►
From YouTube: Kubernetes SIG Apps 20181015
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,
there's
the
agenda.
We've
got
several
things
to
chat
about
today.
The
first
thing
that
I
wanted
to
actually
bring
up
as
an
announcement
was
for
folks.
We
are
running
low
on
seats
for
the
contributor
summit.
Let
me
share
my
screen
real
quick
here
we
go
so
the
contributor
summit
is
coming
up,
kubernetes,
contributor
summit
and
there's
actually
two
parts
to
it.
There
is
a
contributor
summit
and
there
is
a
new
contributor
track.
A
If
you
were
interested
in
this
stuff
sign
up
very
soon,
because
we
will
be,
they
will
be
running
out
of
available
slots
and
they'll
be
a
waiting
list.
So
here
is
the
details
on
how
that
works.
You
can
actually
find
it
in
the
community
repo
and
events
2018
and
then
you
can
find
it
from
there,
but
that's
the
one
announcement
that
I
had
did
anybody
have
any
other
announcements
they
wanted
to
share,
and
this
is
for
Seattle
by
the
way.
A
A
Then
we
can
jump
ahead
to
one
of
the
discussion
topics
and
we
can.
We
can
circle
back
to
you
in
a
little
bit
here.
Just
let
me
know
when
you're
ready.
So
in
the
discussion
topic
ception
you
jump
down,
there
were
actually
two
items
that
I
threw
in.
In
addition
to
the
normal
developer
tools,
they
kind
of
pertain
around
developer
tools
and
then
we've
got
open
discussion.
So
if
somebody
had
something
that
they
wanted
to
throw
in
here,
maybe
something
specific
under
developer
tools
or
any
of
these
things.
A
Please
speak
up,
but
one
of
the
things
you
know
I
know
folks
are
trying
to
to
showcase
their
stuff
and
to
give
feedback
on
their
stuff,
and
you
know
there's
a
lot
of
developer
tools
that
are
out
there
and
even
just
trying
to
find
them.
And
a
while
ago
we
had
talked
about
building
something
in
her
say,
gaps
to
kind
of
showcase,
this
stuff
and
we
didn't
and
the
reason
for
that
was
actually
because
of
the
CNC
F
landscape.
A
Okay,
so
one
of
the
things
that's
been
happening
with
the
landscape.
Lately
is
redoing
the
categories.
So
if
you
come
into
categories,
there's
going
to
be
some
differences
and
they're
pretty
general
here,
but
one
of
the
things
that
is
wanted
to
make
sure
is:
everybody
knew
to
get
their
stuff
included
in
here
if
they
didn't
already
have
it,
because
this
is
kind
of
that
that
search
and
discovery
mechanism
that
we
had
originally
thought
about
doing
as
part
of
this
or
the
appdev
working
group.
A
When
this
came,
we
decided
not
to
do
something,
and
so
we
want
to
make
sure
it
works,
because,
as
the
scene
CF
does
things,
they
can
continually
iterate
and
change
things,
and
one
of
the
things
they
recently
did
was
work
on
the
categories.
And
so,
if
you
come
down
to
things
like
app
you'll
see
the
CI
CD
category
right
and
in
the
CSE
D
category,
there's
a
whole
bunch
of
stuff.
There's
some
member
companies
and
projects
there's
some
who
aren't,
but
even
as
I
looked
at
this
I
thought.
A
In
fact,
if
you
come
into
something
like
the
let's
go
to
the
application
definition
and
image
built
like
category,
and
so
you
see
the
CN
CF
project
so
home
and
a
couple
of
sandbox
projects,
if
you
all
didn't
know,
build
packs
known
primarily
for
things
like
Cloud
Foundry
have
made
their
way
sandbox
projects
into
the
CNC
F,
it's
it's
being
integrated
into
other
toolings,
I
want
to
say,
K
native
there's
foot
ways
of
doing
those
things
I'm.
In
addition
to
that,
there
are
other
projects
that
aren't
under
the
CNC
F
here.
A
But
there
are
still
that
probably
could
be
listed
here
that
aren't
in
this
space
and
as
I
look
at
that
something
I
I
know
of
a
couple
that
I
use
that
aren't
listed
here.
So
if
you're
not
in
here
already,
please
it's
as
simple
as
doing
a
PR
against
the
github
repository,
which
is
github
C
and
C
F
landscape.
If
I
could
spell
it
right,
it's
actually
submitting
a
pull
request
against
this
repo
and
then
it's
generated
out
of
here
a.
C
Mud
with
varying
supposed
to
discuss
the
categories,
what
the
categories
are
and
I
mean
I'm,
not
I'm,
not
intending
to
go
and
make
a
big
change,
but
I'd
like
to
kind
of
have
a
conversation
like,
for
instance,
this
category
we're
looking
at
at
the
moment.
I,
don't
understand
my
telepresence
is
there
and
other
things?
Maybe
I
would
like
to
be
able
to
look
at
tools
that
are
specifically
for
kubernetes,
but
at
the
same
time,
convert
is
another
old
one
here
right,
yeah.
A
C
A
A
Google
Groups
is
not
available
in
China
through
the
web
interface,
and
so
people
can't
sign
up
and
use
it
on
one
of
the
largest
markets
for
the
CN
CF,
and
so
that's
caused
them
to
use
an
alternative,
and
so
it's
you'll
find
a
bunch
of
mailing
lists
that
are
over
here
for
people
to
use
whether
it's
home
or
anything
related
to
the
CN
CF.
The
TOC
things
like
that
the
categories
are
managed
by
the
TOC
that
has
Tuesday
calls
every
other
or
sorry,
the
second
and
fourth
Tuesday
the
month.
A
I
want
to
say,
and
so
it's
it's
that's
the
group
that
oversees
us
one
of
the
TOC
members
oversaw
the
categories.
But
this
is
the
mailing
list
for
the
conversations
happened,
and
so
this
mailing
list
or
the
TOC
one
which
you
can
also
find
on
there
are
places
to
maybe
ask
those
questions,
and
if
you
go
to
this
one
here,
the
CN
CF
reference
architecture,
you
can
ask
questions
about
it
and
the
people
who
are
involved
in
building
it
or
on
this
list.
They
can
respond
to
that.
Okay,.
C
A
A
And
making
changes
is
just
a
pull
request
away
because
that's
how
it's
all
handled,
in
fact,
if
you
get
into
this,
oh
that's
the
it's
been
a
while
since
I've
managed
this.
So
if
you
go
into
things
like
main
specs-
oh
it's
not
in
here!
Okay,
there
is
oh,
yes,
you
will
see
things
like
here's,
many
of
the
linked
files
or
here's
all
the
landscapes
that
are
generated
because
they
generate
PDF
versions
and
all
that
other
detail
from
it
and
then
there
and
there's
a
server
list,
one
as
well.
A
There's
a
trim.
They've
got
a
bunch
of
stuff
I'm
trying
to
remember
where
they
can
fig
is
in
here
right
here.
Sorry,
there's
a
ya
know
file
with
all
of
the
repos
in
here,
along
with
details
on
them,
and
so
you
can
see
things
like
your
category
name,
your
subcategory,
so
even
changing
things
is
as
simple
as
moving
things
around
in
a
yamo
file.
A
A
If
you
have
suggestions,
please
pass
them
along
the
TOC
and
the
CNC
F
is
trying
to
make
this
as
useful
as
possible.
In
fact,
one
of
the
areas
they've
gone
into
recently
is
the
server
list,
one
which
is
the
landscape
just
for
serverless
and
they've
got
a
print
version
of
it,
and
you
see
there's
certain
things
like
you
can
zoom
in.
A
So
you
can
see,
there's
things
like
here's
different
tools
and
security
and
there's
frameworks
for
there's
platforms
where
they're
toasted,
which
is
where
you
gonna,
get
things
like
lambda
or
installable,
which
is
where
we
get
something
like
coulis
and
K
native
and
ref
is
now
on
there.
So
it
gives
you
some
ideas
of
even
what's
in
the
market,
where
they're
at.
A
A
A
B
B
We
are
packing
some
necessary
step
that
are
helpful
for
production
all
thing
all
one
place
the
men
will
tease
about
cube
TV
if
you
know
how
to
deploy
one
database
in
cube
TV,
but
you
will
able
to
deploy
everything
every
they
supported
that
a
dozen
caved
but
keep
doing
an
hour.
Every
database
structure
is
similar
and
everything
work
done
in
similar
flow
here
is
a
list
of
some
supported
database.
We
have
instant
backup
schedule,
their
cup
initialization,
live
on
buildings,
promises,
monitoring
system
and
yeah.
Of
course,
I
have
clustering
to
today.
B
I
will
show
a
short
demo
how
to
deploy
a
put
this
cluster
I
would
give
to
be,
and
if
I
get
time,
I
can
show
maybe
instance
tab
back
off.
Okay
I
have
already
installed
give
DB.
My
cluster,
you
can
see
here
cube
TV
operator
programming
we
do
here
is
install.
He
has
created
some
share.
Do
you
call
that
a
dispatch
on,
for
example,
for
for
Jessica?
B
B
B
This
separating
this
post-conversion,
creating
this
Safari
CIT
for
this.
For
these
images
they
help
us
to
modify
that
docker
images
without
changing
our
operator
without
a
motivational
operator.
It
also
provides
a
job
to
use
their
custom
operator.
True,
ok,
here
we
will
create
a
whole
this
cluster
and
the
cluster
demo.
B
We
are
going
to
use
for
this
button.
This
is
in
a
foliage
partials.
Yet
we
have
shown
just
before
we
are
going
to
create
a
replica,
so
there
will
be
a
clip,
for
one
of
them
will
be
after
will,
act
as
master
and
other
three
will
act
as
a
standby
replica
master
will
be
selected
by
a
direction.
Then
here
we
have
standby
mode,
it
can
be
either
warm
or
hot
one
instantly.
B
Node
means
you
cannot
connect
the
standby
replica
the
system,
but
if
you
go
only
here,
only
he
used
will
be
used
for
any
failover
situation.
But
if
you
use
hot
standby
mode,
then
you
do
able
to
connect
with
the
stem
by
replicas
and
sub
read-only
queries,
but
not
right
course.
Only
read-only
queries
from
this
repeater.
Okay,
then
we
have
here
I
specify
store,
is
required
for
each
each
replicas.
Then.
Finally,
we
have
another
thing
called
termination
policy.
Jesus
testify
when
you
try
to
deploy
a
delayed,
Espada's,
CID
or
Cuba
walk
into
the
window.
B
I
have
a
current.
We
support
for
termination
policy,
do
not
terminate
pause
delay
turn
wipeout
do
not
terminate
if
it
is
Donna
terminate
termination
policy.
When
you
do
try
to
delay
these
possible
buttons
here,
they
cubed.
You
will
reject
a
request.
This
there's
an
accidental
deletion
database.
Okay,
let's
get.
B
B
B
B
B
B
B
B
Yes,
these
are
yeah
from
here.
We
are
16
settings
up
for
about
fully
asleep,
because
we
can
see
her.
We
have
two
replicas
and
they
are
the
Streamy
replicas.
Now
I
will
show
how
to
do
automatic.
Failover
I
will
try
to
force
the
master
not
to
fail
until
itself.
I
will
delete
the
part,
and
one
of
this
replica
will
take
over
as
master
okay.
Our
busy
demos,
zero
was
acting
as
master.
Yes,.
B
B
A
B
B
Okay
sure
cube
TV
watches
for
snapshots.
They
are.
This
is
its
natural
charity.
If
we,
if
we
schedule
said
configure
scheduled
backup
this
year,
due
will
be
automatically
created
by
DB
and
new
snapshot
will
be
created.
But
now,
in
our
case
we
will
take
backup
in
Stanley,
so
we
will
create
this
Eddy
and
keep
the
voltage
a
snapshot
where
we
are
saying
that
we
will
take
questions
database
and
this
database
name
here.
B
Okay,
so
Jeff
is
running,
it's
not
sure
to
succeed,
so
database
already
started
our
cube
TV.
Damn
fine.
B
A
C
With
snapshot,
oh
go
ahead.
Sorry,
somebody
else
equations
to
that
just
curious,
because
I
thought
well
from
what
I
understood
the
Venus
was
the
the
kind
of
the
go-to
solution
for
my
sick
honk
abilities
that
is
backed
by
the
Mexico
community.
I,
wonder
whether
you
have
any
specific
reasons
why
you
diverge
from
from
that
path.
C
B
B
Ocupado
does
not
work
for
only
for
MySQL
and
it
will
work
falling
on
six
database
a
list.
Yet
we
are
ready,
also
yeah,
it's
City,
so
we
are
taking
on
database
into
a
general
plan,
a
general
structure,
general
platform.
So
if
you
can
deploy
my
secret
with
cube
TV,
you
will
able
to
get
all
of
them
other
how
the.
Furthermore
we
have
some
station
facilities
such
as
clustering,
but
it
back
up
and
initialization
from
snik
short
or
pair
additional
ski
and
pole
case.
B
We
have
also
use
a
visualization
from
wall
archive,
so
our
main
goal
is
to
take
all
the
data
base,
all
the
necessary
data
base
into
a
general
platform
and
general
structure.
So
you
don't
have
to
learn
every
blandness
different
operator.
Did
they
don't
need
to
know
different
operator
to
four
different
database.
C
D
D
The
test
and
cockroach
DB
for
MySQL
and
leave
copper's
DB
is
Postgres
interface.
At
this
point,
there
are
very
different
architecture
from
what
could
be
DBS
like
the
tenses
are
scale
out:
sharded
rep,
my
SQL
database
with
automated
leader
election.
It
uses
energy
VM
MySQL
under
the
hood,
but
its
own
distributed
system,
that's
kind
of
architected
from
scratch
to
run
natively
in
a
cloud
native
environment
like
kubernetes,
or
you
could
probably
get
it
to
run
on
these
surfs
same
thing
from
a
copper
TV,
cockroach
TV
is
again
Chardin
replicated
my
scale
out.
D
D
C
D
A
And
I
was
gonna,
say,
I
think
there's
slightly
different
use
cases
for
some
of
this
right,
so
coop
DB
is
kind
of
providing
a
SAS
solution,
essentially
through
kubernetes
for
folks
to
say.
Give
me
a
database
of
this
type
or
give
me
some
clusters
and
the
features
you'd
get
around
it.
If
you
were
using
something
say
in
a
public
cloud
and
I
think
for
folks
who
do
on-premise
or
aren't
in
one
of
the
three
major
public
clouds
and
are
just
looking
to
get
some
of
this
going.
A
C
E
Yeah,
so
this
is
the
model
I
think
I
won't
be
the
amuro's
on
this
project.
Yes,
I
can
answer
this
question,
so
the
answer
to
that
question
is
service
brokers
that
are
available
today
on
various
cloud
providers:
kind
of
running
the
databases
on
that
like
sort
on
the
hosted
cloud
provider
solution,
but
with
cube
TV,
you
were
running
database
inside
your
queries
cluster,
and
you
can
run
that
like
on
your
bare
metal
on
the
car
provider
or
any
cowboy
doc
that
doesn't
have
a
its
own
hosted
database.
So
we
also
have
a
service
broker
interface.
C
E
And
advantage
with
the
managing
with
our
rule
series
is
this
because
they
are
not
like
a
generic.
Well,
there
are
two
advantages:
one
is
that
they
are
not
generic
like
service
broker
interface,
so
like
we
have.
We
use
the
validation
admission.
Controller
validation
will
work
technique
that
like
to
implement
that,
like
do
not
terminate
feature.
So
if
you
try
to
delete
them
like
a
QT,
B,
C
or
D,
it
will.
E
If
you
have
that
field
set,
then
it
will
reject
your
unit
operation,
but
that
kind
of
thing
you
cannot
really
do
with
the
service
broker.
That's
one
problem
and
the
second
problem
with
service
broker
I
think
it
needs
a
it
uses
the
aggregation
service,
so
you
need
to
have
your
own
it
city
and
all
that
I
mean
if
it
moves
to
like
a
CR
D.
At
least
that
problem
goes
away,
but
it's
not
the
case
today.
E
A
E
C
A
A
So
I'm
sharing
my
screen
again
and
here's
the
document
that
I
started.
It's
actually
pretty
rough.
What
I've
broken
it
down
into
is
three
major
categories,
and
so
I
really
want
feedback
here
like
what
do
you
all
think,
and
so
I've
got
three
major
areas
here.
First,
there's
the
application
operator,
then
there's
the
application
developer
and
then
there's
the
kubernetes
workloads,
API
contributor
I,
know
the
contributor
experience
folks
get
into
some
of
this
stuff.
But
since
we
own
the
workloads
API,
it
would
be
nice
to
maybe
have
some
some
details
on
this.
A
But
I
put
that
kind
of
third
down
here
and
for
application
operator
and
developer
a
kind
of
sub
broke
that
down
and
I'm
curious.
What
folks
think
about
that,
because
I
kind
of
get
into
it
like
if
you're
doing
functions
as
a
service
right
as
a
user
like
I'm,
writing
functions
and
I'm
sticking
them
into
my
event
based
system,
okay
native
couplets,
whatever
you're
almost
operating
mini
applications,
and
so
I
thought
about
putting
that
in
as
kind
of
a
sub
area.
A
If
you're
doing
pass
on
kubernetes
what
kind
of
things
matter
there
and
one
of
the
things
that
caught
my
attention,
if
folks
don't
know,
Cloud
Foundry
released
a
new
project
to
use
kubernetes
as
a
scheduler
and
study,
Diego
I
think
it's
in
their
labs
incubating
I'm,
forgetting
the
term
that
they
use
over
there.
But
it's
a
new
thing
that
was
released
and
then,
if
I'll
call
the
traditional
kubernetes
app
operation,
so
I
kind
of
broke
those
down
thinking.
There
might
be
some
slightly
different
workflow
and
details
in
these
areas.
A
C
A
Gonna
argue
no,
as
far
as
roles
go,
I
would
argue,
no
I,
think
in
practical
measure,
probably
not
at
a
lot
of
companies.
You'll
have
maybe
especially
large
companies.
You'll
have
one
group
that
does
the
cluster
operation
and
you
might
even
punt
a
lot
of
that
to
a
public
cloud
or
then
somebody
else
is
doing
the
application
operation
in
the
cluster,
and
so
it's
kind
of
a
different
scope
to
role.
A
And
I
take
a
look
at
this.
Moore's
rolls
right,
they're
separate
roles,
but
the
same
person
may
do
more
than
one
role
or
in
and
that'll
happen
in
some
companies
and
other
companies.
It
might
be
different,
like
in
a
small
startup.
You
might
just
have
a
handful
of
people
who
occupy
all
of
the
roles
and
yet,
in
a
large
corporation,
it's
pretty
easy
to
segment.
Each
of
these
into
different
roles
that
may
even
be
in
different
departments.
A
C
Because
I
mean
I
just
generally
tend
to
you,
know,
I,
don't
exactly
believe
separating
these
things
out,
but
maybe
I
haven't
worked
in
a
large
enough
team.
To
kind
of
you
know
convince
myself
on
that,
but
I
do
think
that
that
to
some
extent
agree
most
recently,
like
all
the
things
to
do
with
DevOps
inside
sheep
was
mostly
about
trying
to
blend
these
lines
even
more
and
I
wonder
whether
that's
a
good
idea
to
kind
of
separate
them
out
but
like
for
what
purpose
again
I'm.
Still,
it's
not
very
clear
to
me
still.
A
A
It's
more
of
a
description
we'd
like
to
capture
the
information
on
folks
and
I'm.
Sorry,
we
were
talking
about
this
while
ago,
but
I
want
to
capture
the
information
on
who
these
folks
are
to
try
to
understand
who's.
Doing
these
different
roles,
because
I
think
it's
important
if
we're
building
things
for
folks
to
understand
them,
because
their
biases
may
not
be
ours
and
and
I
think
that
when
I
realized,
that
was
the
first
time
that
I
was
introduced
to
an
application
that
had
500
engineers
working
on
it.
A
And
it
was
one
application,
and
so
there
were
many
distinct
people
wearing
different
hats
for
the
different
roles,
because
there
were
so
many
people
on
them
and
I'm.
Now
we're
you
know,
there's
single
applications
that
up
a
thousand
engineers
working
on
them
and
they're
complex,
but
a
lot
of
those
you'll
have
different
people
who
do
slightly
different
things,
and
this
was
kind
of
some
of
that
stuff.
That
I
ended
up
dealing
with
with
the
separation
and
here
we're
trying
to
just
maybe
target
the
ones
that
specifically
apply
to
kubernetes
like
where
kubernetes
meets
the
end.
C
Mean
it'd
be
good
to
kind
of
break
in
a
well
I
mean
expand
this
a
little
bit
right
and
start
with
and
like
sort
of
add
some
examples
of
what
those
faults
do
and
what
actual
application
means.
In
these
cases,
I
presume
application
could
be
something
that
consists
of
like
several
different
parts
like
as
in
deployments
in
a
cluster
right.
A
Yeah,
it
could
be
one
micro
service
that
you're
responsible
for
microphone
micro
services.
It's
kind
of
a
squishy
term
totally
realize
that,
in
fact,
when
we
get
into
application
developer,
you'll
see
that
I
even
kind
of
changed.
Some
of
this
up
right.
Looking
at
some
of
this,
because
you
know,
there's
kubernetes
native
application
developers
and
like
there's
folks
who
are
just
trying
to
build
things
using
api's
like
they
may
get
something
in
through
one
deployment
of
an
application,
and
then
that
deployment
goes
off
and
to
do
certain
pieces
of
work.
A
A
Stuff,
like
Maria
or
my
sequel,
that
we
just
saw
or
Postgres
right
that
are
kind
of
more
traditional,
and
in
that
case
somebody
who's
going
to
develop.
Some
complexity
would
be
like
the
coop
DB
developers
or
somebody
building
a
controller
or
operator
really
around
managing
some
of
those
applications
to
add
the
intelligent
business
logic
like
we
saw
today,
demo
right.
C
So,
okay,
so
so
then
I
mean
in
DB
operations,
application
operation
operators
category
would
you
would
you
consider
like
folks
who
use
off-the-shelf
applications
and
make
sure
that
there
aren't
happening
or
well
well
predominantly
I'm
thinking
right
certainly
include
like
in-house
applications.
Yes,.
A
Because
you're
doing
the
duty
of
operating
it
rather
than
necessarily
developing
and
in
like
an
open
source,
we
use
so
many
different
things
that
come
from
so
many
places
right.
Somebody
who's,
building,
a
wordpress
site
might
have
some
custom
business
logic,
and
yet
they
may
be
running
engine
X
in
front
of
it
and
my
sequel
behind
it.
A
Maybe
there's
a
caching
server
somewhere
in
there
all
of
these
things
they
got
from
other
places
and
they're
gonna
operate
it
and
I
use
WordPress,
because
it's
the
example
everyone
uses,
but
there's
all
these
other
components
that
you're
operating
you
got
from
elsewhere.
Yet
there's
some
custom
business
logic
to.
C
It
seems
like
it'd,
be
good
to
expand
a
little
bit
more
like
on
some
of
these
things
right
and
write
out
some
of
the
example
applications
in
some
of
these
cases.
I
won't
be
more
specific
that
then
there'd
be
a
bit
more
to
discuss,
because
at
the
moment
the
older
is
to
discusses,
like
the
question
of
whether
the
East
lines
should
be
defined
at
all,
so
just
expand
it
a
little
bit
more
with
specific
cases
like
where
does,
for
example,
right.
Where
does
who
would
be
running
like
the
service
broker?
A
C
A
C
We're
talking
about
here,
yeah,
because
I
did
I,
did
struggle
with
with
that
being
case.
Sometimes
right,
you
come
to
talk
to
a
few
queries
developers
in
whatever
setting.
That
is,
whether
that's
a
sick
of
working
group
meeting
all
right,
cube,
corner
whatnot
and
when
you
talk
about
something
people
seem
to
think
about
very
kubernetes
specific
cases
right.
So
you
talk
about
build
people,
think
about
kubernetes,
build
system,
win
side
of
kubernetes
repo
right.
C
They
don't
necessarily
think
about
any
specific,
build
systems
that
users
may
be
using
anything
like
that
right
and
that's
something
that
makes
makes
makes
it
hard.
It
makes
it
quite
hard
to
have
a
conversation
about
actual
use
cases
of
the
end-users
with
people,
because
people
think
about
converting
this
project,
but
first
yeah,
yeah
I,
think
this
would
be
very
helpful
for
that.
A
A
C
F
The
definition
of
profiles
and
the
overlap
with
some
one
of
my
goals
when
trying
to
as
we
prowl
through
this
flow
through
this
data,
looking
at
the
patterns
of
usage
and
I,
think
it
might
be
pretty
relevant
to
them
to
help
define
what
is
a
faz
operational.
If
you
have
cluster
that
look
like
this,
that's
going
to
look
more
like
a
fan
or
a
pass,
or
traditional
or
infer
operator,
and
one
of
the
things
that
I'm
looking
at
right
now.
F
As
far
as
data
sources
on
that
is
trying
to
decide
which
applications
best
represent
heavy
usage,
our
community
and
so
pulling
from
there
and
then
being
continuously
pulling
the
newest
versions
of
those
applications
and
creating
audit
logs
audit
logs
contain
all
of
the
API
information
about
the
call,
including
the
application
that
called
it
and
I'm
trying
to
create
some
extensions
so
that
every
API
call
also
includes
the
call
stack
all
the
way
down
through
the
client.
Libraries
and
whatnot
I.
Just
wanted
to
my
partly
to
put
this
on
the
mailing
list
or
something
but.
A
A
A
A
A
Okay,
does
anybody
if
anybody
has
a
demo,
they
would
like
to
have
we've
got
open
demo
slots
in
the
next
few
weeks,
so
please
come
reach
out
to
us.
If
you've
got
something
you
want
a
demo
or
showcase
or
talk
through
next
week,
we'll
be
back
I'm
talking
about
the
workloads
API
stuff
with
that,
does
anybody
have
any
final
parting
thoughts,
words,
ideas,
anything.