►
From YouTube: Kubernetes SIG CLI 20201014 - 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
Good
morning,
good
evening,
good
afternoon,
depending
on
where
you
are
today
is
october
14th-
and
this
is
another
of
our
monthly
buck
scrub
call.
Today,
our
host
is
eddie,
so
eddie
take
it
away
from
here.
B
Cheers
thanks
everyone
for
coming.
We
have
a
few
new
issues
to
work
through
I'll,
go
ahead
and
share
my
screen.
B
It's
not
my
first
rodeo
so
first
off
we
have
two
pull
requests
open
on
cube
ctl.
They
honestly
kind
of
look
a
little
spammy
someone
added
a
period
to
some
output.
So
we
just
need
to
comment
on
these.
Tell
them
to.
A
I
would
close
one
as
the
duplicate
of
the
other
and
then
probably
the
other
as
it
just
I'm
very
frequently,
also
rejecting
the
brs
that
are
fixing
the
static,
type,
checkers
or
govet
fixes,
because
a
lot
of
them
are
blindly
following
go
vet
rules
so
that
particular
dot
can
be
fixed
along
with
something
else.
B
B
A
B
A
B
B
Once
we
get
the
pr
faq
up
there,
hopefully
we
can
have
people,
stop
cool
okay,
matcha
uk,
responding
that
one
yeah
I'll
I'll
take
care
of
that
one
okay.
So
this
is
827.
C
C
So
in
the
code
it
is
possible
to
hook
hook
in
before
and
after
the
apply,
but
from
the
coop
cuddle
command
line.
It's
not.
I.
I
can't
think
of
a
way
that
unless
you
you
bash
it,
I
I
can't
think
of
a
way.
A
Yeah,
probably
if
you
want
to
build
something
on
top
of
cube
cuddle
apply,
you
can
reuse
the
code
that
we
have
and
especially
the
bits
that
sean
was
talking
about.
But
there
is
no
hooks
that
you
can
actually
inject
into
or
reuse
from
this
from
the
cli
itself.
Yeah
to
out
out
comment
it
as
such
and
and
close
it.
So
I'm
gonna.
C
Yeah,
I'm
gonna
put
one
more.
You
know
I'll
I'll
put
a
little
a
few
more
lines
on
here
and
close
it.
My
myself,
it
sounds
like
he
understand
I'll
I'll,
try
to
explain
it
better.
C
So
now
that
I
think
of
it
there
there
may
be,
you
can
run
a
command
before
kubecuttle
and
pipe
the
output
as
long
as
it's
config
into
cuddle
apply
I'll.
Put
that
in
there
as
well
and
customize
build
is
one
of
those
commands
that
you
could
run
to
create
the
config.
C
B
Awesome
thanks
sean
9,
43.,
cube
ctrl
run,
does
not
fail.
Oops
sorry
was
someone
saying
something
that.
B
Okay,
my
43,
when
you
pass
a
oh
okay
sal,
you
opened
this
one.
E
Yeah,
this
came
from
a
bug
that
was
open
against
oc,
but
I
realized
it
was
coming
from
cube
ctl,
so
there's
an
issue
that
was
open
five
years
ago
that
looks
like
it
should
have
been
fixed,
but
I
just
found
that
it.
This
behavior
is
still
there
yeah
there
you
go.
E
So
if
you
don't
pass
the
command,
if
you
don't
pass
the
name
of
the
pod
but
past
that
command
like
I
did
in
the
the
the
pod
will
get
the
name
sleep,
but
then
it
will
never
start
it's
it'll,
just
I
think
it
was
pending
or
I
opened
this
a
few
weeks
ago
yeah.
So
I
don't
remember
exactly,
but
it
is
still
there.
A
Double
hyphen
is
problematic
and
there
are
tricky
issues
around
how
we
treat
them,
how
you
have
to
count
the
arguments
et
cetera,
et
cetera,
if
there's
at
least
the
commands
that
have
options
to
use
this
and
without
the
double
hyphen.
If
I
remember
correctly,
exec
had
both
and
we
we
dropped.
A
I
deprecated
the
double
that
would
the
what
the
version
without
the
hyphens
and
specifically
require
the
hyphens,
but
we
will
have
to
wait
a
little
bit
longer.
It
might
be
the
same
with
the
run.
B
So
if
that's
the
case,
you
should
probably
I
just
dropped
a
an
open
pr
into
the
chat
that
messes
around
with
the
the
the
flag
ordering.
Would
you
be
able
to
just
test
and
see
if
you
can
reproduce
it
on
that?
It's
probably
not
related,
but
it
seems
similar.
A
I
it's
still
my
it's
still
in
my
cube
and
like
I'm
just
throwing.
C
A
It's
like
the
kids
being
pushed
for
the
past
couple
of
days,
every
single
time,
I'm
I'm
getting
into
work
in
the
morning,
and
I
I
have
a
plan.
The
plan
is
like
complete
to
destroy
other
things
and
like
yeah.
E
E
Yeah
yeah.
I
know
that
we
have
a.
We
have
an
open
bug
in
oc
for
that
that
I'm
working
on.
But
this
oh
remember
mate.
It
was
the
local
host
thing,
kubernetes
master
and
ver
defaults
to
localhost,
but
no
see
it
doesn't
yeah.
E
Yeah
and
the
one
that
the
one
that
I'm
working
on
is
in
cluster
config,
it's
it's
from
within
the
pod
that
that
impersonate
doesn't
work.
C
A
Finally,
someone
is
saying
that
if
it
changes
the
act
as
in
the
config
to
ass,
it
worked
just
fine,
but
the
types
in
the
in
the
api
are
pointing
to
act
as
I
will
need
to
check
whether
we
are
be
using
the
same
conflicts
or
we
have
a
different
ones.
It
might
be
related
to
daw1
and.
A
Not
and
then
we
should
probably
just
straighten
things
out
so
that
they
are
in
sync,
that
one
is
on
that.
E
E
We
should
invite
them
to
these
meetings.
C
C
So
I
think
he's
I
think
his
request
has
already
been.
You
know
already
exists
yeah
there.
It
is
coop
cuddle
external
diff
that
doug's
put
in
there
ten
days
ago.
C
A
A
C
That
this
already
exists.
I
think
it
and
I
think
it's
been
answered.
If
we
look
back
at
this,
the
fact
that
doug
has
created
a
pr
to
to
change
how
it
deals
with
args
is.
C
C
B
A
Oh
by
the
way,
eddie
did
you
turn
on
the
needs
triage
for
the
keep
qb
cuddle
repo.
B
B
Spec
I
feel
like
this
is
a
duplicate,
though
I
think
yeah
there.
A
C
That
type
of
a
change
is
may
not
be
feasible
at
this.
B
C
Should
we
actually
have
other
six
cliers
weigh
in
on
this
at
the
60
lie
meeting
just
to
get
more
perspectives,
find
out
how
it
looks
like
it's
something
that
a
lot
of
people
are
interested
in,
but
I
don't
see
how
we
can
change
the
default
location
without
breaking.
C
Basically,
everybody
we
give
everybody
the
opportunity
to
to
specify
you
know
we
give
them
functionality
so
that
they
can
move
it
to
the
directory
that
they
want,
even
if
they
do
have
to
specify
other
flags
or
put
it
in
the
environment
variable
you
know,
config
environment
variable,
but.
A
It's
doable,
it
would
require
a
an
enhancement
where
we
would
describe
the
it's
not
about
the
change.
A
change
is
probably
a
couple
lines
of
code
itself,
because
we
would
have
to
change
the
conflict
loading
mechanism
and,
like
I
said
that,
will
be
probably
a
couple
lines
of
code
where
we
just
inject.
Oh
look
also
at
the
that
other
directory
that
other
location,
but
the
fact
that
we
will
be
changing
the
defaults.
A
Work
to
announce
that
this
this
is
going
to
happen.
What
will
be
the
the
migration
migration
procedure?
How
to?
We
will
probably
start
with?
Oh,
let's
keep
the
current
default
behavior,
but
you
can
opt
into
the
new
behavior,
then
after
release
or
two,
and
we
will
probably
be
doing
some
warnings
and
then,
after
a
version
of
two,
we
would
allow
opt
out
of
the
new
mechanism
to
be
able
and
then
eventually,
after
a
couple
of
releases,
we
would
be
fully
switched
so
it's
doable.
C
So
the
one
of
the
problems
would
be
that
a
lot
of
people
are
are
creating
coop
config
files
and
putting
them
in
the
default
place
and
all
the
tooling
all
the
tooling,
which
expects
it
in
the
you
know
common.
So
I
I
think
it
might
be
a
little
bit
more
involved
than
I
think
it
might
be
quite
involved
actually
yeah.
A
Like
I
said
it's
more
more
announcing
more
evangelizing
than
the
actual
code
change,
the
code
change
is
very
simple.
If
I,
if
I
have
the
the
keep
config
loading
in
my
hat-
and
I
I
was
looking
at
it
some
time
ago-
and
I
haven't
opened
vr
for
for
the
order
that
are.
B
A
I
think
we
have
a
pr
for
that,
one,
that
if
there
are,
if
there
is
no
contact
set
yeah,
there's
open
yeah,
it's
kind
of
better.
I
think
I
I
have
it
on
my
list
to
look
at
or
there's
another
one
that
is
similar,
get
it
out
of
the
queue.
B
8.92
cube
ctl
describe
prefix
match
feature
okay.
This
is
by.
B
E
Yeah
yeah,
I
see
him
around
the
westford
office.
I
think
he
works
in
the
westford
we
office,
but
I
don't
know
what
timmy's
on.
A
C
B
A
C
Yeah,
I
could
see
a
case
for
removing
that,
based
on
you
know,
performance.
B
I'll
leave
that
for
now,
okay,
that
was
the
last
for
this
part
of
the
queue
some
of
these
already
prioritized.
C
We
had
talked
about
this
yeah
go
ahead,
sean
sorry
to
interrupt
yeah.
We
had
talked
about
this
and
this
is
about
making
the
code
more
reusable
by
pulling
out
the
cobra
command.
You
know
the
cli
specific
bits
out
of
the
options
so
that
the
options
are
clean.
You
could,
you
know,
create
the
options
and
then
call
the
complete
and
the
run
on
the
options,
and
it
ends
up
being.
C
You
know,
pretty
involved
for
every
single
one
of
the
commands
we've
kind
of
played
around
with
it
for
the
apply,
so
the
apply
has
in
its
options.
It
has
flags,
and
so,
if
you
want
to
reuse
the
apply
options
you
you
know,
you
have
to
fill
in
the
apply
options,
but
you
have
to
in
a
separate
area,
run
the
you
know,
parse
the
flags
and
then
manually
fill
in
the
options
you,
if
you
look
at
the
apply
options,
there's
some
ugly
stuff
about
like
there's.
C
C
This
is
actually
something
that
I
was
hoping
to
make
a
little
progress
on
in
the
near
future.
B
We
had
this
down
for
120
planning.
Was
there
any
prior
work
for
that
here?
It
is
yeah
separate
cobra
did.
Did
you
have
any
prior
work
that
you
we
could
point
in
reference
to.
C
C
So
I
I
think
that
there
was
a
a
pr
to
clean
up,
try
to
clean
up
or
start
cleaning
up
the
apply
options,
but
I
think
it
was
months
ago.
B
B
B
A
I
think
there
was
a
pr
already
some
time
ago,
but
that
was
sure
if
that's
still
there.
B
B
B
Okay
back
on
track,
so
all
right
I'll
respond
to
that
one.
We
didn't
have
anything
for
this
right.
Nothing
surfaced.
A
A
Find
I
didn't
find
anything
but
yeah
it's.
I
would
bump
the
priority
on
that
one.
It's
definitely
something
that
we
will
be
working
on.
B
B
D
B
Will
also
I'm
gonna
get
this
dashboard
the
config
cleaned
up
and
and
we'll
put
this
somewhere,
because
I
know
the
release
team
and
some
other
folks
are
starting
to
use
this
tool
as
well.
So
it
would
be
good
for
us
to
pull
all
the
other
issues
and
we
can
actually
figure
out
the
sections
we
want
in
here.
I
know
we've
been
saying
that
for
a
while
was
there
any.
A
Discussion
about
deploying
that
in
the
cncf
and
so
that
each
and
every
single
sig
could
have
their
own
conflicts.
B
I'll
talk
to
laurie
because
laura's
running
point
on
that
for
the
release
stuff
so
I'll
see
what
lori
is
thinking.
A
There
yeah
that
would
be
good,
because
I'm
if
there
are
more
six,
actually
reusing
that
functionality,
then
why
not
make
it
more.
A
E
A
E
Because
it's
because
there's
no
way
yeah,
I
have
a
pr,
I
think
it
even
merged.
Let
me
come
up
here.
H
Hey
doug,
hey
guys,
thank
you
for
inviting
me.
E
We're
solid
we
used
to
we
used
to
bug
sally
all
the
time.
A
A
Probably
I
would,
I
would
be
surprised
if
there
would,
if
that
wouldn't
be
failed.
If
I
remember
correctly,
folks,
from
the
storage
team
from
red
hat,
were
working
on
it
to
add
the
storage
class,
so
I
think
the
empty
is
because
it's
empty,
but
I
might
be
wrong.
Maybe
there's
a
bug
or
there's
also
the
option
that
some
time
ago,
when
we
had
a
discussion,
whether
we
want
to
show
empty
or
show
problems,
I
think
we
put
their.
B
A
A
Would
you
mind
also
tagging
this
as
triage
accepted?
Oh
yeah,
we
gotta
start
doing
that.
A
Edit
and
patch
are
doing
the
same
thing,
because
edit
does
a
patch
in
the
back,
I'm
not
sure
about
delete.
A
Delete
remove
something
directly,
but
that's.
B
F
B
B
F
A
So
this
is,
this
is
just
us
adding
a
test
it
it's
about
fixing
the
test.
Currently,
I'm
looking
at
it.
The
test
is
that
we
are
invoking
cube
cutter
version
and
we
are
ensuring
that
client
version
server
version.
Major
miner
did
comment
are
there,
but
without
any
particular
values.
On
the
other
hand,
this
ensures
that
we
do
print
them,
I'm
not
quite
sure
if
or
what
we
would
be
checking.
C
So
when
we
run
a
e
to
e
test
on
something
we've
just
built
and
we
don't
use
the
minus
x
values
to
fill
in
it,
you
know
there's
a
particular
there's
a
particular
thing
we
need
to
do
in
order
to
to
get
a
real
server
version
to
be
the
major
minor,
etc.
C
Usually,
when
you
build
locally
you're,
not
actually
adding
those
values-
and
that's
I
think
what's
going
on
here-
is
that
those
are
just
junk
values
does
does
any
of
that
make
sense.
I
think
we.
A
Do
inject
them,
even
if
you
build
locally
you're
gonna,
have
information
that
the
kid
comet
is
dirty,
because
you
have
a
local
modifications,
major
and
minor.
If
I
remember
correctly,
it
might
be
wrong.
I'm
just.
C
Checking
that
one,
unless
you
actually
do
a
minus
x
flag
when
you
build,
if
we
look
at
the
version
code,
there's
you
you
have
to
actually.
A
A
little
bit
ago,
so
we
do
print
miner,
as
20
plus
major
is
one
always
good
version
is
120
alpha
2,
which
is
basically
information,
there's
attack.
Then
there
is
probably
a
number
of
comments
between
the
tag
and
the
current
one.
There's
the
kid
comet
from
the
current
and
there's
also
the
good
tree
state,
whether
you
build
from
a
clean
or
dirty
meaning.
You
have
local
changes.
A
So
it
has
to
be
some
reasonable,
drag
x
for
each
of
the
items
I'll.
F
A
B
H
Yeah,
I
was
looking
some
bugs
to
see
what
I
can
help
and
I
notice
this
one
which
is
duplicated.
So
I
don't
know
what
you
guys
will
decide.
A
A
I'm
pretty
sure
we
have
it,
because
we've
added
that
one,
because
we
have
lots
of
pods
with
multiple
containers,
so
it
is
done
through
annotation.
A
Pick
and
that
one,
what
would
be
better,
though,
is
instead
of
having
a
very
targeted
annotation.
B
Call
okay,
I
assigned
you
to
that
dog.
Okay,
we
got
seven
minutes
left.
B
H
Yeah
looks
like
what
he
he
wants
is
a
red
done,
so
just
a
matter
of
styling
the
autocomplete
I
tested
here
exactly
what
he
wants.
A
Sometimes
the
autocompletion
takes
a
little
bit
of
time,
I'm
not
sure
if
people
are
aware-
and
it
will
depend
on
the
time
between
your
machine
and
the
cluster,
if
you're,
if
you're
waiting
for
pod
names,
completion.
C
B
A
H
A
H
Was
wondering
if
you
should
say
to
this
guy
like
if
you
want
to
do
it
to
yourself,
you
can
do
a
plugin,
it's
easy
to
do
a
plugin
for
yeah.
That's
the
other
option.
B
A
Yeah
I've
noticed
that
somebody
already
tagged
it,
I'm
not
sure
what
jordan,
who
has
the
power
to
approve
it
did
not,
but
I've
noticed
that
he
was
also
asking
for
a
backboard
to
119.
B
A
A
Driver
on
client
I've
noted
there's
a
apparently
there
is
a
bug
in
our
handling
dry
run
because
in
our
e2es
we
pass
dry
run
with
its
pace
and
the
value
is
then
and
apparently
the
the
way
we
parse
the
value
passed.
This
way,
it
complains
that
there
is
no
value.
So
if
I
do
try
run
slap
space
client
instead
of
equals
client,
it
will
complain
that
I
should
be
using
equal
client,
because
that's
the
new
format
and
not
the
old
one.
A
I
need
to
have
a
look
closer
look
at
it,
but
something's
fishy
with
this
one.