►
From YouTube: Kubernetes SIG CLI 20200219 - 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
B
Awesome
thanks
everyone
for
coming.
This
is
round
two.
Hopefully
we
can
I
think
we
got
through
close
to
a
hundred
issues.
Last
time,
which
was
awesome,
I
think
we
can
power
through
a
lot
of
these
I
think
we
have
some
new
folks
joining
us.
So
we'll
do
a
screen
share
and
we'll
run
through
a
bunch
of
these
open
issues
and
then
folks
feel
free
to
volunteer
step
up
speak
up.
You
know,
there's
a
lot.
There's
gonna
be
a
lot
of
good
things
for
people
to
get
involved
in
and
take
on.
So
let's
do
this.
B
Yeah
so
we'll
just
go
through
issue
by
issue.
What
do
we
got.
A
C
So
there's
I
think
this
is
part
of
a
class
of
issues
that
we
see
on
occasion
where,
where
there's
a
it's,
not
really
an
issue
with
coop
control
as
much
as
an
issue
of
how
the
the
entire
cluster
and
API
server
have
been
set
up.
I
don't
know
for
a
fact
about
this,
but
this
really
feels
like
you
know.
Somebody
set
up
their
cluster
API
server
in
a
way
that
was
difficult
for
them
to
connect
to.
C
B
B
C
B
One
of
the
things
that
we
definitely
realized
last
time
was,
we
shouldn't,
be
afraid
to
close
issues
and
I'm
talking
with
Eric
to
hopefully
get
the
Fate
about
in
here
to
close
out
the
stale
ones
anyway.
So
so
I
took
a
peek
at
this
one
and
users
saying
that
in
yet
you're
allowed
to
not
use
a
shebang.
You're
allowed
to
use
a
double
slash
to
like
de
notice
shebang
same
kind
of
thing.
I,
don't
think
this
is
something
we'd
ever
support.
A
Mean
yeah,
what
we
do
is
I
lives
in
the
and
in
the
Linux
version
we
just
call
an
exit
exit
fee
and
we
pass
the
primary
with
all
the
arguments,
everything
so
it's
up
to
the
partner
binary
how
to
treat
it.
We
are
not
parsing
this
in
any
way,
so
actually,
in
my
mind,
it
might
be
a
problem
from
the
from
the
system
and
not
from
the
cube
plugin
itself.
A
Yeah,
but
we
we've
had
numerous
conversation
around
this
topic:
I'm
not
harvest
insured
that
we
we
will
be
fixing
that
one.
There
were
different
approaches,
whether
you
pass
a
namespace
or
not
whether
you're
hosting
all
or
not,
I'm,
pretty
happy
with
the
current
state
of
world.
So
it's
definitely
to
be
close
as
a
lot
of
they're.
Okay,
Sean.
B
B
A
B
A
A
A
The
way
customizes
kinetic
bill
is
is
billed
and
the
fact
that
it
relies
on
some
of
their
libraries
that
are
exchanging.
We
have
circle
circle
dependencies
when
we're
trying
to
bring
in
your
version
of
customized,
because
it
relies
on
some
parts
of
kubernetes
and
cube
cut'
all
is
inside
of
kubernetes,
so
we're
waiting
for
cube
Carl
to
be
outside,
and
then
we
can
rely
on
the
publisher.
C
E
Oh
so
this
is
saying
this
is
suggesting
using
air
out
standard
out
our
standard
error
for
status
without
affecting
standard
out
I.
Think
is
what
what
they're
saying
I
mean.
I,
don't
know
what
your
guys
thoughts
are,
but
it's
like
I
mean
dude.
Do
we
use
standard
air
consistently
as
that's
what
we
write
to
when
there's
a
problem?
I,
don't
know.
A
E
Air
out
yeah,
the
one
he's
referring
to
I
just
got
rid
of
that
in
another
poor
request
so
because
it
was,
it
didn't
seem
right
to
be
writing
a
message
out
to
standard
error.
Yeah
I
think
I
think
really
just
solved
this
one,
maybe
just
to
get
rid
of
that
message
altogether.
In
the
case
of
that's
in
the
case
of
one
particular
case,
there
I
guess
where,
where
they're
talking
about.
E
Three
lines
that
are
right
here
but
yeah
I
think
if
it's.
If
it's
writing
the
output
to
a
to
an
actual
file,
then
it
makes
sense
to
say
a
cluster
info
dump
to
file
XYZ
or
whatever.
But
if
the
destination
is
standard
out,
we
don't
have
to
tell
them
hey
we
just
we
just
wrote
to
standard
out,
you
know
it's,
it
is
it's
there,
so
I
think
that's
just
if
it's
standard
out
then
just
get
rid
of
and
don't
then
get
rid
of
that
message.
That
says
clustering.
So
don't.
A
I'll
be
curious
to
learn
if
this
problem
still
appears
in
the
latest
version
like
117,
or
something
like
that.
I
know
that
there's
a
lot
of
a
lot
has
changed
and
and
customized
over
some
period
of
time,
and
that
period
of
time
might
be
between
14
and
17,
not
that
we
had
the
problem
of
updating
customize,
which
is
moving
pretty
fast,
we're
startup
2
at
the
current
level,
for
at
least
for
some
time
until
we
get
the
update.
D
B
E
E
A
E
A
I
think
I
recall
some
of
the
discussions,
so
we
did
it
intentionally
when
we
were
going
with
this,
and
the
reason
for
that
was
that,
since
this
was
combined
with
the
ability
to
use
the
selector,
the
selector
can
be
pretty
brutal
and
my
my
fear
was
that
it
might
get.
It
might
select
a
pretty
extensive
number
of
parts,
in
which
case
we
would
be
overwhelming
the
API
server
and
the
cubelets
with
requests
for
logs.
So
I
I
decided
to
have
some
reasonable
limit
to
start
with.
E
E
A
B
D
A
We
had
this
discussion
several
times
and
you
just
need
to
be
very
careful.
I
had
a
stab
at
it
at
it.
I
remember
I
close
at
least
one
or
two
issues
related
to
similar
thing.
It
is
just
one
fix
I'll
comment
and
I'll
find
the
I
find
the
the
reference
PR
where
we
said
that
we
are
not
going.
We
are
not
going
to
do
this.
E
B
D
So
I'm,
looking
at
in
the
code
and
kubernetes,
oh
I
DC
go
line
128,
it
says
all
provider
field,
deprecated
refresh
request,
doesn't
send
scopes.
I
get
oh
I'll
keep
working
on
this
one,
but
if
it
does
turn
out
to
be
deprecated
and
he's
on
version,
117
I
think
it's
a
bug
that
it
didn't
give
him
the
right
air.
B
A
I
did
I
mean
what
what
we
had
so
the
ticket
that
I
was
working
on
was
that
somebody
invoked
cue,
got
a
delay
and
delete
a
particular
namespace,
and
then
afterwards
they
passed
all
in
hope
that
they
are
removing
just
one
namespace,
whereas
they
removed
all
of
them
and
make
the
stuff
worse.
There
were
cluster
admins,
so
they
literally
removed
everything
from
the
cluster
and
that's
one
of
that's
one
of
the
reasons
where
I
come
up
with
that.
A
I
came
up
with
the
with
the
solution
to
this
one,
but
we're
hesitant,
and
we
figure
out
that
it'll
be
best
to
I.
Don't
know
have
some
kind
of
admission,
and
on
top
of
that,
just
be
careful
when
you're
invoking
delete
all
and
you're
a
cluster
on
there.
That's
literally
a
that
kind
of
a
response.
There
are
some
ideas
about
how
we
might
have
want
to
approach
this
one,
but
nobody
actually
took
a
stab
at
it.
A
B
B
A
A
Aware
of
those
okay
but
yeah,
I
know
what
you
mean.
F
Hi
this
is
Karthik
I'm.
Actually,
a
newbie
can
I
have
some
opinions
on
this
random
ordering.
G
A
A
A
That's
not
quite
right,
because
customized
that
would
have
to
be
customized,
it
would
have
to
be
a
plugin
and
it's
not.
It
is
just
embedded
directly
in
some
of
the
commands
through
k,
flag,
I'm,
not
sure
if
we're
gonna
provide
any
particular
information.
That
literally
will
require
us
to
provide
information
about
every
library
that
we
use
soldiers.
A
A
B
A
F
G
B
D
D
B
C
So
sit
lest
apply,
there's
actually
three
commands
that
have
to
do
with
the
last
applied
annotation
and
it's
it's
usually
surgery.
You
have
to
do
inside
when
you've
gotten
into
a
bad
state.
The
last
applied
annotation
is
just
a
version
that
was
applied
last.
It's
used
for
the
three-way
merge
if
you're
using
anything
with
these
last
applied,
annotation
commands
then
usually
you've.
You've,
somehow
gotten
your
objects
into
a
bad
state.
B
A
C
B
A
I'm
inclined
to
close
this
issue.
We
haven't
heard
from
him
for
the
past
two
months
and
he
clearly
stated
in
the
last
comment
of
his
that
it
was
just
on
AWS
because
he
was
not
able
to
reproduce
this
on
OpenStack,
which
basically
means
it
is
not
a
cubical
problem,
but
an
environmental
issue,
as
in
he's
having
some
kind
of
a
process
running
on
the
cluster.
That's.