►
From YouTube: Kubernetes SIG CLI 20220608 - Kustomize bug scrub
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
Hello
and
welcome
to
today's
60
lie
bug
scrub
for
customized.
Today
is
wednesday
june
8th.
A
Today
we
are
going
to
be
going
over
customized
issues
that
have
not
been
triaged
yet
and
if
we
have
time
some
older
ones
that
have
not
had
attention
in
a
while
we're
going
to
be
using
the
customized
triage
board,
which
I
will
paste
in
the
chat
here
for
everyone.
A
This
is
also
a
great
opportunity,
if
you're
new,
to
customize,
to
claim
an
issue
that
you
would
like
to
work
on
or
to
ask
questions
about
about
anything
that
we
discuss
or
if
you
have
an
issue
that
you
want
attention
or
that
you
have
want
to
bring
to
our
attention
or
you
want
to
have
us
discuss,
then
you
can
also
bring
it
up
at
any
time
and
we'll
go
over
that
before
we
get
started.
A
Is
there
anyone
who
is
new
to
this
customized
bug
subscriber
to
six
zli
general,
who
would
like
to
say
hello.
B
A
C
B
A
Good
question:
we
are
using
the
kate's
infra
orgs
infrastructure,
but
whether
that
registry
is
on
the
new
thing
or
the
old
thing.
I
don't
remember
offhand.
A
Yeah,
I
don't,
I
don't
know
what
we
would
have
to
do
to
actually
move
the
images
themselves.
The
ask
is,
I
think,
from
the
other
perspective
like
these
registers
are
being
mirrored,
so
we
can.
If
you
are
using
these
images
on
the
on
the
end
user
side,
then
you
should
use
the
new
endpoint
and
it
will
use
the
best
registry
for
your
your
location
and
save
us
money.
A
B
Yeah,
I
just
no,
I
don't
have
any
info.
I
just
the
email
that
dylan
sent
out
popped
into
my
head.
A
A
So
we
have
this
issue
here,
pasted
in
the
chat
ability
to
template
resource
results.
So
for
starters,
template
is
a
flag
for
me
here,
because
templating
is
an
astute
feature.
Customize
deliberately,
never
ever
ever
does
templating
that's
kind
of
its
originating
design
principle.
So
something
to
keep
in
mind
as
we
read
this.
A
A
As
part
of
a
different
customization,
which
is
something
that
we
also
intentionally
don't
support,
I
don't
know
if
that
one
is
listed
on
these
dude
features,
but
the
templating
certainly
is
you
can
find
that
as
part
of
the
response
and
natasha
do
you
have
any
other
different
thoughts
about
this.
D
I
think
the
second
one
seems
like
kind
of
a
weird
design
to
me
like
one
resource
url
will
pull
all
the
remote
resources
and
then
the
other
ones
will
reference
it
like
it's
weird
to
me
that,
like
how
do
you
decide
which
one
is
supposed
to
be
the
one
with
the
remote
url
like?
I
think
it
should
be
like
symmetric
and
encapsulated.
D
That
said,
I
also
think
we'd
recommend
not
using
remote
urls
like
as
a
best
practice
like
I
think,
jeff
used
to
say
like
it
should
only
be
for
prototyping.
A
Yeah,
since
this
is
a
know
what
length
I
might
save
it
and
write
a
more
careful
response
afterwards,
but
the
the
three
things
seem
to
be
yeah
encapsulation,
the
use
of
remote
urls.
To
that
extent
being
not
something
you
recommend
and
the
templating
not
being
something
that
we
would
ever
accept
for
the
foundational
principles-
reasons
that
are
outlined
here
interesting,
we
don't
actually
use
the
word
templating
in
the
head
in
the
headings.
I
guess
it's
unstructured.
A
Edits
is
probably
where
we
cover
it,
but
it
could
be
something
we
could
clarify,
potentially
in
our
docs
all
right.
So.
A
This
is
an
upgrade
from
the
pre-kml
to
the
post
kml.
So
that's
worth
noting.
D
D
I
can
take
a
look
at
this
one
and
see
if
that
change
caused
their
problem.
D
A
A
And
we've
had
other
issues
opened
when
folks,
don't
like
that.
That
structure.
A
Part
of
it
is
designed
around
security
reasons
so
that
you
have
like
an
encapsulation
of
each
base,
not
just
security,
but
security.
Is
that
came
to
mind?
Is
that
the
file
loader
that
implements?
It
also
has
security
controls,
but
what
you
can
reference
from
within
a
customization,
but
the
idea
is
that,
like,
when
you
have
a
base,
everything
around
it,
everything
it
requires
to
be
expanded
is
local
to
it.
A
D
A
Yeah,
I
guess
like
it's
also
a
question
of
like:
why
are
you
copying
the
files
around
it's
like
you?
You
definitely
have
to
have
this
other
structure
to
begin
with
and
then
you're
having
to
move
it
later.
Why
can't
you
just
you
know
if
you
are
setting
up
a
repo
for
customize,
then
why
don't
you
start
with
that
layout?
I
don't
know.
D
D
I
think
we
can
reject
the
future.
Maybe
we
can
ask
them
like
why
they
need
this
instead
of
reusing,
customization
bases
or
like.
C
C
You
please
provide
more
information
on
why
the
standard
structure
is
where,
with
a
series
of
encapsulated
faces,
that
can
be
reused,
does
not
work
for
you
acquiring.
You
do
have
an
original
structure
with
multiple
faces.
C
C
A
Yeah,
it's
like
they
want
to
actually
be
using
helm
but
they're
using
it
through
customize.
In
order
to
get
various
features
out
of
here
like
this
in
the
overlay,
they
still
want
to
be
held
native,
which
is
not
possible
because
by
the
time
you're
in
the
overlay,
you've
already
generated
the
files
from
the
base.
Then
they
are
files
at
that
point
right
so
like
this
would
be
an
additional
helm.
Chart
that
they'd
be
adding
unless
I'm
mistaken,.
A
So
the
only
way
to
do
something
like
this
would
be
with
composition,
which
doesn't
exist
yet,
but
it
allows
you
to
manipulate
the
generator
config
itself
before
running
the
generator
and
that
would
include
the
helm
generator
in
theory.
It's
not
possible
right
now,
but
well.
A
A
So
right
now
they
could
do
it
with
the
transforming
transformers
feature.
They
would
have
to
use
the
full
expansion
inside
the
transformers
field
and
then
pull
it
in
in
the
overlay
right.
A
A
A
As
resources
and
those
resources
just
get
pulled
in,
you
have
to
have
them
in
their
own
separate
base.
I
don't
know
this
is
a
problem
for
them
and
the
reason,
actually,
maybe
the
best
thing
to
link
to
would
be
the
example
from
the
composition
kept,
because
it
shows
both
that
it
works
and
that
it's
ugly,
because
you
have
to
these
transfers
that
you
want
to
transform.
A
The
resources
themselves
aren't
transformers,
so
you
try
to
run
them
well,
that
doesn't
work,
so
you
have
to
have
like
this
kind
of
separate
tree
pulling
in
and
transforming
the
transformers
until
they're
ready,
which
means
that
you
end
up
with
sort
of
two
parallel
lines
coming
in
the
one
that
is
like
the
dev
version
of
the
resources
and
the
one.
That's
the
dev
version,
the
transformers
and
you
have
to
pull
those
up
together
and
merge
them
and
then
potentially
like.
A
If
you
wanted
to
apply
say,
like
a
dev
label
to
your
to
your
the
results
of
the
transformers
or
generators
in
this
case,
which
is
highly
plausible,
then
you
need
another
layer
to
do
that
because
they
haven't
been
combined.
Yet
now
you
have
that
transformation
defined
in
two
different
places.
It's
just
it's
not
very
elegant,
but
it
does
work
theoretically.
A
Well,
not
theoretically,
it
does
actually
work.
It's
just
an
ideal
structure
which
is
part
of
the
motivation
behind
the
composition
cap.
I.
C
Don't
think
that's
probably
gonna
give
me
what
I
want
these
enhancements.
Should
we
just
keep
links
to
these
things?
It's
the
second
one
yeah,
that's
our
issue
for
it,
but
yeah
that
I
do
wanna.
A
A
Yeah
they
they
it's
not
very
obvious
when
you
have
to
dig
through
the
repos,
but
so
you
have
your
plug-in
space
that
is
going
to
list
a
set
of
things
that
are
actually
transformers
as
resources
so
see
here.
It's
fake,
http,
load,
balancer
thing.
This
is
actually
a
transformer.
That's
going
to
be
run
by
a
docker
image
and
all
three
of
them
are
that,
and
this
customization
is
pulling
in
those
three
things.
Then
you
have
to
go
up
a
layer
here.
A
You
have
a
customization.yaml
that
pulls
in
those
bases
and
then
patches
on
because
they're
being
treated
as
resources.
At
this
point,
this
patch
gets
applied
to
the
http
load,
balancer
transformer,
to
change
the
domain
to
something
else
and
that
hasn't
the
transformer
hasn't
run.
Yet
so
it's
still
just
transformer
config
and
then
you
pull
in.
A
A
So
when
we
hit
this
layer
and
we're
evaluating
the
customization
tree,
we
hit
this
layer,
we'll
actually
execute
the
transformers
at
this
point
and
they'll
be
transformed
through
the
lower
layers
to
whatever
final
shape
we
wanted
them
to
be
in
so
yep,
that's
how
that
works,
and
then
you
have
a
top
layer
that
pulls
that
in
which
now
the
transformers
have
already
run,
and
it's
integrated
the
resources
that
came
from
the
transformers,
since
they
will
have
run
in
the
lower
layer
with
the
ones
that
we
added
from
the
plain
resource
set,
and
then
it
applies
the
common
labels.
A
So
you
have
to
do
all
these
transformations
at
the
higher
level,
because,
prior
to
that,
you
don't
have
the
results
from
the
transformers
available.
So
that's
that's
how
that
one
works
and
in
composition?
It's
a
lot
more
elegant.
You
have
a
more
traditional
structure
with
fewer
layers.
A
Your
composition.yaml
is
able
to
pull
in
both
resources
and
your
transformer
config
like
it
actually
treats
everything
all
the
different
transformers
built-in
and
not
on
the
same
level,
and
you
define
your
own
ordering
with
that
and
then
at
the
higher
level,
composition,
you're
still
able
at
that
stage
to
override
stuff
in
the
transformer
config,
because
nothing
has
been
executed.
Yet
it
has
a
different
execution
model
than
customization.
A
So
you
pull
in
the
you
pull
in
the
base
modules
you
can
override.
Whatever
you
want
with
them,
you
can
add
new
ones
that
you
hadn't
had
before,
like
for
a
specific
environment
and
yeah
it's
at
the
top
level
that
it
will
finally
get
executed,
whatever
you
target
with
customized
build
so
that
works
that
works
differently
anyway.
A
I
thought
I'd
go
over
that
since,
since
it
was
super
relevant
and
like
it's
not
popularly
understood
so
like
that's
the
solution
that
that
will
eventually
work
for
a
use
case
like
this,
and
that
is
something
that
get
the
the
current
version
of
the
weird
customization
tree
to
transform
transformers
is
something
they
can
try
today
if
they
really
want
to
do
it
right
now,.
C
C
C
C
And
is
it
triage
supporter
kind,
support
kind,
sport.
A
Yeah,
oh
and
this
I
didn't
see
this
example
here,
but
that's
correct
like
if
you,
if
you
start
with,
I
don't
know.
If
the
example
there
is
valid,
but
if
you
start
with
your
home
charts
and
then
you
just
you,
have
the
resources
and
then
you
just
customize
native
mechanisms
to
patch
them.
That's
obviously
another
thing
that
you
can
do.
A
I
think
there
is
a
feature,
though:
where
did
I
see
that
that
allows
you
to
tag
the
resources
as
needing
a
suffix.
D
A
No
they're
they're
generating
the
config
map
from
their
their
own
generator.
I
think,
oh,
that
makes.
D
A
A
D
I
think
the
like
built-in,
like
configmap
generator,
uses
this
like
annotation,
like
it
checks
this
annotation
to
determine
if
the
resource
needs
a
hash.
So
if
they
said
it
manually,
I'm
not
sure
it
will
work.
I
don't
know
if
all
resources
go
through
this
code.
A
The
fact
that
we
have
this
example,
though,
suggests
that
it
was
intended
to
work
at
some
point.
A
We
can
ask
them
to
give
it
a
try
unless
we
don't
want
this
to
be
a
thing
that
we
support,
in
which
case
we
should
remove
the
example.
D
A
B
C
Sound
good
yep,
no,
I
didn't
have
a
triosh
label
triash.
I
guess
that's
needs
information
because
we
are
asking
them
to
respond,
and
this
is
kind
support
so
far
anyway,
because
it's
returned
to
an
existing
creature.
A
A
About
the
messaging
around
applied
versus
configured
or
configured
versus
unchanged,
let
me
see
if
I
can
pull
that
up
for
them,
but
this
is
definitely
not
an
issue.
C
A
D
I
was
thinking
the
same
thing
like
if
the
secret
is
already
generated
and
in
the
cluster
there's
nothing.
We
can
do.
A
C
A
D
Versus
this,
I
don't
think
the
secret
template
labels
occur
by
default.
I
think
they
have
to
use
field
specs
for
that.
D
D
Yeah,
that
makes
sense
also
that
pr
is
open
and
it's
on
us.
We
haven't
reviewed
it,
so
we
should
do
that.
Oh.
A
Can
you
ping
me
on
it
make
sure
it
gets
back
into
my
queue?
Simulation
has
escaped
yeah.
I
can
also
get
you
later
this
week.
C
Resources
the
resource
set
yeah,
whatever
that's
honest,.
A
Customize
can
generate
any
resource
kinds.
You
want,
I'm
not
really
sure
what
the
specific
trouble
is.
A
Yes,
customize
can
generate
well
generate.
I
guess
maybe
that's
the.
A
I
see
so
there.
The
problem
is
it's
like
a
question
how
to
make
this
a
a
dry
situation
without
writing
out
like
many
copies
of
what's
essentially
the
same
role
binding
with
one
field
different,
I
think
so.
A
D
I
think
it's
supposed
to,
but
it's
not
well
tested.
A
How
would
that
look
because
somehow
you
need
to
tell
customize
you
know
I
want
all
of
these
name
spaces
to
be
the
ones
that
are
populated,
so
I
think
you
would
still
need
sort
of
an
outrageous
number
of
layers
or
customize,
not
layers,
probably
parallel
customizations,
I'm
imagining
you
have
your
your
base
that
pulls
in
the.
A
Yeah
component,
that
has
the
cluster
wall
in
it,
but
it
pulls
in
that
component.
A
Noisy
you
would
need
like
n
customizations
that
all
have
namespace,
foo
or
whatever,
and
they
all
pull
in
that
components.
You
have
component
at
the
base
things
that
pull
in
that
have
namespace
whatever
they're
all
referencing
the
component,
and
then
you
would
have
another
customization
that
pulled
in
all
of
those
and
the
resources
that
they
all
share
so
that
the
role
binding
was
that
probably
work?
I
don't
think
it's
super
elegant,
but
that's
probably
the
only
possibility,
besides
like
a
custom,
generator
to
actually
take
in
some
more
declarative
parameters.
A
Describing
this
specific
situation,
I'm
gonna,
I'm
gonna,
try
that
to
see
if
it
even
works
at
all,
because
just
for
my
own
knowledge,
I
wanna
know
if
I'm
understanding
components
correctly,
because
I
haven't
worked
with
them
very
much.
So
I'm
gonna
give
that
a
try
and
if
it
works
I'll
say
that
or
the
you
know,
custom
generator
option.
A
Oh
actually,
someone
posted
in
the
chat
here
to
ask
a
favor
at
the
end
of
the
meeting.
We
only
have
three
minutes
left,
so
that's
now
and
yes
absolutely
what
what
do
you
need.
A
Yeah,
our,
if
you
can
maybe
ping
us
in
the
customer's
channel
in
kubernetes
slack,
so
that
we
know
how
to
get
in
touch
with
you,
we
can
find
you
something
to
work
on.
I
don't
know
if
any
of
the
issues
that
we
came
up
today,
like
sometimes
we
get
great
new
contributor
issues
on
during
this
bug
scrub,
but
today
it
was
mostly
things
that
needed
more
information,
so
I
don't
yeah
so
we'll
have
to
we'll
have
to
go,
find
you
something
else.
Does
that
work.
D
D
I
also
do
have
one
issue
that
might
be
good
for
a
first-time
contributor,
I'll
link
it
in
chat,
and
then
we
can
follow
up
on
slack.
Okay,
that's
even
better
thanks,
natasha.
A
Yeah,
it
seems
like
they
should
use
the
configmap
generator,
probably,
but
since
we're
pretty
much
out
of
time
here,
I'll
take
care
of
this
one
as
well
looking
to
see
if
that
is
actually
the
answer.
A
Since
they
have
duplicate
resources,
besides
that
or
you
know,
duplicate
resources
aren't
valid
server
side
either.
So
we
want
to
see
read
the
whole
thing
more
carefully.
All
right.
Thank
you.
Everyone
for
coming
today.
I've
got
through
a
lot
of
issues,
it's
great
to
see
new
faces
here,
and
I
hope
that
everyone
got
something
out
of
this
meeting.
Thank
you
so
much
for
joining
us.
We
have
a
meeting
every
wednesday
of
6th
july
at
this
time,
alternating
bug,
scrubs
and
like
sig
meetings.