►
From YouTube: Kubernetes SIG CLI 20210915 - 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
A
C
Yeah
I
thought
that
antoine
had
said
that
server-side
apply
kind
of
fixed
this
deployment,
replica
behavior
since
the
different
clients
aren't
fighting
for
the
replicas
field.
C
If,
if
I
understand
it
correctly,
there's
some
fighting
for
the
replicas
field,
with
the
with
client
side
apply
and
then
server-side
apply
with
the
field
manager
kind
of
fixes
fixes
it
out.
If
I
understand
it
correctly,.
B
Okay,
okay,
so
let's
call
this
one
looked
at
for
now.
B
Okay,
okay:
this
is
the
one
I
had.
B
B
B
B
I
would
point
at
this
so
there's
potential
of
a
dependency
that
changed.
Maybe
I
was
actually
pretty
stumped
when
I
looked
at
this.
A
B
B
A
That's
something
that
I'm
not
100
sure.
Yet,
okay,
we
would
need
to
have
a
proof
that
this
is
either
I
don't
know
remote
or
a
particular
dependency
where
the
issue
is
and
not
with
the
run
command
itself.
A
D
A
That
I'm
not
sure,
but
we
have
some
dependencies.
I
remember
specifically
for
what
is
the
chain
there.
B
Okay,
if
anyone
wants
to
take
a
look,
feel
free,
I
will
spend
a
little
more
time
on
it
and
potentially
move
it
and
tag
jordan
or
someone
from
api
machinery
that
would
know
that
code.
B
A
Phillip
you're
touching
the
quota
describe
or
get
is
it
possible
that
it
was
fixed
in
one
place
and
we
missed
the
other
one.
B
A
A
E
Still
kind
of
confused
about
what
the
issue
is
to
be
honest,
it's
also
labeled
as
being
described
node's
output.
But
that
looks
like
a
pod
list.
In
the
first
example.
E
Oh,
I
see
it's
a
partial
screen
shot
of
just
that
little
snippet.
A
I'm
not
sure
how
we
are
calculating
quota
within
it
containers.
I
would
have
to
look
it
up.
E
A
Right,
yeah,
because
we
are
caching
only
for
the
time
of
a
single
call
and
whenever
we
are
invoking
a
new
command,
it
is
recreating
everything
or
rereading
stuff
from
the
file.
A
Yeah,
it's
different.
It's
fine
to
reuse
in
a
long-running
application.
It
will
be
slightly
different
in
a
you
know.
Invoke
execute
finish
flow
that
keep
cuddle
follows.
A
B
A
A
So
yeah
I'll
probably
say
it
will
work
fine
for
for
long
running
commands.
I
don't
know
attach,
for
example,
or
exec,
but
not
for
short,
lived.
B
B
A
Private
image
is
private
for
a
reason.
So,
basically
you
need
to
give.
A
The
service
account
a
secret
to
be
able
to
pull
the
image.
If
I
remember
correctly,
yeah
yeah
yeah,
that's
expected
exactly
and
modifying
image
after
you
create
it.
You
can
always
modify.
A
Yeah
we
don't,
we
will
not
allow
any
overrides.
You
can
always
create
pipe
that
into
a
file
and
create
it
manually
if
you
care,
because
there
is
a
reason
why
we
created
this
way.
A
What
it
looks
like
they
have
a
crown
job
and
they
are
on
one
end.
They
are
struggling
with
the
private
images
and
second,
that's
one
problem
and
he
realized
that
we
have
the
image
pull
secret
and
the
second
thing
is:
he
has
a
crown
job
and
he
wants
to
create
a
job
out
of
it.
A
But
during
the
creation
of
the
job
from
the
crown
job,
he
wants
to
tweak
the
command
or
inject
environment
variables,
but
yeah
there's
a
reason
why
we
create
a
job
from
cronjob,
because
we
want
to
have
a
that
copy
running
ad
hoc
without
any
tweaks.
B
B
B
E
B
Yeah,
I'd
I'd
be
against
adding
it
to
and
we'd
have
collisions,
because
these
are
global
flags,
as
so
we'd
have
to
deal
with
the
collision.
There.
A
I
wonder
if
maybe
that's
a
case
of
just
injecting
those
values
properly,
because
when
you
are
invoking
the
config
flag,
you're
you're
only
modifying
your
local
copy.
So
maybe
it's
just
a
matter
of
reading
them
and
supporting
them
in
the
set
credentials.
A
On
the
other
hand,
not
every
field
that
we
have
in
the
yams
are
exposed
directly
through
a
flag
or
a
command.
So
it's
fair
to
to
suggest
that
keep
cuddle
set
a
conflict
set
as
an
alternative.
B
A
I
need
to
double
check,
but
I'm
not
sure
if,
if
we
do
have
that
kind
of
information.
A
So
the
annotations-
or
I
don't
know
else
yeah
this
is
this-
is
showing
that
the
type
of
this
this
field
is
a
map
string
string.
A
It
wants
to
have
and
the
ability
to
present
information
about
specific
keys
that
you
can
use
in
in
a
particular
map,
because
some
of
the
maps-
I
don't
know
he
did,
he
does
mentions
in
pv
spec
capacity.
A
It
might
have
specific
keys
and
if
you,
I
guess,
specify
a
different
one,
it
will
fail
a
validation,
but
that
will
be
very
resource
specific,
so
yeah.
A
I'm
not
sure
if
it's
possible
to
expose
that
kind
of
information
in
the
first
place,
but
yeah
I
haven't
seen
it
yet.
C
But
but
for
the
map
of
string
string
to
strength
so
like
for
integers,
you
could
say
the
range
or
a
min
or
max,
but
for
a
map
string
string,
I'm
not
sure
how
that
would
work
in
the
open
api.
A
B
A
This
okay
yeah
he
he
even
says
that
even
that
in
the
official
documentation,
there's
a
there's
a
mention
that
currently,
the
only
keyword
that
exists
is
storage
for,
for
this
particular
resource,
pv,
spec
capacity
and
there's
not
nothing
else
supported,
but
that
will
be
hidden
in
the
validation
most
likely
right.
A
A
Yeah
I
mean
you
would
basically
end
up.
If
you
try
it,
it
will
fail
and
the
validation
will
tell
you
what
to
do
and
that's
fine.
So
I'm
inclined
to
close
it
as
another
book.
E
A
D
Eddie
is
resource
is
resource,
name,
an
enum.
That's
got
like
different
items
like
storage
and
what
I'm
trying
to
looks
like
a
new
memory.
So
it's
so
it's
coming
in
as
a
string,
but
it's
not
just
like
any
old
string.
It's
got
to
be
one
of
the
few
things,
so
I
think
that's
part
of
the
problem
is
like
it
can
only
be
one
of
those
four
things,
probably
but
there's
nothing.
That
says
that
in
the
open
api
spec.
E
Did
I
just
see
a
cap
recently
about
some
basic
validations
being
able
to
be
expressed
in
the
open
api
section
of
the
crd.
A
A
So
yeah
I
I
could
probably
see
something
like
that
being
added
as
a
a
comment
above
the
field
that
only
key
storage
is
allowed
or
supported.
Currently.
A
Right,
I
would
probably
just
add
it
to
as
a
description
to
the
field
itself
yeah,
but
I'm
sure
there's
more
instances
of
this
possibly.
E
A
It
is
ad
hoc,
it
is
ad
hoc
because
the
capacity
is
defined
as
a
resource
list
which
undercovers
is
eventually
strength
drink.
But
if
you
look
under
the
validation,
the
validation
code
expects
a
particular
resource.
A
A
And
that's
directly.
E
A
comment
into
the
output
seems
like
the
best
solution
we
have
available.
Then
I
did
just
link
the
cap
I
mentioned
in
case
anyone's
curious
about
it,
but
I
don't
think
it's
terribly
relevant.
In
the
end
I
mean.
A
The
description
clearly
points
to
official
documentation
where
capacity
is
explained,
so
you
have
that
pointer
there
and
they
just
figure
out
it'll,
be
easier
to
keep
it
there
rather
than
everywhere,
which
is,
I
think,
fine
as
well.
E
A
The
open
api
spec
right,
but
from
this
type
you
get
information
about
all
the
possible
values
and
there's
four
of
them
right,
and
that
would
be
wrong
because
in
this
particular
context
only
storage
is
allowed
yeah
you're
right.
Okay,
that's
what
I've
meant
before
I
understand:
that's
why
it
has
to
be
explicitly
mentioned,
which
they.
C
A
I'll
do
that
I
did
volunteer,
but
I
of
course
I
was
muted
thank.
B
B
B
D
Yeah
a
little
bit-
and
I
mean
maybe
not
surprisingly
or
whatever
we
don't
follow
exactly
the
spec
anyway,
but
but
I
I
don't
think
it's
part
of
the
spec.
I
think
it's
the
extension
on
on
the
spec.
D
It
did
it
did
make
me
think,
though,
and
I
don't
know
it
seems
easy
enough
to
manipulate
the
output.
However,
you
want
just
by
dumping
it
out
to
to
jason.
I
guess,
but
it
kind
of
made
me
think
it's
like.
I
wonder
how
hard
it
would
be
to
just
like
make
printers
that
are
pluggable,
so
you
can
just
like
build
your
own
custom
printer
and
use
it.
It's
just
a
you
know,
kind
of
like
plug-ins.
A
D
A
Yeah,
exactly
that's
why
we
ship
jason.
C
I
have
a
quick,
related
question,
so
is
there
code
anywhere
within
kubernetes
that
kind
of
does
jsonpath
like
functionality
so
like?
If
you
had
in
code
in
your
go
code,
two
resources
like
in
unstructured
format,
and
you
wanted
to
or
even
just
the
straight
bytes
is
there
any
like
json
path
code
that
allows
you
to
get
at
a
particular
field
with
some
kind.
A
C
C
A
So
as
if
you
look
at
the
help
of
keep
cuddle,
get,
the
output
will
give
you
how
to
do
json
path,
even.
A
Now
you
just
specify
dash
dash
output,
jsonpath,
here's
the
printer.
B
B
A
I
think
we
talked
about
it
last
time
or
some
somewhere
before
that
we
could
theoretically
inject
that
kind
of
information
into
explain
because
it
that
information
is
not
present
on
the
server
so
open
api
cannot
know
that
information,
since
it's
not
an
official
type
in
a
kubernetes
world,
it's
more
like
a
go
type.
Only.
A
A
That's
one
example
and
the
fact
that
cubecarol
works
with
it
closely
is
probably
making
that
impression.
Stronger.
A
Every
every
client
go
user
will
use.
It
doesn't
matter
whether
that's
your
your
go
program,
your
whatever
other
program
written
in,
I
don't
know
in
ruby
python
rust
whatever.
A
As
long
as
there
is
a
kind
of
like
similar
to
our
client
go
library
parsing
available
for
parsing
those
it
will
just
work.
A
A
Reference
do
we
have
anything
under
cube
cuddle
and
there
there's
a
cube
cuddle,
or
is
it
like
entire
reference
for
cube
cuddle
itself
on
the
left
side
above
configuration
api.