►
From YouTube: Kubernetes SIG Apps 20170410
Description
Meeting minutes available at https://docs.google.com/document/d/1LZLBGW2wRDwAfdBNHJjFfk9CFoyZPcIYGWU7R1PQ3ng/edit#heading=h.eul5yofe3hlp
A
So
welcome
everyone
to
the
visit,
the
April
10
to
20
17,
kuber
net
e
cig
apps.
Today,
let
me
go
ahead
and
paste
the
agenda
into
chat.
We
have
a
couple
of
announcements.
We
have
some
stand-ups
just
to
get
status
because
we've
got
a
whole
bunch
of
projects
that
have
been
going
on.
Then
we
have
a
demo
and
and
then
we'll
have
just
some
open
discussion
and
we
can
pick
up
any
topics
as
needed,
so
my
name
is
Matt
Farina.
A
For
anybody
who
doesn't
know
me
normally
Michelle
is
here
too
and
two
runs
about
half
the
time
and
it
looks
like
she
might
be
traveling
today,
so
she
may
not
be
here,
and
so
the
first
two
announcements
we
have
is:
if
anybody
wants
to
jump
in
and
take
notes,
I've
linked
in
the
document
go
ahead
and
feel
free
to
take
notes
as
we
go.
If
anybody
wants
to
take
on
a
normal,
more
formal
note-taker.
Please
go
ahead,
and
let
me
know
if
you'd
like
to
do
that.
A
So,
if
you'd
like
to
do
that,
go
ahead,
if
not
I'll
be
jumping
in
and
taking
notes,
the
other
one
is
Michelle
is
she's,
doing
some
research
on
how
developers
are
deploying
apps
into
coover
Nettie's,
and
if
anybody
followed
her
keynote
from
the
cloud
native
con
you'll
know
that
she
talked
a
little
bit
about
some
needs
and
things
there
and
so
she's
looking
to
have
one
on
one
30-minute
conversations
to
capture
your
experiences,
and
so,
if
you're
up
for
having
those
conversations,
her
email
address,
is
in
the
meeting
minutes.
A
I
will
go
ahead
and
we
stayed
here
as
well.
Please
feel
free
to
reach
out
to
her
she's
trying
to
just
capture
details
around
this
I'm
not
excited
for
her
reason
for
yet
but
gee,
I'm
sure
she'll
come
back
and
share
and
some
goodness
will
come
out
of
it.
And
so,
if
you
have
anything,
please
let
her
know
please
reach
out
to
her.
A
Please
sit
down
and
chat
with
her,
and
so
with
that
we
have
our
updates
on
what
happen
is
happening
in
Coober
Nettie's
courts,
and
these
are
in
the
Cooper
Nettie's
Cooper
Nettie's
project,
rather
than
things
like
helm.
In
those
other
bits-
and
so
we
have
a
number
of
things
happening
in
there
and
so
I
will
just
run
down
real,
quick
and
see
if
anybody's
on
to
let
us
know
what's
going
on,
and
so
the
first
one
I'll
reach
out.
Is
anybody
here
to
talk
about
stateful
set?
That's
that.
B
B
Baz
can
see
is
and.
C
Now
I'm
here
I'm
can
also
talk
with
Clayton
yeah.
B
So
I'll
just
talk
about
a
little
bit
that
I've
been
involved
with
them.
We
had
kind
of
two
functional
level,
things
that
are
being
urgently
discussed.
The
most
important
is
staple
set
up
dates,
so
the
proposal
is
up.
The
feature
is
in
the
features,
repo
and
the
proposals
up
in
the
community
repo
for
being
able
to
do
staple,
set
up
dates
and
have
it
work.
You
know
in
a
similar,
if
not
equivalent
fashion,
to
Damon
sets
in
two
deployments
to
light
actually
changed
staples.
Ed's
there's
been
a
lot
of
good
discussion.
B
D
D
Just
Quentin
here
ones,
one
quick
thing
to
add:
I
just
add
it
to
the
chat
logs
and
the
proposed
federated
state
full
sets
design.
If
anyone
wants
to
chime
in
there,
I
think
we're
be
through
fair
amount
of
review
so
far
and
we're
ready
to
start
developing.
But
if
you
want
to
give
see,
they
can
put
a
piece
feel
free
to
do
so.
A
All
right,
thank
you,
I
didn't
realize
there
was
a
the
proposal
for
the
Federated
state
or
federal
aid
site
design.
That's
that's
great.
To
see
all
right,
I
just
put
all
of
those
into
the
docks.
Is
there
anything
else
on
stateful
sites?
Are
we
good
I'll
take
silences,
we're
gonna
all
right,
the
next
one
is
do
we
have
anybody
on
who
can
discuss?
Damon
sets.
C
So
I
think,
as
a
was
we're
trying
to
finalize,
we
want
to
have
updated
history
and
there's
some
question
about
how
we're
going
to
match
previously
created
pods
with
their
historical
pod
template.
There's
some
question:
we're
going
to
use
hashes
or
template
generation,
integers
confirm
or
I
scan.
That
has
not
been
settled
yet
and.
C
B
B
Think
anyone
has
proposed
a
design
that
works
that
fully
flushes
out
the
the
risks
and
challenges.
Just
some
folks
are
aware,
like
we
have
a
with
deployments,
we
get
it
easy
because
we
create
replica
sets.
It's
really.
Nice
thing
makes
a
lot
of
sense
with
staple
sets
and
damen
sets,
there's
a
lot
of
downsides,
because
we
don't
actually
have
a
smaller
controller
that
we
want
to
use
for
that.
So
we
need
something
to
store
it.
B
There
been
a
couple
proposals,
something
generic
that
works
across
lots
of
resources
and
could
be
used
or
something
specific,
and
the
generic
approach
has
a
lot
of
the
chain
challenges
that
we're
having
with
deployments
right
now,
which
is
dealing
with
hashes
and
how
you
look
these
things
up
and
the
more
specific
version
is.
It
tends
to
be
something
that
doesn't
work
for
us
new
controllers.
People
might
create
so
I
think
that's
the
challenge.
C
B
C
A
A
E
E
Is
sure
it's
related
to
what
the
I
already
said
about
them
on
sets
and
the
problem
at
wave
and
we
need
to
decide
if
we
are
going
to
continue
with
ashes,
which
is
the
way
deployments,
are
create
the
names
for
the
replica
seeds
for
their
history
versus
switching
away
from
that,
because
we
have
had
some
problems
with
fashion
and
switching
to
something
more
stable
right,
like
template
generation,
which
is
basically
burgeoning
there.
Your
templates
and
use
a
number
for
your
names,
but
both
organize
have
proceeded
pros
and
cons.
A
D
B
B
A
lot
of
it
comes
down
to
you
know,
just
practically
speaking
third-party
resources,
as
also
we
made
some
mistakes
and
if
we
fix
them,
that's
going
to
be
somewhat
disruptive
to
people
using
their
pretty
resources,
and
the
hope
would
be
that
we
can
at
the
same
time
either
document
the
process
whereby
the
disruption
is
the
same
as
if
you've
ever
needed
in
the
future,
to
have
a
breaking
change
to
third-party
resources.
Maybe
you've
got
a
third-party
resource.
We
decide
that
you
want
to
change
behavior.
B
How
would
you
do
that
today
see
there's,
there's
some
similarities
between
those
two
and
then,
if
you
know
various
people,
you
decide
that
that's
going
to
be
very
impactful
to
them.
You
know
there
might
be
some
desire
to
get
people
in
the
community
to
sign
up
to
help
support
the
backwards
compatible
pass
as
an
additional
option
is
not
something
we
generally
do
for
also
resources,
but
it's
important
enough
to
the
community
and
there's
volunteers
to
go
support
that
that
might
be
an
option.
So
look
for
that.
A
E
Yeah,
that's
not
a
proposal
of
us
in
the
community
level
and
it
awaits
more
abuse.
Basically,
syllables
save
reports
right
now
exam.
It's
concept
that
having
deployments
and
demo
sets,
which
is
how
many
thoughts
are
running
for
a
specific
amount
of
time,
and
we
are
trying
to
make
this
more
robust
in
the
pod
API.
E
And
that's
the
city
c'mon
otay
of
time
is
an
option
minerally
seconds
that
you
have
right
now
in
deployments
and
demos
it's
okay,
there
are
some
problems
with
the
current
state
of
things.
One
is
clock
skew,
because
ipod
is
marketed
by
cubelets
in
nodes,
and
then
you
have
the
master
components
from
the
controller
which
is
taking
the
time
stamp
from
them
from
could
be
attended.
It's
trying
to
compare
simple
sleep
time
at
the
house.
E
B
I
was
actually
going
to
add,
I
think
this
is
a
topic,
that's
equipped
and
maybe
discuss
here
and
that
a
lot
of
people
have
the
question
often
like
how
do
I
know
that
my
services
are
working
and
we've
had
a
number
of
proposals
over
the
last
couple
of
years.
That
attempt
to
talk
about
how
do
you
summarize
and
roll
up
or
have
more
sophisticated
levels
of
quote-unquote
health
or
quote
unquote,
ready
s
for
a
larger
variety
of
components?
So
you
know
if
you've
got
a
micro
services
application,
it's
got
five
services,
you
know.
B
Sometimes
it's
as
simple
as
well:
hey.
Are
there
any
all
the
deployments
available,
but
then
there's
another
set
of
questions
which
is
well
if
only
half
of
them
are
available?
What
does
that
mean?
And
how
do
we
encourage
and
help
people
build
apps
that
can
not
only
report
their
status
accurately,
but
on
the
flip
side,
someone
can
understand
that
status
by
making
it
to
the
platform
itself
might
be
aware
of
that
and
that
it
was
maybe
a
larger
discussion.
But
if
people
are
people
are
on
their
own
building.
B
Apps
and
struggling
with
this
question
of
I
know
when
something's
ready.
How
do
I
know
that
everything
is
ready
or
everything
is
working,
there's,
probably
a
discussion
that
we
could
have
in
sig
apps
over
the
next
in
a
couple
weeks,
which
is
you
know,
a
summary
of
where
we've,
what
we've
done
historically
and
maybe
some
discussion
about
how
we
could
see
that
happening.
I
know
that
it's
come
up
quite
a
few
tens
of
issues
recently
I'll.
C
Just
had
a
no
clean
centum
that
I'm
excited
about
this
feature.
The
way
look
at
it
lets
people
writing
operators
to
probably
resources
extensions
correctly,
implement
some
so
much
wiki
logic
to
decide
its
pods
are
ready,
yep
and
yeah.
A
lot
of
things
are
going
to
depend
on
that,
so
factors
out
some
tricky
code
to
the
server
side,
yeah.
B
That
that
kind
of
theme,
I
think
reinforces
like
yeah,
we're
we're
at
16
right
now,
and
you
know
so
we're
six
releases
in
and
a
lot
of
people
have
built
real
apps
and
are
starting
to
build
extensions.
There's
a
lot
of
day-to-day
sorts
of
pain
that
we
as
say,
gasps
probably
need
to
do
a
good
job
of
helping
summarize
the
feedback
and
what
people
are
struggling
with,
and
you
know
help
make
it
easier
to
go.
B
C
B
I
well,
I
know
that
it's
not
something
that
was
really
on
the
committed
list
for
17
or
something
then
discussed.
It's
more
of
a
there's
been
a
lot
of
discussions
right,
so
the
component
status
was
clipping
that
the
proposal
that
someone
started
way
back
before
even
10
and
it's
kind
of
the
same
problem.
B
Okay,
if
you
want
to
run
a
bunch
of
things
on
top
of
Cooper
Nettie's,
like
controllers
or
elastic
surge
or
keep
stir,
how
can
you
summarize
that
and
show
that
up
to
people
and
so
I
think
what
might
be
good
is
to
try
and
pull
together,
maybe
a
quick
summary
of
the
high
level
and
see
if
we
can
link
those
issues
like
all
the
issues
that
people
have
brought
these
topics
up,
so
that
we
could
discuss
it.
So
maybe,
in
a
couple
of
weeks,
that
might
be
a
good
time.
I
would
take.
C
B
No
I
think
I
think
stable.
Pods,
you
know,
is
a
is
a
very
practical,
pragmatic
step.
I
was
mostly
I.
Think
it's
useful
for
this
group
of
folks
to
help
help
kind
of
pull
that
together.
I
think
it
fits
into
this
agreement,
and
it's
good
that
it's
good
for
us
to
occasionally
take
the
opportunity
to
do
a
step
back
and
look
at
kind
of
the
broad
problem
and
say
you
know:
are
we
doing
enough?
Are
we
helping
under
SIG's,
etc?
So
yeah
I
would
say
they're
mostly
separate.
A
G
Hi,
this
is
silly
I
added
it.
Last
week
after
our
meeting,
okay.
C
G
The
primary
intent
of
that
proposal
is
to
help
us
run
fatal
application
and
welcome
terminate
more
cleanly.
For
instance,
there
are
a
lot
of
use
cases.
I
thought
I
would
be
no
mention
that
it
opposes
so
it
was
just
recently
proposed
and
it
requires
more
people
to
found
it
and
suggests
crewmen's
on
it
when
we
brainstormed
internally
within
huawei.
What
we
found
was
that
the
only
drawback
we
have
in
their
proposal
is
that
it
cannot
address.
I
mean
that
specific
guarantee,
if
we
don't
crashes
or
the
determination,
is
still
not
then
clearly
no
traction.
G
So
that's
not
something
we
could
find
an
elegant
solution.
That's
still
a
kb
eight,
the
design,
but
other
than
that.
We
feel
that
you
know
it
seems
to
have
this
moose
of
issues,
so
it
it
seems
to
bring
in
some
sort
of
elegant
destructive
ability.
44
c'mon
already
have
done
any
admitted
arbitrary
containers
when
the
power
in
the
part
is
getting
terminated,
so
easy,
we'll
have
a
look
at
it
and
suggest
to
anything
to
it.
Thank
you.
Yeah.
A
No
thank
you,
and
so
anybody
who's
interested.
I
put
a
link
into
the
proposal
into
the
notes
at
the
ferg
containers
and
so
I.
If
you're
interested
in
this,
if
you're
interested
in
the
destructors,
please
go
ahead,
follow
through
and
give
feedback.
Thank
you,
and
so
with
that
we
are
now
on
to
the
demo,
which
is
the
appt
reg,
the
app
registry.
Do
we
have
somebody
on
to
do
this?
Demo?
Stop
hey.
H
H
And
directory
okay,
okay-
and
we
this
success
under
set
of
canal-
to
manage
all
your
FICA
Jubilee,
so
that
is
that
you
can
cool
foosh
and
manage
application
like
you
will
do
with
google
street
tacos
can
list
them,
so
we
did
an
integration.
Also
on
credit.
I
own
is
apparent
to
you
most
with
self
weight
to
lose
it
the
Pangolin
favorable.
We
can
search
all
fine,
quick.
The
table
can
replace
a
home
yeah.
H
H
C
H
D
H
H
H
H
H
More
precisely
because
I
you
take
on
to
see
here
and
will
Emily's
filling
in
academy
and
I'm
registry
push
cause
the
new
jobs
open.
Top
animal
image
is
very
the
paramilitaries
and
the
catfish
suppose
going
to
the
collision
place
and
then
declare
my
prediction:
are
the
cosmos
are
up
here
when
you
say
I
was
because
this
package
em
with
following
them.
C
H
I
H
Here
so
far
for
4
30
inch
pipeline
for
up
here,
the
deployment
is
automatic
and
we
can
just
delete
it
if
you
want
so
here
we
just
create
a
pusher.
Indeed,
okay,
cool,
we,
okay
with
this
in
the
morning,
spain,
resources
and
for
master
the
payment
of
stag.
Do
you
need
to
be
a
Monroe,
but
I
can
disarm
or
picture
from
your
CI
budget?
Give
an
idea
of
how
we
can
manage
our
continued
to
payment.
H
A
G
I
H
H
A
A
Thank
you
for
the
demo.
This
is
this
is
useful,
thank
you,
and
so
with
that
we
move
on
to
the
next
part
where
we
walk
through
the
stand.
Ups
for
the
different
things
that
aren't
part
of
core
Cooper
Nettie's
for
those
of
you
who
aren't
familiar,
we
alternate
weeks.
One
week
we
go
through,
you
know
just
reviewing
a
bunch
of
the
projects
and
then
like
next
week,
we'll
dive
into
deeper
topics
on
a
couple
of
things
and
not
spend
so
much
time
on
just
an
overview
of
our
a
bunch
of
the
team's
rep.
A
A
All
right,
I'm
idiot
silence
me
either
nobody
hears
me
or
there's
nobody
on
from
home
or
day
is
today
which,
for
anybody
who
hasn't
caught
the
news
the
day
is,
has
some
big
news,
and
so
those
folks
may
not
be
available
at
the
moment
it
seems
according
to
their
blog
they're
joining
microsoft,
so
they
may
just
be
indisposed
today
and
we'll
hear
from
them
next
time.
So
with
that
is,
do
we
have
anybody
on
for
appcontroller.
J
Yeah
we
do
I
guess
like
I,
don't
know
yet
so
I
wish
honestly
I
me
enjoy
from
granted.
So
basically
much
is
still
working
on
this.
This
helmet
and
a
cultural
integration
using
those
rudders
right
so
even
is
running
on
is
interested
once
you
can
provide
more
details,
I
guess,
besides,
that
we
are
working
on
this
close
implementation.
We
also
plan
some
improvement
in
documentation
and
and
different
testing
areas
pretty
soon
so-
and
that's
basically,
are
all
from
my
side
here.
A
J
A
A
I'll
take
silence,
as
we
don't
have
anyone
on
from
there.
There
might
be
overlap
with
the
same
folks
working
at
home
and
maybe
they're
not
here
for
the
same
reason,
in
that
case,
app
registry.
Do
we
have
we
just
had
this
this
great
demo
of
the
plug-in
for
home
tied
to
the
app
registry?
Do
we
have
an
update
status
beyond
the
demo.
H
A
K
I'm
here
Adam
ash:
can
you
hear
it
sure
can
yeah
perfect.
So
we
are
keeping
to
work
on
improving
converter
converts,
conversions,
so
more
recently
by
working
on
adding
build
support
and
fixing
some
issues
with
volumes,
because
it's
getting
a
bit
tricky
volumes
and
we
have
also
started
researching
the
compost
version
free.
So
if
you
would
like
to
add
support
for
it
near
future,
and
that's
it
all
right.
K
K
There
is
not
much
a
lot
compost,
we
are
just
basically
trying
to
figure
out
and
verify
if
our
coach,
it
is
manageable
and
good
and
if
it
can
be
really
useful.
So
we
are
having
discussions
about
our
file,
structure
and
format.
If
yellow
is
right,
choice
or
we
should
use
something
else
and
basically
how
to
keep
it
called
project
manager
go
and
bicycle
post
back
and
can
be
manageable.
We
can
do.