►
From YouTube: Kubernetes SIG CLI 20201007
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
october
7th-
and
this
is
another
of
our
six
cli
bi-weekly
meetings-
my
name
is
mate
and
I'll,
be
your
host.
Today,
quick
announcements,
we
have
siam
with
us
and
he
will
probably
talk
a
little
bit
more,
but
basically
he
created
the
new
docs
for
sexy
life,
namely
for
keep
cuddle
and
customize,
and
it
looks
amazing.
A
So
thank
you
very
much,
sam
for
all
for
your
work.
I
know
that
both
you
and
phil
put
together
a
lot
of
time,
and
it
is
just
looking
amazing.
I
was
clicking
through
through
the
docs
before
this
call
and
it's
it
is
mind-blowing.
A
So
if
you
have
any
questions,
topics
or
suggestions,
I'm
pretty
sure
that
siam
will
be
more
than
happy
to
hear
them
out
and
you
can
always
open
some
issues
or
suggestions
against
the
cli
experimental,
where
this
is
happening.
A
The
second
announcement
that
I
had
was
for
for
kubecon,
north
america
virtual.
We
will
be
traditionally
hosting
a
sexy
li
related
session
and
I
link
that
session
in
the
schedule.
If
I
remember
correctly,
that's
on
thursday
19th
around
5
p.m.
A
Okay,
with
with
the
announcements
out
of
the
way,
I
think
we
can
jump
over
to
the
main
topics
before
we
kick
off
to
justin
sam.
Do
you
want
to
talk
a
little
bit
about
your
work
or
siam
and
phil?
Do
you
want
to
talk
a
little
bit
more
about
the
docs
other
than
that
quick
show
off
that
I
did
during
the
announcements
yeah
sure
you
can
hear
me.
B
Guys,
yeah
loud
and
clear,
go
ahead,
yeah
cool,
so
I
think
I
introduced
myself
to
six
cli
this
bi-weekly
meetings
like
a
month
and
a
half
ago
and
as
a
part
of
google
season
of
docs,
I've
been
working
on
this
docs
thing
and
I
think
it
turned
out
pretty
well.
We
myself
and
phil
used
to
get
on
a
weekly
call
and
kind
of
sort
things
out
set
targets
and
meet
goals,
and
I
think
it
turned
out
pretty
well.
There
are
a
lot
of
things,
especially
in
the
reference
section
under
control.
B
We
added
a
lot
of
new
examples
and
not
just
the
commands,
but
also
the
current
state
of
the
clusters,
the
commander
recon
execution.
What
would
be
the
update
instead
and
stuff
like
that?
So
we
kind
of
also
made
it
very
beginner
friendly
and,
and
I'm
hoping
it
would
be
really
good.
I
mean
if
you
really
like
it.
Let
me
know
if
you
want,
if
you
want
to
give
me
constructive
feedback,
I'm
always
open
for
it.
B
You
can
reach
me
on
slack
or
an
email
and
I'm
looking
forward
to
get
feedback
from
you
guys
so
that
you
can
make
the
docs
better
and
currently
I'm
working
on
the
ci
stuff
that
actually
deploys
this
site
to
the
public
and
hoping
we
are
hoping
to
launch
to
the
other
people
who
are
not
in
the
cli
6
eli
stuff,
6,
yellow
growth
pretty
soon,
and
we
are
looking
forward
to
it
yeah.
That's
it.
Thank.
A
D
Just
that,
like
everyone,
should
take
a
look.
I
think
the
plan
is
to
now
replace
the
control
book
and
the
customized
doc
sites
with
this
and-
and
I
think
the
plan
is
to
have
the
control
book
site
now
host
this
and
just
replace
that
and
then
have
customize
redirect
over
here
over
to
that
site.
D
There's
now
some
great
places
to
add
content,
there's
like
a
back
section
for
coop
control
and
a
contributing
section
for
good
control,
and
the
fact
already
has
some
of
the
stuff
we've
discussed.
So
I
know
we've
all
well
the
discussions
we
often
have
about
some
of
these
docs
that
that
aren't
well
easily
discoverable
or
now
have
a
better
location
that
we
can
take
out
can
can
go
fill
in
after
meetings
such
as
this
one.
E
Yeah,
this
looks
absolutely
awesome.
Great
works.
I
am,
I
see
the
faqs
on
there,
so
I
want
to
remind
everyone
that
we
need
to
keep
updating
that
faq
with
decisions
in
historical
context
when
we
surface
it,
so
we
don't
have
to
keep
repeating
ourselves
when
new
issues
come
up,
we
can
just
copy
pasta.
D
Cool
thanks
guys
yeah
in
the
bug
scrub.
We
can
just
have
maybe
a
person
responsible
for
updating
after
each
bug
scrub
just
pick
a
person
like
like
we
would
a
note
taker
or
something
else,
and
then
they
can
go
and
fill
in
the
back
with
whatever
issues
came
up
in
that
bug,
spread
scrub.
F
Yeah,
just
back
up
any
there's
been
a
couple
times
in
the
slack
channel
recently,
where,
like
our
greatest
hits
of
issues,
keep
arising
and
and
yeah.
If
we
could
easily
make
easily
surface
that,
I
think
that's
going
to
save
us
some
time
and
clarify
some
things
for
for
folks.
F
Specifically,
there
was
recently
a
big
long
threat
about
which
particular
version
you
apply
at
versus
the
version
that
comes
back
when
you
do
the
get
without
specifying
the
group
and
version
and
I'll
put
that
in
that
issue
in
the
chat,
if
anybody's
interested
in
digging
into
that.
F
Let
me
let
me
send
it
and
you
could
tell
me
so
this
so
the
issue
I'm
describing
is
actually
it
was
closed
back
in
2018.
So
I
don't
think
it's
the.
F
E
Question
is
this
the
permanent
url,
or
is
this
going
to
live
somewhere
else
asking
so
we
can
get
this
linked
in
the
issue?
Template.
A
D
A
Okay,
I
think
we
can
jump
over
to
justin.
G
Install
pages
forum
show,
I
was
trying
to
think
how
I
would
restructure
this.
I
install
all
of
my
tools
with
homebrew
on
linux,
but
this
specifically
says
homebrew
is
for
max,
and
so
I'm
just
trying
to
figure
out
like
where,
like
maybe
there
should
be
a
homebrew
section.
That
says
it
works
on
mac
and
linux
and
a
mac
port
section
separately,
but
all
three
of
the
tools
under
install
say:
if
you
have
a
mac,
you
know
homebrew
for
mac
and
I'm
like
well.
G
D
D
D
A
D
That's
it
yeah.
I
I
yeah.
D
Okay,
there
you
go
yeah,
we
can
change
that,
just
to
say
homebrew
and
then
pull
macports
into
a
separate
section.
I
think
that's
not
a
particular
problem
and
they
can
be
their
separate
items
in
the
installation
section.
G
And
crew
also
has
a
homebrew
install
which
it
isn't
on
the
page
and
then
also
the
the
bash
zsh
installer
doesn't
if
you
have
an
arm
linux
machine,
it's
in
the
download,
but
it's
hard
codes,
the
amd64.
So
we
may
want
to
add
that
too
I
can
open
a
ticket
if
we
want
just
to
track
it.
I
was
trying
to
think
of
the
easiest
way
to
that's
a
pretty
easy
one.
D
F
G
D
G
D
G
I
I've
been
I've
been
encouraging
a
lot
of
people.
I've
been
using
it
for
years
now,
it's
easy,
it
keeps
things
up
to
date
versus
go,
get
and,
and
it's
more
up
to
date
than
you
know,
yum
and
app.
G
I
A
couple
caveats
to
installing
crew
through
home
brew-
that's
not
like
recommended,
I
guess,
but
it
still
works.
It
just
won't
update.
I
I
think
there's
I
haven't
looked
at
this
in
a
while.
I
think
it
was
because
of
certain
parts
of
the
installation
instruction
in
like
the
bash
slash
instructions.
You
know
that
doesn't
happen
when
you
brew
install
it
if
I'll
have
to
look
into
it,
but.
G
I
ipr'ed
crew
like
a
week
ago
to
say
that
brew
homebrew
works
on
on
it
on
linux.
So
that's
why
I'm
curious
if
there
is
a
problem
that
I
I
added
a
pr
that
was
incorrect,
but
I've
been
using
crew
with
homebrew
on
linux
for
a
little
while,
especially
since
they
switched
to
the
go
binary
versus
versa,
bash
grip.
I
G
A
Okay,
let's
move
on
to
justin's
k,
experimenter
robber
jesse,
you
want
to
share
your
screen
yeah.
Let
me
stop
and
I
think
I'll
have
to
add
you
to.
J
C
G
Mean
it's
it's
in
the
github
repo,
it's
the
main
goals
of
it
was
I
had
a
shell
function
like
I'm
sure,
a
lot
of
people
do
or
a
lot
of
people
just
do
aliases,
but
I
had
a
shell
function
for
a
long
time,
specifically
for
shell
or
yeah
per
terminal
usage
on
different
clusters,
and
so
I
would
set
context
or
namespace
or
something
based
on
a
terminal
itself,
not
in
a
global
context,
with
file.
G
Switching
like
a
normal
cube
control
used
context
would,
and
so
I
like
doing
it,
shell,
based
because
I
can
compare
two
things
side
by
side.
You
know
test
and
dev
without
like
oh,
let
me
switch
my
contacts
back
and
forth
or
those
sorts
of
things.
So
when
I
was
thinking
of
like
how
I
like
to
do
it,
my
shell,
alias
started
getting
just
really
large
or
my
my
function
did
and
with
newer
tools
that
are
deploying
kubernetes
for
me
tends
to
not
use
the
the
cube
config
file
in
a
lot
of
cases.
G
They
use
some
other
file
and
I
like
having
separate
files.
I
love
like
compartmentalizing
some
of
that
stuff,
so
it's
like.
I
can
just
delete
a
file
and
it's
the
cluster's
gone
for
me.
Basically,
so
this
the
goal
of
this
was
one
to
allow
the
easier
terminal
switching
for
clusters,
which
I
see
a
lot
of
tickets
for
in
the
repo.
For
you
know
like
issues
of
hey,
I
want
please
support
cubenamespace
or
something
along
those
lines.
G
So
I
and
I
had
been
using
those
so
since
I
saw
other
people
using
that
it
seemed
to
make
sense
to
have
some
sort
of
wrapper
or
an
optional
way
to
do
that,
whether
that
goes
into
upstream
cube
control
or
whatever
or
in
the
experiments.
I
didn't
know
where
that
kind
of
fits
and
again
this
is
meant
to
replace
my
function
and
alias
rather
than
the
actual
cubecontrol
command,
I'm
not
using
any
libraries
from
clientgo
or
anything
else,
I'm
literally
just
wrapping
cubecontrol.
G
So
that
was
a
big
goal
and
then
also
being
able
to
switch
more
easily
with
what
I
I
think
of
as
ssh
syntax
with
like
hey.
I
want
to
switch
a
cluster
quickly
instead
of,
like
you
know,
list
all
my
clusters
and
all
this
stuff
and
then
like
trying
to
go
back
and
forth
when
you
have
dozens
of
clusters,
it
becomes
very
difficult
to
like.
Oh
let
me
list
my
clusters.
Okay,
it
was
that
one.
G
Oh
let
me
do
this
thing
and
and
cube
ns
and
qctx
work,
but
again
you
have
to
like
bounce
out
of
the
term
out
of
that
command
you're
in
back
and
forth,
which
was
always
difficult
for
me.
It's
just
like,
oh,
like
I
lose
track
of
where
I
am,
and
you
can
do
some
fancy
stuff
to
make
it
easier,
but
it
was
always
just
better
if
I
could
just
like
on
the
terminal
like.
Let
me
tab
complete
something-
and
I
know
like
context
dash
context,
works,
dash,
dash,
namespace
works
and
it'll
tab
complete.
G
But
again
those
were
to
me
seems
just
more
effort
than
like
ssh
at
something,
and
so
that's
where
I
kind
of
came
up
with
these
like
using
symbols
to
do
the
switching
pretty
easily
or
at
least
quicker
and
and
all
again
all
I'm
doing
is
wrapping
it.
So
let
me
actually
just
share
my
terminal.
I
was
really
trying
to
get
my
tab
completion.
I've
been
working
on
this
tab
completion
for
weeks
at
night
and
like
I
cannot
get
it
and
it's
driving
me
absolutely
crazy
for
a
lot
of
things.
G
So,
specifically
tab
completion
in
bash
works,
and
if
you
saw
on
the
install
page,
you
can
brew,
install
it
it's
on
mac
and
linux,
and
and
so
that
that
just
works
and
then,
if
I
do
like
k,
config
get
context
you
can
see.
I
have
you
know
dozens,
not
thousands,
but
I
have
it
was
a
seven
clusters
or
something-
and
it's
always
like
okay.
I
need
to
switch
between
these
things,
and
so
I
can
nor
I'm
on
kind
right
now.
So
I
can
normally
you
know
every
other
command
in
cube.
G
Control
gets
passed
through
so
get
my
namespaces
and
obviously
you
can
see
them
get
pods
all
that
all
that
functionality
just
works,
because
it's
again
just
executing
out
to
keep
control,
but
the
fun
parts
are.
When
I
actually
start
doing
like
give
me
a
cluster
so
get.
G
Okay,
let's
do
so
at
is
a
cluster
and
because
there's
no
dash
dash
cluster
option
in
cube
control,
it
actually
does
a
lookup
for
what
context
that
cluster
is
part
of,
but
I
I
tend
to
remember
things
as
clusters,
not
context
so
being
able
to
do
at
kind
kind
and
well
I
mean
that's
the
same,
I'm
already
on
that
context,
but
if
I
do
at
let's
do
aob,
because
I
know
that
one's
actually
up
so
tab,
completion
and
bash,
like
I
said,
works
and
and
it
automatically
for
that
one
command
switches.
G
My
contacts
runs
the
command,
which
was
again
to
me,
really
handy
and
then
also
things
like.
I
want
to
get
pods
on
a
specific
namespace,
because
I
don't
know
anyone
that
actually
saves
a
namespace
in
their
context.
I
don't
maybe
people
here
do,
but
it
seems
odd
to
me
to
like
permanently
set
a
namespace
as
an
operator.
I'm
sure
developers
and
different
companies
have
specific
things
they
do
so
so
I
have
again
I
have
tab
completion
for
the
namespaces
in
bash
works
and
I
can
get
the
pods
in
that
namespace.
G
G
Cube
system-
and
we-
I
just
did
commas
between-
like
let
me
just
list
these
things
out
as
a
different
name
space,
and
so
the
one
and
I
really
don't
like
bash's,
auto
completion,
because
I
can't
like
cut
like
these
are
now
an
argument
together,
but
it'll
tell
me
like
which
which
arguments
I
can
complete,
and
it
also
doesn't
zsh-
is
so
much
more
flexible
in
some
of
these
things
that
I
just
struggled
so
many
times
with
bash.
But
I
have
other
problems
with
csh
with
the
way
we
do
completion.
G
So
if
I
do
multiple
things
here,
I
actually
in
the
exec.
I
take
the
standard
out
and
I
print
basically
what
I
call
the
k
space
in
front
of
each
one.
That
tell
me
oh,
oh
if
I,
if
I
did
the
whole
command
right.
G
So
I
print
out
which,
which
k
space
I
separate
them
out
by
the
commas.
So
here's
default
here's
cube
system
and
then
I
say
where
they
came
from.
I
wanna
there's
a
lot
to
fix
here.
Obviously
I
would
like
to
take
off
like
the
extra
header.
G
Oh
when
it's
repeated,
and
I
would
like
these-
you
know
columns
to
align
after
like
I
have
to
buffer
it,
and
then
you
know
align
them
all
after
the
facts,
but
I
can
do
that
with
either
multiple
name
spaces
or,
like
I
said
before,
I
could
do
it
with
with
different
clusters.
So
if
I
do
alb
and
then
I
do
where's
my
let's
do
prod
and
kind.
G
Yes,
I
know
there
we
go
get
pod,
so
I
can
have
two
clusters
and
a
and
a
context
here.
You
can
mix
and
match
them,
and
I
can
get
I'm
not
logged
into
that
cluster,
but
you
can
run
multiple
at
the
same
time
and
and
get
all
the
output
it
works
for
any
cube,
control
command
that
isn't
interactive
right
now
I
don't
escape.
If
I
do
like
an
edit
and
I
like
escape
out
to
vim,
that's
another,
it's
another
bug
that
I
have
filed
that
I'm
just
like.
G
I
need
to
figure
that
part
out,
but
I've
been
trying
to
get
this
tab
completion
working.
The
way
I
wanted
and
the
way
cube
control
auto,
generates
tab.
Completion
is
driving
me
crazy
because
it
like
for
zsh
it's
basically
the
bash
output
that
re-wraps
into
said
a
bunch
of
lines
for
zsa,
which
I'm
sure
a
lot
of
people
here
know
and
in
that
completion
is
throw
it.
It
adds
bash
completion
overlays
into
the
csh
and
it
it.
G
I
can't
fix
it
right
now
so,
but
the
zsh
completion
when
it
does
work
for
certain
pieces.
It
does
a
lot
better
at
like
some
of
the
tab.
Completion
on
the
commas
like
I'll,
get
only
defaults
or
and
I'll
remove
old
context
and
clusters
that
I
have
already
specified
on
the
command
line.
So
I
feel
like
I'm
close,
but
I
need
to
learn
more
zsh
first
and
then
the
other
thing,
like
I
said,
was
I
could
do
export
cube,
namespace
default
and
then
I
can.
G
This
will
run
only
my
default
namespace.
The
other
thing
that's
nice
is,
if
I
just
type
k
I
I
did
my
help
with
no,
because
it's
a
wrapper,
my
help
is
just
hit
the
command,
and
I
say
at
the
bottom.
Basically,
if
you
want
cube
control
help,
please
use
k-help,
because
all
those
commands
are
actually
passed
so
k
will
show
you
kind
of
like
how
we're
going
to
combine
some
of
the
the
cube
config
together
and
then
just
give
some
examples
here.
But
if
I
do
want
to
oh
whatever.
G
G
G
G
I'm
trying
to
think
of
some
other
common
things
that
people
do
in
their
aliases
and
functions,
or
even
with
tickets,
to
see
if
there's
other
helpful
things
to
kind
of
experiment
with
and
and
again
I
wanted
to
do
this
outside
of
some
of
the
proper
repos
just
because
it
I
didn't
want
it
ever
to
seem
like
any
of
the
stuff
was
like
I'm
experimenting
here
like
it
says
in
the
title,
but
I
am
curious
what
other
people
are
doing
and
if
something
seems
important
or
gain
some
popularity
like.
C
G
But
that
was
really
like.
I
mean
most
of
what
I
had
just
oh
and
cube
debug,
because
it
drove
me
crazy
to
try
to
figure
out
how
I
was
executing
things
or
kdebug.
G
Whoops,
it
just
has
a
debug
output
for
everything
that
I'm
basically
doing
inside
of
k
to
see.
Okay,
we're
passing
these
arguments.
Here's
the
cube
config
that
I'm
using
and
we
you
know
we're
setting
that
ns
default
because
I
have
those
variables
set.
So
if
I
unset
cue
config
and
then
run
that
again,
you
can
actually
see
like
okay,
here's
my
actual
cube
config
that
I'm
used.
G
I
still
have
default
environment
variable
set,
so
that's
you
know
being
prepended
here
or
appended
to
the
line
so,
but
that's
really
like
for
debugging
and
just
trying
to
see
what
people
are
doing.
I
don't
know
if
there's
other
like
things
like,
I
said
that
people
see
it
are
common,
especially
within
larger
companies
as
an
ops
person.
G
Managing
all
these
you
know
dozens
of
clusters
in
in
things
like
I
want
to
support
things
like
cube
or
k
apply,
and
I
can
say
like
go
to
all
my
context
and
have
wild
card
for
either
matching
or
you
know,
go
deploy
something
to
all
prod
clusters.
So
I'm
going
to
add
some
wild
card
support
just
to
look
at
those
contexts
and
and
run
that
command
over
and.
G
But
that's
really
like
I
said:
that's,
that's
the
for
me.
I'm
solving
my
own
problem,
but
also
seeing
common
use
cases,
especially
at
companies
and
with
multi-cluster,
like
I
think
you
know,
multi-cluster
is
a
thing
that
needs
to
be
solved
on
the
cluster
management
side,
but
the
interaction
side
as
people
are
building.
You
know
trying
to
build
platforms
that
kind
of
abstract
some
of
that
stuff.
I
still
see
everyone
using
cube
control,
and
so
I
was
trying
to
solve
that
for
myself
with
these
few.
You
know
clusters
that
I
that
come
and
go
for
me.
D
It
seems
like
you
got
some
really
powerful
syntax
there.
I'm
sorry
are
you.
A
A
Okay,
well,
phil
is
trying
to
figure
out
his
echo
thing,
a
quick,
a
quick
headers
solution
to
it
problem.
There
is
a
dash
dash,
no
headers
and
you
can
for
subsequent
get
commands.
That's.
G
Exactly
right,
headers
yeah.
As
soon
as
I
run
once
I
basically
just
said
saturday:
yes,
we've,
the
first
command
output
has
come.
So
if
someone
even
specifies
no
headers,
I
can
specify
that
multiple
times
and
it
doesn't
matter
but
yeah.
That's
that's!
The
goal
is
on
subsequent
command
runs.
Setting
that
flag.
A
In
my
day-to-day
basis,
I
I
at
most
have
one
cluster
that
I'm
interacting
and
oc
has
something
which
is
openshift
client
line
right
that
we
we
set
the
default
project
you're
working
on,
so
that
you
don't
have
to
do
it.
It's
like
literally
what
cubectx
is
doing.
We
have
that
one
baked
in
right.
G
I
wanted
to
add
an
option
for
like
exports.
You
know
in
k
export
at
something,
so
I
get
the
tab
completion
and
then
just
that
would
basically
be
set
as
the
environment
variable.
It
could
permanently
switch
the
context,
but
I
can
use
the
standard
cube,
control
context
commands
anyway,
so
I
was
trying
to
figure
out
like
how,
when
do
I
want
to
someone
wants
to
say,
I
want
to
use
this
one
permanently
or
I
want
to
use
this
one
for
the
life
of
this
terminal.
G
D
No,
and
maybe
I
can
see
parts
of
this
being
upstreamed
more
easily
than
others
right
like
if
you
just
like
just
take
the
at
syntax
for
auto
selecting
a
context
based
on
a
cluster
like
that
seems
like
a
relatively
trivial
or
like
it's
just
adding
the
command
line,
option
right,
but
doesn't
change
behavior
versus
adding
like
stars
and
and
sort
of
globbing
of
context
themselves
and
then
sending
a
bunch
of
requests
to
a
bunch
of
different
clusters
or
contexts,
and
then
combining
those
responses
into
a
unified
result
is
is
like.
D
I
think,
we're
going
to
require
more
thought
about
how
we
want
to
do
that.
So
it's
probably
worth
if
you're
thinking
about
upstreaming
it
like
breaking
it
up
into
different.
G
And
there's
some
things
that
yeah
especially
like
if
I'm
running
multiple,
if
I
want
to
stream
logs-
and
I
want
to
you-
know-
do
it
from
multiple
clusters
with
multiple
pods,
like
there's
other
tools
that
do
that
at
least
from
a
single
cluster
and
and
so
that's
a
pretty
big
shift
of
like
how
just
some
of
that
is
structured,
there's
yeah.
Definitely.
C
G
That,
I
don't
think
makes
sense,
the
environment
variables
would
be
simple
and
some
of
the
syntax,
I
think,
would
be
simple
but
yeah.
It's
I'm
trying
to
figure
out
what
people
are
trying
to
do,
especially
at
the
operator
side
with
I
have
you
know,
tens
of
clusters
or
whatever
and
managing
them
is,
is
difficult
and
even
at
a
a
simplified
context.
G
Switching
for
developers
of,
like
I'm
used
to
you,
know
ssh
to
a
machine
and,
like
I
even
thought
of,
like
oh
user,
I
had
initially
of
like
user
at
cluster,
but
I
don't
know
anyone
that
switches
their
user
like
as
soon
as
you
get
a
user.
You
basically
are
always
that
user
for
the
entirety
of
the
cluster
or
your
context
of
it.
So
yeah.
D
The
the
idea
of
like
streaming
logs
from
multiple
pods
and
multiple
clusters,
at
the
same
time,
I
think,
is
interesting
but
yeah.
I
think
that's
an
interesting
idea,
idea
that
it
deserves
thought
like
to
really
like
see
like
how.
How
do
we
optimize
that
experience.
G
Right
and
that's
what
I
mean
again,
I'm
solving
my
own
problems
here
of
of
trying
to
just
experiment
and
see
what
I
think
is
handy
and
what
I
think
makes
sense.
But
you
know
what
I
initially
thought
would
be
good
for
my
k
space
turned
out
to
be
terrible.
When
I
was
like.
Oh,
I
want
user
at
I'm
like
I
don't
every
switch
user.
Why
would
I
want
that?
G
Oh
I
wanted
you
know
a
different
symbol
for
something
else,
and
I
wanted
multiple
I
wanted
at
with
multiple
clusters
separated
by
commas,
but
then
that
made
switching
namespaces
hard
because
I
didn't
know
which
namespace,
which
went
to
which
cluster,
so
I
had
to
separate
that
out
into
you,
can
only
specify
one
cluster
or
context
with
multiple
namespaces,
not
multiple
clusters
with
multiple
namespaces,
because
that
matrix
of
execute
all
these
things
became
too
complex,
and
I
couldn't
tab
completed
either.
F
Hey
justin,
this
is
this
is
sean,
that's
some
cool
stuff
and,
and
I
see
that
you've
contributed
to
ahmet's
plug-in
on
namespaces
and
contexts,
and
I
just
wanted
to
to
bring
that
up,
and
I
was
curious
how
what
you
thought
about
how
well
that
particular
plugin
fit
your
use.
G
Case
but
my
problem
again
with
that
is
just
that:
it
switches
it
at
the
file
global,
lock
level,
and
I
don't
I
don't
tend
to
work
that
way.
I
have
multiple,
you
know:
tmux
open
with
multiple
terminals
and
and
switching
you
know,
cube
ctx
or
cubens
back
and
forth.
I
was
like
I
switched
terminal
one
to
another
and
I'd
have
to
run
the
same
command
like.
Oh,
let
me
go
switch
context
again.
Oh,
let
me
switch
terminals.
Let
me
switch
context
again.
G
G
Oh,
let
me
find
you
know
what
happens
when
I
you
know:
curl,
those
endpoints
and
and
stream,
the
logs
and
and
so
debugging
that
back
and
forth
was
difficult,
doing
it
at
a
global
file,
lock
level,
and
so
that's
why
I
used
to
always
just
have
separate
files
for
my
clusters
and
I
was
just
exported
cubeconfig,
and
that
was
the
way
I
got
around.
It
was
every
terminal
I
would
export.
G
I
would
set
a
cubeconfig
variable,
so
I
had
a
function
that
would
basically
do
what
he's
doing
with
cubectx,
but
I
would
do
it
at
a
cubeconfig
file
level
and
so
I'd
say
interactive
pick.
My
file
for
my
cube,
config
export
was
was
what
I
originally
had
and
it
worked
to
some
degrees,
but
then
it
was
still
harder
to
like.
Oh,
I
actually
want
to
override
this
piece
of
it
and
change.
A
Oops
sorry
go
ahead
watching
my
question
was
whether
so
you're
basically
going
through
all
config
files
within
cube,
conflict
directory.
G
G
I'm
like
yeah,
don't
don't
touch
those,
and
I
could
do
a
better
job
at
like
looking
at
the
file
to
make
sure
it's
proper
because,
like
the
eks
control
adds
a
bunch
of
lock
files
that
are
empty,
like
they
don't
matter
but
they're
in
my
my
actual
config
variable
when
I
run
it,
and
so
it
doesn't
actually
matter.
But
if
someone
started
like
dropping
garbage
files-
and
there
obviously
would
be
a
problem.
E
I
was
just
going
to
say
when
justin
showed
me
this
demo.
I
was
pushing
him
to
come
to
show
it
here,
mainly
the
the
the
app
syntax
for
targeting
multiple
clusters
and
being
able
to
switch
very
quickly
on
the
fly.
I
I
that
syntax
was
really
inspiring
because
right
now
we
obviously
have
tons
of
flags
and
they're
not
all
there,
and
they
don't
all
do
the
same
thing
and
they
don't
play
nicely.
E
I
think
there's
probably
some
coordination
in
the
future
between
sig
multi-cluster,
because
I'm
sure
they're
thinking
about
this
problem
as
well.
So
hopefully
everyone's
brain
is
turning
as
sigmulti
cluster
approaches.
This
problem
one
day
in
cube,
ctl.
A
I
mean
I
mean
I
totally
agree
with
with
what
you're
saying
eddie
and
what
phil
is
saying.
It
would
be
good
to
gather
some
additional
feedback.
I
definitely
like
the
the
ad
syntax,
where
you
can
quickly.
You
know
pick
this
with
that,
or
the
fact
that
you
can
easily
specify
multiple
and
run
a
single
command
against
several.
A
That
is
very
that's.
Pretty
cool,
oh,
like
I
said,
I'm
I'm
literally
inspired
to
reach
out
to
our
sre
team
and
ask
them
the
exact
same
question:
have
they
tried
this
and
what
they
think
how
they
are
dealing
with
this
the
problem
of
talking
with
multiple
clusters?
At
the
same
time,
I
I
would
be
much
in
favor
of
getting
this
thing
or.
A
G
Yeah,
I
don't
we
can
move
on.
I
was
just
I
know
I
dropped
for
a
second
too,
so
I
didn't
hear
all
of
that,
but
I
yeah
I'd
love
to
hear
what
other
people
are
doing.
I'm
trying
to
just
make
this,
like,
I
said,
an
experiment.
Let
people
use
it
and
see
and
then
gather
some
feedback
based
on
what
people
are
doing
and
it's
yeah
I'd
love
to
hear
what
you
know.
Other
sres
are
doing
for
multi-cluster
management
at
the
interactive
side.
G
I
was
like
creating
and
managing
clustering
things
like
that
as
a
separate
problem,
then
how
do
I
interact
and
debug
some
of
this
stuff
and
do
it
quickly
without
you
know
a
lot
of
overhead,
so
yeah
thanks
for
the
time,
because
I
I
have
been
trying
once
I
get
my
zsh
tab,
completion
working
the
way
I
want
it.
I
hope
to
get
a
little
bit
more
feedback
from
the
community
and
stuff
like
that
and
really
see
what
people
are
doing.
A
Yeah
thanks
for
your
time,
just
that
was
very
that
was
very
cool.
I
had
a
topic
for
delete
flags.
We
can
probably
skip
that
one
for
the
next
time.
I
see
that
eddie
has
two
topics
that
are
much
more
pressing
than
the
delayed
ones
I'll
I'll
skip
the
delete.
A
E
To
your
topics,
yeah
mine
are
short
and
sweet,
but
specifically
for
delete
flags.
Oh
no,
that's
a
different
thing.
I
think
there
was
a
problem
where
we
were
looking
at
an
issue
or
pr
where
all
of
the
apply
commands
are
actually
tied
to
the
delete
flags.
So
we
need
to
think
about
refactoring
that
in
the
future,
but
we
we
can
take
that
for
next
time
when
you
bring
up
yours.
A
F
F
It
turns
out
have
store
the
files
that
are
being
either
applied
or
deleted,
or
it's
yeah,
so
it
it
seems
like
they
were
kind
of
put
in
the
wrong
place.
If
you
look
at
the
that
that
it
has
the
actual,
the
reason
that
that
delete
flags
is
everywhere
is
because
that's
where
the
files,
the
file
names,
are
stored,.
K
A
E
Limiting
two
quick
ones
so
in
sig
testing,
yesterday
ben
brought
up
that
the
docker
hub
pull
rate
limiting
is
going
to
effect
in
november
pretty
soon
so,
there's
just
a
action
that
we
all
need
to
confirm
that
we're
either
using
gcr
repos
for
all
of
our
images
or
we
need
to
figure
out
and
sync
with
ben
and
sugg
testing
on
docker
pull
secrets.
A
E
F
Is
that
that
sounds
new,
and
we
we
never
really,
as
far
as
I
remember,
were
in
charge
of
the.
E
A
E
Yeah
well,
if
brian
or
anyone
is
in
there
and
you
notice,
hub.docker.com
or
just
without
gcr
in
it,
just
yeah
we
gotta
get
those
fixed.
Second
thing:
real
quick
was
that
we
rolled
out
a
needs
triage
label
for
all
the
main
kk
issues
that
are
opened.
I'm
gonna
opt
cube
ctl
in
well.
I
would
like
to.
Is
there
any
objection
to
us
doing
that?
Basically,
whenever
an
issue
is
opened,
it
will
add
the
needs.
Triage
label
yeah
go
ahead.
A
Global
flags
yeah-
I
still
have
that
one
I'll
definitely
have
a
look
at
it.
L
F
L
That's
that's
right,
and
I
know
this
is
kind
of
a
long-standing
issue
with
some
history
on
it.
So
maybe
I'm
wading
into
something
a
little
deep
but
hoping
this
is
sort
of
in
the
right
direction
and
happy
to
to
revise
it
as
much
as
you
guys
need.
A
Yeah
I'll
we
we,
we
agreed
on
the
direction
and
you
need
to
have
a
look
at
the
vr
and
I'll
leave.
You
a
comment
there.
A
Short,
basically,
we
don't
want
to
change
the
order,
but
we
do
want
to
have
a
warning
if
I
remember
correctly
that
pr
was
already
heading
that
direction.
H
So
I
think
this
pr
just
to
summarize,
though,
is,
is
to
basically
support,
and
I
don't
know
if
maybe
this
pr
was
before
we
decided
that
what
you
just
said,
but
this
this
would
allow
global
flags
to
be
before
the
command,
but
only
global
flags.
H
A
It
was
decided
because
I
was
talking
with
jordan
and
that
was
initially
written
down
in
the
pr,
but
I
haven't
looked
at
the
current
state
of
the
pr.
So
I
need
to
have
a
look
once
again
and
basically,
the
answer
and
their
path
will
be
not
to
change
the
defaults
for
flag
parsing,
and
I
think,
last
time
or
two
meetings
ago.
L
I
think
I
think
that's
a
different
pr,
so
there
might
be.
A
Yeah,
I
have
two
in
my
queue
and
they
they
are
somewhat
related
to
I'll,
have
a
look
at
both
and
I'll
I'll
be
leaving
comments.
There.
F
Yeah
thanks,
I
just
want
to
say
thanks
colleen
for
for
taking
initiative
on
this,
and
I
think
that
there's
some
history
and
some
subtlety
as
well
that
you
may
have
waded
into.
A
That's
true:
okay,
I'm
not
seeing
any
other
topics.
A
Does
anyone
have
any
other
items
that
I
want
to
share
with
the
group
stand
ups
or.
F
Related,
so
I
so
much
I,
I
had
a
a
quick
look
at
our
recent
goals
for
q4
and
I
was
curious
if
you
what
you
thought
about,
maybe
revisiting
those
goals
on
a
periodic
basis,
just
to
see
what
our
progress
is
towards
those
goals
for
this
quarter,
just
so
that
we
don't
get
to
the
end
of
the
quarter
and
say
you
know
what
about
this
particular
goal.
A
You
know:
okay
yeah.
Definitely,
I'm
I'm
more
than
open
to
half.
K
A
Actually,
I
have
one
topic
to
to
add
to
that
goal,
so
there
has
been
good
moves
around
the
extraction
of
component
helpers.
The
the
component
helpers
is
a
repository
where
we
will
be
allowed
to
move
some
of
the
shared
codes
between
kubernetes.
A
A
A
I
updated
the
the
cap
for
that
one
last
week,
so
it
made
before
the
freeze.
The
lever
burn
actually
updated.
The
cap
for
cube
cattle
debug,
so
we'll
lee
will
be
working
on
promoting
cube
cattle
debug
as
well
within
120.
C
Is
there
any
additional
work
for
that
on
the
customized
side,
for
basically
right
that
that
is
one
blocker
for
getting
the
latest
version
of
customize
moved
back
into
cube,
cuddle,
correct.
A
If
I
remember
correctly,
there
was
there
were
some
issues
with
how
customized
was
vendoring.
Api
machinery
and
jeff
was
working
on
that
particular
bit.
A
M
F
Work
on
the
removing
the
api
machinery
from
the
customize
in
order
to
be
able
to
to
be
able
to
reintroduce
customizing
the
coupe.
A
Basically,
okay,
cool.
A
Up:
okay
hearing
none,
I'm
gonna
give
you
back
about
three
six
minutes
of
your
time.
Thank
you
very
much
and
see
you
all.
In
two
weeks,
bye,
y'all.