►
From YouTube: Kubernetes SIG CLI 20170315
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
morning,
everyone,
this
is
the
cig
CLI
meeting
for
March
15,
seeing
that
we
don't
have
any
specific
topics
in
the
agenda
for
today,
so
I
guess
first
of
all,
then
we
can
go
around
with
introductions
if
there
is
anyone
new
in
the
meeting,
so
it
doesn't
seem
so,
but
if
anyone
would
like
to
introduce
themselves
and
maybe
talk
about
proposals,
anyone
has
anything
like
that.
I.
C
D
A
Okay,
nice
to
meet
you
and
yeah
I
guess
we
we
already
met
on
github
for
requests,
but
anyway
good
to
meet
you
man
and
anyway.
I
think
I'm
going
to
get
started
with
statuses,
then
so
after
me,
if
anyone
else
would
like
to
share
statuses
specifically,
I
would
like
to
do
that
because
the
past
few
days
have
been
working
with
the
Service
Catalog
stuff.
So
it's
specifically
the
support
on
cube
CTL
for
Service
Catalog,
think
so
so
for
anyone.
That
may
not
be
aware.
A
The
Service
Catalog
reports
is
a
turbonetics
into
the
incubator
project,
so
it's
being
led
by
paul
murray
and
basically
the
idea
is
to
have
an
implementation
on
the
cube
server
for
the
the
Service
Catalog
open
API
proposals
and
I
specifically
started
working
on
a
cube,
CTL
command
for
interacting
with
the
Service
Catalog.
So
the
basic
idea
is
that
you
know
third-party
providers
or
even
other
people
in
your
company
etc
could
provide
services
ready
to
be
consumed
by
your
applications.
A
So,
for
example,
I'm
I'm,
a
developer,
I'm
writing
an
application
to
be
deployed
inside
my
company
and
I
need
some,
let's
say,
for
example,
a
messaging
service
and
so
I
would
be
able
to
go
to
the
Service
Catalog
inside
of
our
company
and
check
if
anyone,
if
there's
any
messaging
service
available
for
me
for
me
to
consume
and
I,
could,
for
example,
we
figure
out
that
someone
in
my
company
provides
rabbit
m
mq
service,
for
example,
something
that
they
manage,
but
they
expose
us
as
a
service
in
the
service
catalog
for
anyone
else
inside
the
company
to
be
able
to
consume
their
their
application,
so
media
as
an
app
developer,
I
could
check
the
Service
Catalog
discover
that
we
have
a
RabbitMQ
service
and
I'll
good
right
request
the
service
instance
for
of
that
rabbitmq
and
bind
death
instance
into
my
my
application.
A
So
basically
what
I've
been
doing,
I'm
working
on
a
proposal
so
no
implementation.
Yet,
but
a
written
proposal
about
the
commands.
We
will
have
under
cube
CTL
to
interact
with
the
Service
Catalog,
so,
for
example,
I
could
list
all
services
available
for
me
to
consume.
I
could
believe
the
instances
I
specifically
have
already
and
to
which
of
my
ibanez
is
deployed.
The
vocation:
each
of
my
services
instances
are
bound,
for
example,
so
I'm
consuming
a
given
rabbitmq
and
it's
bound
to
this
specific
application
that
I
I
wrote.
A
So
we
are
likely
going
to
have
commands
like
you
to
tell
Service
Catalog
least
install
things
like
that.
So
that's
not
how
they're
going
to
look
like
I'm
still
working
on
a
proposal
for
the
command,
but
anyway,
I've
been
doing
that
in
the
past
few
days,
and
I
will
pretty
soon
open
up
request
asking
everyone
for
feedback,
especially
about
how
the
command
will
look
like
you
know:
command,
name,
subcommands
and
the
flag
bill
they're
going
to
have
a
bit
available.
So
there
are
a
few
personas
in
this
service
Kellogg
if
forth.
A
So
there
are
the
people
that
you
know
provide
services
and
for
others
to
consume,
so
so
for
these
people
we
most
likely
will
not
have
you
know
and
user
commands
that
you
will
still
rely
on
that,
create
yes
delete
and
the
command.
We
already
have
you
no
more
like
lower
level
comments,
however,
for
end-users
developers,
actually,
people
that
just
want
to
consume
a
service
from
a
Service
Catalog.
We
will
have
those
like
top-level
commands,
like
least
what's
available
in
the
Service
Catalog
bind.
These
requests
requests
a
munitions
and
mind
install
these
into
my
application.
A
In
some
cases,
we
you
could
even
have
different
plans
so,
for
example,
for
third-party
providers,
let's
say
that
we
have
someone
that
provides
MongoDB
as
a
service
and
they
have
different
plans.
You
know
more
or
less
storage
and
things
that
I
can
pay
monthly.
Something
like
that.
So
for
all
of
that,
we
are
going
to
have
like
top-level
comments,
so
being
able
to
leave
the
services
available,
what
the
plans
are
available
and
eventually
request
and
enroll
in
give
them
plan
that
will
play
lex
pay
for
monthly.
Something
like
that
anyway.
A
I
will
open
a
proposal.
Pretty
soon
there
is
the
proposal
work
going
on
and
also
some
actual
development
work
happening
to
make
him
CTL
work
with
the
Service
Catalog
ati
today,
because
Service
Catalog
API
is
basically
another
instance
of
Cuba
natives
with
different
API
resources,
so
it
has
its
own
its
own.
It's
ed,
for
example.
It's
basically
a
cuba
netted
service,
but
instead
of
having
bad
reputation
controllers
things
like
that,
it
has
a
service
instances
binding.
A
You
know,
brokers
and
you
know
other
kinds
of
resources.
So
something
else
I'm
working
on
is
making
sure
that
keep
CTL
can
play
together
with
these
other
kinds
of
resources.
So,
for
example,
a
cube
CT
I'll
get
all
if
something
is
pretty
broken
today,
when
you
try
to
run
against
a
Service,
Catalog
API,
because
all
tries
to
translate
into
into
a
pod
replication
controllers
and
a
given
set
of
commands
that
we
don't
have
in
a
Service,
Catalog
API
right,
so
I'm
writing
to
make
sure
that
all
can
be
understood
by
different
API
servers.
A
The
right
way
you
know
so
forth
upon
the
Service,
Catalog
API
all
actually
means
binding
service
services.
Instances
and
things
like
that
and
in
my
cube
service,
all
actually
means
father,
replication
controllers
deployments
and
other
remember
exactly
the
list
of
resources
so
anyway,
I've
been
working
on
that
and
the
altima
vs.
prediction,
but
that's
it
for
me.
In
the
past
few
weeks,
I
really
didn't
join
the
last
couple
6-cl
anything
because
I
was
on
paternity
leave,
but
that
for
me
in
terms
of
status,
anyone
else
would
like
to
share.
A
Yeah,
that's
a
very
good
question
because
these
are
set
of
commands.
Writer
for
the
service.
Kellogg
is
exactly
the
thing
that
would
really
apply
to
binary
extensions
like
gypsy
child,
plugins
I'm,
not
sure.
If
we
will
have
you
know
a
proposal
around
plugins
learning
any
time
you
know
for
for
having
those
written
as
s
plugins,
it
would
fit
probably
perfectly,
but
if
we
don't
have
a
proposal
for
plugins
on
time,
this
is
probably
going
to
just
be
regular.
You
know
tips
etl,
commencing
the
the
actual
regular,
regular
binary,
probably
other
a
different
section
help.
A
Yeah,
that's
right
at
least
for
now.
There
is
some
somewhere
going
on
in
terms
of
making
them
all
third-party
resources,
but
for
now
at
least
it's
a
completely
different
instance
of
Cuban
edit,
which
it's
not
on
each
CD,
so
you
actually
have
to
switch
on
a
cube
cgl
in
on
cube
config
between
one
and
the
other
API
servers.
You
know,
okay,.
B
A
D
I
also
working
on
couple
of
PR
1y.
If
I
I
made
a
umbrella
issue
to
chair,
checking
all
the
coop
controlled
man
which
returns
0,
but
on
arrow,
so
the
pr
Abby
is
for
2999
and
I.
Just
I
just
searched
all
the
coop
control
command
to
check
all
all
those
command
a
check
error
in
fine.
So
you
guys
can
take
a
look
when
you
guys
also
I've
been
working
on
apply
subcommand,
so
I
will
apply
edit
laughter.
Apply
is
ready
for
review
for
a
couple
months,
but
I
think
you
guys
are
just
too
bearish.
D
A
D
A
A
Okay,
other
than
that
other
than
status
updates.
Last
time,
I
checked
our
blockers
yesterday,
I
think
we
had,
we
still
had
four
blockers
I
think
up
actually
blocker
issues
but
trailing
test
cases.
So
if
anyone
I
really
didn't
affect
how
we
are
looking
today,
but
yesterday
we
had
like
four
to
be
addressed
for
1.6.
A
So
if
anyone
else
has
some
spare
time
and
would
like
to
take
a
look
at
some
of
those
I
think
at
least
10
one
of
them
was
fixed
yesterday,
so
we
most
likely-
you
will
have
three
today,
but
I
really
didn't
check.
So
if
you
guys
would
like
to
take
a
look
also,
we
still
have
some
some
things
to
be
addressed
in
terms
of
test
failures
for
1.6.
We
are
not
looking
that
bad,
but
you
know
we
still
have
stuff.
A
B
A
A
Okay,
yeah
I
think
that's
the
document.
Philip
was
working
on
for
1.6.
It
looks
like
we
are
pretty
good,
except
for
some
documentation
work
that
we
didn't
finish
really,
but
you
started
listening
things
for
1.7
already,
so
maybe
as
an
exercise,
we
could
take
a
look
at
those,
maybe
try
to
assign,
if
there's
something
we
would
like
to
have
a
dress
for
1.7.
That
does
not
have
an
honor.
Yet
something
like
that.
E
Yeah,
I
also
have
I
guess
it
feels
not
here
today,
but
maybe
at
the
teacher
meeting
at
some
in
general
comments
on
the
road
map.
So
just
shared
this
with
me
yesterday,
so
I
just
started
looking
at
it,
but
they're
definitely
some
things
that
would
be
useful
to
be
aware
of
from
the
broader
project,
and
one
is
up.
We
need
apply
in
so
many
places
in
the
system
that
I
want
to
move
it
server-side
sometime
this
year.
E
A
E
I,
don't
know
if
he
saw
the
insta
gas.
That
I
think
is
where
the
open
compose
effort
was
announced,
basically
a
simplified
configuration
format
for
serenity.
So
that's
something
probably
you
know
right
now
separate
tool
and
you
know
for
the
foreseeable
future
is
going
to
be
a
separate
tool.
But
it's
another
command
line
tool
go
to
bed
useful
to
keep
an
eye
on
one
thing:
people
you
wanted
to
have
requested
many
times.
That's
sort
of
the
line
with
apply
is.
E
Declared
of
expression
of
some
other
common
generators
so
in
particular,
create
config
map
and
create
secret
people,
want
to
check
in
and
maintain
the
original
raw
bits
of.
You
know
whatever
form
their
their
secrets
are
their
configuration
data
wads,
but
they
want
to
declarative
workflow
for
updating
the
config
master,
the
secrets
so
I
think
that's
an
area
where
it's
worth
thinking
about.
E
You
know
for
those
popular
generators
like
create
custom,
a
secret
maybe
even
run
whether
we
should
have
some
sort
of
cute
control,
specific
configuration
syntax
for
that,
and
that's
also
kind
of
why
I
brought
up
hoping
composed,
because
you
know
once
we
depart
from
the
literal
API
resources
for
something,
then
more
things
become
accessible.
There
are
some
other
things
people
ask
for
over
the
years
as
well,
like
defaults,
specified
across
multiple
resources,
or
you
know,
with
the
prune
discussion
people
talk
about.
E
Well,
maybe
there
should
be
a
way
of
specifying
declaratively
the
the
selector
for
the
set
of
resources.
Prune
should
apply
to
you
so
that
you
don't
leave
it
off
the
command
line
by
mistaken
to
eat
a
bunch
of
stuff.
You
didn't
want
it
to
eat,
or
you
know
a
specific
list
of
mo
files
in
the
directory
so
that
you
can
put
other
configuration
files
potentially
side
by
side
with
your
benetti's
ones.
Things
like
that,
so
I
think
that's
worth
thinking
about
for
the
years
you
know
keep
control
starts
to
mature.
You
know.
E
Seeing
things
like
extensions
is
great.
One
thing
I
wanted
to
mention
is
version
skew
as
well
a
officially
we
support
one
minor
release
of
version
q
rights,
a
1.5
should
work
with
1.6
if
it
doesn't
work
in
my
understanding,
some
things
don't
work
like
at
it.
That's
a
flat-out
bug
right,
that's
something!
We
should
consider
patching
1.56
yeah.
A
Absolutely
I
totally
agree,
and
actually
we
had
because
cute
end
to
end
tests
were
renamed
recently.
We
actually
had
the
death
finding
a
couple
bug
that
we
got
fixed,
which
was
nice,
but
we
are
totally
aware
that
there
are
something
like
added
that
doesn't
work
pretty
well,
so
yeah
I
totally
agree
in
terms
of
moving
moving
teams
to
the
server
side.
I
know
that
Clayton
has
a
proposal
for
the
jet
sets
set
of
commands
and
I
always
support
that.
You
know
we
hearing
on
open
chief,
that's
something
I
didn't
really
open.
A
A
an
issue
on
to
burnet
is:
what's
hearing
an
open
shift.
Recently
we
had
our
QE
team,
find
a
performance
issue
and
keep
CTL
yet
when
you
handle
large
sets
of
resources,
so
that
was
for
something
like
twenty
thousand
secrets
or
something
like
that.
That
takes
around
30
seconds
the
list
and
that's
basically
because
having
too
many
things
on
the
client
side,
you
know
we
we
use
a.
A
We
use
a
lot
of
JSON
Dixon
parsing
and
basically
you
know
the
the
urging
of
the
those
are
performance
issues
was,
for
you
know
those
30
seconds
around
20
was
issues
related
to
JSON
or
e
mo
parsing
things
like
that.
So
up
and
moving
that
to
the
server
side,
we
will
likely
see
you
know
a
good
part
of
it.
So
for
the
exemplary.
A
I
agree
and
I
mean
before
not
sure
if
you
were
in
the
meeting
already
I've
been
working
on
the
Service
Catalog
at
fort,
where
we
are
having
to
support.
You
keep
CTL
against
a
completely
different
set
of
resources,
and
we
are
broken
in
many
ways,
because
we
have
a
lot
of
things.
Hard
code
is
on
the
client
side
right
so,
for
example,
the
diol
alias
trek
holder
for
the
resources
it
translates
to,
like
all
actually
means,
pods
reputation,
controllers,
etc.
A
So,
whenever
that
against
a
different
set
of
objects
or
server
resources,
etc,
you
are
some
completely
broken.
So
moving
are
willing
to
make
sure
we
have
enough
information
in
the
discovery.
Api
yeah
yeah,
moving
those
to
the
server
side,
making
sure
that
they
are
discoverable
instead
of
hard
coded
on
the
client
will
likely
prove
that
a
lot
right
so
yeah
and
things.
A
I
think
I'll
make
sure
all
those
comments
are
included
in
that
this
document
Phillip
started
to
write
for
1.7
and
the
other
upcoming
releases,
and
we
can
keep
working
on
that
on
them
and
maybe
have
a
further.
You
know
actually
planning
discussion,
the
maxi
6-cl
I'm
meeting
with
Felix,
so
that
we
make
sure
we
have
all
those
things-
and
you
know
it's
very
important
to
it
to
us
and
keep
keep
them
coming,
because
we
not
always
have
some
feedback
from
from
people
in
the
trenches.
A
E
A
Yeah
yeah
yeah,
absolutely
we
even
had
you
know.
A
few
weeks
ago
we
worked
on
a
written
document
for
out,
contribute
to
the
cube
CTL,
specifically,
but
all
the
things
sig
GLI
and
because
I
was
on
paternity
leave.
I'm
really
not
sure
how
well
that
was
a
you
know,
discussed
in
the
community,
so
I'm
not
even
sure
that
was
really
shared
on
the
main
mailing
list.
So
I
didn't
yeah.
We
can
try
to
make
sure
it's
there
and
for
everyone
interested.
E
Yeah,
I
think
that
would
be
good
and
you
know
the.
I
know
that
there
are
q
roadmap
documents,
there's
the
one.
That's
I
posted,
then
there's
the
other
one
which
I
just
started.
Looking
at
I
definitely
think
some
of
the
long-term
direction,
things
like
moving
code
to
the
server
and
make
ensuring
that
we
can
support.
You
know
arbitrary
resources,
making
versions
key
work
things
things
like
that
that
we
have
those
kind
of
basic
long-term
objectives
documented
somewhere,
so
that
people
understand
the
direction
of
what
we're
trying
to
do
make
a
tool.
Fk9.
A
Yeah
that
def
document
deathless
document
specifically
is
more
like
goals,
not
not
actual.
You
know
individual
text,
s
tasks
or
anything
like
that.
It's
meant
to
be.
You
know
more
long-term,
not
really
long
term,
but
goes
for
year,
so
yeah
anyway,
nice
any
any
other
topics.
People
would
like
to
share
or
their
rights.
That's
it.
I
had
anyways,
okay,.