►
From YouTube: 2023-03-09 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
March
9th
2023
cross
plain
community
meeting.
I
will
go
ahead
and
drop
a
link
to
the
agenda
Doc
in
the
zoom
chat
now
for
anybody
who's
still
joining
right
now,
I
might
not
have
seen
a
link
there
yet
so
there
is
the
link
in
the
zoom
chat
to
the
agenda
Doc
and
feel
free
to
come
in
there
and
add
any
topics
or
suggestions
for
things
that
you
all
want
to
talk
about
as
well.
A
The
agenda
Doc
is
open
for
suggestions
all
right,
so
I
think
the
first
thing
to
talk
about
is
recent
releases
and
patches.
So
let's
start
it
chronologically
here,
because
since
the
last
community
meeting
we
did
a
1.11.1
patch
release.
First
Bob
Hamilton
ran
that
one
so
Bob
do
you
want
to
talk
through
the
the
major
stuff
in
there
real
quick
since
you
ran
that
release.
B
I
appreciate
that
let
me
make
the
visual
aid
bigger
so
that
my
eyes
can
actually
read
it.
There
we
go
yeah
I
mean
the
the
big
things.
The
the
main
reason
for
the
release
was
the
composition
function.
It's
actually
the
bottom
entry
on
that
list,
the
API
patching,
applicator
composition
functions
when
you
enable
composition,
functions,
the
alpha
feature
it
and
used
it
with
other.
B
You
know
existing
compositions
that
was
causing
some
issues
with
generating
additional
resources
that
it
shouldn't
have
been
generating,
and
so
this
was
a
a
patch
for
that
which
was
the
main
reason
for
the
release,
and
then
we
also
pulled
in
performance
issue
fix
for
applying
the
poll
interval
to
all
of
the
claim
and
composite
reconciliers
for
cross
planes,
so
that,
if
you
told
cross
plane
to
reconcile
on
a
five
minute
interval,
it
would
actually
do
that
instead
of
defaulting
to
the
one
minute
interval
on
the
claim
and
Composites,
and
then
there
were
some
updates
to
the
output
of
the
composition
and
composition,
revision
columns
to
make
that
a
little
bit
more
useful
and
then
a
bunch
of
the
punch
and
dependency
object
updates.
A
Yeah
and
fun
fact,
if
I
recall
correctly,
for
this
patch
release,
is
that
Bob
was
the
contributor
for
all
three
of
the
main
fixes
for
the
release,
which
kind
of
put
him
in
the
default
seat
for
running
the
release
as
well.
So
Bob
thanks
again
for
your
contributions
in
the
code
in
the
fixes,
as
well
as
running
the
release
too.
So
thanks
a
lot
man
that.
A
All
right,
so
that
was
after
the
last
community
meeting,
so
it
was
since
we
all
since
we
all
met
last
now
recently.
Today
we
have
a
series
of
patch
releases
that
are
going
out
in
response
to
some
security
advisories
that
were
found
during
our
fuzz
testing
efforts
with
the
cncf
and
Ada
Logics.
C
A
Philippe
has
been
running,
those
I
think
he
was
fixing
the
the
issues
and
running
the
patches
as
well.
So
Philippe
go
ahead
and
get
us
updated
on
these
and
which
link
is
going
to
be
the
most
useful
for
you
for
the
visit,
the
release,
tag
itself
or
the
advisories.
D
Okay
yeah,
so
there
we
go.
Vegeta
was
in
actually
in
the
in
the
runtime
package
in
the
field
buff
package
from
from
because
playing
around
them.
Like
a
repository,
we
decided
to
it's.
It's
actually
not
a
big
issue
for
the
library.
It's
just
the
library
is
just
doing
its
job
book.
D
We
decided
to
release
a
cve,
because
it's
actually,
if
a
user
of
this
repo
of
this
package,
as
is
a
cross
plane,
is
using
it
without
validating
inputs,
cool,
it's
called
the
lead
to
one
out
of
memory
and
and
therefore
a
denial
of
service
on
their
on
their
services.
So
we
we
decided
to
release
an
advisory
for
both
cross
parenting,
time
and
crosswind
itself
and
then
from
there
to
propagate
it,
to
to
cross
plane
and
and
all
things
The
Advisory
on
Cross
print
has
a
slightly
higher
severity.
D
And
supercritical
issue,
because
it's
actually
to
be
exploited,
it's
still,
it
required
the
user
to
already
have
a
lot
of
permissions,
so
it
needed
to
be
able
to
update
and
and
or
create
compositions
and
change
the
the
field
buffing
in
both.
So
that
was
already
a
highly
privileged
users
of
severity
is
not
so
so
high,
but
still
we
kind
of
took
this
as
a
chance
to
to
run
through
the
newly
deployed
security
policy
and
procedures
and
and
all
things.
D
So
we
kind
of
used
use
this
as
a
drill
to
to
test
it
out
and
smooth
any
any
detail
in
that,
and-
and
so
yes,
we
released
the
one
11.2
and
19.2
and
1.10
is,
is
on
the
way
we
had
some
minor
issues
with
the
cross-plane
cosplay
runtime
dependency
in
that
version,
which
was
pointing
to
a
different
release.
Candidate
Etc
et
cetera,
so
had
some
minor
hiccups,
but
probably
will
come
tomorrow.
A
Awesome
Philips
effects
for
updating
us
on
that,
thanks
for
making
those
fixes
and
getting
the
patch
releases
out
for
them
and
then
also
I,
don't
know
if
the
adelogix
folks
are
on
here,
but
the
fuzz
testing
effort
that
they,
you
know,
went
through
and
helped
kind
of
build
out
for
us
and
automate
was
really
really
helpful
to
find
issues
like
this
and
then
kind
of
Drive
our
security
processes
as
well.
A
So
getting
this
in
place.
So
now
we
have,
you
know
a
defined
policy
and
disclosure
process
a
vulnerability
advisory
process.
So
all
that
is
in
place
now
and
then,
as
Philippe
was
saying
using
these,
these
vulnerabilities
found
in
the
fuzz
testing.
As
you
know,
our
first
kind
of
pilot
runs
for
the
advisory
security
advisory
disclosure
process
is
really
nice
to
exercise
that
get
things
in
place,
figure
out
where
the
gaps
are
and
fill
them.
So
I
think
we're
feeling
much
much
better
about
it.
A
Like
Felipe
has
identified
a
couple
things
that
we
might
want
to
streamline
to
make
this
process
a
little
bit
better.
But
you
know:
we've
got
the
process
in
place
and
we're
using
it
now
and
feeling
good
about
that.
So
then,
so
they,
you
were
saying
that
the
last
release
that
we
expect
here,
because
we've
got
some
of
the
the
runtime
releases
out,
I,
believe
and
then
so
we
have.
The
last
patch
release
will
be
attached
1.10
as
well
right.
A
Challenges
with
the
crosswind
runtime
dependency
management
there
right,
so
that
the
last
release
we
expect
to
call
these
completed.
Yes,
awesome,
fantastic
yeah,.
E
A
Driving
all
that
food
and
Hassan
as
well
I
know
you've
been
helping
a
lot
too
man
yeah
awesome.
Thank
you
very
much
all
right.
So
those
are
the
recent
packs
releases.
So
now,
let's
turn
our
eyes
back
to
1.12
and
which
is
the
upcoming
release.
We
are
more
than
a
month
out,
I'd,
say
probably
like
six
weeks
out
still
for
the
scheduled
release
date
of
April
25th,
there's
a
bunch
of
exciting
progress
going
on
on
the
release,
support
there.
A
So
we're
not
going
to
focus
on
every
issue
here,
but
let's
get
a
couple
of
quick
updates
to
some
of
the
stuff.
That's
in
progress,
so
Hassan
I
would
like
you
to
start
off
with
the
exciting
update
about
observe,
only
resources
and
if
it's
possible
for
other
folks
to
start
testing
it
as
well.
F
Yeah
sure
so,
basically
we
put
together
a
couple
of
PR's
and
basically
we
have
two
PRS
in
runtime
and
cross-plane
tools,
and
then
we
have
started
implementing
the
feature
in
objects
so
that
we
can
roll
the
feature
out
to
the
like
generated
providers,
one
at
a
time.
So
currently
the
status
is
the
PRS
are
rating
for
review
and
in
the
meantime,
we
have
prepared
a
preview
image
for
upon
slash
provider
AWS
where
you
can
install
and
start
using
the
feature.
F
So
it
is
already
shared
in
the
observe,
only
special
interest
group
channel
on
the
crossplay
slack.
It
also
has
a
procedure
on
how
to
install
the
provided
image
and
how
to
test
the
feature
out
so
yeah.
Basically,
currently,
we
are
kind
of
put
a
small
pose
on
the
picture,
development
and
and
waiting
for
some
reviews
and
feedbacks
for
the
pr
and
the
the
provided
image.
A
A
Got
together
and
tried
that
out
last
week
and
it
was
super
exciting
to
see
the
feature
live
and
in
in
action
and
seeing
you
know
all
the
status
we
get
populated
with
with.
What's
on
the
you
know,
server
side
and
all
that
sort
of
stuff.
So.
A
Cool
to
see
that
thanks
for
making
that
available
all
right-
let's
see
so
then
another
couple
of
things
of
interest
here-
we're
gonna
dive
into
the
ignore
changes
proposal
from
lovro
later
on
today,
as
a
agenda
item,
so
we're
gonna
skip
that
Philippe
and
Esky.
Do
you
want
to
update
us
real,
quick
on
the
compositions
validation,
work,
I,
guess
first
and
then
we'll
go
to
Esky
for
the
plugable
secret
stores,
work.
D
Yeah
sure
so
there
is
a
an
open
PR
with
proposal.
Design
document
for
composition,
validation,
I,
I,
implemented
a
few
I
added.
A
few
recent
changes.
It's
yeah
number
I
added
a
few
recent
changes
after
trying
to
implement
it
in
a
small
POC
to
validate
the
approach.
So
it's
it's
waiting
for
feedback,
the
design
document
and,
in
the
meantime,
we
are
already
kind
of
working
on
a
less
POC
implementation
has
all
the
best
and
resource
describing
the
document
but
yeah.
D
The
document
is
there
so
feel
free
to
to
share
any
feedback.
A
Awesome,
that's
a
great
update
as
well
Philippe,
okay
and
then
Esky
is
Esky
on
on
the
call
as
well.
Yeah
awesome.
G
So
we
have
one
open
PR
on
Cross
plain
repository,
which
updates
the
existing
applicable
Secret
store,
one
pager.
What
what
it
does
is
actually
changing
the
deployment
model
so
instead
of
running
them
as
a
sidecar
next
to
cross
plain
and
providers,
I
propose
to
run
them
as
a
separate
deployments
and
enable
Mutual
TLS
for
Authentication
between
cross
plane
and
grpc
servers.
G
So
there's
that
PR
going
on
I
think
I
have
addressed
most
of
the
issues,
so
it's
ready
to
approve
CeCe
has
on
and
the
second
prism
on
cross-plane
runtime
Repository,
so
I
have
basically
added
to
Proto
called
and
added
some
extension
point
for
running
grpc
servers.
I
think
it
also
got
a
couple
of
reviews.
I
have
addressed
most
of
them.
There
are
some
open
issues,
I'd
like
to
discuss
with
reviewers,
but
yeah
I
think
it's
also
ongoing.
Hopefully
we
can
merge
it.
A
Awesome
thank
you
for
that
update
on
that
as
well,
so
yeah.
So
those
are
some
of
the
major
work
in
progress
that
I
know
of
Jesse.
Did
you
want
to
share
an
update
as
well
about
signature
validation,
work
that
you
had
been.
E
Doing
sure
yeah
yeah,
no
I've
I,
basically
restructured
everything
that
fit
behind
an
alpha
feature
flag,
whereas
before
I
was
naively
putting
it
into
the
a
new
type
in
V1
Alpha.
So
I
believe
that
I'm
very
close
to
having
it
ready
for
review
again
and
I'll
I'll
reach
out
I'm,
just
I'm
in
the
process
of
refactoring
kind
of
how
the
parameters
are
passed.
I
put
a
message
into
the
slack
General
and
kind
of
been
figuring
it
out
myself.
E
A
Awesome
Jesse
thanks
for
that
update
thanks
for
your
persistence
on
that
as
well.
I
would
not
have
guessed
that
you
were
new
to
go.
So
that's
news
to
me:
no.
E
A
Sweet,
okay,
that's
great,
that's
great
news
and
then
I
also
wanted
to
call
out
I,
don't
think
it's
part
of
the
1.12
board
specifically,
but-
and
let
me
see
if
I
can
find
it
real,
quick
on
another
screen,
and
here
it
is
so,
let's
open
it
over
here.
So
I
do
want
to
call
out,
though
the
work
that
is
going
on
to
continue
maturing
composition
functions
right,
so
that
was
released
in
Alpha
in
the
1.11
release.
A
It
is
a
quite
important
feature
and
it
has
gotten
a
lot
of
early
adopters
on
it
and
a
lot
of
feedback
as
well.
So
you
know,
as
Bob
mentioned,
we
already
did
a
1.11.1
patch
release
to
fix
some
functionality.
That
was
that
was
that
was
causing
challenges
with
adopting
it
and
using
it
successfully.
So
we
are.
We
are
keeping
a
list
here
of
all
the
things
that
we
want
to
do
to
mature
composition,
functions
and
improve
the
experience
and
reliability
and
everything
around
there.
A
So
if
folks,
you
know,
as
you're
using
composition
functions,
please
do
you
know
chime
in
on
this
list
on
this
issue.
Here
with
you
know
your
experience
or
things
you
want
to
see
improved
or
whatever,
so
we
can
have
it
all
in
one
place.
So
we
are,
you
know,
tracking
and
continuing
to
you
know,
fill
in
all
the
gaps
and
improve
and
mature
a
composition
function
successfully.
A
Of
course,
I
want
to
call
that
out
and
then
Hassan
I
also
know
that
this
issue
here
about
back
off,
is
not
working.
I
know
that
there
had
been
some
further
talking
and
discussion
and
work
on
it.
So
if
there's
and
I
know
that
that's
something
that
was
highly
demanded
as
well,
is
that
something
you
can
give
us
another?
A
quick
update
on
for
today,
too
man
yeah.
F
Yeah
sure
yeah,
basically,
we
are
having
some
progress
on
the
on
this
PR
in
controller
runtime.
We
kind
of
you
know
agreed
that
this
is
a
viable
solution,
but
Nick
had
some
concerns
around
like
like
there
could
be
some
implicit
assumptions
that
making
a
such
update
actually
causes
a
Reconciliation
of
Riku.
So
basically,
what
we
did,
what
I
did
was
was
like.
I
I
have
conducted
a
set
of
experiments
to
make
sure
that
it
is
not
changing
or
breaking
any
functionality.
So
these
are
the
like.
F
The
different
types
of
experiments
that
I
have
conducted
and
also
I
made
some
like
kind
of
static
code
analysis
in
the
controller
runtimes
managed
to
consile
it
to
make
sure
that
we
don't
have
such
an
implicit
assumption.
So
so
far,
it
looks
like
this
solution
is
fixing
one
of
the
problems
in
the
original
issue.
Basically
the
the
original
problem,
and
hopefully
we
can
get
some
progress
in
this
PR
and
start
consuming
it
from
the
providers
on
the
issue
while
like
I
was
I,
I
was
investigating
the
problem.
F
Actually
I
found
another
case
where
backup
is
not
working,
which
is
the
the
external
creation
is
failing
for
some
reason,
the
the
reason
was
I
was
testing.
Was
the
the
permission
issue
so
to
to
keep
things
clear,
I
moved
that
part
into
a
separate
issue,
and
hopefully
that
there
are
like
couple
of
solutions,
and
hopefully
we
can
get
to
an
agreement
on
on
a
possible
solution
back.
F
A
F
A
C
A
Do
you
think
that
it
is
feasible
then,
to
if
you
have
a
proposal
for
this
particular
scenario
or
case
as
well?
Do
you
think
it's
feasible
to
include
this,
at
least
for
now
in
the
1.12
Milestone,
to
keep
getting
some
effort
on
it
and
attention
on
it.
F
Yeah
yeah,
that
makes
sense
basically
yeah
I
need
some.
Some
input
from
from
I
can
make
here.
A
Awesome
all
right
so
I
will
then
put
this
on
the
I'm
doing
this
on
another
screen,
I'm
putting
that
in
the
1.12
Milestone
as
well.
Just
so
we
keep
continue
tracking
it
too,
okay,
so
that
maybe
should
show
up
there.
It
is
shows
up
there
in
a
may
as
well
assign
you
to
it
as
well.
That's
on
no
good
deed
goes
unpunished.
A
Okay,
sweet
sweet!
All
right,
so
I
expect
to
see
your
face
show
up
there
sometime
soon,
but
all
right
that
is
I,
think
everything
that's
going
on
in
1.12
that
I'm
aware
of
right
now,
oh
sorry!
Well
we
have
an
update
to
get
into
for
for
performance
related
things
as
well.
So
we
will
talk
about
that
specifically
too
I.
Think,
though,
that
that
is,
that
captures
everything
going
on
right
now
in
the
within
the
community,
in
the
ecosystem
for
core
cosplay
and
stuff.
A
So
thank
you,
everybody
for
for
updating
us
with
all
that
activity.
That's
going
on
there's!
Obviously
a
lot
of
things
happening
right
now,
so
this
is
very,
very
exciting.
Is
there
anything
that
folks
wanted
to
call
out
for
as
a
priority?
Or
you
know
a
hot
issue
that
folks
are
seeing
that
is
not
captured
already
within
the
release
board
and
such.
A
Good
good,
so
let's
go
ahead
and
move
on
past
1.12
and
let's
go
into
provider
releases
and
updates
there.
So
I
think
that's
okay,
good
folks
did
update
this
with
some
other
recent
releases.
I
think
there's
a
couple
of
releases
for
the
community
AWS
provider
and
then
all
the
up
unofficial
providers
were
all
updated
today,
I
believe
so
those
are
links
to
the
latest
releases.
There,
provider,
kubernetes
and
provider
Helm.
We're
also
updated
I.
Think
maybe
last
week
too.
A
So
there's
links
to
all
the
release,
notes
there,
where
folks
can
catch
up
with
everything
that's
going
on,
but
we'll
keep
we'll
keep
moving
through
the
agenda,
though,
specifically
here
all
right,
so
performance
concerns
in
the
upbound
official
providers,
the
up,
Jet
and
I'd
say
up
jet
related
writers
in
general.
There
are
definitely
updates
to
share
here.
I,
don't
want
to
steal
the
spotlight
from
anybody
specifically,
so
John
are
you
in
a
position
to
with
your
audio
and
stuff
okay,
cool
I'll?
Let
you
do
it.
H
All
right,
can
you
hear
me
fine,
yeah,
you
sound
great
awesome,
all
right,
so
just
an
update
here.
We've
made
some
some
good
progress
in
terms
of
first
of
all,
we
we
had
a
bit
of
a
holistic
view
of
the
problem
and
and
set
out
some
phases.
The
first
phase
is
all
about.
H
H
H
If
you
can
go
back
to
the
to
the
doc,
so
I
can
reference
the
notes
we,
oh
yes,
we
also
did
a
series
of
load
testing
and
you
can
maybe
open
that
that
issue
there
and
essentially
you
know-
we've
been
trying
to
to
establish
a
baseline
of
where
we
were
with
the
0.30
releases
and
earlier
and
and
then
we,
we
kind
of
like
evaluated
again
the
the
approach
we
that
we
previously
disabled
for
for
memory,
leaks
and
and
duplicate
resources.
H
We
we
use
a
basically
a
a
grpc
server
to
communicate
with
the
terraform
provider
and
the
results.
Look
quite
promising.
H
We've
also
been
able
to
identify
the
Upstream
issues
where,
where
the
memory
leaks
and
and
duplication
happens,
and
so
after
many
hours
of
investigating,
Alper
and
surrogen
have
done
a
lot
of
work
in
this
have
feel
confident
that
they've
got
a
good
path
forward
to
re-enable
the
the
grpc
server
approach,
which
should
give
us
a
drastic
Improvement.
H
So
the
current
phase
that
we're
moving
into
is
is
fixing
those
issues
that
we
that
we
identified
with
this
and
and
then
the
next
release
of
the
of
the
providers
will
will
likely
have
it
turned
on
by
default
or
at
a
minimum
of
a
configuration
option
to
to
turn
it
on
and
then
we'll
we'll
evaluate
where
we
are.
We've
got
a
few
further
optimization
ideas.
H
Once
we
get
to
a
stable
point,
we,
where
things
are
back
to
kind
of
like
a
healthy
scenario.
You
know
we'll
also
release
a
a
bit
of
a
sizing
guide
to
give
folks
just
the
some
guidance
as
to
you
know,
if
you're
looking
to
to
kind
of
like
provision,
so
many
resources
Etc.
H
This
is
more
or
less
the
nodes,
types
and
sizes
that
you
need,
and
then
you
know,
together
with
the
the
kind
of
like
focus
on
on
some
of
the
cre
scaling
issues,
and
you
know
we
might
want
to
just
have
a
look
at
at
how
that
work.
Martin
also
further
have
a
positive
impact
on
on
the
resource
utilization
to
determine
whether
there's
any
further
work.
We
want
to
look
at
optimizing,
but
you
know
in
the
next
few
weeks,
we'll
we'll
be
releasing.
A
Awesome
John
thanks
for
that
update
there
and,
like
definitely
you
know,
albertans
are
going
to
make
great
progress
there
and
there's
been
a
lot
of
help
from
the
community
as
well
with
testing
things
and
you
know
recording
their
observations.
You
know
sharing
and
collaborating
there.
So
it's
been
some
great
work
from
from
a
number
of
people
on
this.
A
One
output
from
this
is
a
much
more
enhanced
series
of
metrics
around
provider
performance
so
that
we
have
those
going
on
into
the
future,
we'll
be
benefiting
from
having
those
available
as
well.
For.
H
Yeah
sorry
I
forgot
to
mention
you
part
of
the
reason
for
introducing
the
Matrix
was
also
to
make
sure
we
can
do
some
regression,
the
testing
and
put
that
in
place,
so
that
we
can
see
if
there's
any.
If
you
know
if
we
consume
a
new
version
of
the
Upstream
terraform
providers,
if
it
has
any
negative
effects,
Etc.
A
Right
yep:
that's
that's
all
all
the
better
to
make
sure
that
we
fix
this
and
then
we'll
stay
in
a
more
healthy
position
as
well.
All
right,
John
I,
believe
that,
with
the
next
section
of
updates
from
what
the
engineers
at
upbound
are,
working
on,
I
might
have
stolen
a
few
of
those.
While
we
looked
at
the
1.12.
A
A
lot
of
actually
years
are
working
on
1.12
stuff.
Anything
else
specific.
You
want
to
add,
though,
that
we
might
have
missed.
H
No
I
I
think
mostly
just
you
know,
on
the
external
secret
stores.
You
know
we're
close
to
wrapping
up
that
design
proposal.
So
if
anybody
has
any
last
minute
feedback,
please
have
a
look
as
well
as
the
composition,
validation,
webhook,
you
know,
so
the
only
kind
of
like
request
is
you
know.
If
there's
any
last
minute
feedback,
please
get
it
in
there,
so
we
can
consider
it
other
than
that
the
the
folks
have
already
covered
the
updates
and
loveru
will
talk
about
the
ignore
changes
on
update
later
perfect.
A
All
right
so
then,
we'll
keep
on
moving
on
here.
I
think
that
yeah,
so
so
in
terms
of
like
content
within
the
community
there's
a
number
of
cool
links
here
to
check
out
well
this
one
only
if
you
should
only
if
you
speak
and
read
German
but
there's
definitely
some
really
interesting
content
here.
I
def
I
wanted
to
call
out
specifically
that
the
AWS
team
is.
E
A
Is
Carlos
is,
though,
yeah
there
Carlos
do
you
wanna?
Do
you
want
to
speak
about
the
disaster
recovery
guide
that
your
team
was
working
on.
I
Yeah,
it's
a
it's
a
one,
additional
blog
post
of
a
series.
Another
one
is
planned
to
go
out,
but
this
one
took
a
long
time
to
get
everyone's
review,
so
it
took
it
took
a
long
time
to
get
it
review.
I
was
written,
so
this
helps
in
the
area
and
We
Gather
feedback
from
different
companies,
including
outbound
review.
This.
I
This
content
of
at
least
give
some
guidance
to
to
customers
on
in
terms
of
disaster
recovery,
and
then
what
are
the
levels
of
This
Disaster
Recovery
like
if
you
lose,
if
you
lose
the
the
cluster
or
if
you
lose
control
plane,
if
you're
doing
upgrades
and
you're
working
with
AWS
resources
that
are
sensitive
in
state
like
RDS
and
S3,
specifically,
how
do
you
recover
either
those
manage
just
basically
using
Valero
and
what
are
the
things
to
to
watch
out
for
and
what
are
the
things
that
you?
I
Maybe
you
need
to
mutate
as
you
restore
so
there's
different.
There's
three
scenarios
in
here
and
related
to
that
I
was
able
to
get
a
I
guess
a
win
for
the
community
was
able
to
get
a
cross-plane
talk
into
GitHub
scan.
Where
I'll
be
discussing
this,
this
blog
post
and
giving
a
talk
there.
A
Awesome
Carlos,
congratulations
on
on
that
talk
that
sounds
great
man
yeah,
and
this
this
content
is,
is
really
good
right.
This
is
this
is
some
thorough
stuff
we're
talking
about
a
number
of
different
scenarios
to
be
aware
of
so
thank
you
very
much
to
the
to
the
AWS
folks
for
driving
this,
and
thank
you
as
well
for
everybody
that
was
contributing
to
it
with
good
information
as
well.
I
think
most
most
of
those
people
are
on
this
call
so
yeah
all
the
authors
here,
like
really
good
work.
I
It
took
a
while
to
get
everyone's
companies
and
marketing
and
all
the
stuff
you
thought
the
technical
stuff
is,
is
hard
right.
I
Welcome
feedback
and
then
there's
a
if
you
happen
to
be
a
AWS
customer
using
cross
plane,
there's
a
link
in
the
blog
post
at
the
bottom.
If
you
want
to
schedule
some
time
with
my
with
my
team
that
we
also
do
engagements
and
we
bring
in
different
parts
of
the
community,
including
outbound
and
others,
so
just.
A
Reach
Out,
sweet,
awesome,
you're,
definitely
a
good
resource
here
and
then
there's
another
another
jump
in
really.
J
Quick
Carlos,
if
you
feel
so
inclined,
I,
think
that
would
be
a
great
thing
to
adapt
into
a
docs
knowledge
base
article
as
well,
and
then
link
back
to
the
original
AWS
post.
Just
so,
we
have
kind
of
two
points
of
discoverability.
So
if
you
find
it
through,
Google
you'll
probably
end
up
on
AWS
and
if
you're
crawling
through
Crosspoint
docs,
you
would
find
it
there
and
I
can
I
can
take
care
of
it
in
the
future.
J
I
J
And
I
think
I'll
just
say
without
totally
sidelining
the
thing
I
think
one
reason
for
having
the
knowledge
base
is
to
allow
us
to
do
something.
That's
a
little
bit
more
say,
vendor
specific
right.
We're
talking
about
a
specific
solution
with
AWS
I.
Think
that's
totally.
Okay.
J
Within
that
knowledge
base,
and
if
Google
wants
to
come
along
later,
do
the
same
thing
about
gcp
great
we
would
love
their
contribution
and
kind
of
separating
like
knowledge
base
might
be,
is
more
opinionated
content
and
the
user
docs
being
more
unopinionated
factual
kind
of
vendor,
neutral
content,
I
see
cool.
A
Awesome
yeah
good
suggestion
Pete
thanks
for
bringing
that
up
and
thanks
again
to
all
the
folks
contributing
to
this
really
good
content.
So
another
another
set
of
links
for
other
interesting
content
going
on
within
the
ecosystem.
A
Victor
had
a
cool
like
live
stream,
I
guess
recently
going
on
here.
For
you
know
the
the
you
choose
a
series
of
like
picking.
You
know
what
what
what
approach
the
the
they
want
the
host
to
take
in
doing
that
live
so
I
think
that's
pretty
interesting
and
here's
a
link
to
to
that
to
catch
up
with
that
too.
A
There
aren't
really
new
kubecon
updates
here
today,
but
these
resources
are
still
useful.
First,
you
know
checking
out
all
the
Crossbay
related
talks
and
then
the
Country
Fest
session
is
still
going
forward
and
I'm
pretty
excited
about
that
I
think
a
number
of
people
are
signed
up
for
that,
and
so
we
will
continue
to
need
help
with
you
know,
moderators
and
and
folks
that
can
help
with
the
live.
You
know
Hands-On
session
there
so.
A
Do
reach
out
to
me
to
be
able
to
if
you're,
if
you're
interested
in
supporting
that
session.
A
A
Actually
so
yeah
do
contribute
to
that
and
add
your
use
case
to
there
if
you're
interested
in
that
we
we
made
a
note,
then
about
you,
know:
composition,
functions
being
matured
to
Beta,
so
here's
just
a
link
to
that
to
be
able
to
easily
get
to
that
issue
and
then
add
your
thoughts
about
how
to
continue
maturing
and
improving
it
all
right.
A
So
with
that
I
wanted,
we
have
lovro
here
today
to
kind
of
walk
us
through
a
proposal
that
he's
making
about
how
ignoring
changes,
specific
scenarios
for
cross-plane
so
lovro.
If
you
want
to
go
ahead
and
share
your
screen
to
kind
of
talk
us
through
the
the
high
points
of
the
experience
that
you're
proposing,
then
it
is
all
yours,
my
friend.
K
Sure
sure
thank
you
by
the
way
hi
everybody,
my
name
is
lobra.
This
is
my
first
time
here,
but
I'm
very
happy
to
join
the
community.
A
K
K
So
this
idea
came
out
from
an
issue
that
Tim
opened
up
and
he
had
a
problem
where,
because
cross
plan
reconciliates
all
the
parameters
on
the
spec
for
provider
in
some
cases,
this
is
not
welcome,
because
some
of
the
parameters
under
it
are
being
also
managed
by
some
external
something
external
right
and
for
usually
it's
Auto
scaling
that,
for
example,
in
this
case
group
The
Design
desired
capacities
being
set
by
a
cross.
I
K
And
it
was
in
the
four
provider
spec,
but
because
it
was
Auto
scaling,
the
desired
capacity
was
being
changed
externally
by
the
autoscaler,
and
that
was
causing
like
ups
and
downs,
because
Crosman
was
trying
all
the
time
fighting
with
data
scaler
trying
to
set
its
value
for
desired
capacity,
and
initially
there
was
a
proposition
to
use
something
like
ignore
changes
in
your
field,
which
has
which,
where
you
can
put
just
like
a
field
Paths
of
the
things
you
would
like
to
ignore
during
updates.
K
So
it
was
an
interesting
idea
right
from
the
start
that
most
people
were
also
had
issues
with
Auto,
scalers
and
yeah,
and
one
more
suggestion
came
up
basically
actually
from
the
observe,
only
conversation
to
use
something
different
than
ignore
changes
to
use
the
something
called
init
provider.
It
basically
will
just
use
the
same
like
schema
like
before
provider,
but
it
would
be
used
only
for
create
okay.
So
this
is
just
the
intro.
K
So,
based
on
this
I
researched
a
bit
and
made
a
proposal,
I
gathered
up
the
examples
that
the
people
were
reporting
and
for
now,
I
had
only
four,
so
I'm,
really
hoping
that
we'll
get
some
more,
so
we
can
decide
which
exactly
direction
to
take.
Most
of
them
are
connected
with
auto
scaling,
but
there
is
also
one
that
is
connected
with
automatic
upgrades
for
node
pools
where
Crosman
later
fights
about
fights
for
the
version
of
the
nodes.
K
So
anyway,
it
seems
that
there
is
a
need
for
a
wave
in
Cross
plane
to
set
some
Fields
remark,
some
fields
or
somehow
that
they
we
don't
want
them
to
be
updated
like
at
all.
Let's
leave
their
updates
for
some
somebody
else,
and
we
can
just
check
the
states
and
like
copy
back
what
the
state
is,
but
don't
update
it
right.
K
Okay,
so,
first
solution.
The
solution
that
I'm
proposing
actually
is
the
one
with
the
ignore
changes.
So
in
this
solution
we
would
add
in
the
spec
and
ignore
changes
field
and
which
will
basically
be
a
sling
array
with
the
file
pads
of
all
the
things
that
we
want
to
ignore.
K
C
K
And
crossband
won't
set
its
desired
size
and
force
it
on
on
the
on
the
external
resource
and
from
from
one
side.
This
is
a
bit
confusing
because
you
still
have
this
desired
charged
here
as
well,
but
in
the
observe
only
design,
we
will
get
a
new
field
in
the
status
called
that
provider,
so
we
will
be
able
to
also
see
like
what's
the
real
real
estate
and
what's
the
real
desired
size.
I
know
it's
not
ideal,
but
maybe
somebody
has
a
better
idea
also
for
this
foreign
okay.
K
One
one
thing
that
comes
up
here
is
also
that:
how
can
we
ignore
the
fields
that
are
required?
Basically,
there's
required
on
updates,
like,
for
example,
you
won't
be
able
to
ignore
region
in
AWS
resources
right,
because,
if
you
don't
set
it
for
updates,
it's
that
won't
work
right.
K
So
there
are
some
fields
that
are
only
for
creates
and
there
are
some
fields
that
are
only
for
create
and
update,
and
for
us
it's
like
it's
not
clear
which
ones
are
which
so
here
we
just
suggest
that
better
not
to
to
be
basically
be
careful
when
putting
required
fields
in
you
know,
changes
and
do
it
only
when
you
have
like
a
sense
of
knowledge
and
how
the
calls
work.
Underneath.
K
Okay-
and
here
we
have
some
implementation
details,
I'll
just
go
fast
through
them,
because
I
think
the
more
interesting
is
the
other
Solutions.
So
for
this
implementation
it
ignore
changes
for
object.
Basically,
all
the
changes
would
be
in
the
providers,
so
the
providers
written
with
objects
for
them.
We
will
just
use
the
terraforms,
ignore
changes,
field
and
set
the
the
field
pads
from
here
into
this
field
in
the
life
cycle
of
the
resource,
and
that
should
work.
K
I
didn't
do
a
POC
yet,
but
this
is
just
like
a
theory
for
now
and
for
other
providers
that
don't
use
objects.
We
could
just
provide
some
helper
methods,
that's
unsets
ignored
fields
on
the
resource
and
they
could
just
call
it
before
updating
resistors.
K
So
it
would
be
like
those
those
fields
are
not
set
in
the
four
provider,
so
that
was
a
bit
help
with
having
as
little
changes
as
possible
in
the
providers
after
implementing
this,
and
we
would
also
release
it
as
an
alpha
feature
with
a
flag
in
the
providers
and,
of
course
ignore
the
fields
when
the
feature
is
not
enabled,
but
still
we
will
have
to
add
the
fields
into
the
spec
of
the
or
the
resources.
There's.
Also,
an
alternative
approach
that
suggested
is
is
a
set
of
changing
API.
K
We
could
just
ship
it
as
an
optional
annotation
in
the
there's,
like
a
first
solution
to
see
how
useful
is
this
change
and
the
first
like
survey
users
and
see
how
many
people
are
using
it
if
it's
first
to
setting
it
like
in
the
spec.
K
C
K
An
alternative
that
was
considered
this
is
this
other
one,
I
was
showing
you
within
the
providers.
K
Well,
it's
we'll
just
skip
this
one,
this
init
providers-
and
so
this
would
be
something
like
it
has
some
connotation
with
like
init
containers
as
well,
so
it
may
be
more
familiar
to
people
used
to
kubernetes
and
basically
it
would
be
like
the
schema
would
be
same
as
the
for
provider
one
and
you
could
just
set,
for
example,
this
desired
size
here,
and
it
would
be
only
used
in
the
creates
of
it's
a
great
step
and
then,
in
the
update
step,
the
providers
who
normally
use
just
for
for
provider
and
this
desired
size
would
never
be
like
fought
fought
with.
K
But
here
the
problem
is
that
first,
you
need
to
set.
You
can
just
ignore
fields
and
don't
worry
about
the
values
here.
You
have
to
actually
set
some
value,
and
the
second
thing
is
that
there
is
a
problem
with
the
late
in
initialization
that
kind
of
blocks.
This
solution,
because.
K
With
latency
Association,
we
could
have
desired
size
being
set
here
again
in
the
for
provider,
even
though
it's
already
in
in
provider
and
then
the
same
thing
would
happen
so
for
now.
We
don't
know
how
to
solve
this
issue
with
the
solution
with
latency
Association,
but
we
are
thinking
about
it
as
as
well,
because
there
is
some
concern
that
this
solution,
which
ignore
changes,
could
be
a
bit
too
powerful
and
could
solve
too
many
problems
that
we
maybe
don't
want.
K
And
this
one
maybe
a
bit
more
like
specific
to
this
problem
that
we
just
want
to
set
something
to
Recreation
and
later.
We
don't
do
worry
about
it,
but,
as
I
said
latest
initialization,
like
kind
of
messes
this
up
and
we'll
have
to
figure
out
how
to
how
to
overcome
this,
and
the
last
solution
would
be
not
to
do
any
generic
solution
and
just
do
the
changes
like
Case
by
case
in
the
provider
code,
because
it's
not
yet
certain
how
many,
how
many
resources
actually
need
this.
K
K
But
basically
there's
like
a
concern
that
we
would
solve
something
that
would
not
be
used
so
much
and
we
would
change
the
API,
maybe
for
not
not
enough
benefit
but
yeah.
But,
as
I
said,
we
don't
have
yet
the
data
on
how
how
much
this
feature
would
be
user.
If
you
would
make
it
generically
so
yeah
see,
we
still
don't
know
if
you,
if
anybody
has
some
additional
like
use
cases
for
this,
please
feel
free
to
give
us
to
be
very
valuable.
A
Yeah
and
I
think
that's
that's
I
would
exactly
Echo.
That
level
is
that
the
one
of
the
most
useful
things
here
is
going
to
be.
You
know
more
like
sharing
people's
use
cases
for.
K
A
This
feature
would
be
helpful
for
them,
because
we've
got
a
classification
or
a
set
of
use
cases
and
there's
some
patterns
with
them,
but
there
may
there
likely
are
ones
out
there
that
we
haven't
heard
of
yet.
So
that
would
be
very,
very
useful
for
folks
to
comment
on
this
proposal
here.
K
Exactly
exactly
so,
I'll
call
everybody
to
try
and
if
they
have
some
some
thoughts
about
it
or
some
use
cases,
please
contribute.
A
Awesome
lovers,
so
thanks
for
walking
us
through
that,
and
then
you
know,
there's
plenty
there's
opportunity
right
now.
Folks
have
some
quick
questions
specific
to
this
proposal.
You
know
I
think
we
could
take
one
or
two
of
those.
Otherwise,
though
you
know,
the
conversation
is
ongoing
in
The
Proposal
PR
that
I.
A
The
agenda
doc
3822
that
folks
can
contribute
there
too.
K
Yeah,
exactly
there's
that
already
feedback
so
feel
free
to
join
in.
C
A
All
right
so
then,
we've
got
one
more
agenda
item
here
that
was
a
holdover
from
last
week.
It
looks
like
we
have
a
quick
question
in
the
chat
as
well.
So
let's,
let's
take
that
quick
question,
real
quick,
so
Rahul
is
asking
about:
can
the
providers
be
namespaced,
scoped
and
so
I
think
that
there's
a
number
of
different
dimensions
to
look
at
that
at?
But
overall
the
general
answer
here
is
going
to
be?
Probably
no
that
you
know
providers
like
manage
resources.
They
are
cluster
scoped
objects.
A
You
can
think
of
it
as
something
analogous
to
in
a
kubernetes
cluster.
You
know
like
the
nodes
of
the
cluster
that
infrastructure
there
is
at
the
cluster
scope.
It
has
the
you
know
the
concern
of
The
Operators,
the
Builders
of
the
cluster,
so
we
that
is
something
that's
come
up
before
you
know
of
namespace
Scopes
providers,
either
they're
managed
resources
that
they
offer
or
you
know,
sharding
to
have
you
know
different
providers
isolate
running
and
isolated
in
a
single
namespace.
A
You
have
multiple
disabled
providers
running
at
the
namespace
level.
So
so,
if
let's
try
to
find
a
link
to
that,
so
you,
if
you
know
Raul,
you
can
add
what
you're
thinking
about
for
your
use
case
to
the
discussion
on
those
issues.
A
All
right
and
Pete
did
you
did
you
want
to
talk
about
docs
366?
It
was
a
holdover
from
last
week's
agendas.
I,
don't
know
if
it
was
actually
something
you
wanted
to
talk
about.
Still.
J
J
This,
oh,
this
is
a
no
sorry,
sorry,
there's
two
of
them.
That's
up!
So
this
one
I'm,
sorry!
Currently
in
the
docs,
we
maintain
each
version
as
well
as
a
master
version,
but
Master
is
just
a
copy
of
whatever
is
the
latest.
J
J
So
as
we
come
up
to
112
a
new
documentary
written,
we
can
either
create
a
branch
or
put
it
as
non-published
for
v112,
but
there
would
never
be
a
master
box
version
that
is
published
and
so
I'm
trying
to
keep
the
concept
of
Master
Branch
as
one
thing
and
get
ignore
that
specifically
talking
about
the
version
master
that
is
published
on
the
docs
website
and
dropping
that.
A
It's
so
Pete.
You
said,
though,
that
like
there
still
is
an
experience,
though,
for
seeing
documentation
essentially
for
the
next
crosswing
release.
That
is
not
out
yet
right,
like
as
new
features,
get
landed
into
cross-plane
master
and
they
are
they're
not
actually
part
of
release
yet,
but
there
are
docs
that
we
want
to
show
for
them,
so
people
can
start
using
them
and
seeing
Etc
before
the
release.
There
is
an
experience
around
that
still
right
that
you're
proposing
yeah.
J
And
I
will
share
very
quickly
just
to
give
you
okay,.
J
Like
400
windows,
theoretically
to
this
one,
so
what
we
would
do
again,
this
is
independent
of
branching.
What
we
can
do
is
we
can
come
in.
So
each
release
has
its
own
folder
and
then
Master
is
a
folder.
J
J
A
J
J
I
just
found
it
back
in
this,
but
in
this
template
it
would
say:
hey,
go
change.
This
parameter
on
the
1.12
folder
to
make
it
publish
and
be
live.
A
Got
it
got
it,
but
still
pre-release,
though
not
like
the
latest
of
hey.
This
is
this
is
what
you
should
use
and
it's,
but
it's
not
actually,
you
know
officially
released
yet.
J
J
And
so
it's
really
kind
of
I'm
happy
to
talk
more
in
the
issue.
But
the
the
thing
we're
solving
for
is
that
we
kind
of
have
two
different
personas.
J
A
Got
it
I
see
this
picture
you're,
making
there
cool
so
yeah
I'll.
Take
a
look
at
that
issue
as
well
to
see
if
you
know,
God
internalize
it
a
little
bit
better
and
understand
it.
J
Thanks
here
and
if
anybody
else,
anybody
else
has
sponsor
feelings.
Dm
me
drop
something
from
the
issue
telegram,
whatever
works
for
you.
A
Cool
and
then
there's
I
guess
to
round
it
out
for
the
day
here.
There's
a
question
in
the
chat
as
well
around
when
I
Cube
control
apply
and
delete.
Crds
crds
are
hung
as
they
have
finalizers,
even
after
deleting
objects.
Associated
any
any
comments
there
good
question:
I
would
that's
not
expected
Behavior,
but
you
know
when,
like
I
the
understanding
or
the
expectation
is
that
the
crds
that
are
exposed
by
a
provider
are
managed
by
the
the
package
that
they're
contained
in
and
the
package
manager.
A
So
it's
expected
to
you
know
like
if
you
you
delete
all
live
instances
of
resources
of
managed
resources,
so
none
of
those
left
and
then
you,
you
know,
delete
the
provider
object
or
the
configuration
object
whatever
it
is
so
that
the
Crosswinds
package
manager
then
goes
and
starts
cleaning
up
all
the
crds
that
are
defined
in
that
package
that
it
should
take
care
of
them.
A
Remove
the
finalizers
get
clean
everything
up
so
I,
don't
know
if
there's
a
path
that's
being
taken
here
outside
of
that,
but
like
that's,
the
package
manager
is
the
one
who
is
expected
to
do
all
that
cleaning
up
and
there
should
not
be
that's
not
expected
to
be
a
manual
action
where,
like
you're,
actually
doing
deletes
on
crds
yourself.
So.
A
Cool
all
right,
everybody,
so
that's
the
end
of
the
agenda,
thanks
all
for
the
participation
in
the
chat
thanks
everybody
for
talking
through
all
the
cool
stuff,
we're
working
on
so
obviously
a
whole
lot
going
on.
I
checked
right
now
and
a
lot
of
exciting
progress
so
great
to
see
everybody
today,
thanks
everybody
for
joining
in
and
we'll
see
you
all
again
soon.