►
From YouTube: Kubernetes SIG CLI 20210203 - 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
I'm
your
host
eddie
zaneski,
just
reminder
that
we
are
abiding
by
the
cncf
code
of
conduct
so
be
excellent
to
each
other,
and
with
that
let's
get
started.
A
I
also
do
want
to
give
a
quick
shout
out.
I
I've
been
struggling
with
some
depression
for
the
past
few
weeks,
so
just
thanks
to
everyone.
Who's
been
pulling
some
extra
weight,
especially
doug
and
team
and
y'all
are
awesome.
So
thanks
for
being
supportive
and
picking
up
my
slack,
but
I'm
feeling
better
now
good
to
be
back.
A
Okay,
so
nick
nick
had
a
pr
that
they
wanted
to
review
it.
I
think
brian,
you
already
approved
this
one.
A
C
Yes,
my
apologies
for
not
having
gotten
back
to
nick
nick
and
I
will
take
care
of
that
today.
C
Awesome
so
so
can
I
brian,
can
I
ask
you,
did
you
have
any
concerns
at
all
about
this
approach.
B
I
initially
had
a
concern
because
I
I
was
unsure
that
observed
generation
was
something
that
was
at
the
condition
level
and
not
just
at
the
status
level,
but
that
is
not
the
case.
It
is
at
both
the
status
currently
it's
at
both
the
status
and
the
condition
level.
So,
as
I
just
did,
I
didn't
want
to
like
introduce
a
new
new
convention
or
something
with
this,
but
I
don't
think
it.
I
think
this
is
fine.
B
A
A
D
I
mean
I
remember
what
it
is,
I'm
not
opposed
to
it,
but
it
is
work
that
someone
would
need
to
own
and
I
think
it's
it's
not
a
trivial
amount.
D
Today,
if
like
we
want
to
from
certain
commands,
have
you
know
set
like
a
part
like
it,
we
should
make
it
explicit
if
we're
going
to
make
it
like
machine
personable.
D
But
I
think
it's
I
think
it's
fine
for
us
to
like
explicitly
opt
in
to
say,
hey
like
this
command
should
have
you
know
an
output
format.
That's
machine
personal
like,
for
instance,
like
like
this
is
looks
like
it's
error.
Message
focus
but
like
an
example
of
where
this
would,
you
know,
make
sense
right
is
like
if
coup
control
get
like
it
emits
a
table
today
you
know
and
if,
if
we
wanted
to
make
that
table
like
in
json
or
something
so
that
it's
parsable
like
that
would
seem
reasonable
to
me.
A
Though
we
actually
have
a
pr
from
knight
42
that
adds
csv
output
support,
I
think
it's
just
waiting
for
a
final
approval
to
control.
Get
you
mean
yeah
yeah,
so
there
you
go.
I
looked
at
it
yesterday.
Actually.
A
Okay,
so
action
for
that
will
be
I'll
type,
something
up
a
little
broader.
That's
gonna
be
bigger
scope
that
we
need
to
think
about
at
some
point
right.
If
we're
gonna
figure
out
what
that
will
look
like
I'll,
probably
need
a
cap.
D
D
Right
because
like
it
might
I
expect
there
to
be
a
good
amount
of
back
and
forth
on
exactly
how
you
want
to
do
this,
and
if
someone
was
like
yeah
like
this
is
really
important
to
me.
I
don't
wanna
I'd,
say:
okay
like
well,
we
shouldn't
just
block
it.
We
should
enable
you
to
do
the
work.
D
My
biggest
concern
is
like
that.
It's
like
someone,
you
know
if
someone's
like.
I
would
like
this
feature.
D
It
would
be
nice
to
have
like
that's
not
going
to
turn
it
into
a
completed
product
right.
So
someone
has
to
be
like
really
committed
to
seeing
this
getting
done
and
and
then
I
think
we
can
have
a
conversation
with
it
at
that
point.
D
D
Yeah,
I
guess,
like
my
question,
was
like
rather
than
saying:
okay
like
here's,
that
protocol,
like
you,
can
just
exact
curl.
That
thing
I
guess
like
so
this
is
like
an
https
right,
but
then
like
what,
if
there's
like
you,
don't
want
it
to
be
public
and
you
want
it
to
be
through
like
something
that
requires
some
sort
of
you
know
authorization
to
get
the
coup
config
or
I
don't
know.
F
F
D
A
D
It
just
seems
like
it's:
it's
a
bit
weird
right
like
like
what
are
we
encouraging
you
to
do
right
like
if
it's
just
https
with
like
you,
can
just
hit
it
with
curl,
then
we're
encouraging
you
to
put
like
some,
like
plain
text
like
world
readable
file
with
your
credentials
for
your
cluster
on
the
internet?
D
Right,
I
I
don't
know
is
that,
like
that
seems
like
what
we're
encouraging
as
like
a
good
pattern
and
like
maybe
there's
like
if
it's
a
dev
cluster
and
you're
behind
sso,
I
I
don't
know,
are
behind
like
a
vpn,
maybe
there's
some
way.
You
could
do
this
and
feel
comfortable
about
posting.
Your
coupe
config
on
you
know
curlable
endpoint,
but,
like
I
I'd
imagine
like
there's,
probably
like
a
lot
of
ways
that,
like
you
could
do
this.
D
That
are
not
that
that
someone
might
want
right,
like
is
like
gcs
bucket
right,
would
be
an
example
where
yeah
at
least
like
it's,
you
know,
echoed
right
and
encrypted,
and
that
sort
of
thing
so,
like
you,
know,
remote,
remote
storage
and
like
how
would
you
get
that
right?
That
would
be
like
you'd
g
cloud.
D
Like
you
know,
pull
it
down
or
something
like
that,
so
those
are
the
patterns.
I'd
see
like
that
would
be
probably
more
canonical
and
a
little
more
secure
and
the
what's
being
proposed
wouldn't
support.
Those
like
we'd
have
to
implement
each
one
of
those
which
is
kind
of
why
I
was
suggesting
like
what
do
you
really
want
to
do
right,
like
I'm
not
like
if
we
were
to
pursue
this
like
if
this
is?
D
If
this
is
a
pattern,
that's
like
valid
and
we
wanted
to
pursue
this-
like
maybe
there's
like
a
middle
ground
between
the
friction
caused
by
exact
plug-ins
today
and
and
what
what's
being
proposed
here,
like
it
says
server
is
that
is
that
just
getting
the
server?
Sorry,
maybe
I
missed?
No,
I
am
this
sorry,
it's
kind
of
early
in
the
morning.
This
is
the
config
that's
being
pulled
down
from
ssh
right.
D
Yes,
so
the
you
know,
could
you
could
we
just
say
like
where,
wherever
like
we
put
https
where's
the
example
of
what
the.
D
Of
where
this
is
being
configured
to
be
pulled
down,.
D
F
D
Yeah
I
mean
why,
wouldn't
we
do
this
like?
Why
wouldn't
we
just
make
it
so
you
could
have
an
environment
variable
that
has
the
actual
contents
of
the
coupe
config
right
and
then,
when
your
shell
starts
up
you
just
like
populate
that
environment
variable
with
whatever
command
you
want
and
if
that's
a
curl,
then
that's
fine.
D
D
I
know
I
think
we
might
be
open
to
saying,
like
you
know,
maybe
making
a
little
simpler
to
populate
it
from
not
a
file
but
from
or
or
pull
the
config,
not
just
from
a
file,
but
maybe
from
an
environment
variable
or
something.
But
that
would
be
probably
the
most.
We
consider.
D
A
F
The
discussion
it
was
basically
yes,
it's
something
reasonable,
so
instead
of
this
is
mostly
for
for
watch.
If
you
start
watching,
the
watch
basically
does
initial
list
and
then
continues
from
a
particular
point
that
it
read,
and
someone
proposed
that
maybe
instead
of
doing
the
initial
list,
we
could
just
ask
for
a
watch
from
a
particular
version.
The
server
supports
it.
So
this
is
only
to
expose
this
functionality
in
the
get
seems
pretty
reasonable
and
it
shouldn't
be
even
that
hard
to
implement.
F
I
remember
that
someone
was
actually
reaching
out
to
me
with,
for
those
and
I've
noticed
that
we
almost
completed
the
only
one
that
we
had,
which
was
changing
the
create
commands
from
the
structured
generators
to
non-generators,
and
that's
almost
done
so
there's
nothing
that
we
could
point
new
people
to
work
on.
A
A
A
B
B
D
Yeah,
what's
the
concern
here,
it's
just
like
that
the
documentation
understand
that
how
to
run
the
command
or
is
it
the
documentation's
wrong?
I
just
think
it
was
a
user.
A
F
Probably
the
simplest,
the
simplest
thing
to
verify
this
thing
is:
ask
the
user
or
you
can
even
do
it
on
your
own,
to
invoke
the
command
with
verbosity,
7,
eight
or
nine.
I
can't
remember
which
one
is
which,
but
basically
at
one
of
the
highest
highest
levels,
you
will
get
the
request
as
a
curl,
that
you
can
literally
copy
paste
with
a
full
body
that
we
are
sending
to
the
server
and
you
can
verify
what
kind
of
fields
and
what
values
are
being
sent
to
the
server.
F
F
Quick,
something
tells
me
this
goes
back
to
the
to
the
issue
that
we
have
about
using
the
list
for
presenting
the
results
from
get
so
I'll
I'll,
probably
point
to
the
other
one
about
documenting
what
list
is
and
we're
not
gonna
we're
not
gonna
fix
that
one.
F
We're
not
gonna,
set
any
artificial
numbers
on
that
list,
because
the
list
being
used,
it's
only
as
a
container
to
present
the
data
to
users.
F
Yeah,
I
think
it
should
be
pretty
easy.
I
think
it's
doable.
Definitely,
although
it
the
change
will
be
server-side.
So
it's
something
that
we
will
expose
to
a
broader
set
of
users,
not
just
keep
cuddles.
F
I
think
the
the
triage
of
party
was
actually
suggesting
that
there
is
a
similar
issue.
If
you
go
back
to
the
triage
party
below.
F
F
A
A
A
A
I'll
see,
if
I
can
dig
it
up,
did
you
have
a
follow-up
for
the
annotation
one
mustang.
F
F
F
If
I
remember
correctly,
stefan
and
nikita
vendered
part
of
the
go
open
api
that
we
just
need
to
switch,
keep
cuddle
to
instead
of
the
current
one.
F
I
you
don't
have
to
tag
them,
that's
something
that
they
already
paying
me
for.
So
I
I
have
it
on
my
list.
I'm
I'm
hoping
to
be
able
to
pick
it
up
in
the
near
future
and
near
future
is
probably
this
month.
I'm
hoping
it
should
be
pretty
straightforward,
so
just
sign
that
one
to.
F
B
F
F
Yeah
it
should
yeah,
it
might
be
possible
because
I.
E
Think
I
I
sent
a
patch
to
fix
that
we
had
a
reg
regex
and
was
not
including
the
dash.
So
if
you
want
to
assign
it
to
me,
I
can
double
check
and
I
will
close
if
it's
already
sold
or
not.
F
F
So
maybe
we
need
merged
858.14
days
ago.
A
A
F
A
A
E
I
was
wondering
if
this
one's
a
real
bug,
because
it's
looks
like
it's
a
watch,
so
it
will
keep
pop
up
the
message.
So
I
wanna
hear
from
you
guys,
if
you
you
wanna,
just
show
one
message
or
you
know,
because
it's
a
watch
type,
it
will
keep
popping
up
the
message
to
us.
E
E
Know
it
basically
goes,
you
know,
so
you
set
wrap,
cut
three,
you
know
it.
You
show
up
like
a
row
out
finish
one
of
three
and
then
two
of
three
and
then
three
of
three
and
then
that's
it.
So
the
user
wanna
just
you
know
instead
of
pop
up
the
message,
but
I
think
that's
the
way
that
watching.
E
F
Yeah,
that's
definitely
correct,
because
what
we're
showing
is
we're
showing
information
about
updates
coming
from
the
server
the
fact
that
a
user
is
not
necessarily
seeing
a
change
doesn't
mean
that
the
replica
hasn't
changed
in
the
background,
so,
for
example,
between
the
rollout
to
finish
two
out
of
three
and
another
one,
for
example,
status
on
the
replica
might
have
changed.
If,
let's
say
the
replica
has
five
parts
running.
A
Yeah
doug,
can
you
close
that
one
with
the
working
as
intended?
Yes,
buddy!
Thank
you.
Okay,
that's
triage
party
mache,
any
follow-up
to
that
other
one
annotations.
F
I
can't
seem
to
find
sami
I'll
I'll
have
a
look
deeper.
I
I
need
to
find
in
the
code
where
we're
doing
that
one
and
I'll
leave
it
information,
whether
we
can
do
it
or
not.
Okay,.
F
A
F
A
All
right,
okay,
so
that's
that
repo,
we
don't
have
jeff.
Do
we
and
sean,
I
believe,
early
phil.
Is
there
anything
you
want
to
go
through
for
customize.
D
A
A
A
A
E
Recorded
yeah
I'm
looking
to
this
one,
I
think
that
one
is
real
and
the
only
the
only
concern
I
need
to
look
is
that
is
from
my
understanding.
Look
in
the
code.
E
And
that
he
has,
he
has
other
patches
related
to
this
topic,
so
I'm
watching.
A
A
A
F
E
If
I
remember
this
one,
he
wants
to
add
us
other
columns
that
are
not
available
like
he.
F
There
that's
weird,
because
I
would
expect
the
custom
columns
or
basically
any
anything
like
jsonpath
json
file,
gold
template
any
of
those
whatever
whatever
don't
bloody
printers
should
work
against
an
actual.
F
F
Oh
there's
another
there's
a
nasty
to
do
in
the
get-go,
with
my
name
on
it.
Currently,
we
don't
support
custom
calls
with
server-side
print
so
in
these
cases
for
the
force,
the
old
behavior,
but
that
basically
means
we
are
forcing
the
entire
resource
to
be
retrieved
and
it
should.
F
A
F
Yeah
he
will
be
working
towards
getting
the
debug
command
to
g8.
So
it's
a
very
good
feedback
for.
F
A
A
A
A
A
D
A
F
F
F
We
might
want
to
consider
bumping
that
a
little
bit
higher
for
cubecontrol.
I
know
that
we
did
that
for
for
the
openshift
client,
because,
with
the
increased
number
of
the
discovered
data,
it's
getting
pretty
congested.
F
F
Keep
the
discovery,
information
cache
from
rare
reading
discovery
information
more
frequently
I'll
look
into
it.
If
there
are.
A
A
A
A
D
Yeah,
the
that's
a
shell
feature
but
like
so
this
wouldn't
work
for
a
couple
of
reasons
like
one
like
what's
printed
here
like
yeah,
I
mean
what's
gonna
happen
is
the
shell
is
gonna,
expand
it,
and
this
can
provide
like
two
additional
arguments
so,
like
the
there's
a
workaround
around,
I
think
you
use
like
there's
a
creative
way
of
using
x
arcs
with
this
or
something
and
there
there
is
like
some
shell
magic
you
can
do
with
maybe
xrx.
D
D
D
D
Yeah
it'd
be
good
actually
to
have
some
like
standard
way
we
can
evaluate,
like
which
features
we
should
be.
D
You
know
accepting
contributions
for
based
on
the
amount
of
folks
wanting
to
see
them
right
like
because
right
now
it
does
seem
just
kind
of
like
if
we
think
it
makes
sense,
but
it
would
be
less
work
for
us
if,
like
we
had
some
standard,
you
know
definition
of
saying:
okay
like
let's,
let's
have
like
30
users
plus
one
this
thing,
and
then
then
we
can
didn't
make
sense
to
talk
about
seriously
or
something
like
that.
D
I
can
I
can
probably
throw
something,
quick
and
dirty
together
on
like
and
then
put
in
the
docs
to
say,
like
here.
Here's
how
you
get
here's
the
first
step
to
getting
a
new
feature
considered,
which
is
find
this
many
people
who
also
want
the
feature.
A
D
B
D
Go
ahead:
yeah
you
mean
a
control
plug-in
that
you
can
like
go
or
something
or
crew
installed.
B
D
Yeah
cool
yeah.
That
would
be
probably
a
good
thing
to
add
on
it
should
be
like.
I
guess
there
should
be
if
there's
a
plug-in
like
we'd,
consider
upstreaming
it
under
wet
conditions
as
well
yeah.
This.
D
It'd
be
great
if
we
could
get
usage
metrics
for
plugins,
I
I
guess
github
stars.
I
don't
know
I'll
run
that
by
chris.