
of
prioritize
and
start
sort
of
figuring
out
where
to
get
started.
A
Awesome:
okay,
then
muafik.
Do
you
want
to
talk
about
provider
ownership
here.
C
Yeah
yeah
well,
so
ownership
may
not
be
the
best
word,
but
last
week
we
met
with
the
maintainers
to
talk
about
something.
That's
been
like.
You
know,
a
topic
of.
In
my
mind
that
is
like
you
know.
C
Sometimes
there
are
issues
open
and
like
your
pr's
open,
but
for
you
know
for
a
while
they
don't
get
response
or,
like
you
know,
some
for
some
reason
they
just
look
through
the
cracks,
so
especially
the
provider
ones,
and
so
we
wanted
to,
like
you
know,
say
a
sign
or,
like
you
know,
have
someone
just
like
you
know,
keep
track
of
them
like
not
actually
working
on
them
or
like
fixing
them,
but
actually
you
know
bringing
them
up
in
the
community
meeting
or
or
like
you
know,
or
upon
planning
meetings,
so
that
we
are
aware-
and
we
are
you
know
we
prioritize
so
that,
like
you
know
those.
E
C
Sorry
so
that
those
issues
are,
you
know,
acknowledge
it
and
kept
track
of.
So
these
are
the
initial
assignments.
So
aws
has
two
me
and
aaron
because,
like
you
know
it's
it's
got
the
most
activity
among
the
providers
and
azure
and
http
azure
as
alpha
and
gcps
hassan.
C
So,
like
you
know,
as
I
said
like
you
know,
these
are
not
like
this
time.
That
assignment
is
not
meant
for,
like
you
know,
hey,
you
are
the
one
that,
like
you
know,
everyone
should
ask
questions
or,
like
you
know,
should
fix
the
pr
review,
all
the
pr's
and
all
that
stuff.
But
it's
just
that,
like
you
know,
keeping
track
of
things
like
you
know
being
on
top
of
the
issues,
so
that,
like
you
know,
the
community
is
aware
of
them.
C
For
example,
if
you
bring
up
some
issue
here,
a
community
member,
could
you
know
take
that
issue,
I'm
like
no
fix
it
but
like
if
we,
if
you
don't
want
like
you
know,
we
can't
go
into
like
every
individual
issue
and,
like
you
know,
triage
them
in
the
community
meeting.
So
these
people
will
do
that,
like
you
know,
interactively
and
then,
like
you
know,
come
up
with
a
list
of
issues
or
prs
that
are
like
you
know
worth
mentioning
in
the
next
community
meeting.
A
C
A
Yeah,
that's
that's
exactly
the
point
I
wanted
to
make
too
much
is
that
you
know
weird
like
the
way
you
scale
the
throughput
or
velocity
of
an
open
source
community
is,
you
know,
scaling
the
folks
that
can
help
pr
shepherd
vr's
through
too
right.
So
you
know
we're
always
happy
to
grow
the
maintainer
team.
If
folks
want
to
get
more
deeply
involved,
you
know
start
reviewing
polar
cross,
making
more
contributions,
we're
happy
to
work
with
folks
and
kind
of
support.
A
You
know
their
growth
as
maintainers
and
scaling
the
project
that
way
so
always
reach
out
to
us
if
you're
interested
in
getting
deeper
into
the
project
and
contributing
we're
happy
to
have
more
people
in
scale
to
keep
continue
to
scaling
to
continue
to
scale
the
community
awesome
yeah
thanks
for
everyone's
efforts
on
that.
Definitely
appreciate
that.
Nick
you
wanted
to
talk
about
an
indesign
column
too.
B
Yeah,
just
a
thought
that
came
up
earlier
in
this
one
thing
I
noticed
when
I
look
at
this
board-
is
that
for
folks
who
you
know,
aren't
as
intimately
familiar
with
the
goings
over
the
crosswind
project
as
some
of
us
maintainers
are,
it's
probably
not
obvious,
there's
a
bunch
of
stuff
that
says
it's
in
progress
like
plugable
secret
stores,
a
lot
of
the
things
about
like
pulling
pulling,
can't
find
it
now
the
cross
resource
references
and
things
like
that,
and
I
think
that
could
it's
very
vague
as
to
like.
C
B
So
I
would
suggest
that
we
might
want
to
actually
break
up
the
sort
of
in
progress
and
review
in
progress
to
reflect
whether
it's
actually
implementation
is
in
progress
or
just
design,
dockers
in
progress,
or,
just
maybe
add,
like
a
design
dot
column
to
be
like
yeah,
where
you
know
indesign
or
something
like
that.
I
don't
know
how
folks
feel
about
that,
but
I
feel
like
it
might
make
it
more
obvious.
B
A
Yeah
good
good
question,
nick
one
of
the
things
that
I
might
be
kind
of
struggling
with,
that
is,
you
know,
if,
like
being
able
to
represent
the
various
states
of
of
the
progress
of
something,
that's
a
design
document
only
like
if
it's
in
design
and
then
does
it
have
like
a
separate
column.
That's
you
know
in
review
as
well
too,
or
is
it
just
stay
in
the
indesign
column
until
it's
it
ships
or
it
was
their
separate
done
column.
So,
like
do
you?
A
Do
you
separate
or
bifurcate
like
designs
and
features
the
entire
way
across
the
board,
or
I'm
not
quite
sure
exactly
how
to
manage
the
lifecycle
yet?
So
I
would
have
to
kind
of
think
about
that
and
see
what
that
experience
is
like
a
little
bit
more.
A
Yeah,
okay,
yeah
yeah-
that
would
make
that
can
make
plenty
of
sense.
I'm
not
against
that
at
all.
One
thing
that
is
nice
is
that
I
think,
at
least
mostly
through
convention's
sake,
is
that
the
things
that
are
designed
design
only
focus
is
that
the
the
issues
tend
to
have
design
document
in
the
title
at
least,
I
think
like
the
like
the
cross
resource
reference.
One
is
one
I
could
think
of
says
design
document
on
it,
so
that
does
help
with
the
clarity.
A
I
don't
think
we
do
that
ubiquitously,
but
that
is
that
is
somewhat
helpful.
Either
way,
though,
having
an
indesign
column,
which
you
know
is
very
clear
about
what
sort
of
focus
is
going
on,
and
you
know
what
what
is
the
maturity
of
of
the
effort
there?
That's
still
in
the
design
phases,
it
could
be
helpful
and
I
definitely
support
that.
A
Yeah
yeah
you've
got
editor
permissions
on
this
google
doc,
so
you
can
just
do
anything
willy-nilly
across
the
project
now
yeah.
That
sounds
good
to
me,
nick
okay,
cool,
I
I
I
want
to
just
call
quick
attention
to
this
item
here
this
item-
I
guess-
and
the
agenda
is
that
I
added
this
one
a
couple
community
meetings
ago
and
I
think
it
has
a
lot
of
potential
too.
You
know
folks
can
like
be
adding
things
to
it.
A
You
know
before
the
meeting
or
as
we're
progressing
is
that
this
I
think,
like
the
intent
of
this
agenda
item
here
is
just
things
of
interest
that
have
gone
on
in
the
community
like
interesting
discussions
on
slack
where
somebody's
using
crossplane
in
a
way
we
hadn't
really
heard
of
before
or
running
into
an
interesting
issue
that
you
know
it
didn't
get
some
some
attention
earlier
in
the
meeting
just
interesting
things
happening
in
the
community.
A
You
know
feel
free
for
upcoming
community
meetings
to
add
stuff
into
here
as
well
too,
to
kind
of
call
attention
to
it
as
a
bit
of
a
catch-all
for
things
that
maybe
weren't
discussed
otherwise,
but
are
interesting
things
that
are
just
kind
of
neat
to
talk
about,
or
educational
or
funds
too,
so
feel
free.
B
A
Add
stuff
into
that
column
or
this
item
for
next
time
too
so
yeah
so
we'll
keep
moving
along,
though,
because
we're
about
two-thirds
through-
and
I
see
a
lot
of
links
here,
but
I
don't
know
if
that
indicates
a
duration
of
them,
but
nick.
I
think
you
added
this
this
item
here
on
new
since
last
community
meeting.
B
Yeah,
so
so
so
I
might
have
misunderstood,
but
basically
when,
when
we've
off
mentioned
this
concept
of
provider,
ownership
I.e
that
people
will
come
and
sort
of
give
a
bit
of
an
update
as
to
where
the
providers
are
at
and
we'll
be.
I
think
there's
two
parts
of
it
right.
One
was
that
they
will
be
responsible
for
sort
of
triaging,
providing
issues,
sort
of
thing
and
keeping
people
honest,
make
sure.
B
There's
a
signees,
I
actually
volunteered
to
do
that
for
crossplay
and
cosplay
runtime,
and
one
of
our
kind
of
forcing
functions
we
discussed
doing
potentially
was
to
effectively
do
a
status
update
in
this
meeting
so
that
at
least
we
have
to
shame
ourselves
by
being
like
there's
a
hundred
open,
pull
requests
or
whatever-
and
you
know
some
people
get
a
bit
of
accountability
there.
So
this
is
just
my
attempt
to
do
that.
We
don't
have
to
walk
through
all
of
them.
This
is
effectively
there's.
B
Perhaps
some
overlap
with
the
pr's
to
discuss
or
something.
But
this
is
more
a
handful
of
issues
that
I
thought
were
of
interest
that
are
new
since
the
last
community
meeting,
and
also
just
the
notes
that
there
are,
you
know,
there's
about
10
other
things
that
have
been
opened
that
I
didn't
think
were
particularly
like.
There
were.
D
B
So
much
from
the
community
or
already
had
action
sort
of
thing,
so
you
can
basically
look
at
this
list
of
things
as
stuff.
That
is
new
in
the
last
two
weeks
that
may
or
may
not
that
probably
warrants
like
attention
sooner
rather
than
later,
that
isn't
just
like
hey.
We
should
go
build
this
sometime
in
the
future
kind
of
issue
or
an
internal
sort
of
accounting
thing
from
the
maintainer
team.
D
B
Out
I
I
do
see
that
this
agent
is
getting
sort
of
pretty
long.
There's
overlap
with
other
sections,
so
you
know
I
don't
know
it
might
be,
might
be
worth
like
a
better
sync
at
some
point,
to
figure
out
how
we
want
to
do
this
or
represent
this.
A
And
nick
are
there
any
specific
issues
here
that
you
want
to
dive
into
or
kind
of
worth
getting
into
at
least
a
high
level
overview
about
them
for
for
educational
sake
here
or
not?
No.
B
Some
of
them
are
already
some
of
them.
We've
already
covered
like
the
races
and
security
one
and
then
the
others
are
kind
of
fairly
fairly
straightforward
things.
It's
more
just
an
accountability
that,
like
let's
say
the
scale
away
thing
right.
Someone
from
the
community
wants
skateboard
support,
there's
a
provider
just
as
a
definitely
we're,
not
probably
not
going
to
go,
build
that
in
the
next
fridge
or
anything
like
that.
We
just
wanted
to
mention
these
things
as
they
come
up.
A
Okay
I'll
tell
my
friend,
dan
and
I'll,
tell
nick's
friend,
too,
okay
cool
all
right.
That
sounds
good
yeah,
thanks
for
bringing
some
attention
to
those
as
well,
nick
all
right
and
then
let's
jump
into
the
pr's
as
well
too,
to
kind
of
get
some
eyes
on
these
nick
here
europe
first
for
this
one.
B
Yeah,
this
is
one
that
we
asked
for
some
community
input
for,
and
I
think
we've
already
got
a
good
amount
of
community
input
that
is
at
least
tangentially
around
what
we
were,
what
we
were
saying
so,
a
while
ago,
one
community
member
added
a
a
bit
of
functionality
to
all
crossplane
resources
once
once
the
new
cross,
plane,
runtime
release
goes
out
where
in
the
debug
logs,
if
we
determine
that
a
particular
resource
is
not
up
to
date,
I.e
that
we
need
to
go
and
do
something
to
the
cloud
to
change
it
to
match
your
declarative
state.
B
We'll
now
include
in
our
debug
blog
about
that
a
little
diff
of
why
everything
is
not
up
to
date,
and
I
think
that's
a
great
idea.
This
is
a
tricky
part
of
the
logic
and
a
lot
of
the
comments
on
this
pull
request.
So
far,
are
people
calling
that
out
so
when
you're
developing
cosplay
resources,
definitely
and
often,
if
you're,
like
debugging
you're,
trying
to
you
know
if
you're
deployed
cross
playing
and
you're
like,
but
there's
one
resource
that
updates
all
the
time.
B
We
should
also
do
that
in
the
events
that
are
emitted
by
crossplay-
and
this
has
been-
I
guess,
a
little
bit
more
controversial
for
for
a
lot
of
folks,
so
well,
the
the
idea
there
is
that
you
could
do
a
coupe
cuddle
describe,
let's
say
my
rds
instance
or
something,
and
if
it
was
needing
an
update,
it
would
actually
include
a
diff
in
its
events
that
say,
here's
why
it
needs
an
update.
You
know
this
field
changed.
B
Some
of
the
concerns
that
we
have
around
this,
mostly
from
the
maintainer
team,
are
just
that.
We
already
have
a
hunch
that
crossblade
is
over
using
events
going
back
to
the
scalability
thing.
I
think
a
lot
of
the
reasons
we
do
see.
Slow
downs
in
high
scale
clusters
is
because
often
on
each
reconcile
every
30
seconds
each
resource
will
emit
like
three
four
events
or
something
like
that.
B
So
if
you
include
a
diff
in
the
event
and
it's
a
large
diff,
the
size
of
those
events
which
are
just
kubernetes
objects,
written
to
the
api
server
gets
a
lot
bigger
and
it's
quite
noisy
in
the
updates
as
well.
There's
also
a
lot
of
logic
to
de-duplicate
events.
So
if
the
event
sort
of
changes
a
bit
in
the
diff
from
time
to
time,
then
then
it'll
have
less
deduplication
of
the
output
there.
B
So
my
personal
opinion
on
this
is
that
we
probably
the
sweet
spot,
is
having
this
enabled
debug
logs,
but
not
having
it
enabled
in
the
actual
events
that
are
output
there.
But
the
main
reason
I
bring
this
up
is
just
to
bring
it
to
folks
attention.
I
would
love
it
if
people
could
open
this
issue
up,
have
a
read
through
and
just
sort
of
weigh
in
on
what
you
think
here.
B
I
think
one
of
the
big
themes
that's
come
out
so
far
is
just
that
that
I've
read
there's
a
lot
of
folks
waiting
in
like
people
who
have
are
either
like
operators
of
cross
plain,
like
are
the
platform
team
deployed
cross
who
have
built,
managed
resources
and
it
to
my
mind,
you
know
that
those
those.
B
Where
they
do,
they
get
cosplayed
to
expect
them
to
go
like
look
at
the
logs,
whereas
events
are
usually
targeted
more
to
end
users
of
cosplay,
where
I
think
this
might
be
a
little
bit
overkill
amount
of
information
for
them.
A
If
there
are
any
other
opinions
on
the
on
events
in
debugging
and
verbosity
there,
then
we
can
move
on
to
the
next
pull
request,
for
this
is
a
pr
is
the
issue.
I
think
it's
an
issue.
B
A
No
yeah-
this
is
this
is
fine
right
here,
because
you've
been
making
some
progress
on
this
right.
B
Not
a
lot.
Apologies
I'm
slightly
distracted.
This
is
the
time
of
day
when
my
puppy
just
goes
mad
and
like
barks
and
jumps
around
and
whatnot
she
sort
of
detects
when
I'm
having
the
most
meetings
of
my
day
and
gets
the
most
fussy.
B
This
is
an
issue
that
someone
from
sap
brought
up
where
he
has
discovered
the
crossplane
at
scale
for
certain
classes
of
resources,
which
mostly
is
actually
aws.
Networking
related
resources
by
can
be
by
coincidence,
at
the
moment,
crossbank
can
actually
leak.
These
managed
resources.
B
I've
been
doing
a
lot
of
back
and
forth
debugging
with
I
don't
actually
know.
The
person's
name
d
is
what
he
goes
by
from
from
sap,
and
they
had
noticed
they're
spinning
up
clusters
up
and
down
a
lot.
They
use
aws,
vpcs
and
one
of
the
resources
they
use.
There
are
route
tables
at
aws
and
they
noticed
they
couldn't
delete
their
vpcs
because
there
were
unexpected
route
tables
created
in
those
vpcs.
B
We've
triaged
that
and
figured
out
that
it's
most
likely
the
crossblade
is
actually
sometimes
very
rarely.
When
you
ask
for
one
route
table,
you
actually
get
two,
you
get
a
duplicate,
so
part
of
the
problem
that
the
the
theory
as
to
why
that's
happening
is
certain
resources.
It's
kind
of
an
edge
case,
but
again,
especially
at
aws,
have
what
we'd
say
is
a
non-deterministic,
unique
identifier
right
so
without
going
into
too
much
detail
about
how
the
cross
plane
like
reconcile
logic
works,
it
is
much
better
for
us.
B
If
we
could
tell
the
cloud
what
the
unique
identifier
should
be
for
a
resource.
So
it's
much
better
for
us.
If
we
could
say,
look
create
me,
a
creamier
database
called
bob,
and
then
we
can
query
on
the
next
thing.
Is
there
a
database
called
bob
and
then
we
can
figure
out
whether
whether
that
database
actually
exists
or
not?
B
B
When
we
observe
the
world,
we
basically
diff
what's
stored
in
the
api
server
for
the
real
world,
and
we
when
we
create,
update
or
delete
we
write
to
the
api
server,
the
kubernetes
api
server.
Unfortunately,
if
we
happen
to
have
a
stale
view
of
a
resource,
we
write
to
the
api
server.
That
right
is
rejected.
We
return
and
we
come
back
so
we
basically
by
and
large,
can't
really
we
have
to.
B
E
B
B
Server
and
then
because
of
optimistic
concurrency
control,
the
api
server
will
say
no
by
reject
you're
right.
You
can't
write
this
we'll
error
out
and
we'll
return,
and
we
will
like
not
have
persisted
that
id
to
the
api
server.
What
we
realize
is
that
also
potentially,
when
we
get
an
event
saying
hey,
go
reconcile
this.
B
Sometimes
we
actually
have
a
stale
view
from
our
cache
of
the
resource
that
we're
supposed
to
reconcile
so
we've
already
added
code
to
try
really
hard
to
persist
to
these
external
resources,
rather
than
just
giving
up
after
one
error
and
moving
on
we'll
actually
sit
there
in
a
loop
for
a
couple
of
seconds
I'll.
Try
it
really
hard
to
persist
to
the
database
before
we
return,
but
what
we
do
is
we
persist
it
to
the
api
server
as
an
annotation,
and
the
presence
of
that
annotation
dictates
whether
we
should
create
a
resource
or
not.
B
The
problem
that
we're
seeing
here
is
sometimes
you'll
see
two
reconciles
coming
at
around
the
same
time
and
they
both
will
appear
to
not
have
that
annotation
on
them.
That
says,
we
have
created
this.
That's
our
gate
that
says:
we've
created
this
so
we'll
just
create
the
thing
twice
and
then
we'll
record
it's
one
friend.
B
Whichever
gets
in
first
will
record
its
id,
the
other
one
will
lose
and
be
gone
and
we'll
basically
have
created
an
extra
resource
without
knowing
it
so
so
far
I
have
just
been
going
through
d
gave
like
a
lot
of
debugging
information
that
the
to
the
way
through,
like
a
lot
a
lot
a
lot.
So
so
I
tried
to
recreate
this
myself.
I
believe
I
have.
B
I
actually
ran
a
script
that
just
spun
up
and
down
vpc's
route
tables
all
night
last
night
and
two
route
tables
leaked
after
about
12
hours
of
constantly
recreating
stuff
every
every
two
minutes,
so
it
doesn't
happen
very
much,
but
it
does
happen,
and
I
have
an
idea
that
is
proposed
in
that
issue
as
to
as
to
how
to
address
it.
B
Next
steps
after
after
you
know
verifying
that
I
have
been
able
to
recreate
this
will
be
taking
that
idea
and
formalizing,
probably
a
little
bit
of
a
like
a
really
small
one
page
or
something.
But
the
very
short
version
is
that
we
can
tell
if
we
have
the
the
reliable
way
to
tell
if
you
have
a
stale
version
of
a
resource
coming
into
the
reconcile
is
to
try
and
write
that
back
to
the
api
server
and
then
the
api
server
will
reject
it
and
say
sorry.
B
This
is
a
stale
resource,
so
I'm
considering
adding
basically
an
annotation
before
we
do
the
create
that
says,
we
need
to
go
like
I'm
about
to
do
a
create
effectively
and
then
that
would
basically
mean
that
we'll
change
the
failure
mode
from
one
resource
is
created.
Another
one,
silently
leaks
that
you
just
don't
know
about
it
to
either
one
resource
is
created
or
one
resource
cannot
be
created
and
that
you've
got
to
go.
Do
some
manual
cleanup.
C
B
My
I
I'm
not
sure
if
this
is
exactly
what
you're
asking
my
understanding
is
that
the
the
problem
is
most
likely
not
that
we
are
seeing
multiple
reconciles
at
the
same
time
in
so
much
of
a
race
situation.
So
more
as
there
is
maybe
the
best
way
to
think
about
it
is.
There
might
be
like
a
race
between
the
stat
of
the
state
of
the
cache
which
is
hydrated
by
a
watch
on
the
api
server,
and
the
events
somehow
are
potentially
coming
in.
B
B
The
the
the
folks
at
sap
are
really
eager
to
to
get
this
fixed
asap.
I've
been
pushing
back
a
little
bit
saying.
We
really
need
to
understand
this
a
little
bit
better.
So
d
has
done
a
lot
of
debugging,
but
he
comes
from
a
cmx
boston
java
world.
So
like
his
sort
of
understanding
of
the
go
code
is,
is
maybe
not
as
great
as
those
who
work
with
it
every
day.
B
So
we
do
need
to
come
in
and
I
think
you
know
get
a
slightly
better
understanding
of
what's
happening
before
we
actually
go
and
do
anything
here.
But
my
my
hunch
is
that,
like
the
best
way
to
guarantee
that
this
won't
happen
is
to
use
basically
a
marker
before
we
do
the
create
so
that
we
can
make
sure
we
don't
create
twice.
C
Yeah
and
also
like
you
know
so
so
I
see
the
this
leak
test
composition.
You've
got,
you
said,
like
you
know,
you've
seen
two
leakages
in
12
hours.
Right
yeah.
Does
he
see,
like
you
know,
a
similar
frequency
or
like
more
than
that.
B
My
understanding
is,
he
posted
an
update
last
night
and
he
said
that
I'm
guessing
he
said
yeah
around
around
the
same.
Well,
I
can't
say
exactly
the
same
frequency,
but
he
does
say
it
happens
very
rarely.
It's
not
like
a
you
know,
one
in
ten
times
or
something
like
that.
It's
more
like
a
one
and
a
hundred
times
or
one
in
two
hundred
times.
E
C
Yes,
yes,
I
was
surprised
that,
like
you
know,
there
is
the
there's
like
no
possibility
that
we
we
would
get
like
non-up-to-date
version
of
the
cr
and
possibly
like
you
know,
true,
go
routines
working
on
the
same
cr
like
these
are
kind
of
like
some
of
the
foundational
things
that
we
assume
from
controller
on
time.
Yeah.
B
At
the
same
time,
my
hunch
is
that
it
is
we're
getting
them
sort
of
we're
getting
maybe
the
updates
in
order,
but
then
we
have
to
like
the
way
that
the
way
that
the
code
works
is
basically,
as
you
know
above
knows
you
just
get
told
hey
this
thing
changed,
then
you
have
to
go
fetch
it
from
the
api
server
and
that
fetch
the
apso
is
backed
by
a
cache
and
that
cache
is
hydrated
by
like
a
list
watch
on
the
api
server.
B
I
I'm
curious
to
see
whether
the
events
are
coming
from
that
cache
or
coming
from
something
else.
The
the
machinery
in
the
code
down
there
is
actually
quite
hard
to
follow.
To
be
honest,
it's
there's
a
lot
of
layers
of
abstraction,
so
I
need
to
do
my
traditional
once
a
quarter
deep
dive
into
the
controller
on
time,
caching
and
eventing
code
to.
C
C
Yeah,
just
just
just
a
small
detail
regarded,
I
remember
we
fixed
root,
specific
root
table
bug
regarding,
like
you
know,
status
updates
that
it
is
doing,
and
it
was.
It
turned
out
that,
like
composition,
controller
updates
were
actually
causing
the
resource
to
be
stale
in
the
managed
reconciler.
So
that
could
be
an
avenue
that
you
can
like.
You
know
investigate
like.
Maybe
that
controller
invalidates
the
cache
and
manager
that
doesn't
know
that
hey
a
new
change
is
there.
B
Yeah,
I
didn't
quite
follow
that,
but
let's
let's
maybe
take
this
one
offline
and
going
to
chat
about
later.
A
Yeah
definitely
a
worthwhile
investigation
to
continue
getting
to
its
end,
so
thanks
for
the
time
spent
on
that
we're
pretty
much
almost
out
of
time
here.
So
any
of
these
issues
here
can
folks
that
wanted
to
bring
those
up.
Can
we
or
any
of
those
that
need
to
be
brought
up
right
now
or
can
we
take
them
to
an
async
format
like
you
know,
either
github
or
slack,
or
postpone
them
to
next
community
meeting
any
any
of
those
available
for
folks
that
added
issues
here.
D
D
The
second
one
I
am
going
to
try
and
fix
this
for
1.4,
and
basically
just
I
have
the
proposal
for
how
this
would
work
on
the
the
issue
and
there's
been
some
back
and
forth
on
it.
So
would
love
folks,
look
at
that
nick.
I
know
you
opened
the
two
at
the
bottom
and
I
kind
of
weighed
in
on
those
I
think
locke
were
probably
pretty
good
aligned
on
promoting
that
one
to
v1
beta
1
for
1.4.
D
I
have
concerns
about
controller
config
that
really
are
difficult
to
back
up
because
it
is
being
used
successfully
by
a
lot
of
people,
but
I
just
have
concerns
about
this.
Like
catch-all
resource
sort
of,
I
guess
so
I
don't
know
exactly
what
you
wanted
to
cover
for
those
nick,
but
I'm
fine
to
take
those
offline
too.
B
B
I
imagine
a
few
people
have
opinions
about
lock
but
control
the
config.
I
I
would
like
to
get
community
input
on
how
they
feel
about
the
api.
What
they'd
like
to
see
changed?
If
anything,
I
think
the
the
broader
thing
here
is
that
I
we
have
these
two
resources,
which
are
the
only
two
core
resources
at
the
moment
that
are
v1
alpha
one.
As
I
mentioned,
I
would
like
to
have
alpha
resources
be
off
by
default
in
future.
B
I
don't
think
we
really
want
to
put
these
behind
a
feature
flag
and
turn
them
off,
given
that
people
are
using
them
now,
so
the
other
alternative
is
to
promote
them
to
be
what
they
want.
So
that
was
kind
of
my
motivation
here,
but
then
the
other
thing
in
reality
is
we
don't
have
a
policy
about.
You
know
how
long
does
something
get
to
stay
in
v1
alpha
one?
B
How
do
we
make
a
decision,
so
I
do
feel
a
little
uncomfortable
with
the
idea
that
we
have
a
resource
that
we
kind
of
don't
love
that
we're
not
sure
about,
but
the
people
are
using
and
we're
just
like
not
promoting
it
to
beta,
so
that
I
I
want
us
to
just
start
spinning
the
wheels
of
being
okay.
What's
the
plan
to
get
these
things
debated
like
you
know,
and
we.
B
To
like
an
option
would
be
to
just
drop
them
or
you
know
you
know
their
alpha,
there's
no
there's
no
contract.
We
can
get
rid
of
them.
We
can
reimplement
them
a
different
way,
not
an
urgent
priority
for
v
1.4.
I
just
kind
of
wanted
to
track
that
we
don't
want
things
to
just
sit
in
alpha
indefinitely
and
never
get
promoted.
So
we
should
have
a
tracking
thing
to
be
like
how
do
we?
How
do
we
get
the
debater
at
this
stage?
B
A
I
wonder
if
we
should
follow
up
to
get
like
solicit
more
feedback
from
the
broader
community,
like
put
it
on
slack
like
tweet
about
it
or
something
as
well
too,
to
kind
of
get
that
get
that
out
to
a
broader
audience
for
more
feedback
and
how
their
house
being
used
or
how
they
want
to
see.
It
continue
to
mature.
B
Yeah
definitely,
my
hunch
is
that
you
know
for
the
people
who
use
controller
config,
they'll,
probably
like
great.
Yes,
it's
valuable.
We
need
this
like
you
know,
so
I
don't
imagine
everyone's
gonna
come
back
and
be
like.
Oh
no,
I
hate
this
functionality
or
whatever
the
question
is
more
architecturally.
You
know.
B
B
So
so
you
know
maybe
there's
a
better
way
to
frame
those
settings
that
we
just
haven't
landed
on
yet.
D
D
A
Okay
cool,
so
let's
continue
following
up
on
that,
then
all
right!
So
that's
everything
on
the
community
meeting
here,
sorry
for
going
over
a
little
bit
here,
but
definitely
appreciate
everyone's
participation
and
attendance,
and
so
we
will
see
you
all
on
slack
in
github
in
two
weeks.
The
next
weekly
meeting
thanks
everybody.