►
From YouTube: Kubernetes SIG CLI 20210303 - 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,
good
morning,
good
evening,
good
afternoon,
depending
on
where
you
are
today
is
march
3rd-
and
this
is
another
of
our
monthly
six
cli
box
crop
meetings,
our
host
is
eddie,
so
eddie
take
it
away
from
here.
B
B
Okay,
so
let's
start
you
wanna
yeah
pr's
are
a
good
idea.
Do
you
have
do
you
want
to
go
through
the
dashboard
or
look
at
the
labels?
Maybe.
A
A
Yeah,
I
think,
there's
there's
basically
yeah,
there's
a
create
well
close,
there's
a
create
noteport.
I
think,
there's
something
missing:
there's
a
create
service
nor
noteport.
I
think
something
along
those
lines.
So
yeah.
B
A
Yeah,
I
should
probably
fix
them.
My
comment:
there's
a
cube:
keep
cattle
create
service
noteboard,
so
the
functionality
is
there
yeah,
I'm
trying
to
save
expose
from
from
from
when
we
ended
up
with
keep
cuddle
run
so.
A
B
A
B
Anyone
have
other
prs
to
look
at
brian
has
one.
I
got.
D
One:
it's
not
super
urgent,
it's
the!
When
you
do
coop
cuddle
version.
It
gives
you
a
a
warning.
If
the
version
skew
is
more
than
plus
or
minus
one
between
the
client.
D
A
Finally,
and
then
this
should
be
pretty
good
gotcha,
so
I'll
update
it
yeah
just
just
fix
that
one
thing
and
I'll
I'll
leave
I'll
leave
the
tab
open.
So
I'll
have
a
look
at
it
right
after
right,
after
it's
rebased.
B
A
Yeah,
we
we've
talked
about
it
so
many
times
that,
and
I
remember
that
I
wanted
to
get
it
in,
and
I
forgot
about
it
in
the
midst
of
other
issues,.
A
A
So
I
I
spoke
recently
with
the
the
thing
is
not
sure
if
that's
for
pod,
but
basically
the
the
problem
is
that
client-side
validation
for
the
built-ins,
mostly
work
works.
Okay,
the
problem
is,
with
the
with
the
crds
most
most
frequently,
where
the
the
logic
for
validating
how
we
are
using
on
the
client
side,
a
different
library
than
on
the
server
side.
A
I
spoke
shortly
with
with
antoine
last
week
and
with
stefan
about
it,
we
want
to
switch
the
open
api
implementation
that
we
currently
have
from
cube
open
api
to
go
open
the
api.
I
can't
remember
or
the
other
way
around
basically-
and
I
don't
know
I'll
if
I
make
it
for
121,
it
will
be
there.
If
not,
then.
A
A
Yeah,
basically,
just
like
antoine
said,
we
want
to
make
sure
that
the
client
translation
can
be
weaker,
but
we
can't
have
stricter
validation
on
the
client
side
than
the
server.
A
A
B
A
B
C
E
Yeah,
the
server
is
rejecting
that
they
seem
to
be
saying
like
in
general,
it's
okay!
If
the
if
the
server
the
client
doesn't
catch,
everything
that
the
server
considers.
C
E
A
Oh,
this
is
the
the
value
being
an
integer
where
it
should
be
a
string.
A
Yeah,
it's
a
valid
issue,
it's
something
that
we
could
look
into,
but
I
think
it
might
be
related
with
with
a
open
api
implementation,
because
that's
that's
where
we're
basing
the
validation
from.
A
B
B
B
A
A
A
Yeah
he
he
even
admitted
that
the
code
gets
messy,
which
basically
means
and
confirms
my
I'm
not
too
super
happy
about
adding.
A
A
I
would
want
to
see
this
to
go
through
the
process
that
phil
was
talking
last
time
as
in
get
sufficient
number
of
people
interested
in
interested
in
that
feature,
and
we
can
consider
that
one
until
then,
the
answer
is
no.
A
B
This
is
never
easy
to
find
awaiting.
B
B
That
on
this
repo,
no,
but
I
can
add
myself
as
a
right
owner.
B
A
B
B
Validate
certificate
format?
Okay,
so
I
actually
looked
at
this
one
and
I
implemented
it,
but
it
broke
all
of
the
tests
it
broke.
It
literally
broke
almost
every
test
because
of
how
a
lot
of
the
older
tests
were
written
that
weren't,
using
like
the
testing
factory
and
all
that
jazz.
So
this
would
take
a
bunch
of
tests
to
be
rewritten
to
implement.
B
A
Yeah,
I
think
it's
reasonable.
We
can
start
with
with
a
warning
and
open
a
open,
an
issue
with
help
wanted
where
we
could
get
someone
interested
parties
to
actually
fix
all
the
tests
that
are
failing.
B
A
A
I'm
betting
that
his
cube
config
is
this
big
that
the
loading
is
taking
significant
amount
of
time.
I
would
check
if,
and
that
would
be
probably
my
first
question:
have
you
tried
actually
how.
A
Hold
on,
can
you
scroll
back
a
little
bit
first
time
it
works
as
expected,
but
when
I
do
second
call
again
takes
more
than
40.
A
A
Yeah,
but
that's
only
for
for
the
discovery,
it's
possible
that
there
are
a
couple
things
in
place.
We
have
a
pretty.
A
A
B
What
is
this
in
spain?
This
is
in
spain,
okay,
they're
in
the
eu
time
zone.
I
let
me
see,
if
they're
on
slack,
because
this
would
be
easier
to
just
dig
in
together.
A
B
B
F
I
need
to
check.
He
says
he
says
that
if
he
uses
common,
it
doesn't
work
the
environment
variable
and
you
open
a
bug
in
the
control
need
and
it
become
a
shell
script.
F
So
I'm
I
was
wondering
if
it's
something
that
we
should
handle
or
not
but
looks
like
he
replied
and
he
wants
to
that
working
the
environment.
F
It's
about
yeah,
it's
it's
about
that!
He
wants
to
use
a
super
command
in
the
control
div.
B
B
F
I
will
try
to
solve
this
one
this
week.
If
you
guys
to
share
about
this,
one
will
be
nice.
F
Yeah
I
was
working
one,
so
that's
why
I
grabbed
this
one,
and
I
was
surprised
that
he
wants
to
use
us
to
be
comment
because
so
far
I
never
you.
I
never
saw
anyone
using
it.
A
D
Ago,
yeah,
I'm
trying
to
remember
who
worked
on
that.
I
thought
it
was
gabby
fisher,
but
it
wasn't
going
up.
No
okay.
That
name
doesn't
sound
like
it.
I
want
to
say
it
was
a
couple
months
ago
and
it
was
on
it
was
on
the
overall.
A
Yeah,
I
remember
no,
we
didn't
merge
that
one.
I
think
I
I
complained
on
on
something
on
wpr.
I
can't
remember
where
that
ended
up
being
yeah.
I
think
there
was
some
edge
cases
with
which
are.
A
That's
from
clayton
I'll,
take
it
I'll
I'll.
Take
it.
A
B
B
F
B
B
A
And
the
way
it
is
watching
rollout
happening,
so
I'm
not
sure
if
that's
feasible
and
now
that
quickly.
B
C
Yeah
yeah,
what
you
know
what
much
I
said
so
they
they're
getting
customized
into
coop
control
is
a
massive
pr.
That's
been
ongoing
with
feedback
for
the
last
couple
weeks.
C
I
had
a
quick
question
while
we're
looking
around
so
mache
and
brian
were
really
generous
in
giving
me
time
to
work,
to
look
at
the
kep
859
pr
to
implement
control,
headers
and
and
now
there's
I
had
to
when
when
the
ci
ran
again,
I
had
to
run
update
import
internal
imports
or
something
like
that,
and
it
changed
a
file,
a
gosum
file
in
code
generator,
which
is
outside
of
sig
cli,
and
then
I
had
to
update
another
one
which
had
like
27
other
go.
Some
changes.
Has
anybody
seen
that?
C
Is
anybody
seeing
that
recently.
A
It's
a,
I
think,
it's
a
today's
issue.
I
could
just
wait
then,
and
then
get
it
right
right.
I
was
poking.
I
was
poking
into
sig
release
channel.
It
seems
that
it's
related
with
with
a
pr
that
was
bumping
one
of
the
recent
dependencies.
C
That
time,
okay,
yeah
I'll
just
I'll,
just
sit
back
then
and
wait
to
see.
If,
once
it
resolves
itself.
A
Let
me
let
me
quickly
see
if
sig
release
managed
to
come
to
the
conclusions.
If
I
remember
correctly,
jordan
was
suggesting
that
we
need
to
have
some
test
image
be
rebuilt
with.
A
B
A
A
Hold
on
are
we
building
cube
cutoff
from
yeah.
Keep
cuddle
is
already
out
of
tree,
so
if
you
take
a
dependency
and
or
if
you
take
the
the
keep
cuddle
sources,
it
should
build.
B
A
B
A
B
A
B
Okay,
so
proxy
reject
paths
can't
be
empty.
There's
a
pr
for
this
one
awesome.
A
A
Talk
about
including
server
version,
client
and
server
version.
Well,
I
think
they
are
already.
D
A
B
The
other
thing
I
want
to
do
as
soon
as
we
can
build
cube
control
out
of
tree
is:
we
can
divert
all
cube
control
related
issues
to
that
repo.
So
we're
not
split
like
this.
A
I'm
I'm
even
inclined
to
do
it
right
away.
We
don't
even
have
to
because
we
we.
C
B
Ryan,
do
you
want
to
include
that
in
your
change
for
the
version
warnings
or
do
I
sorry
to
misunderstand,
are
we
not
making
a
change?
I'm
sorry
include
what,
in
the
morning
in
the
morning
directing
people
to
open
cube
control
related
issues
in
the
cube
control,
repo.
A
I
would
I
would
I
would
do
the
template
change
in
a
separate
vr
that
will
have
to
go
most
likely
through
for
different
approvers
and.
A
Basically,
it
would
require
you
to
go
through
the
entire
cluster
and
figure
out
that.
Maybe
you
wanted
something
else,
which
is
it's
simple
when
you
are
doing,
for
example,
if
you
make
a
typo
in
the
get
command,
for
example,
cobra
does
it
because
it
knows
all
the
possibilities,
so
it
suggests.
A
A
No,
it's
11
on
this
levenshtein
distance,
that's
what
it's
called
and
it
basically
looks
at
the
different
level
letter
variations
in
a
word.
A
But
you
have
to
build,
you
have
to
have
a
dictionary
as
an
input
right
and
it
it
basically
calculates
how
far
two
words
are
from
each
other
and
where,
when
they
and
you
specify
what
the
distance
is,
what's
the
accepted
distance
between
the
two
words
are,
and
it
will
propose
you
just
that
cobra
does
it
for
us
automatically.
If
I
remember
yeah,
if
you
do,
for
example,
keep
cuddle
kit,
it
will
say
that
git
is
unknown
command.
Did
you
mean
set,
get
edit
wait.
A
Oh
just
try
keep
cuddle,
get
get
in
your
cli
and
you'll
see
it
in
action.
A
Yeah,
so,
basically
to
make
it
work,
it
would
have
to
you
invoke
the
command
where
you
put
a
wrong
name,
space
upon
getting
an
error,
and
the
error
can
be
for
multiple
reasons,
because
you
might
not
have
the
the
necessary
access
rights
to
a
namespace,
which
is
one
possible
issue.
A
You
actually
made
a
typo
in
the
name
or
a
couple
others,
and
on
seeing
the
error,
we
would
have
to
build
a
dictionary
of
possible
values
which
basically
means
yeah.
We
would
have
to
have
a
list
of
all
possible
main
space
that
you
have
access
to,
and
then
we
would
be
trying
to
match
something
and
yeah.
It's
it's
not
something
that
we
want
to.
A
Do
I'm
not
sure
if
any
of
the
commands
that
I'm
aware
of
does
it
for
you
yeah.
B
A
A
Random
attacks
against
your
cluster,
because,
if
you
knew
the
difference,
you
could
basically
do
a
dictionary
search
against
all
possible
namespaces
and
you
can
build
what
the
cluster
looks
like.
This
is
similar
to
rejecting
your
login
attempt
with
saying
your
login
and
password
is
incorrect.
Instead
of
saying
that
your
login
or
password
is
incorrect,
which
is
just.
E
Yeah,
I
agree
it
that
has
both
security
and
scalability
issues,
even
though
it
sounds
on
the
surface
like
a
ux
nice
to
have
especially
like
it
almost
sounds
like
they
want
it
to
suggest
like.
Oh,
that
was
a
valid
pod
name,
but
that
part
is
actually
in
a
different
name
space,
which
would
be
even
worse
because
that'd
be
totally
not
feasible
at
all.
A
We
are
already
getting
we're
already
getting
blamed
for
privileged
flag
and
the
cube
color
run.
Young
jan
coldwater
is
having
a
blast
after
immersed.pr,
but
you
still
have
to
be
explicit
about
saying,
dash
dash
privilege
when
you're
invoking
kukato
run,
and
that
would
be
so
much
worse.
A
B
C
A
Yeah,
but
that's
that's
only
kyu
karodev.
What
I'm
talk,
what
what
eddie
is
asking
is.
C
A
Yeah
predefined
error
codes.
We
don't
have
anything
like
that
for
cube
cuddle,
so
it
would
have
to
be.
Someone
has
to
write
a
a
cap
and
then
we
would
slowly
apply
that
across
the
entire
cutter,
so
that
whenever
you
get
a
an
error
code
doesn't
matter
which
command
you're
working
with,
and
you
see
one
two,
five
whatever
it
is
clear
because
it
will
be
consistent
across
the
across
the
cube
cuddle.
A
A
It's
more
about
somebody
has
to
write
a
cap.
We
need
to
have
an
agreement,
how
it
will
look
like
and
then
the
implementation
it
will
take
a
good
couple
of
months
and
it
can
be
driven
by
multiple
people.
A
Similarly,
to
what
what's
happening
with
the
generators
in
the
create
world,
we
have
at
least
10
people,
if
I
remember
correctly,
working
on
getting
rid
of
the
generators
in
the
crate
and
I'm,
I
think
we're
almost
there.
Ricardo
can
prove
me
wrong,
but
I
think
we
tacked
last
pr
yesterday
is
that
correct.
G
A
Yeah
and
I
and
I
have
some
generator
related
and
dropping
run
unrelated
flags
which
somehow
connect
with
all
that
and
I
started
building
putting
together
apr,
but
I
didn't
finish
that
one
I'll
try
to
do
it
today
tomorrow
will
be
nice
too.
G
A
Yeah,
that's
basically
it
it's
just
about
putting
together
a
document
with
proposed
codes
that
will
apply
across
the
board
for
all
the
commands,
so
it
will.
It
will
require
a
little
bit
of
digging
through
all
the
commands
so
that
it
you
know
it
does
make
sense.
Then
we
will
have
to
agree
on
the
on
those
error
codes
and
if
we
get
the
agreement,
then
it'll
take
some
time
for
for
implementation.
A
Currently,
we
we
only
state
that
there
is
a
zero
for
success
and
non-zero
for
any
error,
so
it
won't
be
breaking
any
api,
but
we
would
have
the
ability
to
specify
a
little
bit
more
concretely
what
the
error
means
instead
of
just
saying.
Yes,
it's
an
error.
It's
not
an
error.
G
A
Okay,
thank
you
very
much
y'all.
As
usual,
it
was
very
productive
and
nice,
seeing
you
all.