►
From YouTube: Kubernetes SIG CLI 20220928 - 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
Okay,
welcome
everyone.
This
is
a
customized
monthly
bug
scrub,
so
we'll
be
going
through
the
customized
triage
board,
which
I'm
pasting
in
the
chat
and
trying
to
clean
out
some
of
these
old
issues.
I
have
this
in
order
of
when
issue
was
last
updated
from
oldest
to
newest,
so
hopefully
we
can
get
back
to
some
of
these
people.
A
So
we'll
start
with
this
one,
because
I
tried
the
first
one
this
morning,
configure
Json
patch
apply
options.
Can
everyone
see
the
issue
on
my
screen?
Yep
all
right,
cool
I'll,
also
paste
it
in
the
chat.
A
A
A
A
Okay,
thanks,
I'll,
just
type
that
up
and
since
we
have
other
customer
ization
fields
that
handle
the
use
case,
you've
described
and
we'd
like
to
limit
the
complexity
of
the
customization
file,
foreign.
A
If
you
have
a
use
case
that
is
not
currently
supported
by
any
current
customization
Fields,
please
feel
free
to
open
the
issue
with
PR
example.
A
A
A
A
A
Anna,
you
read
about
components
right.
Do
you
know
if
that
would
help
I
think
they're,
trying
to
use
Replacements
to
reference
resources
that
are
like
in
the
parent
customization
instead
of
within
the
base.
A
Yeah
I
that's
kind
of
I
was
thinking
too,
although
you
said
it
a
lot
more
eloquently
than
what
was
going
on
in
my
head.
So
do
you
mind
writing
up
what
you
just
said
on
this
issue
and
then
also
asking
for
like
any.
A
Any
other
thoughts
on
this,
or
should
we
move
on
not
hearing
anything
so
we'll
move
on
customize
edit
set
image
should
be
item
potent
this
sounds
very
familiar,
but
I
guess
we
never
commented
on
it
when
running
customize
edit
set
image
multiple
times
with
the
same
parameters.
The
result
is
multiple
image
blocks.
A
A
I
think
that,
even
if,
even
if
it
was
just
creating
the
one
list
because
of
the
way-
because
it's
not
like
before
it's
not
like
a
dictionary
and
so
yeah
keyword,
do
you
duplicate
on
so
it's
just
gonna
have
multiples
of
all
the
injuries
yeah,
that's
her,
but
yeah
that
doesn't
match
what
you
just
showed.
A
A
A
A
Yep,
that's
definitely
a
bug
so
I'll
accept
this.
A
A
A
Were
so,
there
were
four
people,
but
there
was
someone
else
here
before
so
they
must
have
came
and
then
left.
A
A
All
right
by
the
way,
Katrina
I'm,
sorting
by
last,
updated
in
case
you
wanted
to
follow
along
okay.
Next
one
add
an
extra
kind
of
patch
up
location
place
this
into
the
chat.
A
B
Interesting
so
they're
using
the
base
directory
not
as
an
actual
base,
but
as
a
common
like
a
shared
directory
that
is
allowed.
A
A
Oh,
this
sounds
like
a
really
similar
issue
to
another
one:
we've
just
looked
at
where
they
were
also
doing
like
Json
patch,
and
they
were
saying
this
doesn't
work
if
there
is
no
like
if
the
field
doesn't
exist,
yet
it
can't
create
it
which
I
closed,
because
there
were
like
other
other
ways
for
them
to
do
the
same
thing.
A
So
I
don't
know
if
you
were
here
for
that
issue,
I'll
repasted
in
the
chat.
It
sounds
like
a
very
similar
thing.
A
B
Right
I'll
go
looking
for
the
Jason
patch
spec,
but
maybe
we
can
send
that
because
that
that's
one
reason
in
my
mind,
like
a
big
reason,
that
we
should
never
Implement
our
own
bespoke
stuff.
On
top
of
that,
for
these
patches
I
wonder,
are
they
clear
at
all
on
why
they
wouldn't
use
SMP
in
this
particular
case.
A
A
But
they
do
also
mention
that
the
Json
library
that
we
use
supports
the
option
to
add
the
path
if
it
does
not
exist,
but
we're
just
not
using
that.
So
they
specifically
say
like
there's
a
line
of
code
where
we
use
apply
and
we
could
use
apply
with
options
instead
to
like
support
this
type
of
use
case.
B
A
What
it
sounds
like
I
mean
the
spec
obviously
is
up
to
interpretation,
but
it
does
say,
add
to
an
existing
object.
So
if
it
doesn't
exist,
then
it
sounds
like
you
can
kind
of
like
it's
not
really
specified.
A
A
A
A
B
Yeah,
that's
that's
good!
If
we
had
a
more
time,
then
I
we
could
even
try.
This
was
a
patch
that
we're
trying
to
apply
but
I
think
it's
more
valuable
to
move
on
to
the
bug,
scrub
and
leave
it
to
them
to
to
try
the
SMP,
because
they
are
asking
for
an
alternative
technology,
and
that
is
that
is
the
one
that
customized
already
supports.
B
A
A
A
So
update
allowed
to
use
the
new
label
format
with
a
flag
to
enable
disable,
selectors,
so
I
think
the.
B
A
Just
accept
it
unless
there
are
like
other
concerns
with
supporting
it.
B
And
it
looks
like
we
didn't
have
a
tracking
issue
just
appears,
like
you
said,
okay,.
A
A
B
I,
don't
I'm
gonna
try
to
look
for
the
issue
if
any,
it
seems
to
me.
We
might
have
just
even
discussed
this
when
you
were
implementing
it
and
said
like
keep
it
simple
and
see
if
it
actually
needs
this,
given
that
you
have
to
Target
a
specific
object,
yeah
the
point
that
you
can
already
Target
like
an
entire
kind,
I
I
think
it's
consistent
to
to
be
able
to
use
labels
for
filtering
at
least.
A
I
agree,
especially
like
we
throw
an
error
if
there's
more
than
one
source
regardless,
so
it
doesn't
really
matter
how
we
get
the
source.
A
Okay,
I
think
wait
can
accept
this
I
want
to
double
check
because
I
think
the
type
we
use
for
First
Source
does
actually
like
I.
Think
it's
a
selector
type
which
has
labels
on
it.
So
I
want
to
double
check,
but.
A
A
A
A
Fields
annotation
selector
and
the
label
selector.
B
No
I
was
just
still
trying
to
dig
up
the
annual
discussion
of
this,
and
the
only
thing
I
found
was
me
asking
like:
do.
We
want
to
restrict
selector,
more
or
less
or
and
Jeff
basically
saying
we
are
okay,
with
basically
accepting
a
wild
card
on
this,
but
we're
the
discussion
was
kind
of
oriented
around
this
particular
structure
that
we're
going
to
use
or
not
use.
So
it's
not
like
it.
It
wasn't
so
much
of
a
conversation
about
the
feature
per
se.
A
B
In
the
original
design
issue,
though,
the
the
type
is
shown
to
use
the
same
selection
fields
for
Target
and
Source.
If
that
says
anything
to
us,.
A
A
Okay,
I
think
that's
fine,
for
whatever
reason
when
I
implemented
it
I
have
like
separate
like
Source,
selector
and
Target
selector
types.
Maybe
that
was
just
an
accident
I,
don't
really
remember,
but
I'm
happy
just
making
them
consistent.
B
A
B
B
Interesting
because
that's
kind
of
like
the
opposite
of
what
we
do
with
remote
right,
remote
customizations
people
complain
about
the
opposite
because
we
have
they.
They
face
performance
issues
because
we
never
cache
it.
A
B
A
I
agree
and
I
think
for
remote
resources,
I
still
like
localized
being
the
solution
for
that,
like
customized
localized
way,
Anna's
working,
on
which
we
could
also
theoretically
extend
to
localizing
helm,
charts,
which
we
currently
stated
that
we
wouldn't
support
with
customized
localized.
But
we
could
eventually
do
that
like
handle.
All
of
this
like
remote,
like
downloading
remotes
and
customize
localized,
but
for
a
customized
build
state
that
it
shouldn't
like
locally
cache.
Anything.
A
It's
very
complicated
because
helm
plugin
uses
some
environment
variables
to
like
tell
you
where,
like
some
caches,
are
and
then
also
like
because
of
the
way
customized
build
is
currently
implemented,
it
has
its
like
local
cash
right
now
and
so
like
handling.
All
of
that
with
localizes
is
a
lot,
so
we
said
we
wouldn't
support
it,
at
least
in
Alpha
I.
B
A
It
does
do
it
because
it's
forked
off
of
customized
code,
at
least
the
one
we
have
right
now
it
does
it
for
like
exec,
because
it's
just
running
but
yeah
for
containers.
It
can't
possibly
do
that.
B
A
B
A
B
A
Update
the
version
customized
does
not
we
pull
the
chart,
and
then
you
are
stuck
with
the
old
version.
Unless
you
manually
delete
it
and
the
bug
scrub.
We
discussed
that
on
the
right
fix,
for
this
would
be
to
instead
download
the
remote
chart
to
a
temporary
directory
where
we
use
it
only
for
the
current
itself
and
clean
it
up
after
that
means.
Each
customized
build
will
download
the
chart
well
freshly
download
the
chart.
A
Okay,
hello,
customize
to
create
parameters
from
non-customized,
created
resources.
A
B
Right
I
think
this
came
up
before
and
there
was
some
useful
input
from
the
community
on
some
issue.
That
I
hope
will
be
able
to
dig
up
here,
but
about
what
they
were
doing
like
someone's
saying.
Oh
you
sealed
secrets,
so
I
do
this
so
I.
Do
that
because
yeah
they
shouldn't
have
to
commit
it.
What
they're
trying
to.
B
Get
the
secret
into.
We
can't
tell
what,
because
they've
only
got
the
target
Source
here
so
like,
if
they're
just
trying
to
get
a
data
field
into
say,
an
environment
variable
of
a
workload
which
is
the
classic
thing
to
do,
then
they
should
be
using
the
the
you
know
and
from
feature
not
Earth-like
yeah
and
from
Secret
in
their
workload
that
they
need
to
get
that
password
somewhere
arbitrary,
then
yeah.
That's
that's
going
to
be
difficult.
A
A
Just
one
best
practices
for
handling
secrets:
we
use
you
give
your
own
nice
response.
A
That's
so
good
I
think
I
saw
that
and
didn't
know
it
was
new,
so
I
thought
I'd
just
like
never
seen
it
before.
Okay.
A
A
A
A
B
Yeah
I
agree:
triage
needs,
information,
kind,
documentation.
A
A
B
A
A
A
B
I
think
I
wrote
a
partial
response
to
this.
If
you
wanna
take.
B
Yeah,
it
was
30
days
ago,
so
I
think
we
should
to
kind
support
as
resolved
and
close
it
and
ask
them
to
reopen.
If
we
haven't
actually
address
the
use
case
and
are
you
able
to
use
the
triage
labels
to
yourself,
I
haven't
tried
but
I'm,
assuming
no.
A
Yeah
we
can
talk
about
that
today.
You
might
need
to
be
a
maintainer
to
be
able
to
like
try
out
your
shoes,
so
we
can
figure
that
out.
A
We
are
saying,
oh,
if
Hannah
has
a
response,
doesn't
work
too.
Please
reopen
more
details.
All
right.
A
I'm
gonna
say
yes,
so
yeah
the
locker
removes
no
values
from
that.
I
think
we
looked
at
this
last
week,
so
I'm
gonna
skip
it.
A
Customize
localized
repetitive
base
is
an
issue
filed
by
Anna,
so
I
will
also
skip
it.
Placement
Transformer
does
not
trigger.
We
looked
at
this
one
last
week
or
last
month
or
two
didn't.
We.
B
A
Yeah,
it
looks
like
Katrina
you
assigned
this
to
yourself
last
time,
so
whatever
you
get
a
chance,
yeah
yeah,
okay,
create
a
common
customized
base.
A
B
A
Oh
okay,
we
can
just
end
this
here
and
revisit
this
issue
or
I
can
look
at
it
offline
all
right
thanks!
Everyone
I'll
go
ahead
and
stop
recording
thanks.