►
From YouTube: Kubernetes SIG CLI 20221109 - 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
If
you're
trying
to
talk
or
on
mute,
thank
you
Marley
hello
and
thank
you
for
joining
us
for
the
November,
9th
6cli
monthly
bug
scrub.
My
name
is
Eddie
Zane
I'll,
be
your
host
today,
quick
reminder
that
this
meeting
is
recorded
and
this
meeting
at
all
kubernetes
meetings,
abide
by
the
cncf
code
of
conduct,
so
please
be
excellent
to
each
other.
Do
we
have
any
folks
who
are
here
for
the
first
time
that
would
like
to
introduce
themselves.
A
B
C
C
C
Wait
so
it
thinks
something.
C
B
C
B
B
C
And
in
theory,
we
should
be
able
to
look
at
the
the
flags
in
Cobra
defined
in
Cobra
in
the
Cobra
command
and
say
oh
well,
namespace
is
a
is
a
value
flag.
Therefore,
I
know
that
it's
allowed
that
I
need
to
skip
over.
Whatever
is
the
next
token
in
so
it
seems
like
this
should
be
solvable.
That's
so
I'm
hesitant
to
just
like
dismiss
it.
D
D
Sorry,
but
what
if
the
select
is
Boolean,
that
is
the
problematic
part
we
cannot
understand.
It
is
belonged
to
the
flag
or
the
Standalone
real
plugin
name.
B
B
F
Yes,
that
was
based
on
doing
a
PR
review,
where
someone
was
applying
a
strategy
that
were
presented
a
few
loses
ago,
I
think
to
label.
They
were
adding
the
same
thing
for
annotations
and
I
found
a
piece
of
the
implementation,
so
I'm
not
fishy,
where
it
basically
wasn't
accommodating
for
the
fact
that
you
can
remove
an
ad
at
the
same
time
and
it's
producing
inconsistent
messaging.
If
you
did
that,
just
it
was
taking
the
length
of
the
object
overall
as
a
signal
of
what
happened,
which
is,
is
not
actually
reliable.
F
So
I
opened
this
issue
to
show
that
the
behavior
is
already
incorrect.
In
label
and
the
pr
author
is
actually
going
to
change
the
behavior
for
label
at
the
same
time
as
doing
something
better
for
annotate,
so
they
did
update
their
PR
and
I
I.
Think
that's
all
good.
They
are
actually
proposing,
instead
of
saying
or
actually
I,
think
I
I
propose
it
and
ended
up.
Agreeing
that,
like
maybe
we
shouldn't
be
trying
to
say,
is
labeled
and
unlabeled.
F
F
So
I
guess,
if
anyone
has
any
objection
to
that,
then
maybe
comment
on
the
pr.
D
I
I
I
I
totally
agree
because
the
court
itself
to
distinguish
the
label
and
unlabel
is
a
little
bit
tacky
and
I
I
would
be
very
happy
if
we
just
modified
it.
Foreign.
F
A
E
G
E
E
So
it
seems
reasonable
if
we
have
it
there,
I
don't
have
any
strong.
F
E
E
No,
because
that's
that's
not
part
of
the
yet,
if
I'm
sure
that
we
supported
it's
just,
it
will
be
no
matter
what
how
you
will
try
to
edit
those
fields.
They
will
be
always
rejected
by
the
server
I.
The
only
part,
I'm,
not
sure,
is
whether
there's
a
a
reject
or
a
silence
ignore.
If
you
try
to
modify
those.
F
Right,
so
what
would
be
the
value
of
actually
making
the
slide
work
if
it's
not
going
to
be
possible
to
actually
edit
the
fields
that
the
flag
wouldn't
enable
purpose.
E
Theoretically,
yes,
but
if
you're
editing
something
would
you
be
willing
to
have
two
windows
open
in
one
where
you'll
be
doing
a
get
and
another
where
you
would
be
doing
edit
or
if
there's
an
option
to
do
both
you
would
use
both
I
think.
That's
it's
more
about
the
the
fact
that
we
never
supported
the
show
Inc
and
the
showing
of
The
manage
fields
in
edit
just
tripled
when
we
added
this
capability
into
get.
C
Yeah,
it's
kind
of
a
common
problem
where
I
think
we
we
see
some
flags
that
are
common
across
commands
and
we
bring
in
a
whole
bunch
of
unrelated
flags
that
we
don't
necessarily
want
this
feels
like.
Maybe
that
is
that
is
that
what
you're
saying
Eddie
or
yeah.
C
Yeah
I
just
did
a
PR
that
got
rid
of
a
whole
bunch
of
flags
that
are
deprecated,
I
should
say
a
whole
bunch
of
flags
that
came
in
on
run
and
I
guess.
Unfortunately,
we
probably
do
the
same
thing
here.
If
we
wanted
to
get
rid
of
it,
deprecate
it
and
then
three
releases
later
take
it
out.
I
mean
it.
We
can't
get
rid
of
the
flag
where
it's
at,
because
it's
it's
being
used
by
some
other
command,
but
we
can
at
least
take
it
out
of
this
command.
A
A
B
C
Unless
so,
I
guess,
because
you're
right,
it's
not
going
to
break
a
CI
pipeline
or
something
unless
somebody
had
some
like
automated
or
tools
that
they
had
developed
that
that
used
edit.
And
so
it's
unlikely
that
somebody
would
be
using
this,
especially
if
it
doesn't
really
work,
and
it
is
interactive.
So
yeah
I
could
probably
remove
this
one
I'm
guessing.
A
Enrique,
do
you
want
to
take
a
look
at
this
one?
This
might
be
a
good
first
issue,
all
right,
Yes.
B
B
C
A
That's
what
exactly
what
it
is,
that's
exactly
what
I
think
is
the
problem.
Is
it's
not
in
this
person's
path,
and
it's
also
not
named
correctly
I.
Don't
think
the
binary
needs
to
be
named.
I
had
to
answer
this
for
somebody
the
other
day,
I
think
it
needs
to
be
named,
like
Cube
Kettle
underscore.
C
F
B
B
C
This
isn't
really
a
cube
cuddle
plug-in
though
right.
It's
a
client,
credential
I,
don't
know
what
the
right
term
is.
I,
don't
know
if
it's
a
plug-in.
The
word
plug-in
I
think
is
overloaded.
Here
is
that.
B
A
C
Yeah
that
could
get
really
confusing
too,
because
if
you're
like
I,
mean
you're
changing
your
directory
it
unless,
unless
somehow
you
have
that
directory
in
your
path,
I
think
I
think
your
answer
is
right:
it
needs
to
be
in
their
path
and
I.
Don't
know
I'm
curious,
why
they
think
they
need
to
do
it
that
way
like,
but
or
why
the
install
script
for
that
client,
credential
plugin
doesn't
put
it
in
their
path
for
them.
B
B
A
Guys
apply
what
oh,
my
God
Eddie
I'm,
not
sure
why
this
guy
expects
Json
to
not
be
a
part
of
this
foreign.
G
F
F
C
B
C
Do
you
think
that
it's
so
the
error
message
makes
it
sound
like
it's
trying
to
read
their
git
keep
file
and
it's
saying
I
recognize
I,
don't
know
I,
guess
I'm
just
trying
to
interpret
this
differently,
but
I
I
think
you're
right.
It's
probably
working
as
intended.
I
mean,
but
should
the
error
message
say
no
files
found
in
in
this
directory,
but
it's
not
really
no
files.
It's
just
I
think
the
error
message
is
probably
fine.
It's
it's
saying:
hey
I
couldn't
find
any
files
that
I
care
about
in
this
directory.
A
C
A
bit
logged
in
it's
that
error
message:
it's
the
one
in
the
it
says,
error,
reading
directory
name
that
I
have
and
then
it
says,
recognized
file
extensions
are
Json,
yaml,
yaml,
okay,.
B
A
F
Jason's
definitely
explicitly
supported
it's
right
in
the
in
the
main
reference
stock.
It's
actually
the
first
example
weirdly
enough.
B
C
I
mean
I,
guess
the
error
message
could
be
improved
a
little
and
say
couldn't
find
any
files
to
process
or
something
like
that
instead
of
saying
I
mean
because
it's
not
really
that
we,
we
couldn't
read
the
directory,
it's
that
we
read
it
and
there
wasn't
anything
due.
C
B
A
Yeah
I'm
kind
of
curious.
If
what
would
happen
if
there
was
a
Json
file,
but
it
wasn't
like
one
of
our
Json
files
in
there
or
a
yaml
file,
because
I
could
see
that
I
mean
at.
A
C
B
F
A
B
A
B
F
F
B
A
B
One
later
but
foreign.
F
So
roulette
restart
is
really
really
naive.
It.
What
it's
doing
is
adding
a
time
stamp
somewhere
in
the
and
I
think
maybe
environment
variables,
if
I
remember
correctly,
and
that
causes
the
spec
to
be
new,
which
just
triggers
a
rollout.
It's
we're
not
actually
doing
anything
mechanical
as
part
of
the
role
it
restarts.
So
it's
going
to
follow
the
same
strategy
that
would
be
used
for
any
other
rollout.
So
this
feature
request
wouldn't
actually
be
a
rollout
restart
feature.
It
would
be
a
workload
API
feature.
A
Also
fun
fact:
the
show
managed
Fields
flag
is
present
on
rollout,
restart.
B
E
E
A
A
control
plane
to
follow
the
final
rules.
These
stocks,
you
should
address.
F
C
A
B
B
F
There's
so
many
like
I,
don't
know
if
we
have
one
that's
actually
canonical,
but
they're
too
open
like
two
new
ones
right
now,
but
it
looks
a
bit
about
this
and
then
any
number
of
of
closed.
C
Yeah
there's
kind
of
two
cases:
I
mean
one
of
them
is
like
I'm
doing,
crew
cuddle,
logs
and
I
totally
lose
my
network
connection
temporarily,
and
then
it
comes
back,
but
the
things
that
I'm
logging
are
still
there.
But
then
there's
the
other
case
of
I'm
logging.
Some
things
and
the
Pod
goes
away
and
comes
back
like
or
goes
away,
and
maybe
another
one
comes
up
in
its
place
and
then
I
don't
know
how
or
like
I'm
I'm
can
you
do
coupon
logs,
like
on
a
on
a
on
a
deployment
itself,
yeah
yeah.
E
We
would
deployment
you
would
have
to
set
up
the
entire
connection
afresh,
because
we
are
basically
opening
a
connection
to
cubelet
to
a
particular
pod
on
a
particular
node
which
is
basically
done
by
the
API
server,
because
you
are
requesting
logs,
so
soft
endpoint
sub
resource
of
a
particular
pod.
E
E
We
run
the
entire
run
log,
because
we
need
we
would
need
to
read
the
current
deployment
state
figure
out.
What's
the
current
first
container
and
only
that
stream
over
to
that
one.
E
A
D
This
is
a
deep
collect
request,
combining
with
other
flags
I
didn't
close
it,
but
maybe
because
I
might
receive
something
and
it's
beautiful.
C
D
C
A
A
A
F
Yeah
they're
asking
for
one
that's
going
to
be
aware
of
what
the
server
can
do.
I
think
I
mean
I,
understand
why
they
don't
want
to
contact
a
server,
but
ultimately,
what
they
want
is
100
accuracy
which
would
require
full
duplication
of
what
is
actually
going
like
they're,
never
going
to
get
that
full
accuracy
without
actually
talking
to
the
server
right
and
I.
Think
we
we
don't
want
to
make
copies
of
the
logic
client
side.
Even
if
we
could.
G
G
B
B
B
A
A
C
A
Okay,
that
is
it
we
are
up
on
time.
Thank
you
all
for
coming
and
we'll
see
you
next
week,
we
got
through
a
lot
today
great
job.