►
From YouTube: Kubernetes SIG CLI 20200916 - 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
september
16th-
and
this
is
another
sexy-
live
blogger
book,
scrubs
meeting
that
we
are
holding
every
four
weeks.
A
My
name
is
mache,
but
your
host
today
is
eddie
and
eddie
will
be
taking
it
away
from
here
before
I
do
so
welcome.
First
timers.
We
are
very
happy
to
have
you
on
our
meetings.
Do
you
want
to
introduce
yourself
and
tell
a
little
bit
about
you,
your
background,
what
your
expectations
are
and
so
forth.
B
My
name
is
pamela,
I'm
based
in
dublin
in
ireland,
I'm
a
devops
engineer
at
workday
and
I
have
been
wanting
to
get
more
involved
in
open
source
for
the
last
year.
They're
really
proud
in
workday.
So
I
thought
the
cli
group
was
a
good
place
to
start.
I
kind
of
had
a
few
chats
that
could
come
virtual
people
and
this
seemed
like
a
good
group
cool.
A
I'm
in
the
same
time
zone
because
I'm
based
in
poland,
so
it
will
be
much
easier
for
you
to
think
with
me.
Sooner
than
later,
do
we
have
anyone
else
that
is
new
to
our
meetings.
C
D
D
E
I'll
go
next,
hey
everyone!
I'm!
E
F
F
A
G
Yep
I
lost
my
chat
window
there.
It
is
okay,
so
here's
the
dashboard
we'll
be
going
through
if
anyone
wants
to
follow
along
at
home,
also
welcome
to
everyone
new
before
we
dive
into
those
issues.
I
have
a
couple
that
phil
has
tagged
me
on,
but
of
course,
phil's
not
able
to
make
it
today.
So
hopefully
we
can
take
a
look
at
them
this
one.
That
is
specific.
We
have
sean
at
least
yep.
I've
seen
chad
on
the
list.
G
H
I
H
H
H
But
jeff
reagan
who
whose
handle
is
monopole
who's
leading
customize,
should
be
made
aware
of
this
at
the
very
least.
Okay.
Is
there
any
other
labels
we
can
add
to
this?
Is
there
a
customized
label?
Maybe
we
should
make
that
if
we
haven't
already
so
this
is
actually
in
the
customized.
G
A
Adnan
probably
correctly
pointed
to
a
duplicate
of
1510,
which
is
kind
of
related,
because
the
1510
is
discussing
a
broader
term,
including
strategy.
Strategic,
merge,
patch
for
native
cube
objects,.
G
Okay
I'll
tag
jeff
on
here.
If
anyone
on
the
call
wants
to
dive
in
and
maybe
triage
and
see
if
they
can
reproduce
and
actually
nail
down
like
a
nice
case,
feel
free
to
toss
your
name
on
here.
H
It
looked
like
the
initial
description
was
actually
a
pretty
straightforward.
G
All
right,
I
will
tag
jeff.
We
might
need
phil
to
actually
go
through
this
one
at
some
point,
because
I
know
why
phil
just
didn't
tag
jeff
on
it.
G
So
I
will
keep
that
on
the
backlog
for
now
the
other
one
is
this
one.
G
A
G
Awesome,
could
you
slash,
assign
yourself
on
the
issue,
please
yeah
sure
and
awaiting
more
evidence,
yeah.
If
you
can,
if
you're
able
to
reproduce
it
and
stuff,
then
we
can
change
the
priority
cool,
thank
you
and
if
you
need
help
reach
out
to
mache
or
myself
or
someone.
Thank
you.
G
G
G
Okay,
that's
easy
so.
A
My
my
backlog
is
a
little
bit
messy,
but
yeah
I'll
definitely
add
it
to
my
list
for
either
today
or
tomorrow
for
sure.
G
G
A
That
we
probably
are
missing
so
the
majority
of
the
in
the
code
behind,
if
I'm
correct,
responsible
for
setting
this
up,
it's
capable
of
having
a
callback,
I'm
not
sure.
If
usually
there
is
a
sick
term
call
back,
not
sure
if,
if
the
pot
dies,
we
can
catch
that
one,
but
it's
probably
something
we
could
look
into
and
if
so
then,
just
close
the
connection.
On
our
end,
I'm
not
100
sure
I
do
know
that
we
have
a
callback
but
for
client-side
changes
and
that
callback
is
being
used,
probably
in
the
debug.
A
G
G
A
It
might
require
some
some
rewrite
in
the
in
the
top.
H
Corner
so
so,
since
it's
it
would
be
adding
a
flag,
that's
not
there.
We
might
want
to
get
some
consensus,
so
this
isn't
a
bug
so
to
speak
right.
This
is
a
feature.
A
Where
I
would
I
would
I'm
okay,
I
can
argue
in
both
directions,
but
it
would
be
nice
if,
if
across
all
of
the
commands,
we
have
a
consistent
behavior.
H
Should
should
we
put
that
in
the
in
the
issue,
then,
because,
right
now,
if
it's
not
already.
H
Just
to
keep
that
that
that
context
going
forward
that
we
want
this
consistency
and
that's
why
this
particular
new
flag
is,
is
a
good
idea.
G
For
context
for
those
that
aren't
plugged
in,
we
have
recurring
issues
and
pull
requests
and
feature
requests
that
come
up
that
have
a
historical
answer,
and
so
we
have
started
to
capture
those
answers
now.
So
we
have
something
to
point
to.
A
G
G
One
okay,
920
port
forwarding
for
services,
should
favor
choosing
newer
pods
over
older
pods
this
one.
We
brought
this
up
last
week.
I
think
we,
it
was
a
bit
of
a
conversation
we
needed
to
have
about
expanding
this,
to
give
people
the
option
to
sort
those
or.
A
G
H
H
So
I
think
the
flatten
automatically
will,
whenever
there's
a
list,
it'll
it'll
spit
out
the
entire
list.
I
believe.
H
So
so
as
it
stands
now
in
the
apply
we
we
we're
grabbing
every
single
one
of
the
objects
before
we
iterate
through
and
apply,
but
off
the
top
of
my
head.
I'm
not
sure
how
we
would
not
flatten,
if
you
want
to
put
me
on
that
I'll
I'll
reach
out
to
phil
and
get
a
better
feel
for
what
he's
asking
for
here.
J
G
G
G
A
H
Just
just
so,
we
know
this,
this
part
of
the
code
can
get
pretty
opaque
and
byzantine,
the
printing
part
it's
so
I
bring
this
up.
I
I
don't.
I'm
not
sure
that
this
would
be.
A
great
first
issue
is
what
I'm
trying
to
say.
Yeah.
This
part
of
the
code
may
be
hard
to
garage.
A
G
A
A
A
G
Okay
support
multiple
endpoints
for
a
cluster.
A
We
had
a
similar
discussion
back
at
our
for
open
shift,
although
that
was
for
something
else,
and
there
was
a
similar
ask,
basically,
where
someone
was
asking
to
add
the
ability
to
have
multiple
ips
being
configured
so
that
the
client
side
code
would
do
load
balancing
in
case
of
a
failure,
and
basically
all
of
us
agree
that
that's
not
quite
how
you
should
be
doing
this,
because
if
you
have
a
aj
setup,
I
would
expect
you
have
a
proper
load
balancer
being
put
in
front
of
it.
A
G
J
Age,
it's
not
the
same
that
we
there
were
some
age
problems
with
having
to
do
with
jobs
that
had
to
get
started
multiple
times
or
something
like
that.
This
is
this
is
a
problem
of
the
age
column.
J
So
when
you
define
additional
printer
columns
in
your
crd,
spec
and
age,
so
if
you
don't
define
any
additional
printer
columns,
then
you
get
age
by
default,
but
as
soon
as
you
put
one
additional
printer
column
in
there,
you
no
longer
get
age,
so
you
have
to
add
age
back
in
yourself,
like
I
showed
in
the
last
that
last
list
item
there
under
additional
printer
columns.
A
A
J
G
A
A
You'll
have
a
current
state
in
your
in
your
repo
and
you
might
if
you
modify
one
particular
element
of
of
your
deployment,
you
do
want
to
have
that
particular
thing
being
reflected
in
your
gear
as
get
as
well
and
then,
if
you
apply
the
entire
directory
once
again,
only
the
changes
will
be
applied.
A
So
I
think
the
brian's
answer
is
is
a
good
middle
ground
as
well,
but
I'm
not
100
sure
we
want
to
add
some
additional
filtering
capabilities
to
file
reading
that
we
currently
have
in
the
builder.
G
Yes,
totally
agree:
sean,
you,
okay,
writing
that
one
up
and
closing
this.
Yes,
thank
you
number.
G
G
A
A
Yeah,
if
you
have
one
particular
that
needs
to
be
approved,
send
it
my
way
and
I'll
make
sure
that
it's
approved
I'll
sort
through.
G
A
Yeah,
that
seems
like
a
pretty
simple
thing
to
fix,
because
when,
when
commands
such
as
cluster
and
cluster
info
was
clustering,
dump
was
written
and
that
containers
did
not
exist,
so
it'll
be
pretty
simple,
you'll
just
copy
a
particular
for
loop,
but
instead
of
iterating
over
containers,
you
will
have
to
iterate
over
in
its
containers.
Additionally,
do
you
first
iterate
over
in
the
container,
then
you
turn
it
over
containers?
G
Anyone
want
to
take
this
one
on
yeah.
Okay,
could
you
please
slash
assign
it
to
yourself?
Would
you
say
this
is
backlogged
or
important
long-term
priority.
I
G
J
Yeah,
that's
a
good,
that's
a
good
question.
I
can
follow
up
on
that.
J
J
A
A
J
Yeah,
I
might
I
mean
I
might
be
wrong
about
that,
but
I
think
yeah
it's
definitely
seems
to
be
yeah.
I
don't
know
it
is
so
what
I'm
wondering.
As
I'm
saying
this
is
I
mean
shouldn't.
We
have
something
in
the
auto
completion
that
is
able
to
detect
that
there's
a
problem
and
not
and
not
at
least
not
hang.
G
A
We
probably
already
passed
this
over
not
to
cobra,
because
the
command
hasn't
kicked
in.
Yet
it
is
the
bash.
A
A
A
If
not
maybe
there's
some
flag
that
we're
just
missing
setting
for
the
completion
coming
from
cobra
to
ensure
that
there's
a
some
kind
of
a
default
timeout
being
set
so
it,
I
would
say
reasonably
simple
thing
for
first
time
and
for
first
timer
to
dig
in
it's
a
little
bit
of
a
bash
or
zsh.
A
You
can
probably
fix
both.
At
the
same
time,
they
should
have
a
similar,
similar
possibilities.
J
J
J
J
J
H
J
J
G
G
So
this
is
I've,
read
this
one
before
so.
This
is.
We
are
showing
that
you
can
wait
while
doing
a
delete,
but
it
does
not
actually
support
it.
So
it's
getting
the
inherited
flags
from
somewhere
but
does
not
actually
implement
the
flag,
so
we
either
should
implement
the
flag
or
we
should
figure
out
where
it's
coming
from
and
not
output.
That
here.
H
So
there's
been
a
recent
change
to
the
delete
flags,
or
at
least
there's
a
pr
to
do
that
in
order
to,
instead
of
making
it
a
boolean,
there's
a
a
string,
whether
it's
orphan
cascade
foreground
background
that
pr
may
affect
this.
H
G
Happy
this
dashboard,
if
you
haven't
looked
at
this
yet
we
can
write
a
bunch
of
filters
to
basically
show
us
whatever
we
want,
like
issues
that
haven't
been
commented
issues
that
need
attention.
It's
like
the
who
wrote
this
thomas
thomas
wrote
this.
It's
really
cool,
maybe
one
day
we'll
figure
out
what
we
want
to
show
in
here.
I
G
H
A
Whether
this
this
will
help
the
the
original
author
with
the
issue,
because
he
clearly
states
that
he's
using
that
inside
of
the
cluster.
And
if
he's.
A
Of
the
cluster
to
talk
with
the
cluster
api
and
that's
what
the
in
in
cluster
config
is-
and
it
is-
and
it
is
basically
the
first
authentication
to
be
used
within
the
cluster,
because
that
automatically
is
using
the
service
account
capabilities
that
you
have
that
you
can
easily
inject
into
a
pod.
So
that's
that's
the.
A
Way
of
doing
this,
rather
than
manually,
passing
the
the
cubeconfig
tokens
and
whatnot
and
that's
what
he
clearly
was
doing
in
in
his
examples.
G
G
This
like
rotates
the
token
I
don't
know
I
haven't,
used
those
all
right,
we'll
see
what
he
replies.
I
I
I
messaged
you.
I
talked
to
my
friend
on
the
tecton
team
about
this,
because
this
guy
also
worked
on
tecton
and
this
the
need
for
this
again
like
last
week,
we're
going
to
have
to
have
a
story
for
ci
city
eventually,
because
we
just
keep
getting
issues
related
to
ci
cd
for
cube
control.
G
A
I
mean
as
long
as
this
service
account
token
volume
projection
is
injecting
those
into
the
same
place
where
you
can
expect
the
default
service
count.
Tokens
then
you're
basically
set
and
from
his
examples.
A
G
Okay,
we'll
wait,
you'll
see
what
he
says.
I
I
understand
we
don't
want
to
end
more
flags,
I
get
it.
I
do.
I
mean.
A
If,
if
for
some
reason,
the
this
mechanism
is
injecting
them
at
a
different
location,
honestly,
I
don't
see
any
particular
reason
why
we
would
not
allow
this
flag
and
it
should
be
pretty
straightforward
to
do.
G
Also
we
talked
about
this
last
week.
I
got
this
part
where
the
token
output
was
shown.
We
didn't
think
it
would
like
show
the
command,
but
this
apparently
comes
from
doing
a
set
dash
e
in
your
bash
script,
yeah
and
that
actually
yeah
that
echoes
the
full
command
so
yeah.
He
was
right
in
this
case.
G
That's
what
my
my
buddy
assumed
he
was
doing,
okay
cool,
so
we
are
basically
up
on
time.
D
A
A
A
But
the
pr
is
already
approved
so
as
soon
as
we
as
we
open
the
gate
for
120.
A
A
For
yes,
no,
it
just
means
that,
probably
if
we
don't
speak
up
and
then
we
will
end
up
applying
force,
forcing
everyone
to
apply
the
label
for
the
entire
of
120,
but
I
might
be
wrong
I'll
I'll
wait
for
the
answer,
because
vojtec
has
some
questions,
whether
it's
for
120
only
or
if
it's
for
for
for
an
extra
an
extended
period
of
time.
E
G
We
made
it
through,
so
we
didn't
have
this
last
month,
which
is
why
we
had
a
bit
of
a
backlog
of
new
issues,
but
hopefully,
next
month
we
can
start
digging
through
the
backlog
a
little
bit.