►
From YouTube: Cloud Custodian Community Meeting 2021-07-20
Description
Our community meeting is public and we encourage users and contributors of Cloud Custodian to attend! You can find the notes for this meeting on our github repo: https://github.com/cloud-custodian/community/discussions
To get an invite to the meeting join the google group and you'll receive one via email: https://groups.google.com/g/cloud-custodian Read the minutes for this meeting
https://github.com/cloud-custodian/community/discussions/15
Join our community!
https://github.com/cloud-custodian/community
A
And
welcome
everybody.
The
date
is
july,
20th
2021,
and
this
is
the
cloud
custodian
community
meeting
I'll,
be
your
your
host
george
castro.
Thank
you.
Everyone
for
joining.
We
have
people
joining
the
meeting
as
we're
going.
This
is
only
our
second
meeting,
so
this
is
the
agenda
introduction.
We
had
one
last
time
we're
gonna
take
some
quick
some
time
for
us
to
quickly
introduce
ourselves.
A
If
you
haven't
been
here
before
some
kickoff
items,
and
then
we
have
some
agenda
items
today,
we'll
let's
revisit
the
enhancement
feature
process.
Kapil
mentioned
last
time.
Marco
wants
to
talk
about
policy
testing.
A
I've
got
a
straw
person
proposal
about
slack
and
then
liz,
and
I
were
thinking
about
doing
a
weekly
doc
session
that
we'd
like
to
get
some
ideas
and
feedback
on
and
then
I
think
we
wanted
to
talk.
Gcp
provider,
maintainer,
update
question
mark.
I
forgot
what
this
one's
about
we'll
we'll
cross
that
bridge
when
we
get
there
so
welcome
everyone.
This
is
only
our
second
public
meeting,
we're
planning
on
making
this
a
regular
thing
as
we
get
through
the
agenda
items
when
we
get
to
the
end.
A
If,
if
no
one
has
any
extra
agenda
items
that
they
want
to
whack
onto
the
agenda,
we'll
just
go
ahead
and
have
a
pr
party.
We
can
talk
about
some
of
the
pr's
burning
issues,
anything
that's
happening
in
github.
We
can
discuss
as
a
community
the
idea
there
being
we
can
get
through
like
a
formal
agenda
and
then
get
people
used
to
talking
about
how
new
stuff
is
coming
in
and
the
processes
that
we're
doing
that.
A
So,
as
you
write
things
like
a
a
brand
spanking
new
contributor
guide
and
all
this
friendly
kind
of
documentation,
we
can
start
to
get
that
feedback
and
get
that
tight
feedback
loop
from
the
community
that
we
want
with
that.
Is
there
anybody
new
that
wants
to
introduce
themselves
or
someone
from
last
time
who
wants
to
introduce
themselves?
This
is
optional.
You
don't
have
to
introduce
yourself,
but
if
you've
been
using
cloud
custodian,
we'd
love
to
get
to
know
you
get
your
feedback
and
things
like
that
and
I'll
go
first.
A
So
you
don't
have
to
george
castro
started
a
stacklet
a
few
weeks
ago
to
kind
of
help
get
things
like
this
rolling.
He
him
his
and
I
live
in
ann
arbor,
michigan
and
my
hobbies
are
f1
and
nba.
Anybody
want
to
introduce
themselves.
B
Yeah
I'll
introduce
myself
as
well:
hey
marker
chappie,
I've
also
at
stacklit,
and
I've
worked
on
some
of
the
cloud
custodian
oss
stuff
recently,
mostly
around
some
of
the
c7
and
terraform
pieces
and
gcp
cloud
pieces,
and
I'm
he
him.
C
C
I'm
at
capital
one
he
him
and
we
are
contributors
to
the
aws
provider.
E
F
Hey
everyone
sunny.
He
him,
I'm
out
in
l.a,
excited
to
have
this
a
second
meeting.
G
So
I
work
at
capital,
one
two
and
I
live
in
the
tysons
area.
Close
to
reston
have
been
around
on
the
project
for
a
couple
of
years
now
excellent
welcome.
H
Okay,
I'll
jump
in
so
I'm
liz
pronouns.
Are
she
her?
I
am
a
developer
advocate
at
stacklit.
I
am
in
the
city
of
san
francisco
in
california
and
let's
see,
hobbies
include
lance.
I
Hey
everybody,
david
glick.
I
work
at
capital
one.
I
am
the
product
manager
for
the
really
the
internal
platform
of
cloud
custodian
at
capital,
one
and
he
him
live
in
richmond
virginia
and
very
much
into
international
travel.
Just
got
done
with
my
second
trip
a
few
days
ago
to
europe.
Since
restrictions
were
lifted,
was
it
fun?
Was
it
I
had
an
amazing
time,
I'm
a
big,
foodie
and
and
and
the
restrictions
really
weren't
that
bad
at
all.
But
apparently
right
after
I
left
things
started
to
get
crazy.
A
Okay,
all
right,
let's
move
on
all
right,
so
kickoff
items
I
copied
this
over
from
our
last
meeting.
I'm
gonna
keep
these
in
the
meeting
notes
here
for
a
while.
I
opened
a
community
repo
where
we
can
file
meta
issues
about
like
either
agenda
items
and
as
well
on
there.
I
have
a
community
project
board
as
a
community
manager.
I
like
to
do
all
my
work
out
in
the
open.
So
if
you
look
at
that
board,
it's
like
a
little
stripped
down
kanban
board
of
the
things
I've
been
working
on.
A
So
if
there's
any
friction
or
anything
that
you
find,
you
have
any
cool
ideas
and
something
that
you
want
to
work
on.
Absolutely
100
percent
feel
free
to
follow
an
issue
there
and
I'll
I'll
take
a
look
at
it
and
we
can
triage
it
appropriately
like
to
be
proud
of
being
affiliation-less.
A
When
doing
my
community
work
so
feel
free
to
just
ping
me.
If
you
have
any
good
ideas,
my
job
is
to
help
cloud
custodian,
oss
100
of
the
time
so
feel
free
to
reach
out
to
me
for
any
of
that
kind
of
stuff,
and
if,
if
you
hadn't
noticed
last
time,
I
take
the
notes
on
everything
in
the
video
we
post
a
video
to
youtube
every
week.
A
If
you
missed
the
early
beginning,
just
a
heads
up,
this
meeting
is
being
recorded
and
we
are
putting
this
on
youtube
and
then
the
notes
and
everything
in
that
same
community
repo
in
the
discussions
area
I
file
in
all
the
these
meeting
notes.
I
take
them
out
of
this
hackmd
and
I
put
them
in
there.
I
very
much
believe
in
github
as
a
source
of
truth
for
a
lot
of
things,
so
I'm
gonna
be
shoving
as
many
things
into
github
and
not
in
google
docs
or
different
things.
A
A
All
right
enhancements
feature
process
kapil.
We
talked
about
this
a
little
bit
before
and
I
think
we
had
as
a
group
mostly
nodded
towards
google
docs
for
drafting,
and
then
I
I
believe
you
were
going
to
look
at
a
template.
Do
you
want
to
fill
us
in
on
what
you
got
going
on
here?
Well,
I
switch
over.
J
To
upload
she
said
did
not
spend
much
time
on
the
proposal
like
the
template
for
the
first
proposal,
the
first
enhancement
proposal
being
the
definition
of
the
enhancement
process.
I
was
spending
my
time
on
trying
to
get
through
the
backlog
of
prs,
but
in
general
for
people
that
didn't
catch
last
week.
This
is
basically
the
notion
that
larger
changes
that
potentially
have
impact
to
use
your
exposed
syntax
would
have
some
sort
of
longer
form
proposal,
taking
into
account
consideration
and
things
like
backwards
compatibility,
etc.
J
I
think
there
there's,
probably
the
the
biggest
open
question
on
it.
That
probably
needs
to
be
defined
is,
what's
the
what's
the
step
up
between
something
that
doesn't
need
one
and
something
that
does
need
one
and
definitely
open
for
feedback
on
that.
But
hopefully
we
can
cover
that
in
that
first
proposal
itself,
but
no
no,
no
active
changes
this
week
on
that.
B
A
All
right
moving
on,
oh,
the
agenda
is
still
open,
so
if
you
have
something
that
you
want
to
whack
at
the
bottom
here
feel
free
to
do
that
and
if
we
have
time
we
get
to
it,
if
not,
we
put
in
the
backlog
for
next
week,
marco,
you
added
this
one
policy
testing
and
then
you
have
a
spec.
Let's
take
a
look.
B
Go
ahead,
yes,
so
speaking
of
enhancements
in
feature
process.
This
is
something
that
I
kicked
around
for
a
little
while,
and
I
eventually
opened
up
an
issue
and
a
google
doc
with
a
policy
testing
proposal,
I
haven't
been
able
to
really.
I
don't
feel,
like
I've
gathered
really
enough
interest
or
feedback
from
people
yet
other
than
again.
B
Thanks
john
for
the
plus
one
and
another
person
who
commented,
and
so
I
am
interested
in
pushing
policy
testing
forward
as
a
topic
and
potentially
kind
of
carving
out
a
small
group
of
people,
either
secular
community
that
are
interested
in
driving
this
piece
forward.
But
the
gist
of
this
is
the
the
thing
I've
been
noticing,
especially
now
more
and
more
as
I
get
deeper
and
deeper
into
the
sprawl
of
policies.
That
kind
of
collect
over
time
is
changing.
B
Policies
can
be
pretty
nuanced
in
that
a
change
could
effectively
potentially
wipe
out
whole
swats
of
infrastructure,
so
having
a
means
in
which
you
can
easily
describe
tests
for
policies
and
express
them
in
something.
That's
not
any
one
language
specific.
B
So,
in
this
case,
I've
gone
with
a
somewhat
yaml
dsl
in
the
proposal
to
just
to
be
able
to
run
policies
through
a
ci
and
validate
that,
given
a
policy
it
does
what's
expected
of,
it
is
something
I
think
that
more
and
more
I'm
finding
a
need
for
in
my
day-to-day
and
I'd
like
to
potentially
push
this
forward
from
proposal
into
an
accepted
proposal.
Given
that
there
isn't
much
of
a
process
today,
that
seems
a
little
stalled
until
we
get
the
future
enhancement
process
workflow
in
place.
B
It's
back
from
january
now
and
there's
a
larger
google
doc
that
outlines
kind
of
the
workflow
process
for
it
and
see
if
anyone
had
any
feedback
or
any
follow-up
items
that
they
may
have
come
across
here
in
the
community
and
if
not
I'll,
probably
be
looking
to
assist
or
once
there
is
an
enhancement
proposal
process
pushing
this
into
the
pipeline
as
one
of
the
first
candidates
so
that
we
can
start
iterating
this
on
iterating
on
this
inside
of
the
c7n
repo.
B
That's
that's
pretty
much
it
just
more
of
this
exists
that
I'm
looking
to
find
anyone
either
here
in
the
community,
call
if
you're
watching
the
video
later
please
come,
find
me
to
kind
of
get
a
small
group
of
people
across
different
custodian
experiences
of
different
scales
of
policies
to
make
sure
that
what
we're
authoring
or
what
we've
designed
at
least
isn't
anything
that's
going
to
clash
or
is,
is
missing
huge
chunks
of
things
that
would
solve
problems
for
people.
K
H
K
Great
job
with
the
proposal,
I
think
it's
I
know
there
was
some
discussion
about
flight
recording
and
how
that
plays
in.
So
I
know
that's
that's
a
bit
of
an
open
question
and
and
aside
from
that,
just
what's
a
useful,
the
most
useful
first
step
beside
like
yes,
spec
looks
cool,
let's
roll
with
it,
and
maybe
that's
after
we
go
through
and
and
this
gets
approved
and
you
have
a
small
group
of
people
we
can
roll
with.
Maybe
that's
a
discussion
for
that
group.
B
Yeah,
I
don't
know
I
every
time
I
come
back
to
this
and
I
think
you
know
I'll
just
start
doing
I'll.
Do
a
real
small
straw
man.
I
end
up
like
halfway
through
like
okay.
This
is
a
very
large
straw,
man
yeah,
and
it's
just
like
what
what
there's
so
much
groundwork
that
needs
to
kind
of
be
sketched
out
it's
hard
to
either.
You
need
to
build
something
very
shallow.
B
That's
not
representative
of
what
this
would
look
like,
or
you
end
up,
starting
to
architect
the
whole
system
out,
and
that's
why
I
always
get
stalled
in
that.
I
don't
want
to
invest
a
ton
of
time
where
I
may
end
up
taking
an
approach.
That's
not
really
scales
well
for
a
lot
of
people.
So
that's
why
I'm
kind
of
interested
in
getting
with
a
few
more
people
from
the
community.
I
know
john,
you
expressed
interest
and
I
may
think
you
and
get
her
later
to
run
through
some
things
with
you.
B
But
a
more
formal
approval
process
and
review
process
would
be
helpful
here.
C
A
Start
one
all
right,
so
your
first
action
item,
marco
you
and
I
will.
A
B
H
B
H
A
J
Seem
to
be
well,
you
know
single
qrr
too,
like
just
to
be
aware.
You
can
like
this
is
going
to
be
a
separate
tool.
Sub
directory
it
doesn't,
it
can
be
incremental
per
se,
and
there
is
at
least
one
item
for
this,
which
is
and
requires
some
supporting
custodian,
which
I
think
I
saw
somewhere
in
the
agenda
with
regards
to
policy
destinations
for
lambdas.
B
J
It
does
doesn't
need
to
land
as
one
genuine
pr,
and
I
think,
there's
and
actually
it's
probably
better
that
way,
but
especially
what
there's
a
spec
that
actually
lays
out
sort
of
directionality
on
where
it
wants
to
go
and
then
the
other
part
was,
I
think,
there's
at
least
one
backlog
item
against
custody
itself
and
the
aws
provider,
at
least
for
the
functional
testing
on
lambda
event-based
policies
with
regards
to
destinations,
although
and
flushing
out
the
spec
or
or
beyond
spec.
B
Cool
great
I'll
go
hunt
through
that
pr
and
add
it
to
the
list
of
things
to
chat
about
in
this
follow-up
session.
A
J
Sorry,
I
think
it's
already
in
the
notes.
I
think
it's
if
you'll
search
for
destinations
a
destination,
singular
you'll
see
it.
A
And
policy
support,
destinations,
5250.
A
All
right,
I
wanted
to
bring
up
another
straw
person
about
possibilities
of
using
slack
as
our
real-time
chat
as
opposed
to
gitter.
I
know
for
me.
I
found
a
tool
that
was
kind
of
uncommon
compared
to
other
ones,
was
kind
of
weird
for
me,
so
I
logged
this
and
then
I
was
performed
by
lots
of
people
that
slack
might
be
blocked
in
certain
corporate
situations,
and
things
like
that,
so
I
thought
I
would
put
this
on
here.
A
Is
anybody
feeling
strongly
about
this
because,
as
kapil
started
to
rattle
off
users
that
have
are
using
cloud
custodian
but
are
blocking
slack,
it
became
clear
to
me
that
this
is
a
straightforward
thing
and
that
gitter
is
been
around
for
five
years
now.
At
this
point,
does
anyone
have
super
strong
opinions
on
here?
A
I
do
from
an
adoption
perspective,
but
I
also
understand
that
we
gotta
kind
of
listen
to
users
requirements
there.
D
I'd
say:
slack
is
much
easier
to
use
and
search
the
history
of,
but
from
my
prior
company
it
was
blocked.
I
know
that
I've
heard
of
some
other
companies
that
block
it
as
well.
So,
ultimately,
I
do
like
it
better
as
a
tool.
It's
just
yeah
the
fact
that
not
everybody
would
be
able
to
use
it.
H
So
I'm
not
as
familiar
with
getter,
I'm
more
familiar
with
slack
and
I'm
wondering
if
there
are
some
ways
that
getter
can
be
like.
We
can
modify
it
so
that
it
is
so
that
it
has
more
of
the
aspects
of
slack
that
make
slack
easier
to
use
because
yeah,
I
would
plus
one
jameson
on
getter.
H
K
I
I
learned
only
within
the
past
week
that
there
is
another
channel
and,
if
so,
because
everything's
coming
in
through
one,
we
end
up
trying
to
keep
discussion
very
focused
and
short,
and
it
has
kind
of
a
stack
overflow
effect
where,
if
it's
not
right
to
the
point,
it's
a
bit
of
a
distraction
and
there's
an
understandable
resistance
to
that
which
I
don't
know
how
much
that
runs
against
what
you're
trying
to
do.
George
here.
A
Yeah,
well,
you
know,
I
I
think
it's
having
there's
also
a
benefit
right
to
saying
like
I'm
on
aws,
I
want
all
the
aws
stuff
here,
the
gcp
stuff.
Here
you
know
that
kind
of
my
my
thinking
was
from
like
kind
of
an
adoption
perspective.
When
you
look
at
you
know,
kubernetes
with
like
150
000
users.
You
know
cncf
with
like
in
the
multiples
of
thousands,
but
I
also
want
to
be
cognizant
of
that.
We
we
need
to
be
where
the
users
are,
and
I
didn't
want
to
propose
something.
A
That's
like
okay,
we'll
just
introduce
a
bunch
of
new
tools.
Now
I
know
jameson
and
aj.
I've
seen
you
both
in
getter
constantly
kind.
You
know
answering
user
questions
and
doing
that
stuff,
and
I
would
hate
to
basically
triple
your
workload,
but
I
figured
you
know
it
was
worth
discussing
and
getting
feedback
so
far.
It
feels
like
not
a
lot
of
people
are
a
f
fan
of
it,
but
it's
getting
the
job
done
and
it's
better
than
not
than
cutting
off
access
to
a
large
swath
of
users.
What
do
people
think.
J
A
Well,
I
mean
the
world
hasn't
fallen
apart
because
you're
using
gitter,
you
know-
and
I
have
yet
to
run
into
an
officer's
project
that
has
either
failed
or
succeeded
because
they
chose
the
wrong
chat
tool.
You
know,
but
just
in
my
brain
you
know
I'm
always
thinking.
What's
the
best
way
to
reduce
friction,
but
if
slack
is
a
problem
for
enterprises
because
they're
not
using
it,
then
we
have
to
take
that
into
account
and.
J
B
H
Oh,
I
just
think
it's
so
I
think
it's
worth
investigating,
just
because
I
I
kind
of
am
always
of
the
belief
that,
like
there
is
a
way
to
make
everything
as
frictionless
as
possible
and
like
now
that
I've
heard
more
of
the
argument
for
or
like
some
of
the
cons
of
slack,
which
it
sounds
like
slack,
would
lead,
would
then
kind
of
lead
it
in
more
of
like
then.
It's
not
you
know
it's
not
as
inclusive
and
there's.
H
There's
got
to
be
a
way
to
make
getter
like
onboarding
or
adoption
like
not
like
more
frictionless,
because
like
it
could
be
as
easy
as
like
having
something
like
somewhere
easy
to
discover
where
it's
like
hey.
We
acknowledge
that
getter
might
be
a
little
bit
intimidating
or
like
difficult
to
adopt
if
you're
not
used
to
it
like
here's.
H
Some
like
tips
right
because
I'm
not
just
like,
I
think,
like
even
first
the
acknowledgement
is,
is
good
right
because
yeah
it
gives
maybe
people
the
feeling
that,
like
or
lets
others
know
like
yeah,
we
get
it
it's
a
little
weird
at
first,
but
trust
us
yeah,
so
yeah.
I
think
that
at
the
very
you
could,
at
the
very
least,
do
that.
D
And
so
I'd
say
the
biggest
pain
point
for
getter
is
the
search.
I
mean
the
post
stuff
and
reply
it's
fairly
easy.
Just
some
of
the
replies
can
get
a
little
wonky
if
you're
doing
code
snippets
or
trying
to
edit
a
reply
or
something
like
that.
But
honestly,
the
the
main
issue
that
that
I
personally
have
with
getter
is
just
the
search
history
is
is
real.
It's
not.
It
doesn't
display
the
results
in
a
very
efficient
manner
and
it
just
makes
finding
previous
posts
on
a
topic
painful.
H
D
Easier
to
do
yeah,
though
that
would
be,
then
you
have
a
separate
system
that
you'd
have
to
go
search
which
wouldn't
be
ideal,
but
that's
kind
of
the
main
pain
point.
From
from
my
point
of
view,.
C
J
I
mean
we
technically
have
a
stack
overflow
I
think
like.
Currently,
if
we
looked
at
all
of
our
different
channels,
we've
got
getter
a
mailing
list.
Reddit
and
stack
overflow
and
part
of
the
question
is:
are
we
fragmenting
ourselves
too
much
and
the
more
channels?
We
add
the
more
things
we
have
to
monitor
and
maintain,
and
I
think,
there's
a
value
to
having
a
critical
user,
a
critical
mass
on
a
user
population
in
one
place
so
that
people
can
actually
answer
each
other's
stuff,
whereas
creating
additional
channels
is.
J
I
don't
know
same
time
like
I
don't
want
to
like
be
inertia
to
trying
out
something
new
in
that
context
of
the
beginner
discussions.
You
know
it
to
me,
it'd
be
like
hey.
You
want
to
have
a
trial.
We
want
to
try
this
out
for
a
little
bit
and
see
if
it's
worthwhile,
especially
if
we
could
do
like.
J
I
think,
there's
a
lot
of
common
questions
that
come
up
in
chat
and
to
jameson's
point
part.
The
reason
they
get
asked
over
and
over
again
is
just
search,
not
great
and
having
a
more
threaded
discussion
forum,
where
some
of
that
stuff
can
surface
up
is,
has
a
good
amount
of
value.
B
B
The
one
thing
that
keeps
coming
up
is
search
is
painful,
but
I
think
the
the
underlying
issue
there
is
that
people
aren't
able
to
google
the
questions
they're
looking
for
and
have
to
come
to
chat
constantly.
This
might
pair
well
with
the
documentation
stuff
that
you
were
talking
about,
george,
but
it
may
be
worthwhile
pulling
up
like
what
are
the.
B
What
are
the
last
x
questions
that
were
asked
in
getter
this
last
week
and,
let's
just
go,
write
doc
pages
for
those,
because
the
sphynx
stocks
will
generate
much
better
search
ability
than
what
we
have
in
gitter
and
see
if
measure.
If
that
cuts
down
on
duplicate
questions
being
asked
over
and
over
again.
J
Yeah,
my
my
number
one
pain
point
with
getter
is
just
lack
of
anything
useful
on
mobile
it
just
it
I'm
constantly
on
mobile,
and
it's
though,
like
you,
can't
it's
very
hard
to
threaten
to
reply
and
if
I
do
it
in
matrix
versus
the
web
point
versus
the
native
client,
it's
all
different
results.
So.
C
A
Okay
yeah,
I
was,
I
opened
discussions
for
cloud
custodian,
but
then
I
locked
it
because
I
didn't
want
people
to
start
posting
stuff
there
and
be
like
hey
who's,
checking
that
you
know
to
create
another
thing
with
the
intention
of
only
posting
like
the
meeting
notes-
and
you
know
stuff
like
that-
maybe
we
could
use
it
grabbing
useful
things
from
getter
and
maybe
you
know
cleaning
it
up.
A
M
J
A
Yeah
you
could,
we
could
easily
create
a
faq
or
you
know,
top
q
a
from
chat
or
something
like
that
liz,
and
I
will
do
that
because
I
didn't
want
to
take
up
the
bulk
of
the
meeting
talking
about
slack,
because
I
know
how
we
all
love
to
talk
about
our
tools
but
yeah.
Why
don't
liz?
Why
don't
we
investigate
discussions?
Because
that
gets
us
mobile
for
free
we're
using
it
for
meeting
notes
anyway?
Maybe
a
curated
set
of
useful
facts,
a
question
for
the
group
of
the
people
that
are
here.
A
Is
there
anyone
actively
scrubbing
the
cloud
custodian
tag
on
stack
overflow,
because
what
we
used
to
do
in
the
past
is
just
tell
everyone
to
go
to
the
tag
and
then
you
create
the
tag
and
then
the
kind
of
top
10
faq
generates
itself
based
on
traffic
votes
and
things
like
that.
But
that
also
takes
a
level
of
commitment
that
I
don't
know
if
everyone's
willing
to
take
but.
D
K
G
J
I
H
H
A
F
Yeah
on
the
topic
of
like
searchability,
specifically
for
discussions,
do
you
have
you
found
that
it's
like
the
seo
is
like
good
for
finding
answers
to
stuff
like
usually,
if
I
google,
something-
and
I
mean
maybe
it's
just
other
communities-
don't
really
use
discussions
as
much,
but
I
don't
think
I've
ever
seen
a
discussion
link
pop
up
in
google.
F
That
would
be
my
only
concern
because
I,
like
I
don't
want
people
to
just
ask
like.
Why
isn't
my
policy
running?
That's
like
a
github
issue
like
that?
That's
not
the
appropriate.
F
But
typically
that
is
where
I
seem
to
find
answers
a
lot
of
the
time
for
other
open
source
projects.
So
that's
that's
my
main
concern
with
discussions,
but
maybe
some
I
mean
stack
overflow
is
great
for
seo
for
sure
it's
just.
We
would
have
to
curate
that
and
be
on
it
all.
The
time.
A
Yeah
and
kapil
just
mentioned
he's,
never
googled
something,
and
you
end
up
on
a
github
discussion,
github
issue
sure
but
yeah,
never
a
discussion.
Okay.
That
gives
us
liz
that
gives
us
stuff
to
think
about
and
do
research
on
and
that
kind
of
sort
of
relates
to
liz
was
pinging
me
and
said.
I
really
want
to
work
on
docs
a
little
bit
every
week
and
we
got
to
talking.
We
decided
that
her
and
I
could
just
hop
on
a
call
like
this
once
a
week.
A
We'll
probably
pick
I
don't
know
a
thursday
or
friday,
but
do
it
open
like
this
just
hop
on
a
thing
like
this
open
a
browser
share
the
screen
find
a
paint
point
in
the
docs
and
then
just
bust.
It
out
spend
an
hour
a
week
on
that
and
we
figure
we
could
leave
that
open
invite
if
people
want
to
join
in
to
learn
something.
Maybe
you
ran
into
a
bug
and
the
docs
there
confused
you
and
you
can
like
bring
it
up
to
us,
you
know,
and
then
we
can
maybe
quickly.
A
You
know
sprint
on
that.
Little
topic
there
and
kind
of
get
the
docs
into
a
place
where,
like
wouldn't
it,
be
neat
if,
when
people
report
new
bugs
and
then
you
find
that,
oh
no,
you
know
it's
because
of
the
this
section
of
the
docks
wasn't
specific
enough.
You
can
kind
of
fix
that
much
quicker,
and
then
that
gives
you
that
little
open
source
dope
I
mean.
So
if
anybody
has
any
ideas
on
that,
I
don't
think
I'm
gonna
put
it
on
the
calendar
or
anything
like
that.
A
A
You
know
we're
going
to
fix
the
getting
started
page
today
or
whatever
it
is,
and
I've
pinged
this
on
twitter
with
other
community
managers.
Other
projects
see
if
they
do
something
like
this.
Anyone
have
any
ideas
or
comments
or
things
that
they'd
like
to
to
see
us
working
on
there.
A
F
F
I'm
sorry
anything
top
of
mind
specifically
on
the
box.
H
Yeah,
no,
I
don't
have
a
specific
page.
I
it's
been
my
experience
around
because,
like
I'm
kind
of
I'm
coming
into
this
from
like
kind
of
a
different
trajectory
right
and
so
like,
I
have
found
that
some
of
the
things
that
maybe
seem
like
obvious
to
someone
who's
more
familiar
with.
It
is
not
like
as
obvious
to
me,
and
so
then
it's
I
always
find
myself
like.
I
can't
think
of
a
specific
example
right
now,
but
then
I
find
myself
like
hung
up
on
little
things.
That
then,
could
also
like.
H
Sometimes
it's
just
like.
Oh,
I
just
need
to
be
explained
that,
like
oh,
I
know
what
it
was.
It
was
the
custodian
mailer,
I
think.
Maybe
it
was
the
way
the
information
was
arranged
or
there's
something
about
it.
That
made
it
for
some
reason.
I
didn't
realize
that
then,
like
final
step
is
like
run
it
and
like
was
like,
went
to
aj,
and
I
was
like
why.
H
What
do
I
do
next,
and
I
can't
remember
specifically
what
it
was
about
that,
but
it's
just
like
it's
just
such
a
it
would
be
just
like
a
minor
thing
like
maybe
even
like,
like
a
summary
at
the
top
of
like
you
know,
this
is
how
this
works
and
then
like
going
into
the
different
like
the
different,
like
notifications,
that
you
can
use
yeah,
so
that's
kind
of
more
or
less
like
what
I
was
experiencing
was
like.
Oh,
I
would
like
kind
of
like
this
bigger
picture
view
of
it.
J
J
This
is
what
it's
like
to
run
it
and
it's
definitely
super
helpful
for
someone
new
to
the
project
as
they're
getting
up
to
speed
to
do
that
because
they
have
this
mystical
thing
called
beginner's
mind
regards
to
it
that
a
lot
of
us
have
lost,
and
so
that
would
be
a
fantastic
contribution,
because
I
think
that's
something
that
when
I
we've
actually
we've
got
some
community
blog
posts
that
actually,
I
think,
do
better
jobs
than
our
official
docs.
J
F
I
would
also
know
we
do
have
a
few
videos
like
tutorials
and
walkthroughs
on
like
running
event-based
policy,
running
writing
your
a
poll
based
policy
and
stuff,
like
that.
I
think
you
could
probably
take
the
bones
of
those
and
build
out,
like
you
said,
like
a
narrative
style
like
doc
page,
because
people
sometimes
still
prefer
to
watch
a
video
and
having
it
all
on.
All
in
text
also
makes
it
easier,
so
you
don't
have
to
like
transcribe
from
a
video
feed
into
like
you
know,
a
yellow
document.
A
A
I
think
this
can
help
us,
but,
as
I've
said
before,
I'm
a
I'm
a
stickler
for
cadence,
so
you
know
if
liz-
and
I
are
just
going
to
commit
to
one
hour
a
week
if
you
want
to
show
up
if
you
got
the
hour
to
do
it
with
us,
I
think
just
one
hour
a
week,
we
could
start
to
at
least
saying
down
a
lot
of
the
paper
cuts
and
get
to
where
we
need
to
be
videos.
That's
a
great
idea!
Sonny!
That's
that's
awesome.
Anybody
else
have
any
other
ideas.
F
Yeah
I
saw
I
saw
someone
on
gitter
mentioned
that
they
watched
the
event-based
one,
which
is
the
one
that
I
did
for
kubecon
eu.
So
I
was,
I
was
surprised.
I
didn't
think
anyone
was
actually
like
looking
that
up,
but
apparently
so
so
I
think
you
should
probably
you
know
do
something
with
that
information.
A
F
A
That
is,
that
is
a
george
item.
Definitely
I'm
gonna
put
investigate
video
section.
A
You
know,
and
you
just
click
on
it
and
you
subscribe
to
it
and
there's
everything
you
could
ever
want
to
know
for
the
past
three
years
of
cube
contacts
or
something
I'm
sure
we
can
start
to
at
least
build
something
like
that.
That's
a
good
idea,
especially
if
someone's
told
you
that
they
watch
the
video
and
learned
something
useful
from
it.
A
All
right,
gcp
maintainer
update
kapil.
Do
we
want
to
talk
about
this
one
or
or
next
week.
A
And
then,
with
that
that
kind
of
handles
everything
from
the
agenda
items,
anyone
want
to
add
anything
new
or
any
questions
about
what
we
discuss.
If
not,
if
you
want
to
bail,
you
can
bail,
we
spend
the
rest
of
the
the
time
of
the
15
minutes
kind
of
going
over
outstanding
prs.
So,
if
you're
looking
to
dive
into
a
pr
or
looking
for
a
place
to
help,
stick
around.
A
All
right,
moving
on
to
the
pr
review
kapil,
I
think
you
had
good
news
on
app
autoscaling.
J
Yeah
looks
like
all
the
feedback
got
addressed
and
just
a
matter
of
merging
it,
so
it
was
great,
it
looks
like
you
did,
merge
it
yeah.
I
was
quite
a
lot
about
worth,
but
and
I
think
it
came
out
pretty
well
overall.
A
A
Here
there
we
go
all
right
this
next
one
gcp,
iam,
filter.
J
Yeah,
I
think
this
one,
I'm
gonna.
I
think
I've
got
some
feedback
for
the
author.
I
was
newly
on
it.
I
think
the
challenge
is
we
don't
want
to
name
space
squat
on
something,
that's
just
doing
raw
doc
evaluation.
We
want
to
actually
keep
the
filter
name
and
push
the
value
filter
again.
So
this
is
sorry
background
context
effectively.
Every
gcp
lots
of
gcp
resources
have
I
embedded
ion
policies,
basically
a
json
blob
in
a
particular
format.
This
basically
lets
you
do
this
filter
lets.
J
You
do
jvs
path,
expressions,
value
filters
against
that
json
doc
in
practice,
that
is
actually
a
really
poor
way
of
expressing
any
semantic
intent
with
regards
to
that
json
doc,
like
I'm,
not
exposing
this
to
users
outside
of
my
org,
is
actually
really
hard
to
express
against
rajesh
on
doc,
with
jamie
s
pat,
so
I
think
the
goal
here
would
be,
let's
at
the
same
time,
there's
a
lot
of
useful
stuff
here,
like
it
fetches
jamie
the
policy
docs
for
lots
of
different
resources.
J
Let's
go
ahead
and
push
the
value
filter
portion
of
that
to
like
a
sub
attribute,
instead
of
taking
over
the
top
name
space
on
the
filter,
so
it'll
be
more
like
I
am
policy
and
as
the
filter
name
and
then
doc,
colon
and
then
the
sub
for
the
sub
mapping
for
the
actual
value
filter
seems
like
a
good
compromise
to
keep
this
feature
and
capability,
but
still
allow
for
a
future
expansion
into
more
semantic-oriented
use
cases.
J
A
All
right
and
then
you
have
the
gcp
folder
one,
which
is
a
different
one.
Any
opinions
on
this
one.
A
Nope
yeah,
I
don't
think,
there's
movement
on
this
one.
So
we'll
do.
Let's
see
no
change
this
week,
anyways
contrib
docs
code
coverage
status
failed
marco
volunteered
to
help
out
on
this.
One,
though
I
think
he
had
the
bail.
K
Yeah,
so
I've
got
a
comment
on
the
the
pr,
not
the
issue,
because
yeah
that
that
comes
a
little
older
but
on
the
the.
D
K
I
added
an
example
which
now
makes
it
clear
to
me:
it
may
be
clear
to
zero
additional
people.
So
that's
why
discussions
like
this
are
useful.
So
anybody,
even
now
or
later,
if
you
happen
to
look
through
that
and
say:
okay,
you
added
words,
but
it
doesn't
actually
help.
K
That's
that's
totally
fair,
but
what
I
think
is
going
on
there
is
that
they're
looking
for
for
a
way
to
match
metrics,
to
kind
of
bin
metrics
over
a
period
of
time
and
then
match
any
individual
bin,
but
the
way
that
the
pr
is
structured
seems
to
be
just
kind
of
re-implementing.
The
client
filtering
on
the
client
side,
which
is
what
I
think
kapil
had
mentioned
previously.
So
I
think,
there's
a
like.
K
It
seems
like
there's
a
valid
use
case
like
a
bit
of
an
edge
case,
but
a
valid
one,
and
but
I
don't
think
this
pr
is
going
to
fix
it.
So
that's
that's
it
for
now.
K
I
guess
my
question
to
the
group,
and
we
can
we
can
punt
on.
This
is
right.
K
Taking
if
we
look
at
metrics
we're
giving
it
a
certain
amount
of
days
and
we're
just
setting
the
period
to
that
whole
span
of
days
and
pulling
back
one
data
point,
does
it
make
sense
to
allow
a
custom
period
where
we
can
say,
let's
look
back
two
weeks
and
break
that
into
chunks
of
one
day
each
and
then
match
if
any
one
of
those
days
matches
a
given
set
of
criteria
like
is
that
a
valid
case
you
want
to
support
so.
K
It
should
be
the
one
api
call,
but
we'll
get
back
instead
of
one
like
a
data.
What
we've
been
assuming
and-
and
the
code
focuses
on
this-
it
tries
to
make
a
call
and
get
get
metrics
back
where
there's
an
array
of
data
points
with
only
one
item
in
it,
and
we
would
instead
want
to
get
you
know
to
match
any
item
in
that
array.
J
Seems
reasonable.
The
other
consideration
concern
is
that
we
currently
store
the
metrics
as
an
annotation
and
that
potentially
gets
fairly
unwieldy
at
large
volumes
on
the
metrics
data.
It
can
cause
issues
downstream
for
notifications
and
yeah
I'll
just
be
sensitive.
To
that.
The
other
thing
I
mean
I've
had
a
pipe
dream
for
a
while
on
at
least
on
notifications
of
instead
of
having
statically
sized
buffers
doing
dynamically
sized
buffers
where
sorry
notifications
are
effectively
how
we
send
messages
out
for
to
the
mailer.
There
are
static
lease.
J
There
are
static
limitations
to
how
many
what
the
size
max
size
of
a
packet
is,
and
we
currently
hard
code
that
as
far
as
how
many
resources
we
will
buffer
or
batch
to
try
to
fit
generally
within
that
window,
obviously
resources
size
very
widely
by
type.
So
there's
a
question
there.
J
K
That's
true,
that
is
a
good
concern.
Okay,
that's
worth
noting,
maybe
on
the
pr2,
so
you
don't
lose
track
of
it.
That's
a
that's
a
good
one!.
J
Yeah,
the
there
are
one
or
two,
it's
very
piecemeal.
I
want
to
say:
there's
like
two
of
them
out
there
remember.
One
of
them
was
like
ec2
user
data
where
we
right,
he
keep
the
cache
an
annotation
cache,
and
then
we
purposely
pop
it
out.
But
it's
a
pretty-
and
we
could
do
that
from
this
case
too,
if
we
enabled
that
but
the
when
we
do
that,
we
also
yeah
that
we
can
do
that.
For
this
case
too,.
J
A
J
This
is
fairly
fundamental
and
just
trying
to
make
sure
that
we
limit
the
number
of
api
calls
that
we
participate
fully
with
dry
run
and
that
that
we
don't
break
any
existing
custom
workflows
with
regards
to
layers
and
packages
where
the
things
I
think
there
was
a
separate
notion
of
having
some
something
being
able
to
run
and
look
at
all,
the
policies
that
are
gonna
be
run
in
a
given
execution
such
that
it
doesn't
have
to
do
all
the
layer,
checks,
sort
of
per
policy.
D
A
A
Or
maybe
a
little
summary
on
the
reopening
yep
that'd
be
dope.
Thank
you.
It
doesn't
have
to
be
that
detailed
anything
else
on
this
one.
A
A
B
J
J
N
Yeah,
I
think
it
was
copy
related
resource
tag
because
some
of
the
resources,
the
way
that
it's
presented
from
aws,
is
in
different
format,
in
key
value
format
versus
list
format
and
depending
on
what
I
ran
into
this
doing
something
I
don't
remember,
the
exact
resources
now
and
which
got
me
down
the
world
of.
I
should
make
a
function
that,
like
automatically
convert
from
one
to
the
other,
so
that
we
can
simplify
this
and
and
it
just
kind
of
blossomed,
into
a
much
larger
pr
thinking.
J
So
all
the
tag
actions
accept
the
same
inputs
and
I'm
just
trying
to
yeah.
I
was
trying
to
figure
out
what
the
actual
problem
was,
but.
J
C
A
All
right
cool
and
then
we
have
one
on
converting
tags
that
we
had
talked
about
last
week.
A
I'm
sorry
this
one
aws
code
deploy.
Thank
you.
A
G
Yeah,
I
just
wanted
to
get
some
thoughts
on
this
with
regards
to
implementing
different
resources
of
good
deploy.
So
I
have,
I
have
code,
deploy
applications
and
deployment
implemented,
not
sure
if
there
is
something
on
top
that
would
have
to
be
in
here.
J
Yeah,
I
I'm
not
fully
familiar
with
code
deploy
every
time.
I've
looked
at
for
my
own
personal
stuff
or
whatnot.
I
I've
always
felt
like
it's
very
yeah
and
looking
at
it
for
this
one.
J
I
think
the
weird
thing
to
me
was
like
generally,
we
don't
operate
on
data
like
it
seemed
like
it
was
application
deployment
group
deployment,
but
deployment
group
was
optional
in
the
middle
or
you
could
directly
over
an
appointment
for
an
application
without
going
through
the
deployment
group.
J
I
think
the
dated
use
case
was
trying
to
make
sure
sort
of
what
the
sources
and
destinations
were
were
valid
ones
yep
the
the
challenge
with
the
deployments
is
it
almost
felt
like
a
data
plane
activity,
and
I
just
like
at
the
prediction
I've
been
chatting
back
and
forth
a
little
bit
on
gitter
as
well
about
this.
The
deployments
themselves
are
immutable
once
they
reach
a
final
state,
but
you
potentially
can
stop
them
in
the
middle,
which
makes
me
wonder,
is
that
safe
versus
attaching
like?
J
If
we
can
do
it,
the
deployment
group
level,
then
we
can.
You
could
stop
it
before
the
application
actually
gets
into
a
weird
state
through
an
actual
deployment.
That's
using
that
config.
What
was
unclear
to
me
in
that
context
was
there's
a
it
wasn't
clear
that
a
deployment
always
needed
an
appointment
group.
J
If
that
was
the
case,
then
we
could
just
target
deployment
group
instead
of
deployments,
because
targeting
deployments
were
potentially
at
a
very
different
data
volume
level,
although
potentially
not
be
meaningful,
it
might
be
hundreds
or
a
thousand,
as
opposed
to
you
know
billions
in
s3
or
something,
but
the
more.
The
other
question
would
be
leaving
applications
in
inconsistent
states.
If
we
stop
at
mid
deployment
was
the
other
part.
That
was
a
little
unclear.
G
Yeah,
I
think
I
think
that's
a
good
idea
to
have
the
deployment
group
resource
as
well
in
that,
but
the
only
downside
with
regards
to
implementation
would
be,
would
it
be
able
to
trigger
based
off
an
event,
since
I
think
deployment
group
would
have
to
be
a
child
resource
of
application
type,
because
if
that
doesn't
happen,
then
I
I
think
it
would
be
the
purpose.
J
J
Okay,
so
if
that's
the
case,
then,
if
we
target
deployment
groups,
then
we
can
avoid
the
application
having
to
deal
with
commercial
deployments
that
works
out
and
it's
probably
a
lower
cardinality.
So
it
sounds
like
it
would
be.
A
child
resource
with
maybe
some
needing
some
a
custom
described
resource,
augmented
function,
yeah.
A
Goes
awesome
and
with
that
we're
at
the
top
of
the
hour.
So
if
you
have
a
meeting
you're
a
minute
late,
but
thanks
everyone
and
we'll
see
everyone
next
week,
open
agenda
and
I'll
get
the
notes
out
to
the
list
as
soon
as
possible.
Thank
you.
Everyone
great.