►
From YouTube: Kubernetes SIG CLI 20180523
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
Okay,
good
morning,
good
afternoon,
good
evening,
depending
on
where
you
are
it's
May
23rd-
and
this
is
our
another
60
line
meeting
I-
think
we
don't
have
done
many
topics,
but
let's
go
are
shortly
through
all
of
them.
So
two
announcements.
So
in
between
the
last
time
that
we
spoke,
and
today
the
sick
release,
actually
managed
to
extend
the
code.
Slosh
and
code
freeze
dates
by
a
week,
so
they
moved
to
May,
29th
and
June
5th.
Accordingly,.
A
Which
means
we
have
additional
week
to
crank
some
code
into
one
11
release,
so
I
hope
that
extended
period
of
development
will
will
become
useful
for
everyone,
and
but
that
also
means
that
the
entire
queue
is
so
stable.
So
congratulations
and
thanks
to
every
single
person
that
devoted
their
time
to
maintain
the
tests
flakiness
at
a
very
low
level,
which
brings
us
to
the
next
topic.
The
test
grid,
the
person
that
was
looking
after
the
tasks
over
the
past
two
weeks
was
one.
So
one
can
you
give
us
a
short
update
on
that
sure.
A
B
For
the
most
part,
that's
been
a
few
flakes,
mostly
due
to
time
outs
happening,
so
recording
those
I've
either
linked
a
few
issues
on
the
log.
If
you
take
a
look
at
the
testing
playbook,
there
have
been
two
incidents
where
we've
actually
just
had
highly
recurring
test
test
lakes,
and
so
those
have
been
noted
in
being
basically
in
the
failures
section
at
the
bottom
of
the
test.
B
I
will
paste
a
link
to
the
playbook
itself.
I,
don't
pasted
in
the
chat
as
well.
Also
I'm
gonna
had
an
update
of
the
test
coverage.
We've
gained
three
or
four
new
packages
for
the
most
part,
coverage
has
remained
the
same
or
improved
a
little
bit
and
yeah
the
tune
or
the
single
item
that,
but
is
recurring
at
the
moment,
is
the
e
to
e
cops
age
of
Leo's
failure?
A
Ok,
the
next
person
taking
over
the
test
cop
duties
will
be
actually
me
so
yeah.
That's
pretty
much.
It
I
think
about
that
at
the
top
of
the
testing
playbook
doc
that
you
just
saw
a
minute
ago.
We
have
a
sign
up,
April
Jackie's
gonna,
be
a
cop,
a
testing
cop
for
one
of
the
next
time
frames
feel
free
to
add
your
name
to
that
list.
A
So
far
we
have
people
sign
up
all
the
way,
til
til,
beginning
of
July,
which
is
very
good,
and
thank
you
every
thank
you
a
lot
to
every
single
person
that
sign
up
there.
Ok,
so
now
we
can
go
actually
to
the
main
topics
we
have
3.
Let's
start
with
the
simplest
one
that
I
put
here,
because
I
want
to
start
the
discussion
about
the
exit
codes
for
two
minutes.
I
have
lender
two
issues
that
brought
me
there.
A
Yes,
on
qtd
roth,
can
I
the
person
complaining
would
like
to
see
a
exit
code
non-zero
exit
code
whenever
he
is
not
allowed
to
perform
some
action.
The
same
applied
to
keep
CTL
apply,
which
is
the
other
issue
there.
A
person
would
like
to
see
a
non-zero
exit
code,
some
sundries
exit
code
for
for
the
cases
where
apply
did
not
perform
any
changes,
I'm
guessing
that
particularly
for
apply.
That
will
be
a
little
bit
harder,
given
that
we
are
currently
in
the
works
of
moving
apply
to
the
server.
A
A
B
So
I
think
that
so
I,
like
it
I'm
bored
of
it
and
I,
don't
think
we'd
have
to
worry
about
backwards.
Compatibility
that
much
because
the
people
are
already
parsing
the
output,
then
I
would
not
really
expect
them
to
be
also
checking.
You
know
for
a
zero
exit
code
status
of
your
currently
see
so
adding
a
nexus
plateau.
Somehow.
C
C
Yeah
you
do
with
that.
This
isn't
one.
If
we
look
at
what
happens
in
like
tools
in
physics,
we
usually
have
grabbed,
for
example,
which
fails
if
nothing
is
found.
We
have
thief
which
is
going
to
fair
leave.
A
difference
is
detected,
so
it's
very
common
for
these
tools
to
later
on
exit
us.
If
something
specific
happens,
which
is
not
an
hour
so
I
think
it
makes
sense
for
cube,
co2h
or
Nancy.
Oh
well,
it
makes
sense
not
not
necessary
if
an
L
happens.
A
Okay,
I'm
hearing,
none
so
I
think
that
slowly
oxygen
and
we
will
be
able
with
as
soon
as
we'll
be
done
with
cleaning
cubes
video.
We
can
slowly
pursue
this
option
of
returning
nonzero
codes
and
that
will
also
simplify
the
scripting
and
something
that
I'm
always
looking
towards
is
the
UNIX
philosophy
off
the
command
for
for
CLI,
so
I
think
that
was,
it
was
pretty
fast.
A
Let's
move
and
the
to
the
second
topic.
Last
time,
Matt
Rowena
mentioned
survey
results
from
from
the
slick
apps
that
they
performed
some
time
ago.
So
who
won
went
through
the
results
and
we
extracted
only
those
that
are
applied
for
the
cube,
CTL
command,
specifically
and
it'll.
Let
who,
on
summarize
the
docks
itself
sure.
B
So
the
second
question
after
that
is
why
do
people
use
queue
control
and,
for
the
most
part,
the
the
things
I
expected
from
there?
The
most
we're
managing
config
managing
a
cluster
people
also
use
the
to
troubleshoot
a
cluster,
particularly
for
obtaining
logs
for
deployments
or
pods.
They
also
use
debug,
as
well
as
monitor
their
applications.
B
They
use
it
to
deploy
and
run
apps.
They
also
use
it,
and
this
was
something
I
saw
frequently
was
they
use
it
for
cue
control,
apply
as
well
as
for
proxying
and
portfolium,
and
lastly,
people
use
it
for
our
back,
particularly
for
managing
permissions,
so
those
items
are
the
ones
that
stood
out
the
most.
It
applies
for
this
question,
so
if
anybody
has
any
questions
or
comments
feel
free
to
interrupt
me
at
any
time
by
the
way.
B
The
next
question
after
that
is
what's
good
about
Q
control.
The
main
point
I
spit
out
is
that
people
like
its
simplicity
and
that
it
just
looks
the
question
after
that
said
or
asked
what
people
would
like
to
see
added
or
encrypting
q
control.
The
main
points
for
that
were
that
people
wanted
better
consistency
and
predictability
across
you
know,
options
for
commands.
B
People
want
better
support
for
old
versions,
the
one
better
filtering
and
sorting,
particularly
not
using
JSON
path,
so
they
want
to,
for
example,
provide
a
column
name
and
be
able
to
sort
in
that,
and
so
this
is
a
good
point
since
matcha
and
I
have
briefly
talked
about
this
as
well.
I
think
it'd
be
a
good,
including
not
to
do
this.
B
People
want
predictable
output
formats
for
all
comments,
particularly
yeah,
Moe
and
Jason,
so
they
basically
expect
that
any
command
that
provides
output
is
able
to
provide
it
in
Yamaha
Jason
as
well.
People
also
want
command
aliases,
and
by
that
they
mean
similarly
to
get
they
want
to
to
the
gift
man.
They
want
to
be
able
to
alias
describe,
for
example,
so
they
want
to
be
able
to
say
cube.
A
B
De
se,
for
example,
and
then
whatever
else
I
believe
this
will
be
possible
regardless,
once
we
once
we
improve
our
support
for
extensions
and
plugins
people
might
be
able
to
make
an
extension.
That
is
simply,
you
know,
a
shorter
version
of
describe
so
to
lend
the
extension
de
se
and
it'll
just
really
to
the
description
of
people
also
want
the
ability
to
switch
between
the
in
spaces
without
having
to
specify
the
namespace
for
every
single
command.
So
they
want
something
like?
U?
B
Controller
can
Vega
set
in
space
currently
they're
using
a
tool
called
cube,
CTX
I
believe
which
does
this
for
them,
but
I
don't
see
any
reason
for
us
not
being
able
to
include
this
sub-command
people
want
better
piping
across
commands
as
well.
So
if
there's
any
cases
where
the
output
of
one
command
cannot
be
piped
in
and
into
another
control
command,
we
should
identify
those
and
basically
make
the
whole
thing.
Consistent.
B
People
are
also
asking
for
a
who
am
I
command.
So
open
shift
has
something
like
this
you're
able
to
specify
Who
am
I
as
a
sub
command,
and
it's
able
to
tell
you
your
username,
the
current
server
that
you're
talking
to
I
took
kimi.
If
you
have
a
token,
so
I
think
to
control
could
benefit
from
this
as
well
so
I
agree.
People
want
better
third
party
a
resource
support.
They
want
specifically
CR
DS
to
be
treated
as
first-class
citizens
or
as
close
to
it
as
possible.
B
B
Also,
people
want
better
documentation
and
organization
of
supported
out
books,
so
I
believe
this
fits
in
with.
Basically,
they
expect
that
every
single
command
that
I'd
put
something
supports
an
Lea
Jason
and
last
people
want
colors
as
well
as
plugins,
better
plugin
support
for
that
we're
keeping
cool.
So
that
takes
care
of
that
one
basically
for
the
template
engine
people
just
either
said
yes
or
no.
B
For
the
most
part,
there
was
some
disagreements
as
to
how
people
feel
about
the
current
and
putting
engine
just
go
to
Yale,
complaining
that
it's
difficult
to
use
arcane
art
to
get
into
some
said
it's
easy
so
that
there's
mixed
feelings
on
the
graph,
but
overall
I,
don't
foresee
us
changing
that
template
engine,
at
least
in
the
near
future,
for
the
next
two
questions
where
people
are
rating,
basically,
the
the
ability
to
visualize
subjects
that
make
up
an
application.
The
answers
to
this
we're
mostly
varied
between
sometimes
and
usually
so.
B
B
D
Is
Matt
friend
I'll
jump
in
real
quick?
Thank
you
for
going
through
all
this
I'm
glad,
there's
some
useful
nuggets
you
could
pull
out
of
there.
I
also
wanted
to
say
when
it
comes
to
grouping
things
and
seeing
it
one
of
the
things
tick
apps
is
working
on
is
the
application
CRD.
In
addition
to
that,
I
can
dig
it
out.
D
There's
a
pull
request,
I'm
coming
up
with
to
document
labels
and
annotation
conventions
to
be
able
to
tie
things
together
as
well,
and
so
there
may
be
some
some
to
use
the
enterprise
word
synergy
between
these
efforts
and
maybe
what
coop
control
eventually
gets
into
and
how
something
like
that
could
come
together.
I,
don't
know,
I
just
wanted
to
make
sure
to
cross
pollinate
information.
B
A
D
Disagree:
I
guess.
The
one
thing
that
will
jump
in
is
a
tool
is
part,
functionality,
part
Convention
and
a
whole
lot
experience,
and
so,
if
folks
want
this,
one
ways
to
say
is
to
use
one
of
these
other
convention
and
then
here's
or
one
of
these
other
tools
and
then
here's
how
you
can
use
that
alongside
of
Kubb
CTL,
it's
kind
of
an
explainer
because
documenting
and
sharing
and
make
it
easy
for
people
to
discover
how
to
do
these
things
is
part
of
a
tool
when
stuff
isn't
really
documented
or
communicated.
D
It
becomes
a
pain
point,
and
so
that's
what,
when
I
think
of
something
like
this?
Maybe
the
applications
here
deed
does
get
into
good
shape,
and
it
is
one
of
those
things
you
could
say.
Here's
how
you
use
that
with
Kubb
CTL
to
accomplish
this
task,
or
maybe
here's
how
you
can
use
the
recommended
labels
and
annotations
to
accomplish
those
tasks
with
what
you
already
have,
but
it
helps
people
accomplish
what
they're
trying
to
which
is.
You
know
it's
not
always
just
about
coding
functionality
and
I.
A
We
have
so
many
different
sub
commands
within
the
cube
CDL
they're,
currently
without
proper
completions
and
I'm.
Honestly
I
I
can't
imagine
anyone
using
Kip's
appeal
without
completions,
but
it
looks
like
some
people
are,
but
even
then
it
is,
it
is
quite
cumbersome
to
figure
out
where
some
commands
might
be
might
be
hidden
under
which
main
sub
commands,
because,
for
example,
a
lot
of
people
complain
that
there
is
no
option
for
switching
contexts
or
verifying
or
updating
contexts
where
actually,
there
is
a
cube,
see
I'll
contact
command
which
allows
to
modify
the
context.
A
The
other
thing
that
people
complained
and
I
think
is
worth
adding
is
the
ability
to
manage
the
namespace
you're
currently
working
on,
because,
for
example,
if
you're
working
for
a
extended
period
of
time
with
a
namespace,
the
requirement
to
pass
the
name
every
single
time,
you're
invoking
a
command
it's
becoming
cumbersome.
So
we
actually
talk
with
with
who
on
shortly
before
this
meeting,
and
maybe
we
should
extend
the
config
command
with
the
ability
to
set
the
current
namespace
and
that
should
simplify
the
managing
and
working
with
resources.
A
That
will
also
simplify
the
case
where
you're
working
with
VR
AB,
because
usually
in
your
application,
will
will
be
within
a
single
namespace
and
that
should
simplify
and
then
you
know
improve
the
user
experience,
because
a
lot
of
people
complain
that
they're
invoking
some
commands
within
QC
deal
requires
a
lot
of
typing.
This
was
the
the
requirements
for,
or
a
suggestions
for,
good
type
of
fatty
acids
for
some
commands.
Obviously,
we
are
not
managing
the
aliases
for
resources,
because
these
are
coming
from
server
and
I.
A
Honestly,
don't
want
to
work
with
that,
because
that
falls
under
the
siga
api
missionary
area,
but
definitely
a
aliases
is
something
that
we
might
be
considering
to
I.
Don't
know,
for
example,
to
to
invoke
cube
CTO
get
comments,
go
all
the
way
down
to
K
G
and
then
Pio
that
will
could
be
a
shortcut
from
cube.
Cto
get
pause
while.
D
You're
on
aliases
I'm
gonna
thrown.
If
you
look
at
like
get
aliases
right,
it's
not
always
just
a
shorter
version,
a
name
like
instead
of
yet
you
have
the
letter
G,
it
can
also
be
a
whole
length
of
multiple
commands
and
information
passed
together.
Yeah
could
be
something
so
I
could
custom
define
G
to
be
get
pods
and
maybe
even
with
some
flags,
and
then
that
would
then
work
as
an
alias
and
so
there's
more
than
one
way
to
think
about
aliasing.
Here,
oh
yeah.
A
Definitely
so
that's
that's
an
open
issue.
I
think
that
at
some
point
in
time
it's
worth
it's
worth,
adding
such
a
functionality
inside
of
the
inside
of
cube,
CTO
I'm,
not
sure
about
the
installation
process,
though,
because
usually
the
installation
process
is
a
thing
that
is
owned
by
the
distributions
or
by
the
vendors
providing
YouTube
CDL
I
know,
for
example,
that
OpenShift
provide
whenever
OpenShift
provides
the
binaries
for
for
the
server
along
the
way.
There
are
client
binaries.
Similarly
GCE
when
you
download
the
G
cloud
suit,
we
are
getting
a
cube,
CDL
embedded
within
it.
A
Other
tools
that
provide
more
more
rich
capabilities
is
also
a
good
one,
because
I
honestly,
don't
see
any
and
the
reasons
that
keep
CTL
would
be
similar
to
cube
CTX
if
I
remember
correctly,
the
name
which
is
rappelled
type
of
CLI,
which
maintains
the
current
context
up
to
the
current
pod
that
you
work
with,
so
that
after
you
invoke
get
pod
and
you're
invoking
logs
it,
it
will
autofill
on
the
pod
name
from
the
previous
command.
That's
definitely
something
that
I
I'm,
not
gonna,
I'm,
not
gonna,
be
happy.
A
Okay,
so,
as
a
closing
remark,
I
totally
encourage
every
single
of
you
to
have
a
look
at
the
dock.
You
know
maybe
you'll
figure
out
some
other
opinions
out
of
these
dogs
so
feel
free
to
share
with
us
next
time
during
the
meeting.
The
last
topic
I
believe
Shawn,
you
added
I
want
is
the
six
you
like
governing
charter,
so
take
it
away.
Yeah.
E
The
steering
committee
member
who
will
be
guiding
our
Charter
is
Jo
beta
and
so
yeah.
Please
have
a
look
at
what
is
a
a
formalization
of
the
roles
and
the
technical
decision
making
process
for
our
sig,
and
let
us
know
your
feedback
within
the
next
few
days,
as
we
move
to
the
next
phase,
to
to
actually
formalize
this
Charter
to
to
get
it
accepted
and
to
to
commit
it
into
into
kubernetes
community.
A
Okay,
that
make
sense.
So
if
you
haven't
seen
the
governing
charter,
it's
not
that
long,
it's
it
outlines
the
main.
The
main
effect
objectives
of
our
artistic,
introducing
some
small
changes
that
we
will
announce
as
well
once
the
the
Charter
gets
fully
published,
feel
free
to
look
at
it.
If
you
have
any
objections,
comments,
suggestions
feel
free
to
leave
a
comment
and
whatever
you
feel
make
sense
will
be
taken
into
consideration
when,
when
approving
this,
this
charter
are
there
any
questions
with
regards
to
the
Charter
itself,.
A
E
Instead
of
actually
doing
a
real
apply
where
we
just
want
to
get
a
simple
and
to
end
working
and
then
iterate
on
that
and
we've
just
yesterday
got
a
simple
end-to-end
it.
It
actually
returns
an
error.
There's
still
what
defaulting
issue
I
mean
that
yet
so
yeah
there's
there's.
Definitely
you
know
quite
a
ways
to
go,
but
but
it's
it
feels
good
to
be
able
to
actually
see
this
command.
This
new
command
Coupe
cuddle
alpha
apply,
which
is
how
we
think
we'll
start
off
with
that,
actually
connecting
to
an
endpoint
and
doing
something.
E
So
what
else?
So
we
will
be
I
think
publishing
a
document
to
describe.
There's
the
one
of
the
more
complicated
areas
that
we're
gonna
have
to
deal
with
is
conversion
and
defaulting,
and
how
that
happens
throughout
this
pipeline
and
it
sounds
like
daniel
smith
will
be
publishing
that
I'll
bring
that
to
2r
cigs
so
that
so
they
can
get
a
wider
audience,
and
I
think
that's
about
it.
Unless
anybody
has
any
questions.
B
This
is
us
part
of
an
ongoing
pattern
to
basically
provide
a
sort
of
standalone
repository
with
CLI
tools
that
we
dog
through
it
ourselves,
but
also
provide
to
third-party
clients
and
extensions
so
that
we
have
a
unified
way
of
talking
to
both
week.
We
have
a
basically
a
way
to
enforce
a
unified
way
of
talking
to
the
yeah
so
yeah.
We,
we
should
hopefully
be
completely
decoupled
from
the
factory
in
the
next
week
or
two,
and
what
that'll
mean
is.
A
F
Sure
hi,
my
name
is
Michael
Gugino
I
work
for
a
Red
Hat
and
work
on
the
open
shift.
Cluster
lifecycle
team,
it's
hoping
to
become
a
contributor.
Remember
here,
doing
some
moonlighting
on
the
CLI
and
just
want
to
introduce
myself
I
have
a
small
unit
test
PR
out,
it's
kind
of
like
a
flare
send
up
so
I
just
want
to
say
hello.