►
From YouTube: Kubernetes SIG CLI 20201209 - 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
december
9th-
and
this
is
another
of
our
monthly
blocker
box
crops,
our
host
is
eddie,
so
I'll
pass
the
baton
to
eddie
to
run
the
show
for
us
today.
Take
it
away.
B
C
No
just
a
little
bit
of
an
announcement
if
anybody
any
questions.
I
was
here
to
answer
questions,
but
the
one
thing
I
want
to
mention
is
next
week:
we're
going
to
talk
about
open
api
integration
into
customize
and
also
provide
an
update
on
the
reintegration
of
the
cube.
Cuddle
plan
made
a
lot
of
progress
and
I'm
excited
to
talk
about
it.
Given
the
opportunity.
B
Okay
cool,
so
I
guess
I'll
just
share
my
screen,
which
I
need
the
baton.
Mr
schulich.
B
Okay,
so
first
one's
up,
we
talked
about
this
at
the
meeting
last
week
and
then
my
I
signed
you
to
respond
to
this
mache.
This
was
structured
log
output
for
cube
cuddle,
so
it
can
be
parsed.
B
A
B
B
Official
documentation
for
kind
list,
so
this
came
up
a
few
times.
I
brought
this
up
in
the
api
machinery
weekly
meeting
and
I
think
they
they
deflect
it
to
us
because
we're
the
ones
who
use
this
improperly
according
to
them,
and
then
we
need
to
decide
where
we
want
to
document
it.
A
A
We
allow
packing
several
different
kinds
of
resources,
so,
for
example,
if
you
request
spots
and
services,
the
pots
will
be
returned
as
pot
list
services
as
service
list,
but
we
can't
combine
both
into
a
consistent
list.
That's
why
we,
we
pack
everything
into
generic
lists,
and
this
way
we
can
get
away
with
having
single
lists
instead
of
several
there's
alternative,
which
would
be
that
we
would
have.
A
Yum
list
or
json
list
not
sure
if
that
will
be
something
we
should
do
or
the
biggest
worry
of
mine
is
that
it
might
be
a
breaking
change.
So
probably
it
would
be
reasonable.
B
So
sig
api
machinery
said
that
kind
list
is
an
internal
type
that
they
never
intended
to
be
exposed
and
we
are
using
it
for
the
reasons
mache
said
of
you
know
mixed
resources:
do
we
create
a
mixed
resource
list
and
use
that
and
document
it.
A
Or
I
think
that
that
goes
back
to
the
original
days
when
the
internal
and
external
types
were
mixed,
and
probably
we
never
fixed
this
properly.
Generally,
the
the
list
was
removed
from
the
majority
of
places.
Apparently,
this
is
where
it
was
left.
Maybe
converting
to
unstructured
could
be
a
possible
solution.
I'd
have
to
have
a
closer
look
a
maybe
we
need
a
long-term
solution,
but
the
quick
one
for
now
is
probably
to
document
that
this
is
what
keep
cuddle
get
will
will
return.
A
A
B
A
We
could
probably
some
I
don't
know,
introduce
flat
and
as
the
as
the
default
as
a
default,
false
we'll
keep
it
as
that
for
a
couple
of
releases,
then
we
will
switch
and
then
eventually
we'll
get
rid
of
the
old
output.
But
it
will
be
like,
like
usual,
extended
period
of
time
needed
to
perform.
That.
A
D
Does
it
make
sense
if
I
move
that
issue
over
to
this
repo?
Will
it
be
helpful
yeah,
that's.
A
That's
definitely
a
reasonable
okay,
but
which
is
here
talking
about.
D
D
Sure
so,
just
a
quick
introduction,
hi
guys,
I'm
sagar.
I
work
on
the
cappy
side
of
things.
This
is
my
first
six
cli
meeting
so
hi
thanks
for
having
this
one
and
for
having
me
here
so
quickly
about
this
issue
like
we
in
the
cappy
code
base,
actually
need
some
code
from
the
cube
ctl
drain
drain
side
of
things.
D
So
I
think,
probably
a
year
back
since,
like
this
code
wasn't
rendered
at
that
time,
we
kind
of
just
copy
pasted
the
drain
code
in
the
capi
code
base
and
kept
on
using
it
in
during
the
year
that
we
were
using
it.
We
made
a
change
wherein,
like
we
are
passing
a
context
to
to
a
patch
call
where
we
patch
the
status
of
the
node
before
we
try
to
cordon
or
uncoordinate.
D
So
now,
now
that
the
code
has
been
rendered,
I
kind
of
checked
that
this
context
that
we
expect
is
kind
of
missing,
so
does
it
make
sense
to
basically
introduce
a
context
which
can
be
used
while
updating
the
status
of
the
node.
So
this
is
kind
of
a
rough
pr
that
I
added
around
the
thoughts
that
I
had
and
I
had
a
slack
thread
in
the
channel,
but
I
think
I
had
it
on
a
friday
and
that's
why
probably
it
kind
of
did
not
work
so
yeah.
A
So
I
wouldn't,
I
wouldn't
move
the
the
issue.
You
already
have
a
pr
which,
from
a
quick
glance,
looks
okay
to
me.
I'll,
have
a
look
at
the
at
the
what's
the
usual
way,
with
with
adding
extra
method
with
context,
but
I'm
pretty
sure
that
what
you
did
here
is
exactly
what
we
should
have.
A
So
I'll.
Add
this
one
to
my
review
list
for
today
or
somewhere
towards
the
end
of
this
week,
and
I
I
think
it
should
be
pretty
straightforward
and
it'll
be
just.
A
B
B
A
B
B
F
Yeah
I
mean
I
can
I
mean
just
check
whether
the
comment
is
accurate,
yeah
yeah.
I
can
do
that
or
you
have
thoughts.
A
The
grace
period
seconds
is
then
passed
to
delete
and
if
I
remember
correctly,
we
are
trying
to
straighten
up
delete,
I'm
not
sure
if
we
have
a
pr,
but
there
was
a
discussion
some
time
ago.
Maybe
there's
even
pr
that
currently
our
delete
command
has
different
options
and
it's
very
confusing
and
it
maybe
it
was
discussed
during
one
of
the
six
cli
calls.
It
is
very
confusing
how
to
use
the
delete
command
properly.
A
Okay,
so
this
is
delete
options,
but
not
from
the
command,
but
the
server
side
delete
options.
A
B
G
They
provide
the
version
they
are
using
because
I
have
been
using
this
command
essentially-
and
I
don't
see
this
output,
at
least
in
my.
H
Slow,
so
so,
if,
in
fact,
that's
the
issue,
it's
not
clear
what
what
the
fix
could
be.
A
Yeah
hold
on
eddie.
If
you
look
at
the
command
what
it
actually
does,
it
requests
full
json
output.
If
you
recall
the
issue
with,
we
create
the
big
list
resource
and
then
we
throw
it
to
the
output.
A
H
H
Sorry
to
interrupt
so
in,
in
addition,
it's
impossible
to
do
any
any
sorting
of
the
objects
when
you're
streaming
right,
so
you
don't
know
which
ones
are
which
and
so
sorting
is
impossible.
Unless
you
see
all
of
the
items
before
you
send
them
back.
A
B
A
The
list
was
984.
yeah.
I
mentioned
that
in
up
above
wait,
984
yeah
984
is
the
official
documentation
on
the
use
it
kind
list.
Oh,
you
said
the
list,
one:
okay,
yeah,
the
last
one.
A
What
did
you
hear?
I
was
thinking
the
streaming
one.
Sorry,
the
streaming.
One
is
another
topic,
but
they
will
be
they.
They
are
interacting
with
them.
One
relies
on
the
other
one
heavily.
Okay,.
A
Yeah
actually
being
able
to
stream
resources
could
be
a
argument
for
getting
rid
of
the
list
as
well.
B
B
Type
that
up
this
one
is
976.
A
A
A
That's
a
looking
under
the
covers
of
how
keep
works
and
it's
fine,
but
we
don't
want
to
extend
cue
pedal
describe
indefinitely
and
pack
it
with
as
many
information
as
the.
A
A
Doesn't
it
yeah
because
it's
part
of
the
meta
you're
right?
But
if
I
remember
correctly,
the
code,
we
have
a
a
helper
that
prints
the
meta
information,
so
we,
I
doubt
we
will
be
printing
finalizers
anywhere
anywhere
else.
A
F
F
I
I
agree
that
it
shouldn't
be.
We
shouldn't
be
adding
more
things
to
describe
necessarily
for
specific
use
cases,
but
it
sounds
like
there's
some
underlying
problem
they're
trying
to
solve
for
themselves
by
looking
at
things-
and
I
wonder
if
I
don't
know
just
thinking
about
like
maybe
there's
some
missing
status
information
or
something
that
that.
A
A
Two
persistent
volumes
and
persistent
volume
claims.
These
are
the
only
two
that
print
finalizers,
for
I
don't
know
why,
but
they
do
and
I'm
inclined
to
remove
them.
B
Okay,
that's
definitely
where
I
saw
it,
then,
because
I
just
remember
trying
to
delete
name
spaces
and
it
was
stuck
because
of
volume,
so
it
also
looks
like
it
shows
up
under
the
the
full
get
yaml
which
is
expected
right.
Yes,.
A
A
But
that
that's
a
good
suggestion,
brian
I'll
I'll,
ask
them
what
kind
of
a
issue
they're
trying
to.
B
A
G
Can
you
can
you
assign
this
one
to
me
as
well?
I
can
I
can
take
this
one.
B
B
B
F
F
B
B
E
E
B
B
Okay
offer
a
silent
flag.
Okay,
so
I
I'm
on
this
one,
so
we
can
move
on,
I'm
just
waiting
for
the
user
to
get
back.
So
I
can
close
this.
We
talked
about
this
at
some
point.
I'll
push
them
towards
apply
yeah.
Well,
it
turns
out
helm,
has
the
flag
they're
looking
for
so,
hopefully
it
solves
everything.
B
B
A
When
we
do
watch,
we
first
list
the
resources
and
then
we
continue
watching
for
updates
to
resource.
Currently
we
always
start
with
a
zero,
so
you
first
get
resources
and
then
only
start
watching.
A
E
A
B
A
B
B
A
B
B
A
That's
a
massive
token
yeah.
Whatever
he's
doing,
we
would
have
to
get
the.
A
Also,
I'm
surprised
that
he's
using
two
different
version
on
on
mac
he's
mentioning
118.8
on
linux
116.1.
A
It
would
be
good
if
he
could
provide
the
output
with
the
latest
120
on
both,
ideally
with
minus
v,
eight
or
nine,
because
I
first
of
all
I'm
the
tokens
that
I'm
working
with
are
not
that
big
and
I
don't
have
an
ending
proxy.
A
B
B
A
A
Yeah,
especially
the
one
export
cube,
config
keep
config
or
he
could
halfway
through
the
post,
where
there's
keep
cuddle,
get
no
dash,
o
there's
an
export
beforehand,
so
either
he
has
to
export
the
keep
config
or
pass
the
cube
contact
manually
to
the
command.
It's
a
user
error,
not
a
bug.
A
A
So
there
are
a
couple
things
here:
yeah
first
one
is
the
user,
but
cube
cattle
doesn't
have
a
notion
of
a
user
per
se,
it's
always
connected
with
the
authentication
token
or
whatever
the
authentication
method
you're
using
and
that's
basically,
usually
based
in
the
cubeconfig,
the
part
where
he
mentions
inside
a
container.
A
H
Yeah,
it
has
nothing
to
do
with
options,
though,
like
an
options
for
a
command.
It's
just
straight
up
using
the.
H
B
A
B
B
A
That
too,
but
but
the
part
where
he
says
a
hundred
plus
users
of
clusters,
I
remember
that
justin
had
it
nicely
figured
out,
because
he
he
literally
is
saying
about
having
a
single
user
for
all.
But.
A
F
A
Yeah,
that's
that's
literally
what
what
justin
put
together
with
k,
where
he's
working
with
this
many
clusters
and
the
k
is
capable
of
figuring
out
which
cluster
you're
either
trying
to
work
with
based
on
the
parameters
you
pass.
A
A
I
I
looked
at
it
in
the
past.
It
might
be
a
duplicate
request.
The
reason
that
I
want,
I
don't
have
logs
for
cron
job-
is
that
with
the
current
way
cron
jobs
are
are
working.
Is
that
it's
hard
to
figure
out
first?
A
There
might
not
be
I'm
not
sure
if
you
want
to
have
the
current
crunch
up.
That
is
running
what
if
none
is
currently
running,
what
if
there
are
multiple
running
at
the
same
time,
because
you
allowed
multiple
concepts
being
run
and
so
forth
and
so
forth.
There's
this
many
edge
cases
and
this
many
options
that
no,
I
don't
want
to
do
it.
C
A
F
All
right,
so
that
was
the
last
in
the
triage
queue
hey
eddie
there
was.
There
was
an
email
about
one
that
that
got
that
about
closed
755
that
somebody
wanted
to
have
reopened.
A
Yeah
when
we
add
that
ian
coldwater
will
immediately
yell
at
us
for
it,
just
like
she
did
with
the
cute
cuddle
run
dash
dash
privileged.
A
I
wonder
if
their
insecure
skip,
delay.
Tls
verify
is
already
in
the
client.
Go
because
that's
basically
what
we
would
have
to
do,
we
would
have
to
pass
pass
it
all
all
the
way
down.
F
H
B
H
Just
send
it
to
the
api
server,
okay,
as
a
patch
with
a
different
content
type,
I
mean
that's
what
the
rest
api
looks
like
at
the
bottom.
You
see
the.
H
But
this
I
mean
I
I
I'm
not.
I
don't.
You
know
completely,
have
my
mind
wrapped
around
what
what
they're
asking,
but
it
does
feel,
certainly
like
something
very
dangerous
and
and
would
be
painful
to
to
try
to
get
through,
not
sure,
cycles
on
this.
A
To
answer
the
question
we
need
to
have
a
sick
api
machinery
sign
off.
I
can
already
tell
you
that
api
machinery
signed
it
off.
If
you
look
at
client,
go
tools,
client
command
api
types,
go,
there's
a
cluster
structure
which
has
insecure
skip
to
it
as
verifiable,
and
when
we
are
talking
with
the
server
we
properly
set
them,
and
if
I
remember
correctly,
we
just
sell
that
to
never
allow
that.
So
it's
a
matter
of
ex
with
us
exposing
that
option
as
simple
as
that.
A
Because,
like
I
said,
we
always
force
the
tls
verification.
A
A
The
server
sends
you
a
certificate
and
then
there
are
two
options:
either
you
have
a
the
ca
from
the
certificate
which
will
say.
Yes,
this
is
a
valid
certificate
or
if
you
don't
have
a
certificate,
you
basically
allow
to
use
the
insecure.
H
So
so
somehow
some
way
you
would
be
turning
off
the
the
off
handler
in
the.
A
H
A
H
So
it
in
the
bug
he's,
he
mentions
this
particular
option
and
claims
that
it's
not
it
doesn't
help
or
does
it
doesn't
fix
his
problem.
A
B
Yeah,
okay:
I
can
test
this
one
I'll
I'll
test
if
this
works
and
and
I'll
reply
or
reopen
or
whatever.
F
So
it
could
just
use
that
same
flag
that
already
exists
and
just
use
it
not
to
verify
I
mean
I
I
don't
know
if
we
don't
even
want
to
do
that,
I
don't.
I
certainly
don't
want
to
make
things
less
secure,
yeah,
maybe
it's
already
there.
We
can
just
use
it.
A
It's
already
there,
I
think,
he's
just
he.
He
looked
at
the
cube.
Cuddle
apply
help
which
shows
the
options
for
apply,
but
apparently
a
lot
of
people
are
not
aware
of
the
keep
cuddle
options
which
shows
you,
the
global
flags
yeah.
I
had
no
idea
this
existed
till
just
now.
So,
okay
I'll
start,
oh
I'll
start
taking
input
for
the
next,
keep
cuddle
tips
and
tricks.
A
All
right
I'll
test
that
with
the
soft
side,
sir,
but
basically
that's
that's
what
he
needs
to
do.
He
just
has
to
use
that
option
with
the
keep
cuddle
option,
because
he
thinks
that
this
is
only
the
server
side.
A
Yeah,
the
sir.
The
server
probably
has
the
same
option,
but
this
we're
talking
about
the
the
client
side
part:
it's
not
bypassing
the
the
authentication
on
the
server
that
parts
is
left
unchanged.
We're
only
ensure
we
are
only
removing
the
verification,
which
is
another
layer
of
security.
A
Obviously,
on
the
client
side,
even
if
the
client
side
will
be
able
passes
this,
it
will
go
through
any
certificate,
but
if
there's
authentication
on
the
server
that
still
will
will
apply
no
matter
what
you
do
on
the
client
side.
B
A
B
Ones-
okay,
we're
over
on
time,
so
I
don't
want
to
keep
anyone
who
has
to
leave
feel
free
to
jump
off,
but
someone
just
posted
this
in
slack
a
few
minutes
ago
that
they
have
an
open
pr
that
needs
eyes
on.
I
just
wanted
to
bring
it
up.
Looks
like
you
held
this
two
weeks
ago,
mache.
A
A
B
Cool
all
right,
well,
thanks
everyone
for
coming.
I
this
is
our
last
bug
scrub
of
the
year
so
see
you.
A
Before
before
holiday
so
see
you
around
next
week
and
have
a
good
one,
thank
you
all
bye,
bye,
thanks
team.