►
From YouTube: Kubernetes SIG CLI 20210519
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
may
19th-
and
this
is
another
of
our
bi-weekly
sexy-
live
calls.
My
name
is
mate
and
I'll
be
your
host.
Today,
I
don't
have
any
particular
announcements.
Matrix,
eddie
or
sean
have
anything
that
they
want
to
probably
announce
to
the
entire
stick.
A
Okay,
sean,
okay,
so
we're
clear
on
that
one:
do
we
have
anyone
new
on
this
call
that
hasn't
been
here
before
with
us
and
wants
to
introduce
themselves
speak
a
lot
about
background.
What
they
are
interested
in.
B
Hi
everyone,
I'm
vinit,
reddy,
I'm
a
final
final
year
under
undergraduate
student
at
idi
jodhpur
india.
I
was
an
intern
at
layer
5,
which
is
a
community
with
open
many
open
source
service
service
mesh
projects
I
wanted
to.
I
would
like
to
contribute
to
the
kubernetes
project.
It
was
tough
for
me,
tough
for
me
to
choose
an
area,
as
this
is
a
huge
project,
so
I
thought
of
picking
some
issues
related
to
cli
and
get
started
with
it
so
yeah.
This
is
why
this
is
why
I'm
here
welcome.
D
C
My
name
is
noel
rodriguez,
I'm
one
of
the
two
who
volunteered
as
the
liaisons
for
this
fig.
I
believe
the
other
one
is
bigger,
so
yeah.
This
is
my
first
time
on
attending
one
of
these,
I'm
very
new
to
the
whole
community.
C
So
I
thought
of
volunteering
as
a
leaders
and
just
to
get
in
so
hello.
D
Yeah
it's
funny,
my
name
is
hamdi.
I
know
on
this
call
it's
my
first
time.
I
I
I
produce
something
in
the
community
of
capernaum
and
customized,
I'm
from
paris
from
france
and
we
we
start
using
kubernetes
and
customizing
the
two
years,
and
now
we
we
are
really
in
production
and
we
have
some
idea
to
improve
customers.
This
is
why
I'm
here.
A
Okay,
hearing
none
welcome
everyone,
I'm
very
happy
to
have
you
on
our
call.
Okay,
we
can
move
on
to
the
main
topics
and
I'll
pass
it
back
on
to
eddie.
I
think
he
has
the
most
topics
to
talk
to
to
walk
through
eddie.
You
want
to
share
your
screen.
E
Yeah
I
I've
been
fighting
with
the
indentation
in
google
docs,
which
I
always
do.
Okay,
let
me
stop.
E
Okay,
a
little
bigger
people
see
that
okay
yeah
looks
great
all
right.
So
I,
the
big
ticket
item
for
today,
is
I
wanted
to
go
through
our
enhancements
cap
backlog,
there's
a
bunch
of
issues
that
got
migrated
that
don't
really
have
any
status
or
on
track.
So
I
push
that
to
the
end,
because
I'll
make
sure
it
gets
for
everyone
else's
stuff.
E
Outside
of
that,
I
just
took
the
topics
that
have
been
sitting
at
the
top
of
future
topics
for
like
probably
six
months
now
and
move
them
down,
so
we
can
get
through
these
and
remove
them
from
the
agenda
once
and
for
all.
Hopefully,
so,
the
first
one
up
was
graduation
criteria
for
plug-ins,
and
this
was
something
we
wanted
to
talk
about
a
while
ago.
E
I
think
this
ties
in
nicely
to
how
we're
gonna
be
prioritizing
work
now
by
community
response
with
thumbs,
ups
and
emojis
and
stuff,
but
the
tl,
dr,
is
that
we
have
some
really
nice
functionality
that
people
have
built
with
cube
control,
plugins
that
might
make
sense
to
land
in
upstream
cube
control.
E
F
Them
so
eddie,
so
just
just
so,
I
understand
this
particular
topic.
You're
you're,
asking
for
what
great
graduation
criteria
would
be
for
specific
plugins
already
created
that
we
may
end
up
pulling
into
coupe
control,
not
into
the
plug-in
code.
That's
already
in
code
control
itself,.
E
A
I'll
probably
start
with
the
odd
the
author's
interest
in
being
part
of
the
main
kubernetes,
because
if
a
particular
author
is
not
interested
in
being
a
building
command,
which
has
some
limitations
with
regard
to
how
frequently
released
or
or
the
fact
that
it
it
will
live
under
kubernetes,
kubernetes
or
granite
state
in
the
near
future,
we
we
don't
want
to
force
anyone
to
do
it,
and
on
top
of
that,
we
we
only
have
limited
resources
to
handle
the
current
load
of
command.
A
So
that's
something
that
I
would
start
with.
If,
if
the
author
is
interested,
that's
the
preliminary
step.
On
top
of
that,
I
would
probably
try
to
measure
the
the
statistics,
and
I
think
the
crew
has
these
kinds
of
statistics.
I
think
chris
can
can
prove
me
wrong.
G
I
have
like
pretty
basic
metrics
around
plugin
downloads.
It's
really
just
github
downloads.
So
it's
hard
to
it's
kind
of
hard
to
say,
like
you
call,
debug
was
kind
of
a
lot
a
little
more
obvious
just
because
it
was
pretty
popular
and
it's
easy
to
see.
But
there
are
some
smaller
ones
that
are
harder
to
tell
with
the
metrics
that
we
have
available.
A
Yeah,
so
that's
one
side
of
the
metric,
the
other
one
might
be.
The
authors
can
actually
track
their
their
download
statistics,
because
not
everyone
will
be
using
crew,
but
they
might
be
going
directly
and
they
might
have
a
direct
statistics
on
their
head,
which
would
be
something
that
we
would
consider.
A
And
thirdly,
the
overall
request
from
coming
from
a
community
to
have
a
particular
command
be
in
the
in
the
corporaties,
would
be
probably
my
three
metrics
that
I
would
start
with.
A
Based
on
that
we
would.
We
would
start
the
the
bringing
in
and
it
would
yeah
go
ahead.
E
Probably
so
that's
what
we
told
d
tyler
with
the
cube
control,
cuddle,
color
plug-in,
was
to
start
with
the
cap.
F
Just
from
my
perspective,
I'd
actually,
I
think,
it'd
be
interesting
for
us
to
enumerate
a
couple
of
the
plug-ins
that
that
you
know
we
we
think
are
good
plug-ins,
that
are,
you,
know,
plug-ins
that
really
meet
a
need
and
that
we
recommend
to
others
just
just
to
get
an
idea
of
of
you
know
what
what
those
are.
You
know
what
the
universe
of
of
candidates
may
be.
A
I
mean
it'll
be
always
on
a
per
command
basis
for
an
example,
eddie
mentioned
already
debug,
which
is
a
in
a
good
example
and
a
good
initial
command
that
we
went
from
a
plug-in
to
a
built-in
and
how
it's
rolling
can
be
easily
tracked
through
both
cap
and
implementation
and
what's
going
on
with
the
command.
A
E
Yeah,
I'm
gonna
I'll
write
all
this
up
in
in
our
docs,
I'm
just
taking
notes
here.
As
far
as
sean's
comment,
though,
something
like
cube
control
get
all
right.
Like
that's,
been
a
thorn
for
a
very
long
time.
We
see
tons
of
issues
for
it
there.
We
could
proactively
seek
out,
like
the
cat,
all
plug-in
out
there
and
see
if
they're
interested
in
replacing
that
functionality
right,
there's
a
plug-in
that
actually
does
get
all
from
all
the
api
resources
and
crds
right.
E
H
Another
example
that
I
think
has
come
up
on
these
calls
in
the
past
is
keep
control
lint,
where
we
said
that
we
would
accept
a
winter,
and
we
asked
somebody
to
implement
it
potentially
from
scratch
as
an
experimental,
but
there's
also
the
fact
that
there's
a
lot
of
linters
that
already
exist
as
plug-ins.
So
this
approach
could
be
an
alternative
for
for
something
like
that
as
well.
H
And
I
guess
that
that
can
also
point
out
that
there
can
be
cases
where
the
there's
something
that
exists
as
a
plug-in,
but
the
way
that
we
integrate
it
into
upstream
might
not
be
just
like.
Take
what
exists
and
and
integrate
it.
It
could
be
like
build
a
framework
around
it
or
take
pieces
of
functionality,
or
so
that
I
think,
highlights
the
importance
of
putting
it
through
kepp
to
discuss
exactly
what
would
look
like
and
how
well
it
matches
what
already
exists
versus
what
we'll
need
to
evolve.
A
Right,
I
I
updated
the
dock
that
we
will
not
start
with
a
cap,
but
rather
start
with
an
issue
and
keep
cuddle
repo
to
gather
the
plus
plus
minus
ones.
And
then
we
will
actually
do
the
final
decision
during
one
of
our
calls.
G
I
A
I
started
I
started
wondering
maybe
that's
slightly
related.
A
The
discussion
is
about
stuff
like
graduation
criteria,
our
approach
for
future
topics
that
we
will
be
working
on,
and
so
for.
I
wonder
if
the
dogs
are
a
good
place
for
it,
or
maybe
we
could
have
something
else.
It's
definitely
not
a
cap,
because
it's
not
an
enhancement,
it's
more
like
a
directions
for
how
to
work
with
us.
A
A
E
Yeah,
that's
a
good
point.
We
can
always
put
it
in
the
developer
guide
or
we
do
have
our
own
docs
site
too,
which
I've
been
talking
to
the
sig
docs
teams
about
figuring
out
how
to
consolidate
all
that
too.
E
E
E
The
way
that
it
was
originally
designed
was
basically
for
one
repo
really,
but
I
like
the
idea
of
having
all
of
our
individual
repos
broken
out
at
the
top
here.
These
are
what
they
call
collections.
So
you
can
see.
Customize
is
up
here,
broken
out
with
all
the
customized
issues,
and
so
I
have
a
I
have
a
pr
that
adds
basically
all
these
rules,
these
individual
rules.
So
this
is
like
new
cci
issue
in
each
triage
right.
So
it's
a
type
of
issue
with
some
filters
on
here.
E
The
original
idea
was
that
rules
would
be
scoped
to
repos,
so
you
would
have
to
create
a
rule
for
each
repo.
The
pri
open
adds
repos
to
the
collections,
so
this
will
pass
down
the
actual
repos
you
want
and
the
rules
will
apply
it.
That
makes
it
a
lot
cleaner
because
we
originally
had
to
have
a
copy
pasted
rule
for
every
single
repo.
So
basically,
if,
if
you
want
to
play
with
this,
you
have
to
use
my
fork
for
now
until
it
gets
merged
upstream,
but
yeah
so
take
a
look
at
trish
party.
E
A
A
E
A
H
E
Yeah
and
again
I
don't
know,
what's
going
to
be
helpful
right,
so
this
definitely
like
there's
a
ton
of
options
for
this
thing,
I'll
put
the
options
for
filters
in
here
too,
but
feel
free
to.
E
E
Has
it
works
so
well
yeah,
any
other
thoughts
for
triage.
E
Party,
okay,
real
quick-
I
want
to
get
through
this
issue.
It's
been
open
for
a
minute.
It's
about
to
close
itself.
This
was
drain
helpers
defaulting
to
delete
immediately.
J
Yeah,
I
I
I
didn't
manage
to
to
look
into
this
one.
Yet
to
be
honest,
I'm
a
little
behind
on
my
tv
and
if
anyone
wants
to
complete
the
desk,
feel
free
to
get
my
kiwi
clean
up.
You
know
and
close
everything
I
need.
F
E
Please
feel
free.
I
think
it
was
a
reasonable
request.
We
just
never
got
her
on
to
doing
it.
So,
okay,
mache,
you
have
one
in
here
deprecation
of
record
flag.
A
Yeah
I
quickly
checked
what
I
meant
with
that,
I'm
not
100
sure,
but
from
what
I've
noticed
is
we
have
this
record
flag
which
allows
you
to
record
what's
going
on
with
any
particular
command?
A
It
is
wired
in
some
of
the
commands,
but
I
think
it's
not
everywhere,
but
you
need
to
explicitly
use
it
for
the
change
cause
annotation
to
be
recorded
on
a
on
any
particular
resource.
A
The
thing
is
for
oc
apply,
we
do
it
automatically
and
the
annotation
is
obviously
different
and
we
stored
the
previous
state
of
the
object
for
change
cost.
I
can't
remember,
will
be
exactly
a
recorder,
but
there's
it's
a
rather
short
information
would
actually
change
more
or
less
so,
at
least
from
quick
glancing,
the
code.
It
looks
like
it's
a
just
a
text
you
can
put
in
almost
anything
over
there,
but
given
that
it's
you
need
to
be
explicit
about
pointing
it
out.
I
wonder
how
many
people
are
actually
using
it?
A
Oh
and
the
consumers
for
that
are,
for
example,
rollout
history.
So,
theoretically,
you
can
record
why
you
change
a
particular
resource
and
then
you
can
record
that
review
that
through
deployment,
history
or
safe,
whatsapp
or
basically
any
controller
history,
and
I
started
wondering
how
many
of
you
are
actually
using
that
and
whether
it
would
just
make
sense
to
get
rid
of
it.
If
there
are
not
that
many
users,
I'm
definitely
open
to
sending
out
an
email
to
kate
staff
and
6
eli.
A
With
that
question,
if
let's
say
I
don't
hear
any
objections,
I'll
start
the
deprecation
clock
in
122
and
in
over
a
year
we
will
be
able
to
drop
it.
If,
if
you
know
about
anyone
using
it
on
purpose
or
explicitly,
let
me
know
and
then
we
will
drop
the
idea,
but
other
than
that.
I
think
it's
just
yeah.
H
The
documentation,
the
inline
documentation
anyway
for
the
record
flag,
says
that
it's
specifically
recording
the
cube
cuddle
command
that
created,
that
was
that
was
used.
So
I
wonder
if
there's
an
intersection
here
with
the
fact
that
we
now
have
a
better
like
that
with
the
sean's
work,
we
now
can
admit
actual
metrics
about
that,
so
it
doesn't
depend
on
scraping
annotations
or
enabling
it
when,
on
every
client
indication.
A
Exactly
that's,
you
need
to
be
explicit
about
using
that
to
actually
benefit
from
it.
So
if
at
least
from
what
I
was
checking,
there's
not
that
many
people
using
it
then
most
likely
we
could
just
probably
drop
it,
and
I
know
I'm
known
for
removing
stuff,
and
I
then
complain
that
for
different
stuff
being
removed.
But
I
don't
care.
I
I'm
here
to
clean
some
of
the
old
stuff
that
that
were
added
and
nobody
is
using,
because
it
will
just
make
our
code
easier
to
reason
about.
H
I
guess
my
point
is
that
it
might
be
worth
mentioning
the
new
functionality.
That
is
because
it's
behind
an
alpha
flag
and
there's
not
much
documentation
on
it.
So
people
might
not
know
that
that's
an
upcoming
alternative
and
their
if
their
use
case
relates
to
trying
to
figure
out
how
their
stuff
is
being
created.
That
would
be
like
a
better
way
that
they
could
do
it.
F
So
there'll
be
some
documentation
for
when
we
moved
to
beta
and
we
we
specifically
decided
against
putting
some
documentation
for
alpha
just
because
it
you
know
it
is
alpha
and
not
clear
that
it's
going
to
be
actually
something
that
people
can
rely
on,
but
as
it
loses
beta,
we're
definitely
going
to
be
adding
documentation
to,
at
the
very
least,
explain
that
there
are
these
extra
cuddle
flags
and
how
to
turn
them
off.
If,
if
you
need
to.
A
And
especially
if
n122
sean's
work
will
be
on
by
default,
it's
a
good
starting
point
to
replace
the
record
flag
and
start
the
deprecation.
F
I
had
always
wondered
if
anybody
ever
used
that
record
flag,
and
I
know
that
as
a
lot
of
teams
have
moved
to
get
ops
it
kind
of,
is
you
know
it's
kind
of
a
vestige
that
isn't
necessary.
A
E
A
E
Okay,
hamdi,
I
think
you
are
here.
Yes,
you
need
to
share
your
screen.
D
Because
I
mean
ipad,
so
I
have
to
connect
with
my
pc.
I
Yes,
yep
hi,
this
I
came
up,
came
across
an
issue
for
adding
extra
column,
support
to
the
cube,
ctl
command,
and
I
think
it's
like
midway
between
using
the
default
columns
and
using
custom
columns,
which
is
like
defining
every
column,
and
it
looks
like
we
will
have
to
I
mean
if
you
want
to
support
this,
we'll
have
to
add
additional
call,
custom
column,
support
to
the
server
side
print,
and
I
found
a
to
do
in
the
code
base
saying
that
custom
columns
is
currently
not
supported
in
server-side
brain.
I
A
So
custom
columns,
since
this
will
be
something
that
will
vary
from
resource
to
resource.
It's,
not
something
that
we
need
to
do
on
the
server
side.
Server
side
printing
has
an
option
to
request
the
full
resource.
There's
a
separate
parameter
that
you
pass
when
requesting
such
that
you
get
the
server
side
of
printing
columns,
which
is
a
table
and
explicitly
you
need
to
request
a
full
json
of
the
resource.
A
A
I
think
it
was
about
being
able
to
add
additional
columns
to
the
to
the
belt
built-ins
yeah.
If
I
remember
correctly,
yeah.
I
The
issue
is
basically
asking
for:
if
you
use
custom
columns
right
now,
people
have
to
define
all
the
columns
right.
So
the
idea
behind
the
issue
is,
I
don't
want
to
define
all
the
columns.
I
just
want
the
default
columns,
plus
an
additional
set
of
columns
that
I
can
define
so
basically
extra.
I
I
So
is
this:
an
issue
is
that
of
issue
that
needs
follow-up.
Should
we
work
on
this,
or
is
that
something
that
is
not
of
interest.
A
I
think
it's
worth
picking
up.
I
see
that
it
is
getting
the
necessary
feedback
that
we've
that
we've
mentioned
in
the
past.
The
original
issue
has
more
than
40,
plus
41
plus
ones.
The
new
one
already
has
15
15
on
it.
A
So
it's
definitely
something
that
we
could
include
and
if
there
is,
if
you're
interested,
particularly
in
implementing
that
give
it
a
shot
sure
yeah
it
most
likely
will
be
what
I'm
saying
it
will
be
very
similar
to
custom
columns,
because
custom
columns
should
be,
although
actually
custom,
columns,
it
will
be
requesting
full
resource
and
just
parsing
the
additional
ones,
but
it
should
be
similar
to
sorting
such
that
you
should
be
able
to
request
the
full
resource
in
addition
to
to
the
tables.
A
A
D
That
yeah,
so
we
I
will
speak
about
this
this
issue.
So
the
idea
is
to
improve
big
time
for
remote
players
and
customers,
and
I
have
a
full
request
and
then.
D
Yeah,
so
in
this
pro
request,
the
idea
is
to
improve,
as
I
say,
prove
the
customized
build
performance
so
the
time
to
be,
and
so
we
consider
get
resources
in
customized
like
code
dependencies
or
other
private
deficiencies.
D
So
we
use
this
same
mechanism
to
to
cash
defendancies
in
the
local.
This
the
idea
is,
if
we
visit
a
repository
twice
in
the
same
bill,
we
use
a
hidden
version,
so
casual
version
in
the
localities
and
also
we
can
use
reuse
between
two
builds.
So
the
the
same
questions
we
different
with
def
set
between
meetable
and
emittable
reference,
and
we
consider
that
the
branch
reference
are
mutable
and
commit
id
or
tag
are
emittable
and
then
so
we
start
meetable
and
so
branch
or
tag
reference
in
different
directories.
D
And
when
we
visit
a
reference,
we
decide
to
store
or
not
the
surface.
So
if
it
is
a
mutable
one,
maybe
we
don't
start
is
if
it
is
an
immutable
imitable
one,
we
store
it
in
the
local
cache
and
we
have
an
option.
I
I'm
I'm
adding
an
option
to
decide
whether
or
not
accept
a
branch.
H
I
think
in
the
pr
we
discussed
having
this
behind
a
flag
as
well
right
that
that
was
an
idea
that
we're
in
favor
of.
D
H
And
is
that
flag
generally
applicable
to
the
git
cloner,
or
is
it
specifically
specific
for
the
cache
functionality
like?
Does
it
also
block
mutable
references
if
you're
not
using
the
cash.
D
H
D
D
K
Hey
thanks,
this
is
jeff.
Thank
you.
Thank
you
for
sending
out
the
pr
I'm
looking
at
it
today,
thanks
for
also
rebasing
and
squashing,
the
commits
I'm
going
to
ask
for
some
more
tests
on
this
thing
to
confirm
things
like.
Can
we
blow
away
the
cache
and
have
normal
operation,
because
I
don't
want
customized,
it's
it's
shouldn't
necessarily
be
in
the
business
of
managing
a
complicated
catch.
So
this
needs
to
be
really
simple,
close
the
repo
put
it
into
the
caching
directory.
K
If
you
blow
it
away,
it
still
works
and
to
the
extent
that
you
can,
with
the
testing
for
this,
please
use
the
fake
file
system
so
that
we
don't
have
more
tests
that
are
actually
depending
on
a
real
file
system.
So
you
can
put
some
stuff
in
the
cache
that
looks
like
a
that
looks
like
a
that
looks
like
a
downloaded
repository
for
example,
but
anyway
everybody
there's
a
lot
of
people
who
want
this
feature,
so
I
think,
as
long
as
it
stays
simple,
we
can
go
ahead
and
add
it.
K
A
Okay,
cool
and
back
to
you,
eddie.
E
So
last
thing
I
want
to
spend
some
time
on
actually
do
you
want
to
do
stand-ups,
first
jeff
and.
K
Sure
katrina
and
I
we
can
do
a
customized
stand-up
briefly,
we've
been
talking
about
the
roadmap
and
we've
been
talking
about
how
to
bring
in
more
people
to
work
on
customize
and
at
the
next
cli
meeting.
I'd
like
to
put
on
the
agenda
a
discussion
of
how
to
do
that.
K
So
we're
setting
up
some
projects
and
going
through
a
lot
of
the
issues
to
try
to
find
places
where
people
can
contribute
and
organize
them
in
the
projects
that
have
well-defined
endpoints,
because
we're
really
eager
to
get
some
more
people
familiar
with
the
code
base,
familiar
with
the
release
process,
familiar
with
how
this
all
works
with
kubecuttle,
etc,
etc.
So
I'm
hoping
that
we
can
get.
You
know
like
15,
20
minutes
next
six
hour
meeting
to
present
some
of
that
stuff.
A
Jeff,
I
would
encourage
you
to
send
out
an
email
to
six
cli
and
maybe
even
k
death,
that
there
will
be
a
discussion
like
that
happening
in
two
weeks
during
our
sixth
july
and
okay,
great
sure.
Maybe
you
can
get
additional
additional
people
interested
in
this
topic
to
show
off
as
well
great.
A
L
So
yeah,
as
I
mentioned
on
the
on
the
in
the
blog
there
we're
about
to
release
a
new
version.
Mostly
it's
bug
fixes.
We've
actually
had
a
couple
of
community
feedbacks,
some
not
so
kind
of
embarrassing
bugs
that
we
fixed
they'll
release
them
pretty
shortly
yeah.
I
guess
they
always
sneak
through
the
cracks.
L
So
that'll
come
out
over
the
next
few
days,
we'll
get
those
embarrassing
bugs
fixed
and
released
we've.
Also,
a
bunch
of
security
fixes.
You
know
rolled
in
dispersion,
bumps
all
the
standard
stuff.
We
have
a
bunch
of
stuff
happening,
sort
of
behind
the
scenes
for
improved
support
for
kubernetes
jobs.
Some
of
it
has
already
landed
in
kui,
but
some
will
land
in
10-4,
hopefully
in
a
few
weeks.
L
So
that's
just
generally
improvement
for
a
more
serverless
approach.
You
know
to
kubernetes
focus
on
jobs,
dashboarding
getting
status
about
the
ongoing
jobs,
that'll
all
land
in
full
force.
Hopefully
in
a
few
weeks
excuse
me,
I
did
have
one
quick
question
if
you
don't
mind
for
around
log
streaming
with
respect
to
jobs.
So
if
I
do
a
log
follow
with
a
label
selector,
let's
say
I
want
to
follow
a
set
of
jobs
that
all
have
a
common
label
of
all
the
logs
of
a
set
of
jobs
at
all
account
label.
L
For
example,
see
we
have
five
jobs
running
and
they
finish
five
more
jobs,
startup
and
so
on,
and
so,
but
they
all
have
a
common
label
so
in
cuckoo
cuddle,
at
least
as
far
as
I
can
tell
now,
if
I
say
logs
logs
dash,
f
dash,
l,
the
follower
will
exit
as
soon
as
that,
instead
of
j,
first
set
of
five
jobs
finishes,
and
it's
up
to
me
to
respawn
the
logs
follower
again
with
the
same
label.
Selector.
It's
sort
of
up
to
me
to
manage
the
outer
loop
of
jobs.
A
A
So
what
we
would
have
to
do
is
for
job
itself,
because
that
most
likely
or
well
even
during
a
rollout
of
a
deployment
or
any
kind
of
long-running
workload,
the
same
problem
will
appear
because
during
a
roll-up
there's
a
change
of
pots
which
are
previously
running
and
are
currently
running.
That's
a
good
point.
Yeah,
it's
not
just
jobs
yeah!
Yes,
it
will
also
affect
rollouts
of
other
controllers,
so
yeah.
I
guess
we
just
need
to
fix
it.
It
just
nobody
complained
about
it.
Before
I
mean.
L
A
A
At
least
for
deployments,
although
even
for
deployments
the
the
rollout,
if
you're
running,
I
don't
know
five
replica
deployment,
the
rollout
will
be
gradual,
similarly
for
stateful
set.
So
whenever
one
stream
ends,
you
would
have
to
close
all
re
re-list
the
parts
and
set
up
new
ones.
L
A
L
Yeah
definitely
non-trivial
of
a
degree
yeah.
In
some
cases
like
jobs,
you,
you
have
a
sense
of
of
the
the
countdown
latch
right.
You
know
what
how
many
pods
to
expect
in
other
cases,
like
you're,
mentioning
it's
more
of
an
infinite
loop,
yeah
there's
a
lot
of
nuances
here.
A
Even
for
a
job,
it
can
be
an
infinite
loop,
because
there
is
an
option
for
job
to
you,
don't
specify
completions
and
but
you
can
specify
parallels,
which
means
there
will
be
this
many
thoughts
running
a
particular
job.
As
long
as
at
least
one
succeeds,
and
then
the
entire
job
is
marked
as
completed.
That's
one
of
the
possible.
A
Working
stuff,
if
anyone's
interested,
I'm
I'm
open
to
seeing
the
pr,
but
I
will
require
some
some
additional
thought
into
how
to
efficiently
work
this
out,
yep,
we
agreed.
L
E
Okay,
so
jumping
into
this,
we
have
about
nine
minutes
left.
There
are
a
bunch
of
these
caps
that
are
open
and
a
lot
of
them
were
migrated
from
the
old
template.
So
we
don't
have
any
context
or
conversations
around
the
discussion
that
was
happened
previously,
so
I'd
like
to
go
through
these,
the
ones
that
are
kind
of
migrated
and
not
really
commented
on
katrina.
I
think
you
said
that
you
would
take
care
of
the
customized
ones.
Do
you
want
to
do
those
here
or.
H
I
started
looking
into
the
kep
like
text
itself
yesterday,
but
I
didn't
get
very
far.
I
was
planning
on
taking
care
of
some
more
of
that
today
and
submitting
a
pr
with
updates.
E
Cool
the
so
these
issues
that
are
opened
should
these
be.
E
That's
a
good
question.
I
see
we
have
jeremy
on
the
call
dude
should
these
should
these
be
closed
if
they're
implemented
should.
M
Yeah,
if,
if
a
cap
has
been
implemented,
then
whatever
tracking
issue
is
associated
with
it
can
be
closed.
That's
that's
the
point
at
which
they
should
be
closed.
A
And
I
think
we
should
update
the
description
on
the
issue
such
that
it
reflects
that
it
is
actually
implemented
because
currently,
if
you
look
at
this
one,
it
ends
at
alpha
yeah.
Maybe
we
can
drop
that
entirely
and
just
mentioned
that
it
was
stable
in
a
particular
release
and
just
leave
it
as
it
is
yeah.
That's
pretty.
E
Got
it
okay,
so
I'll
skip
over
the
customized
ones?
Now
we
can
always
come
back
to
these.
If
you
want
christina.
A
Yeah
that
that
looks
like
phil's
idea
from
the
past,
but
I
I
think
it
was
just
created
and
nothing
else
happened
it.
It
might
be
an
idea
that
was
just
written
down,
but
we
never
continued
on
that.
One.
E
A
Yeah,
but
that
was
just
for
the
migration,
so
I
think
we
can
close,
because
we
are
done
with
that.
One.
A
What
I
wanted
to
see
is
merge
the
cap
as
rejected,
and
then
we
can
close
the
issue
as
well,
so
that
for
a
future,
if
anyone
will
be
looking
for
this,
he
can
find
the
cap
and
and
the
capital
will
write
that
this
was
rejected.
That's
a
valid
statement.
A
Yes,
we've
allowed
it
as
such
for
for
a
future
generation
to
know
why
we
went
this
route
cool.
A
Yeah
it's
on
for
the
next
release,
so
I
think
cool
you
can
leave
it
at
this.
A
E
Still:
okay,
I'll
customize
users,
okay,
so
we'll
take
out
that
one's
current
cube
cuddle
the
bug,
that's
current,
being
tracked:
yeah
cuddle
events,
oh.
A
It's
actually
not
being
tracked.
Debug
is
pushed
over
to
the
next
release.
I
spoke
with
lee
about
it
last
week
or
something
like
that.
It
will
be
pushed
to
the
next
release.
A
Yeah,
we
basically
want
to
have
the
thermal
containers
first
initiated
and
then
we
will
be
finishing.
The
cube
cut
with
the
milk.
A
That's
an
ongoing
topic.
I
have
someone
who
who
created
this,
I
I'll
I'll,
give
it
a
look,
but
definitely
not
something
that
we
want
to
track.
Yet
I'm
hoping
for
the
next
release,
yeah
I'll,
be
able
to
return
to
this.
E
F
So
so,
while
we're
well,
we
just
went
over
the
server
side
apply.
I
just
had
a
quick
question
which
is
kind
of
on
a
tangent.
I,
this
won't
take
too
long,
but
we
removed
antoine
from
the
cli
maintainers.
F
A
Was
a
separate
issue
that
was
in
the
org?
We
didn't
remove
him
from
the
reviewers
and
approvers
yeah.
That
was
admin.
A
Got
it
so
now
I
I'm
on
a
watch
out.
I
would
not
remove
one
from
from
approvers
from
60i.
Definitely
not.
E
Okay
and
that's
it
then
so
the
all
right,
so
the
migrated
ones
are
mostly
the
customized
ones.
So
if,
if
katrina,
if
you
want
to
go
through
those
as
a
group
at
a
next
meeting,
we
totally
can
do
that.
I
wonder
why
they
were
all
migrated
like
that.
A
There
was
a
big
migration
effort,
someone
from
I
think
probably
jeremy
can
speak
about
it,
but
I
was
paying
by
voytech
when
the
template,
for
the
sake
for
a
cast
change
a
couple
months
back.
A
There
was
a
group,
oh,
we
need
to
fix
them
and
and
have
all
the
caps
migrated.
I
think
the
the
folks
from
the
product
readiness
were
implementing
the
automation
on
the
caps
and
for
that
they
needed
to
have
all
the
pre-existing
caps
not
to
fail
so
that
that's
what
basically
happened.
I
was
approving
majority
of
those
caps
boy
they
reached
out
to
me
for.
M
Yeah,
that's
exactly
what
happened
as
there
was
more
stuff
going
into
the
validation
process
for
the
cups,
which
is
goodness
all
around
right
checks.
A
lot
of
things
makes
makes
it
easier
for
like
reviewers
to
make
sure
that
things
are
good
makes
easier
for
the
release
team
makes
it
easier
for
the
proud
readiness
team.
There
were
a
whole
bunch
of
things
that
were
like
out
of
date
and
there
have
been
like
four
or
five
revisions
of
that
chem
template
over
time.
M
So
there's
a
whole
there's
some
other
work
happening
that
will
make
that
easier
going
forward.
But
to
kind
of
get
over
that
hump
there
was
a
a
big
push
to
just
move
them
all
and
shaycard
did
like
90
of
them.
H
Yeah
I'll
just
do
my
best
to
update
the
with
the
text
and
the
issues
for
the
customized
ones
and
I'll
touch
base
with
jeff.
When
I
I
can't
figure
stuff
out
and
if
there's
anything
outstanding,
we
can
go
over
it
at
the
next
meeting.
H
Can
I
edit
them?
Oh,
do
you.