►
From YouTube: Kubernetes SIG CLI 20230412 - 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
Sig
CLI
customized
bug
scrub.
My
name
is
Katrina
very
my
pronouns.
Are
she
her
and
I'll
be
leading
the
bug
scrub
today
today
we
don't
have
that
many
issues
pendings
without
triage.
So
if
we
have
time
we
might
also
go
over
the
draft
roadmap
that
we've
been
working
on,
but
before
we
dive
in
I,
just
want
to
remind
everyone
that
this
meeting
is
subject
to
the
kubernetes
code
of
conduct,
which
in
short
means
be
nice
to
each
other.
A
A
A
Trying
to
prevent
import
to
prevent
Cycles
basically
and
you
can
override
it
on
the
command
line,
but
that's
not
recommended
so
here's
a
direct
link
to
the
issue.
Let's
see
what
this
one's
about.
A
Interesting
suggestion,
so
there's
only
two
modes.
What
I
was
referring
to
is
that
you
can
bypass
the
command
line
flag,
I'm,
just
just
not
use
the
load
restrictor.
So
it's
an
all
or
nothing
kind
of
policy
right
now
and
they're,
proposing
that
they'd
be
able
to
configure
it.
A
It's
interesting
because
I
think
having
a
blanket
policy
of
like
load,
restrictor
limited
or
something
like
that-
wouldn't
be
particularly
helpful
for
security,
because
you
don't
know
where
the
customization
directory
is.
Maybe
it
is
two
directories
done
from
something
you
really
don't
want
folks
to
be
accessing.
So
if
we
were
to
do
this,
I
think
it
would
have
to
be
not
a
certain
number,
but
a
specified
number
like
that
that
is
passed
in
as
part
of
the
butter
sugar
flag.
A
A
Yeah,
okay,
so
the
flag
is
just
like
Dash
load
restrictor
and
currently
takes
value
exactly
like
what
they've
pasted
here,
so
load
restrictions
and
I'm,
like
that's
the
actual
override
value
Natasha.
What
do
you
think
about
the
suggestion.
B
I
also
think
it's
interesting
I
guess
I
don't
have
a
strong
reason
against
it.
I
could
I
I
agree
with
what
you
said
that
it
should
be
a
specified
number
yeah.
B
Another
thought
that
I
just
had
was
for
customized
localized.
We
sort
of
have
a
similar
notion
or
you
can
specify
the
scope
directory
that
is
outside
of
the
customization
directory.
You
can
say,
don't
go
past
this
directory,
so
that's
another
option.
Instead
of
a
certain
number
specifying
a
particular
Parent
Directory.
A
Okay,
yeah,
that's
super
interesting
I
think
we
might
have
a
really
old
issue
so
I.
If
we
wanted
to
to
flip
that
same
idea,
you
could
say
the
scope
is
always
the
current
directory,
but
you
can
specify
which
customization
You
Want
to
Build
by
passing
a
path
to
the
customization.
A
A
Of
course,
everything
mentions
that
build
targets,
sure.
A
It's
just
not
used.
The
current
director
isn't
using
the
case.
Okay,
I,
really
like
that
slot
Natasha
that
we
had
that
analogous
concept
for
localized.
A
A
A
I
would
say,
even
if
we
accepted
this,
it
would
probably
be
priority.
Backlog
and
we'd.
Look
for
a
contributor
since
there's
they're,
not
volunteering,
would
you
agree.
C
C
A
A
And
they're
saying
that
name
change
works
with
the
container
name.
Change
doesn't!
Oh.
That
makes
sense,
because
the
name
change
you
don't
need
the
schema
to
figure
out
what
type
of
list
containers
is-
and
you
do
made
this
game
to
figure
out
this,
but
absolute
one
isn't
old
enough
that
we
wouldn't
be
picking
it
up.
Surely
and
5.0.1
I'm
going
to
try
to
reproduce
this.
A
Shouldn't
need
to
do
an
overlay
because
it
it's
just
a
regular
patch.
So
if
I
put
this
just
in
resources
and
then.
C
A
A
B
Am
a
little
confused,
I'm
trying
to
find
it,
but
the
merge
key
for
the
containers
list,
I
think,
is
the
name
in
which
case
isn't
this
expected.
A
Oh
I
see
because
aligning
change.
True
applies
for
the
resource
name,
not
the
container.
B
A
Like
we're
we're
using
a
really
old
version,
but
maybe
go
through
the
ones
that
were
dropped
in
116
like
a
really
really
long
time
ago,
yeah
you're
totally
right,
though
that
is
the
answer
to
this
question.
It's
that
renaming
containers
isn't
supported
and
the
reason
that
it
appears
to
work
in
these
old
versions
is
that
customize
doesn't
support
them
anymore
and
therefore,
all
the
containers
are
being
replaced
wholesale.
B
You
might
also
want
to
recommend
Replacements
or
a
Json
patch.
A
Yes
very
good
point:
the
alone
name
change,
true
option.
A
Would
somebody
be
able
to
look
up
when
extensions
view
on
beta1
was
dropped
from
supported
versions.
C
B
C
B
C
A
A
Google
that
looks
incorrect,
I'm,
not
sure.
If
what
they're
trying
to
do
will
work
anyway,
maybe
it
will
it
should,
because
it
will
let
us
I'm,
just
thinking
of
the
implementation
I
believe
all
we're
doing
is
taking
the
field
path,
identifying
the
target,
R,
node
and
then
replacing
the
value.
So
we
shouldn't
notice
there.
C
B
Expect
it
if
they
used
correct
syntax
but
they're,
trying
to
use
nested
list
elements
to
select
something
and
I,
don't
think
that
works.
A
Replacement
syntax
can
be
tricky
for
sure,
because
it's
it's
more
advanced
than
our
other
Transformers
really.
C
A
A
Because
okay
I
just
copied
in
exactly
what
they
have,
and
it
is
complaining
that
the
field
can't
be
found,
it's
not
actually
generating.
So
that's
good,
something
we
might
have
wanted
to
fix
if
it
actually
succeeded
with
incorrect
output
that
actually
could
be
related
to
the
change
that
we
made
in
5.0.
A
We
made
The
Replacements
Transformer
a
bit
more
strict
in
that
respect,
so
maybe
they're
using
a
four
point:
X
version.
So
if
I
just
change
back
at
autographs.
C
A
A
A
Her
context,
our
Windows
support
is
fairly
limited
because
we
don't
have
any
contributors
who
have
a
Windows
setup
and
we
don't,
we
haven't,
had
anyone
contribute
to
Windows
related
features
in
a
long
time,
we've
asked
sync
windows
for
participation
and
got
a
little
bit
of
interest
about,
haven't,
had
much
work
happen
in
this
area
in
a
long
time
so
yeah
we
have
a
bunch
of
problems,
including
that
a
whole
section
of
our
test,
Suite,
doesn't
run
against
windows.
So
we
unfortunately
don't
have
super
high
confidence
that
everything
works
as
expected.
A
So
we
definitely
need
a
lot
of
help
with
with
Windows,
and
if
anyone
works
on
windows,
then
there
are
plenty
of
issues,
opportunities
for
for
helping
customize
with
that
anyway,.
A
A
Yeah
I,
don't
think
we're
doing
it
on
purpose
for
specific
compression
purposes,
I'm
guessing
the
reason
that
it
is
the
way
it
is
is
based
on
the
release
tool
that
we're
using
we're
using
a
tool
that
just
generates
all
these
artifacts
for
us
and
yeah
there
they
are
so
that
is
configured
in
the
releasing
directory
here.
Is
it
Cloud
build
that?
Has
that
config
or.
C
A
This
this
looks
relevant
right,
like
I,
just
zip
on
Windows,
that
that
would
be
very
simple
if
that
works.
So
more
direct
link
to
this.
C
A
It's
a
little
bit
annoying
to
test
the
changes
to
the
release
pipeline,
but
if
it's
that
simple,
the
only
hang
up
that
might
happen
despite
this
existing
assuming
this
does
what
we
want
it
to
would
be
that
we're
running
a
really
old
version
of
go
releaser,
there's
an
issue
about
that.
A
They
we
have
a
multi
module
repo
and
that's
very
important-
to
be
able
to
release
through
all
the
modules,
and
that
became
a
premium
feature
in
some
version
and
they
changed
a
bunch
of
stuff
in
the
new
terms.
So
the
we
weren't
able
to
upgrade
and
yeah
it's
definitely
a
problem
anyway.
I
I
think
we
should
accept
this.
B
A
A
Yeah
I
think
we
could
use
some
help
with
this.
It
wouldn't
be
a
party
for
the
maintainers,
given
all
the
amount
we
have
on
our
plate
right
now,
but
it
seems
very
reasonable.
C
A
A
Oh
right,
you're,
not
a
GitHub
board
member,
yet
hopefully
we'll
get
that
settled.
C
A
A
So
that
is
all
for
the
bug
scrub.
We
were
thinking.
We
could
take
a
look
at
the
roadmap
if
you're
not
interested
in
that
feel
free
to
jump,
but
everyone's
welcome
to
stick
around
we're,
trying
out
a
new
format
for
this
this
year,
so
yeah.
Thank
you.
Thank
you
all
for
coming
and
I
guess
I'll
keep
the
recording
going.
It's
could
be
interesting
for
folks
who
are
interested
in
customers
Natasha.
Actually
do
you
want
to
direct
this
part
of
the
conversation.
A
Yeah
or
we're
talking
about
making
sure
that
the
priorities
seem
right
and
that
the
effort
seems
bright
so
that
it's
it's
ready
for
people
to
sort
of
pick
up
on
all
right,
I
think
it's
useful
context
for,
for
everyone
would
be
that
the
purpose
of
this
roadmap
is
to
help
us
align
with
contributors
and
potential
contributors
about
what
we
think
is
the
most
valuable
of
the
project,
because
customize
has
pretty
limited
bandwidth
at
this
point
in
time.
A
So
we
want
to
make
sure
that
folks,
who
are
open
to
helping
the
project
in
a
variety
of
ways,
have
a
way
to
see
what
will
actually
get
the
most
to
maintain
our
attention
and
new
to
this
year.
We
have
this
owner
column,
for
which
of
the
maintainers
and
reviewers
would
have
the
bandwidth
and
the
context
to
help
guide
a
contributor
who
is
interested
in
taking
on
one
of
these
efforts
that
we
think
are
really
really
important.
A
B
Yeah
awesome:
okay,
yeah,
we
can
get
started
so
the
first
one
Helm
support
long-term
plan.
This
one
is
about
the
fact
that
we
have
a
couple
of
built-in
Helm
fields
that
we
were
hoping
we
would
be
able
to
externalize
into
krm
functions.
The
krm
functions
is
its
own
huge
project
and
it's
still
an
alpha
with
a
whole
bunch
of
work
to
get
it
into
beta
and
GA
and
until
then
deprecating
the
built-in
home
Fields
doesn't
make
sense.
B
B
A
A
B
Yeah
prioritize
friction
with
migration
from
deprecated
fields
to
new
Fields,
we're
trying
to
get
ready
for
customization
V1
about
customization
object
itself,
it's
still
on
V1
beta1,
and
we
have
a
few
plans
for
fields.
We
want
to
deprecate
and
replace
with
new
Fields.
So
this
item
is
around
making
sure
that
the
new
fields
support
all
of
the
important
use
cases
that
the
old
field
supported
and
that
migration
is
easy
for
people
who
want
to
upgrade
yeah.
A
C
B
Well,
next
one,
so
this
one
is
related.
B
We
have
three
fields
that
handle
crds,
there's
the
open,
API
field,
the
crd
East
field
and
the
configurations
field
that
all
sort
of
do
the
same
thing,
but
slightly
differently,
and
it's
confusing.
So
when
we
reach
customization
V1,
we
want
to
have
a
good
story
around
handling
crds.
C
B
So
this
one,
we
actually
don't
think
that
an
external
contributor
will
be
able
to
help
efficiently,
so
we're
we're
planning
to
prioritize
it
ourselves.
B
If
someone
was
really
interested
and
willing
to
take
a
lot
of
time
to
understand
the
context
and
put
in
the
effort
that.
C
A
Yeah,
this
is
actually
related
to
the
comment
I
made
when
we're
looking
at
that
Windows
binary
issue.
There
is
actually
it's
probably
linked
here.
Let
me
see
yeah,
so
there
is
not
enough
Automation
in
it
and
it
actually
depends
on
GitHub
administrators
to
be
running
the
process,
which
is
not
great.
We
want
at
least
maintainers
to
all
be
able
to
do
it.
A
That's
the
same
list
of
people
at
this
point
in
time,
but
we
would
like
for
that
to
be
able
to
be
different
and
we
are
not
using
any
of
the
same
tooling
as
the
rest
of
the
kubernetes
project,
we're
using
outdated,
Go
releaser
version.
So
these
two
points
that
are
remaining
could
be
like
Alternatives.
A
A
Then
I
think
that,
like
MVP,
is
to
find
a
way
to
get
off
of
this
ancient
Go
releaser
version
and
that's
just
a
matter
of
investigating
what
the
possibilities
are
and
that's
something
that
we're
thinking
is
good
for
somebody
other
than
the
maintainer
team
to
take
on,
because
we
don't
actually
have
negative
any
context
on
the
say,
release
tools
or
on
other
tooling
other
than
exactly
what
we're
using
today.
So
that
that
research
would
be
would
be
new
to
us
as
well
and
yeah.
I.
A
Think
that
one
is
opportunity
for
high
impact
in
the
LA
in
the
long
run.
But
it's
not
as
urgent
or
as
impactful
to
end.
A
Three
I
think
it's
a
reasonable
rating
for
it
and
yeah
I
guess
probably
a
lot
of
work
so
take
that
makes
sense.
A
It
would
be
great
if
somebody
who
is
familiar
with
Sig
release
as
well
or
who
knows
folks
who
could
help
from
Sig
release
to
make
sure
that
they
make
good
decisions
there
to
take
it
on.
A
You
know
the
next
one
too.
That's
an
issue
that
I
opened.
We
are
as
maintainers
whenever
something
makes
a
PR
that's
related
to
dependencies.
A
We
have
to
go
and
manually
check,
kubernetes
kubernetes
to
make
sure
that
it
will
be
allowed
in,
because
we
don't
like
to
get
surprises
when
we
go
and
try
to
update
customize
and
keep
control,
and
then
we
actually
have
to
go
through
the
whole
release
again
with
the
dependency
change,
because
we're
trying
to
bring
in
something
that
they've
removed
or
that
they've
screened
out
or
even
just
the
wrong
version.
So
to
eliminate
that
sort
of
risk.
We
should
make
a
CI
check
that
actually
works.
A
The
same
way,
the
way
that
the
Upstream
one
that
we're
subject
to
works
and
that's
something
that
I
think
somebody
could
jump
in
and
help
us
with
fairly
clear
and
would
yeah
would
be
helpful.
A
It's
probably
medium
low
medium
effort.
I
could
like
go
either
way
on
that,
just
like
replicating
a
check
that
already
exists,
probably
a
fair
amount
of
code,
trivial.
A
A
Of
it,
it's
unfortunately
not
super
straightforward
to
fix.
There
have
been
a
number
of
attempted
fixes
for
specific
situations,
but
it
it's
certainly
a
painful
issue
for
our
users,
so
yeah,
it's
it's
difficult.
Definitely
at
least
medium
effort,
very
valuable,
yeah
yeah
this.
This
could
be
anywhere
from
medium
to
high
I'm,
not
sure.
A
The
next
one
is
a
KML
issue,
I
think
it's
probably
fairly
straightforward.
It's
I'm
actually
for
this
one
I'd
like
to
change
the
value
down
to
three,
it's
kind
of
a
niche
case
that
we've
never
handled,
but
that's
strictly
speaking,
we
should
it's
not
a
regression
or
anything.
It's
just
a
non-compliance
with
a
specification
that.
A
Ammo
is
a
little
tricky
to
to
work
in
it
and
partly
in
the
patching
code
in
particular,
it
can
take
a
while
to
get
your
head
wrapped
around,
but
so
that
would
be
where
the
effort
is
not
so
much
the
actual
implementation
of
the
patch
strategy,
I
think,
but
the
resulting
model
code
is
probably
going
to
be
pretty
low.
Thank.
C
A
Removing
quotes
is
very
similar
to
the
yeah.
It's
going
to
be
underlyingly
a
KML
issue,
most
certainly
as
well
similar
to
the
integer
problem.
So
it's
commonly
reported
issue
value
five
makes
sense
to
me.
A
Yeah,
all
of
these
that
I'm
torn
on
between
medium
and
High,
these
ones
that
are
sort
of
in
the
bowels
of
KML,
that
the
problem
is
originating
from.
B
I
think
it
might
require
some
some
scoping
to
give
it
an
accurate
effort
label.
A
I
guess
one
thing
so
we're
this
is
a
new
new
rating
system
that
we're
trying
out
this
year
and
it
seems
like
it,
it
might
be
very
effortful,
but
in
comparison
to
something
like
making
a
health
support
plan
and
standing
up
the
care
and
functions
registry
I
mean
that's
enormous,
and
it's
definitely
not
that
big,
so
maybe
maybe
I
should
think
of
it.
That
way,
instead,
in
which
case
medium
makes
total
sense.
We.
B
B
I
can
do
the
next
one,
so
the
yaml,
the
three
Fork
maintenance
story,
we're
currently
maintaining
an
internal
fork
in
the
customized
repo
of
the
go
yaml.
The
three
Library,
which
is
not
great
so
we're
in
the
middle
of
creating
an
official
Fork
under
the
Sig's
kubernetes
six
yaml
Library,
which
fingers
crossed
will
hopefully
be
done
in
the
next
couple
of
weeks,
and
then
we
can
get
rid
of
our
internal
Fork
I,
put
effort
as
medium
kind
of
taking
into
account
the
whole
thing.
B
We're
currently
embedding
one
single
version
of
the
kubernetes
open,
API
data,
so
it's
very
hard
to
retain
both
removed
resources
and
newly
added
ones.
But
if
we
could
embed
multiple
versions,
then
we
wouldn't
have
that
problem.
B
We
already
have
some
support
and
customize
code
for
multiple
versions.
I
think
the
challenge
would
mostly
be
talking
to
people
from
Coupe
cuddle
to
make
sure
that
they
would
accept
multiple
versions
of
the
open,
API
data.
Last
time
we
tried
they
didn't
want
to
vendor
in
multiple
large
files,
with
duplicate
data.
A
B
B
A
No,
that's
okay,
okay
yeah!
That
would
be
it
would
involve
like
a
less
straightforward
answer
to
what
version
of
the
open
API
is
this.
It
would
be
a
range
and
it
would
involve
some
sort
of
just
wrangling,
of
the
the
schema
documents
which
might
be
a
little
annoying.
But
I
think
this
is
a
kind
of
a
nice
independent
task
for
that
someone
to
pick
up
and
it
could
have
a
big
impact
on
our
ability
to
like.
We
get
a
lot
of
support
burden
from
people
trying
to
use
version
like
API.
B
That
would
be
awesome
if
someone
can
pick
that
up.
I
guess
the
scripts
are
currently
in
bash.
So
if
someone
wanted
to
do
that
and
Bash
that's
fine,
otherwise
they'd
have
to
convert
it.
Everything
to
some
other
language.
A
We
generally
only
use
bash
or
go,
but
we
do
have
a
precedent
for
tools
being
written
as
I
forget
where
exactly
the
repo
they
are
but
like
go
release.
They're
not
go
really,
sir.
Some
of
our
release
tools
are
like
packages
somewhere
in
our
repo
that
get
builds,
go,
go
mod,
something
what's
that.
A
One
that,
like
scrapes,
all
of
the
dozens
of
modules
that
we
have
in
like
adjust
their
dependencies
yeah
yeah.
That's
essentially
like
a
ghost
scripting
tool
that
that
is
built
from
our
own
repo
there's
a
handful
of
things
like
that.
This
could
be.
You
know
if
it's
more
complicated
than
we
should
be
doing
using
dashboard,
and
we
could
do
that.
B
Okay,
great
so
effort,
I
I,
put
medium.
It
might
be
high,
depending
on,
like
it'll,
be
a
lot
of
parsing
like
string
parsing
file,
parsing.
B
A
We
are
on
a
very
like
it's
becoming
increasingly
urgent,
I
guess
it
would
be
very
nice
to
have
this
done
sometime
this
year
because
we
so
we
just
checked
what
version
we're
embedding
and
it's
very
old
at
this
point
and.
C
B
B
There
are
some
specific
requirements
that
we
set
for
graduating
it
to
Beta
And.
There
are
also
some
additional,
like
user
feedback
user
requests,
that
we
should
take
a
look
at
as
part
of
that.
So
I
think
this
one
is
fairly
approachable.
Most
of
the
design
is
already
done.
It's
just
about
implementing
some
of
the
last
few
features.
B
A
And
Anna
is
looking
for
help
with
that.
A
One
minute
left
this
next
one
is
another
good,
like
anyone
can
can
help
with
with
this
problem,
it's
a
little
tricky
to
figure
out
how
to
improve.
B
A
But
our
error
messages
in
a
lot
of
cases
are
terrible
and
in
particular,
they're
terrible
because
they're
always
talking
about
this
long
path
to
the
root
directory
as
part
of
the
message,
and
that
might
be
totally
irrelevant
to
what
the
error
message
is,
and
it
really
clogs
it
up
and
makes
it
difficult
to
understand
and
yeah.
It
causes
a
great
deal
of
user
pain.
A
So
that's
an
interesting
small
project
to
figure
out
what
approach
we
could
take
to
being
better
about
that.
In
light
of
you
know,
there
is
a
reason
why
that's
happening
related
to
remote
resource,
Discovery,
I,
believe
not
knowing
exactly
what
the
problem
is.
So
we
need
to
teach
customize
to
be
better
aware
of
what
the
actual
problem
where
it
really
is
coming
from,
so
that
it
can
eliminate
that
part
of
the
error
message
when
it's
not
relevant,
but
that's
a
that's
a
nice
little
project
that
would
be
really
impactful.
B
A
A
We
could
see
if
there's
time,
to
continue
this
in
a
future
6cli
meeting,
because
the
bug
scrub
won't
happen
again
for
another
month.
Otherwise
we
could
see
about
like
scheduling
a
dedicated
meeting
to
go
over
it
together
to
have
time
to
finish
it
as
a
larger
group.
We
don't
want
to
leave
it
pending
for
too
long,
so
we
might
explore
other
options
like
sending
it
out
to
the
mailing
list.
Once
we're
happy
with
the
last
few
items.