►
From YouTube: Kubernetes SIG CLI 20200429 - 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
Actually,
today
the
house
will
be
Brian,
so
I'll
give
you
the
boys
go
ahead.
Brian.
B
B
Yeah
so
I
mean
I
went
through
and
looked
at
a
little
bit
so
I
mean
we
do
have
like
33,
open
books
in
cute
cuddle
and
then
I
also
looked
in
kubernetes,
which
I
know
people
aren't
supposed
to
open
them
there,
but
they
do
sometimes
and
just
for
reference.
We
have
71
there.
A
lot
of
those
are
kind
of
older,
though
so
I
guess
we'll
just
do
what
we
did
kind
of
last
time,
but
try
to
knock
out
some
more.
B
Yeah,
so
cute
cuddle
does
not
keep
cuddle,
get
all
does
not
list
all
resources
in
a
namespace.
So
this
this
one
has
been
opened
a
while
I
think,
but
it
really
it
sounds
like
going
down
to
the
end.
It
sounds
like
that
we
might
just
be
able
to
close
this
one.
It's
not
going
to
ever
return
all
of
the
literally
all
of
the
resources
it
sounds
like
so.
The
question
here
is:
should
we
just
go
ahead
and
close
it
yeah.
A
All
is,
is
Jessen
a
yes,
let
me
find,
which
one
is
that
that's
151
I'll
declare
a
comment
about
why
and
how
all
is
being.
It
is
a
little
bit
confusing.
That's
true,
because
all
is
just
an
alias
on
the
server
and
a
lot
of
people
think
that
actually
all
is
is
all,
but
it's
not
I'll,
leave
there
a
comment
and
I'll
close
it
accordingly.
A
A
It
is
not
definitely
hard-coded
on
n
cube
cuddle,
so
we
are
asking
whatever
the
server
returns
it.
It
also
gives
the
possibility
to
define
your
own
aliases.
It's
just
a
little
bit
confusing
for
a
lot
of
folks.
That
o
is
not
all
it
used
to
be
all
when
it
was
introduced,
but
it's
different.
That's
why
we're
not
going
to
fix
it,
and
there
were
several
issues
where
people
wanted
to
add
additional
resources
to
the
OL
alias.
A
B
B
Cue
cuddle
returns,
pods
in
various
states,
now
I
know
I,
don't
know.
If
we
covered
this
one
last
time
or
not,
I,
don't
AG.
Does
it
as
a
bug.
Ql
returns,
pause
and
bury
States
when
only
those
running
state
state
are
selected.
What's
happening
is
maybe
we
did
talk
about
this,
but
what's
happening
is
the
pot
is
actually
in
a
it's?
When
you
do
a
kook
I'll
get
pod,
it
says
that
it
is
terminating,
but
under
the
hood,
the
it
is
actually
still
running
while
it's
terminating.
A
I
had
a
PR
that
was
partially
expanding
on
this
one,
although
it
wasn't
for
terminating
I,
was
fixing
the
the
pre-staged
of
running.
So
until
all
of
the
containers
in
a
pod
are
running,
we
will
not
show
running
currently
I
think
I've
added
they're
waiting
or
something
like
that.
I
can't
remember
the
exact
state,
but
it
was
expended.
No,
it's
not
ready.
C
B
Right
this
one,
we
talked
to
I,
went
back
and
looked
at.
We
did
talk
about
this
one
last
time,
I
think
it
can
probably
be
closed.
We
were
going
to
reach
out
to
somebody
at
the
end
and
see
if
they,
if
they
would,
if
they
were
satisfied
that
it
was
fixed,
so
I,
don't
know
that
we
need
to
talk
about
this
one
today.
B
This
one
so
what's
happening
here
is
that
there
go
template
is
referencing,
something
that
doesn't
exist
so
in
this
case,
there's
no
certificate
authority
data,
even
though
the
go
tablets
referencing
it
and
what
comes
back
is
a
string.
This
angle
bracket
no
value
that
comes
back
and
I
looked
into
this
a
while
to
act
like
a
month
ago,
and
it
actually
is
inside,
go
template,
processing.
There's
this
line
here.
That
Spacely
says
so,
return
no
value.
A
This
is
a
default
behavior
and
just
close
it
I
don't
want
to
I,
don't
have
special
handling
for
no
value,
because
there
no
value
could
be
coming
as
an
actual
value
and
I.
Just
don't
think
it's
it's
worthy
to
have
special
handlink.
If
it's,
if
this
is
how
it's
being
returned
from
the
go
template,
that's
that's
perfectly
valid
answer.
C
B
Okay,
additional
printer
columns
for
CR
DS
doesn't
work.
This
is
a
kind
of
an
old
one,
but
let's
see
what
this
is.
Oh,
this
one
actually
I
was
yeah.
This
one
I
I
have
a
PR
for
I.
Looked
at
this
one
the
other
day,
so
I
don't
know
that
we
need
to
talk
about
this
other
than
unless
you
guys
want
to
I
guess
we
can
sold.
Look
at
it.
Okay,.
A
B
A
A
Return
codes
and
then
once
we
have
that
one
and
then
we
can
start
implementing
it
across
the
board.
I
just
don't
want
to
I,
don't
want
to
end
up
in
a
state
where
each
command
will
have
different
set
of
return
codes
and
whatnot.
Having
a
consistency
on
being
written
down
in
a
cap
would
just
make
it
simpler,
because
obviously
it
will
take
some
time
for
us
to
go
from
the
current
state
to
every
single
command
supporting
it,
but
every
time
somebody
will
be
asking
about
it,
we'll
just
point
yes.
This
is
this.
A
B
A
Okay,
a
general
cap
that
will
describe
a
set
of
reasonable
return
codes
that
we
can
sit
then
implement
across
the
board.
If
I
remember
correctly,
you
cuddle
off
can
I
convert.
That's
the
one
yeah
if
I
remember
correctly,
off
can
I
returns
nonzero
code?
If
you
don't
have
access
to
something,
so
it
does
implement
this,
but
I
think
that's
the
only
one.
C
D
D
B
D
A
A
Issue
if-
and
this
is
about
cubic-
don't
apply,
I'm
inclined
to
close
it,
because
the
path
currently
is
towards
service
I'd
apply.
A
B
I
think
what
they're
saying
is
that
they
tried
to
deploy,
or
they
try
to
apply
one
yeah
no
file,
but
then
the
prop
for
summary
and
I'm
not
sure
what
would
cause
this,
but
for
some
reason
a
different
deployment
was
causing
the
problem.
One
that
wasn't
even
in
that
ya
know
file
is
that
is
that
what
they're
saying
I.
A
A
Unnecessarily
I
think
the
problem
is
that,
first
of
all,
he
is
applying
a
pre-existing
deployment.
That's
why
you're
seeing
the
warning
that
cube,
gotta
apply
should
not
should
be
used
on
a
resource
created
by
create
or
with
safe,
config
or
apply,
which
would
create
the
special
annotation
with
the
current
state
of
the
resource.
So
when,
when
the
actual
apply
logic
kicks
in,
it
compares
nothing
against
the
current
state
of
the
deployment
that
is
embedded
in
the
file
and
it
notices
the
changes
which
basically
means
a
deployment.
A
A
A
B
So
so
probably
his
interpretation
of
the
problem,
he
said
nothing
in
the
error
message
indicates
the
prom
is
with
an
existing
deployment,
but
it's
not
really
that
the
existing
deployment
had
a
problem.
It's
it's
that
he
was
trying
to
do
something
that
was
invalid
and
he
got
an
error
message
that
so
yeah.
B
C
B
A
I
mean
was
I
right
person,
I
mean
it
looks
like
the
g-cloud
generated.
A
config
is
somewhat
different
and
that
it
doesn't
work
with
the
Burke
you
cuddle
it
should
we
don't
have
anything
in
the
cube
cuddle
that
relies
on
any
kind
of
G
cloud
or
any
kind
of
provider
they
unless
something
like
that
is
being
embedded
in
the
coccyx.
A
A
A
A
I
thought
that
was
exactly
what
I
was
loading
I'm,
pretty
sure
that
if
you
do
you
cuddle
get
F.
The
result
will
be
similar
to
that
k
because
you'll
get
a
list
of
resources
and
then
we'll
react
to
that
list
either
applying
them
one
by
one
or
getting
one
by
one
and
you'll
get
the
result
that
look.
That
will
look
like
this.
A
D
B
B
So
this
is
pointing
the
code,
so
I
don't
know
if
somebody
else
can
reproduce
it.
I
didn't
really
feel
comfortable
working
on
it,
since
I
couldn't
find
I
didn't
try
too
hard
I
guess,
but
I
didn't
easily
reproduce
it,
but
I
think
I
found
where
maybe
it
could
be
fixed.
I
don't
know.
Is
that
something
I
guess
we
want
to
probably
support
I,
don't
think
this
thing,
I,
don't
think
that
does
it.
A
B
A
Set
image
under
the
covers
knows
a
certain
set
of
resources.
We
are
using
sadiq
approximate
or
something
like
that.
Basically,
and
we
just
don't
support
coins
that
there
are
not
registered
and
by
default
in
Cuba
in
Cuba
kado,
so
set
image
will
not
work
and
similarly
a
couple
of
others.
But
the
image
is
as
long
as
annotations
are
simpler,
because
that's
part
of
the
method
that
I,
usually
I,
think
this
is
using
dynamic
line
set
image
is
not,
and
it
is
extracting
that
information
from
pod
spec.
D
C
D
A
C
Don't
so
we
could
change
that
image
to
just
continue
gracefully
if
it
doesn't
know
the
type
and
assume
that
it
doesn't
have
a
pots
back
start
to
clean.
Yes,
what
I
think
it
seems
like
the
wrong
way
to
be?
Thank
you.
There's
there's
a
lot
of
coop
control
commands
of
this
class,
which
word
I,
think
moving
away
from
people
using
anyway
I.
A
A
C
Like
three
or
four
things
going
on
there's
at
least
two
things
they're
here,
the
first
is
the
as
written.
This
is
not
actually
trying
to
set
image
on
the
server
it's
trying
to
set
image
on
local
thing
big,
so
this
command,
like
the
local
and
dry
run
yeah.
I
guess
command
is
not
really
optimized
for
that,
as
bar
likes,
Kayla's
uses
the
stop
resource,
so
I
do
I.
C
Think
most
of
these
commands
are
pouring
it
towards
modifying
that
resource
on
the
server
which
is
kind
of
an
anti-pattern
anyway,
but
maybe
good
for
development,
and
then
the
other
thing
is
it's
hard
coding.
You
know
what
where
can
set
the
image,
and
we
already
have
actually
libraries
like
this
falls
into
the
setter
stuff.
C
So
I
think
I
think
we
need
to
at
some
point
decide.
Thank
you.
If
we
want
like
how
much
we
want
to
go
forward
with
this
yeah
I
think
we
have
the
commands
available,
but
do
we
want
to
put
them
in
control
because,
as
like
as
a
replacement
for
this,
where
as
a
as
a
another
thing,
in
addition
to
this
right,
like
a
different
set
of
commands
for
manipulating
local
configuration,
we
could
do
that.
C
Do
we
wanted
to
tell
people
that
we
are
not
actively
increasing
the
surface
area
of
control
or
at
scope,
and
this
is
sort
of
fun
without
a
scope
and
it
works
because
it
happens
to
work
for
some
things,
but
doesn't
isn't
fully
supported
and
go
use
a
different
tool
or
developing
whatever
you
want
to
just
defer.
I.
C
A
C
Actually
so
the
setters
are
kind
of
you're
right,
but
it's
actually
different.
That's
the
repo
they're
in
but
they're,
actually
in
a
different
Gill
module,
so
the
setters
were
written
such
that
they
have
no
dependencies
on
anything
in
staging
like
I
was
pretty
careful
about
that.
So
that
means
we
can
actually
bender
them
as
cleanly
as
in
like
cobra
or
any
other
library
which
we
vendor
in.
You
want
happy
challenges
around
and
she's
in
there
and
well
they
happen
to
be
in
the
customized
repo.
C
A
So
maybe
we
could,
for
starters,
guide
them
how
to
use
customize
with
the
setters
to
solve
their
particular
use
case
right
away,
and
then,
in
the
long
run
we
definitely
need
to
have
a
conversation
around
set
image
and
probably
similar
settlements.
If
I
remember
correctly,
Dursleys
has
over.
C
C
A
Won't
walk
this
way,
so,
first
of
all,
I'm,
just
looking
to
it
scale,
does
not
have
dry
run
nor
local
right,
so
it
can
definitely
only
work.
So
there
are
I
think
there
are
couple
problems.
One
is
commands
such
as
set
and
the
ability
for
them
to
work
with
non
built-in
resources,
and
that's
one.
This
is
where
the
dynamic
commands
probably
comes
into
play,
and
then
there
is
the
other
problem
that
you
mentioned,
of
imperative
commands
where
they.
C
C
B
C
A
A
B
Right,
a
little
parameter
is
show
kind
combined
with
Oh
sometimes
is
ignored
this
one.
So
I
did
look
into
this
somewhere
far
enough
to
comment
or
not,
but
I
looked
into
this
and
basically
what
what
they're
doing
is
so
like.
Normally,
if
you
say,
show
kind
you
get
like
pod
slash,
but
if
you
do
Oh
name
show
kind,
you
probably
show
kind
fault.
This
OH
name
always
shows
kind.
I
think
is
the
problem
they're
saying
but
they're
they're
expecting
just
the
names,
but
they
can't
get
just
the
names.
A
That
means
that
there's
a
bug
and
name
printer
that
it
does
not
respect
that
it
doesn't
respect
the
show
kind,
flack
I'm,
pretty
sure
that
we
just
haven't
wired
that
bit
into
the
name
printer.
Maybe
we
could
that
that
shouldn't
be
a
big
problem
of
the
name.
Printer
is
part
of
the
CLA
runtime,
so
you
could
look
there
if
you're
interested
in
fixing
this,
let's
see
alright
run
time,
name
printer.
B
So
there's
another
another
part
to
this.
So
that's
I
totally
agree
with
that.
There's
another
part
where
they
have
another
another
after
this
line
here
where
it
says
they're,
trying
and
I
think
this
was
a
little
more
complicated
to
fix,
but
they're
trying
to
do
like
a
custom
column
of
metadata
name,
but
they
one.
A
That
won't
work,
custom
columns
with
show
kind
it
just
it
just
doesn't
make
sense,
because
we
don't
know
which
column
does
why
he
can
name
the
meta
that
the
name.
That's
my
name,
my
fancy
call
or
whatever,
and
we
don't
want
to
guess
this.
One
so
show
kind
with
custom.
Columns
just
doesn't
make
sense
the
first
one
yeah
we
just.
We
should
probably
fix
the
name
printer.