►
From YouTube: Kubernetes SIG CLI 20200401 - 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
Brian,
you
can
hear
me
all
right:
okay,
all
right,
machi,
we'll
figure
that
out
we'll
get
started.
I!
Think
the
last
time
we
said
that
we
were
gonna
start
from
the
back
this
time,
because
we
have
a
ton
of
issues
from
like
2017,
so
I'll
just
get
started,
sharing
my
screen
and
let's
see
who
do
we
have
in
the
chat
hanging
around?
B
A
This
actually
is
related
to
what
Sean
was
recently
talking
about
last
time
on
sick
CLI,
and
that
work
has
happened
and
we
will
be
encouraging
people
to
work
on
this.
This
is
about
ensuring
that
first
first
step
was
that
we
extracted
the
option
structures
for
each
of
the
commands,
and
the
second
option
will
be
to
ensure
that
the
options
and
complete
functions
are
not
coupled
to
Cobra
itself,
but
the
other
can
be
said
separately.
So
it's
definitely
something
that
is
happening
all
the
time.
Yeah.
C
C
C
E
E
B
B
E
A
E
B
E
E
It's
it's
not
done,
and
it's
not
something
we
it's
something
that
you
know
we'd
love
to
have,
but
it's
not
gonna
get
staffed,
likely,
there's
a
lot
of
other,
probably
more
urgent
issues
and
the
the
incoming
rate
of
work
is
not
less
than
the
completion
velocity.
A
This
already
has
a
label
feature
and
it's
in
a
frozen
life
cycle,
that's
sufficient
and
it
was
already
moved
from
the
main
cube
repo
dubiously.
E
The
main
thing
is,
we
want
to
get
it
off
of
the
books
right.
Thank
you
like.
We
don't
want
to
see
it
in
the
list
of
issues
that
we
do
with
the
bugs
from.
We
could
also
do
that
through
a
milestone,
it's
possible
to
searches
that
don't
have
milestone
to
search
for
issues
without
milestones.
We
could
put
it
in
a
milestone.
F
D
B
A
As
I'm
reading
this,
this
is
already
implemented
because
we
have
the
unstructured
type,
maybe
not
necessarily
that
you
are
working
with
this.
As
with.
C
A
E
E
A
E
H
A
B
A
A
A
F
A
G
A
B
C
E
B
E
E
Think
or
at
least
this
looks
yeah
the
lot
of
the
there's
quite
a
bit
of
logic
in
the
bash
RC
or
CS
hoc,
or
you
know,
whatever
your
shell
of
choice
is,
and
so
that
means,
when
you
have
taken
control,
that
it
doesn't
just
naturally
get
whatever
the
latest
greatest
shell.
Completion
is
and
there's
an
alternative
way
of
doing
this,
which
is
to
control
itself
or
the
binary
itself
can
provide
the
shell
completion.
E
So,
instead
of
putting
like
all
these
these
fun
of
these
bash
functions,
for
instance,
you
just
have
one
which
invokes
to
control
itself
and
ask
for
the
shell
completion,
which
we
have
a
example
of
for
customize.
So
we
may
want
to
not
invest
heavily
in
doing
with
this
approach
and
switch
to
the
the
other
approach
which
is
to
have
COO
control,
provide
the
completions.
B
A
E
So
yeah
is
actually
so
what
to
control
completion
does.
Is
it
writes
actually
quite
a
bit?
Let
me
get
the
word
count
on
this
thing.
Okay,
soku
control
writes
close
to
nine
thousand
lines
of
she'll
completion
in
what
I
was
suggesting
was
having
it
write
three
lines
of
she'll
completion,
an
instead
of
implementing
all
the
she'll
completion
has
or
she'll
specific
functions
that
the
actually
are
implemented
in
go.
B
B
All
right
so
on
this
issue,
I
looked
at
this
issue
a
little
bit
and
it
wasn't
really
straight
for
I.
Don't
remember
exactly
what
what
I
ran
into,
but
I
did
look
at
it
when
I
was
working
on
some
others
detail
stuff,
and
it's
really
a
problem
with
like
it's.
It's
a
weird
problem
like
if
you're
typing
a
long
path
and
you
type
part
of
the
path
and
complete
it
it
puts
on
their
space.
And
then
you
can't,
like
you
can't
like
do
successive
completions
after
that.
I
guess
is
the
problem
and
I.
B
Don't
know
that
I'm
kind
of
I
didn't
really
look
too
closely
at
this
PR,
but
I.
Don't
really
in
right
away,
see
how
this
fixes
it
just
it
looks
like
it
cuts
a
whole
bunch
of
stuff
out.
It
looks
like
it
gets
rid
of
the
completion
actually
and
I'm,
not
sure.
Maybe
that
does
fix
it.
I
I
guess
it
just
needs
to
be
tried
out.
B
A
You
kid
thought
about
it:
a
couple
weeks
back
I
think
that
was
rough
right
after
our
initial
buck
scrub.
The
thing
is
that
since
coop
cuddle
is
being
published
as
a
stage
and
repo,
they
are
not
giving
away
those
higher
peripheral
itches
to
anyone.
Okay,
yeah,
I
loved
and
lost
permissions
on
the
project.
C
F
B
B
If
it
still,
it
is
I,
looked
at
it
at
the
bottom
I
could
what
I
found
so
there's
so
there
people
are
saying
they
want
running
pods,
but
let's
see
I'm
trying
to
member
with
this
exactly
was,
but
there
it's
not.
There's
not
like
a
state
like
on
when
it
generates
the
table.
It's
looking
at
the
status
and
saying
that
it's
ready,
but
it
there's
not
like
a
state
that
for
the
overall
pod
that
has
ready
it's
like.
If
all
the
containers
in
the
pod
are
ready.
B
A
A
A
I'm
not
sure
how
possible
will
benefit
that
one,
but
with
pots
it's
complicated,
because
there
was
an
issue
that
I
was
strung
a
couple
days
weeks
back
where
containers
were
reporting,
readiness
each
container
on
its
own.
If
the
pot
overall
was
not
ready
and
that
he
can
get
out
of
conditions
on
a
on
a
pas
level
rather
than
a
per
container
I
think
we
present
some
basic
information
about
which
or
the
number
of
pots
that
is
ready
versus
the
overall
number
but
I
convert
in
where.
A
B
If
I
remember
correctly
now,
I
think
the
confusion
for
this
particular
issue
was
all
of
the
all.
The
containers
are
technically
still
ready,
but
the
overall
pod
is
in
a
terminating
state
which
is
compete.
Well,
it's
it's
not
when
I
say
terminating
state.
We
show
a
status
of
terminating,
but
all
that
really
means
is
that
the
pod
has
a
deletion
timestamp.
It
doesn't
mean
that
the
containers
are
not
themselves
still
still
ready
for
that
short
period
of
time,
while
it's
terminating.
A
B
I'm
not
sure
that
the
use
case
for
this
has
been
really
fully
explained,
like
I,
think
it's
more
of
a
confusion
thing
for
understanding
what
what
it
means
when
it
says
terminating
and
I'm,
not
sure
why
somebody
needs
to
know
that
particular
status
I
mean
I,
guess
there
I
mean
this
was
just
my
opinion.
There's
a
workaround
for
it.
So
I
think
that
probably
makes
it
a
little
less
urgent
to
fix,
but
I
don't
know
long-term,
whether
it
makes
sense
I
mean
I'll.
Leave
that
up
to
you
guys,
oh.
F
A
Wanted
to
go
all
the
way
that,
but
we
we
didn't
go
as
far
we
went
with,
we
will
allow
basic
being
without
links
and
all
those
additional
whistles
and
whatnot
on
top
of
it,
the
basic
copy
of
the
file.
Yes,.
B
A
B
A
B
The
pots
that's
the
container,
stick
around
or
one
is.
How
often
does
it
get
cleaned
up
after
it's
done
running
it?
Is
it
even
technically
possible
to
go
on
the
cubelets,
Anna
I
think
somebody
said:
there's
a
workaround
if
you
farther
up
and
the
tour
the
top
of
the
or
like
I
guess,
maybe
in
the
middle
there's
some
kind
of
weird
workaround
where
you
you
go
on
the
node
and
you
can
actually
go
and
find
the
bit
file.
It's
it's
a
real
happy
way
of
doing
it.
B
B
B
B
Oh,
the
reason
I
want
to
copy
something
from
there
is
not
because
I
mean
there's
two
use
cases
for
copy,
I
guess
number
one
is
I,
have
a
running
pod
and
I
want
to
copy
something
from
there,
but
this
is
like
the
job
finished
and
produced
some
sort
of
output
and
I
want
to
be
able
to
get
that
output
and
I.
Don't
know
if
that's
like
feels
like
if
we're
building
some
workaround
to
copy
things
from
a
completed
pod.
B
It's
right.
It's
really
sort
of
like
a
new
type
of
thing.
I,
don't
know
if
it's
like
another
resource
or
something
that
says
it's
like
a
it's
like
a
job
product
or
something
a
job,
artifact
or
something
that
can
I.
Don't
know
I'm
just
thinking
out
loud,
I,
guess
but
I
wonder
if
this
is
more.
But
there's
like
two
ways:
two
approaches
number
one
is
like
yeah
we
could.
We
could
somehow
figure
out
a
way
to
make
copy
work
or
number
two.
B
E
E
This
is
very
general,
like
you
could
imagine,
hey
I
want
to
copy
in
some
configuration
into
this
pod
so
that
the
pod
behaves
the
way
I
want
it
to
and
when
it's
running
in
production,
which
would
be
like
the
exact
wrong
usage
for
this
command
and
then
there's
like
hey.
Why
is
what's
the
log
file
or
what
weird
thing
is
happening
in
my
pod
in
developing
those
as
part
of
the
debug
command
and
making
them
maybe
a
little
higher
fidelity
and
make
sense
as
well.
B
B
B
E
E
A
A
copy
like
felt
said,
with
this
use
case,
I
didn't
see
any
value.
This
clearly
says
that
the
workflow-
it's
not
appropriate,
but
I
would
I
would
argue
that
the
debug
and
getting
the
core
them
like
fellow
said,
is
perfectly
valid,
but
does
for,
for
that
case
the
back
would
be,
but
the
best,
probably
because
the
core
them
will
be
just
one
of
the
things
that
you'll
be
looking
at.
Who
are
you
Paul
failed?
E
Maybe
the
right
thing
to
do
is
refine
the
issue
to
copy
out
data
from
your
pod.
That's
to
copy
out
data
out
of
a
successfully
completed
pod.
You
should
probably
write
that
to
a
volume
or
for
storage
bucket
or
some
persistent
location
and
to
get
like
debug
information
out
of
a
failed
job.
We
should
create
a.
A
I
I'm,
pretty
I
I'm,
pretty
sure
that
you
lost
it
the
same
at
the
same
time
as
I
did
when
we
started
staging
cube
cuddle.
Last
year,
I
swear
yeah
I
asked
Nikita
about
it,
and
she
says
that
for
every
port
that
were
staging,
oh,
they
don't
allow.
This
I
had
to
list
the
exact
same
rights
as
you.
Okay,
you
can't
do
massage
clothes
either.
No
slash
clothes
works.
Fine,
okay,
/,
the
warden
works
right,
yet
another
incentive
to
to
move
out
of
and
stop
staging.