►
From YouTube: Kubernetes SIG Testing 2018-04-03
Description
A
Hi
everybody
happy
Tuesday,
my
name
is
Karen
Creek
and
burger.
Today
is
Tuesday
April
3rd
and
you
are
attending
the
weekly
communities,
safe
testing
meeting,
which
is
being
recorded
and
will
be
posted
to
YouTube
publicly
shortly
thereafter
on
today's
agenda,
it's
going
to
talk
real
briefly
about
plans
for
coop
con.
If
folks
have
them,
we
got
a
couple
talks
there,
Matt
Fisher,
if
you're,
here
or
other
folks,
there
was
discussion
about
sort
of
a
developer
tools,
working
group
and
we
wanted
to
discuss.
A
B
B
A
So
I
figure
like
there's
a
dot
I
saw
and
Jeff
looks
contribute
to
you
that
sort
of
documented
the
history
behind
moving
from
a
bunch
of
Jenkins
jobs
to
a
bunch
of
Jenkins
job
builder
stuff.
To
bootstrap
to
you
know
where,
where
we
went
ahead
and
yes
I'm
gonna
see
if
I
can't
bring
my.
This
is
fine
plushie
dog
on
stage
mascot,
so
I
wanted
to
give
just
sort
of
an
intro
of
and
tell
some
interesting
stories
and
then
hand
off
to
you
send
for
the
deep
dive
later
in.
B
My
career
plan
is
that
I
can
talk
about
like
in
depths
of
a
lifecycle
if
they
proud,
like
how
proud
works
like
what
like
each
component
of
words
and
what
what
happens
to
be
like
hold
on
the
PR
or
you
slash
tree
past,
and
then
Co
can
talk
about
type
in
general
I,
believe
we
are
going
to
pipe
or
the
main
take
a
repo,
hopefully
before
tubecon.
So
we
should
let
people
know
more
about
italic
works
versus
occurrence
McHugh.
How
much
time
do
you
have
before
this
talk
a.
A
B
A
C
D
D
D
B
C
D
B
A
Totally
okay
I
want
to
keep
us
moving
along,
so
look
for
more
from
Center
and
I,
actually
ya
know
have
exact
some
slides
done
today
soon,
so
somebody
named
Matt,
you
know
Jews
how
about
you
memory
good
I
heard
there
was
a
discussion
at
cig
apps
on
Monday
around
the
concept
of
a
developer
tools.
Working
group
and
I
know
that
folks,
in
this
sake,
have
also
expressed
some
some
ideas
that
maybe
some
of
what
that
working
group
was
proposing
my
overlap
where
we
were
or
what
what
our
interests
are.
E
I
guess
yeah
Mad
Men
Dan.
Here
we
were
on
developer,
tooling,
mini
cube,
scaffolds
container
death,
a
bunch
random
container
tools.
So
we
are
team
originally
made
the
proposal.
There
was
a
lot
of
interest,
obviously
some
pushback
from
say
gasps,
who
felt
like
there
was
a
bit
of
overlap.
I
mean
we
proposed
a
working
group
because
we
knew
that
there
would
be
a
lot
of
overlap
with
a
lot
of
SIG's
and
this
would
kind
of
span
a
bunch
of
projects,
but
I
mean
basically
where
we're
at
right
now
is
trying
to
figure
out.
E
E
So
there
was
no
clear
consensus
from
the
sig
after
meeting,
but
we're
kind
of
taking
the
time
this
week
to
meet
with
a
few
different
SIG's,
including
you
guys
just
to
see
you
know
where
you
know
how
we
could
not
duplicate
effort
and
how
we
could,
actually,
you
know,
produce
some
good
stuff
out
of
it.
So
yeah,
that's
kind
of
where
we
stand
right
now.
We're
planning
to
I
mean
focus
on
a
lot
of
developer.
Tooling,
so
I
mean
we're
we're
the
maintainer
of
mini
cube.
E
So
I
know
that
there's
it
sounds
like
you
guys
have
been
working
on
something
different,
but
it's
similar
for
local
clusters.
You
know
I,
guess
we
would
love
to
hear
about
some
of
the
effort
going
on
there.
You
know
people
have
used
mini
cube
and
things
like
Travis,
CI
and
very
basic
CI
systems
to
kind
of
smoke
test
their
apps,
but
I
mean
I.
Think
we
would.
We
would
love
to
have
a
better
story
there
for
something
like
mini
cube
and
if
that's
not
a
virtualized
solution.
E
F
E
E
Don't
think
we
get
the
word
out
enough,
but
we
actually
have
to
kind
of
main
interfaces
and
mini
cube
to
allowed
this
sort
of
you
know
hiding
behind
many
cubes,
so
one
being
the
the
bootstrapper
interface
which
we
use
to
kind
of
transition
from
local
cube,
which
was
our
single
binary
way
of
we
basically
rendered
in
all
of
kubernetes
and
and
ran
all
the
components
as
go
routines,
which
was
quite
a
hack
that
lasted
quite
a
long
time,
but
it
was
good
because
it
was
fast.
It
was
a.
E
It
was
a
good
development
tool,
but
it
wasn't
great
for
kind
of
looking
like
a
production
cluster,
so
we
we
transitioned
to
cou
Badman.
So
that's
what
we're
using
now
to
set
up
the
cluster,
but
this
this
bootstrapping
interface
is
pretty
simple.
It's
just
kind
of
ways
to
start
the
cluster.
Stop
the
cluster.
You
know.
Do
some
specific
operations
on
the
cluster,
but
I
could
imagine
there
being
other
implementations
of
setting
up
the
cluster
and
it
could
possibly
be
what
you
guys
are
working
on.
E
You
know
it
could
be
a
lot
of
different
things:
the
other
interface
being
kind
of
our
live
machine
interface
that
we've
we've
taken
from
docker
machine
because
remember
that,
but
basically
that's
responsible
for
not
provisioning
the
cluster
but
provisioning,
the
the
kind
of
environment.
So
in
this
case
it
handles
on
kind
of
cross
platform
virtualization
so
using
the
native
hypervisors
IH
platform.
But
we
also
have
a
nun
driver
if
you
guys
are
familiar
with
it,
which
basically
skips
it's
this
step
so
on.
E
G
We're
interested
in
local,
like
multi,
node
and
also
definitely
running
on
clusters,
one
of
our
requirements,
which
is
why
Clinton's
been
looking
at
the
DVD
stuff,
but
just
more
generally
Cinemax,
but
I've
been
here
on
engineering
productivity
at
Google,
working
on
the
tester
for
stuff.
A
lot
I
see
some
overlap
with
this
work
group,
just
in
building
things
on
kubernetes
in
20
through
Burnett
use,
because
it
turns
out
what
all
the
infrastructure
that
testing
does.
G
E
So
I
mean
one
thing
that
so
we
had
a
few
people
approached
us
with
a
multi
note
proposal
for
mini
cube,
and
so
we've
been
meeting
a
lot
kind
of
autumn
band
around
those
efforts.
But
part
of
the
reason
for
starting
this
working
group
is
that
you
know
something
like
that.
Is
it's
not
a
fundamental
shift
in
mini
cube,
but
it's
gonna
require
a
lot
of
work
to
kind
of
support.
E
You
know
our
current
users,
but
it's
something
that
we
definitely
want
to
do
is
just
a
matter
of
making
sure
that
we
we
match
all
the
features
with
people
who
are
going
to
be
able
to
work
on
them
and
maintain
them,
and
that's
kind
of
the
motivation
behind
getting
a
working
group
just
so
that
we
can
connect
all
the
people
that
are
interested
in
these
efforts,
because
I
mean
there
are
a
lot
of
people
interested
in
these
things.
We
just
don't
have
a
way
of
you
know
kind
of
coordinating
that
effort.
A
So
speaking
with
my
steering
committee
hat
on
working
group
starts
supposed
to
be
entities
that
own
code
SIG's
are
supposed
to
be
entities
that
own
code
sub
projects
are
supposed
to
be
things
that
are
well
contained
within
sinks
that
can
sort
of
determine
their
own
destiny,
so
to
speak
so
I
think
like,
for
example,
the
reason
many
cube
may
have
ended
up
within
clustered
life
cycle
is
I.
Think
clustered
life
cycle
is
the
sig
that's
responsible
for
how
you
sort
of
deploy
and
stand
up
and
manage
the
life
cycle
of
various
kinds
of
clusters?
A
Be
that
a
single
node
cluster?
That's
running
on
your
laptop,
that's
easy
for
development
purposes,
or
you
know
a
comp
cluster
that
runs
in
AWS.
Or
what
have
you
so
my
my
reaction
is
one
of
like
a
is
one
of
meeting
an
overload
kind
of
so
I
do
I.
Do
I'm
I
hear
definitely
that
there's
some
overlapping
interests
with
with
mini
cube
and
ways
of
like
whether
or
not
we
want
to
try
and
have
one
canonical
way
of
developing
and
testing
against
them.
A
F
Well,
the
testing
common
stuff
is
both
it
kind
of
it.
It
crosses
over
into
that
space,
because
the
primary
use
case
is
to
develop
common
infrastructure
for
both
upstream
kubernetes,
as
well
as
external
providers,
who
want
to
have
apparatus
that
they
can
dig
into
that's
common
as
the
repo
gets
fort
in
many
directions
right
so
in
it
needs
to
be
tooling
that
spans
repositories
and
that's
part
of
the
goal
and
it
kind
of
overlaps.
It's
definitely
not
it's
a
Venn
diagram,
it's
not
total
overlapping,
but
it's
there
is
overlap.
F
A
A
So
as
an
example,
there's
like
the
application
definition
working
group,
which
was
formed
because
it
seemed
like
there
are
a
bunch
of
disparate
interests
across
SIG's
and
I
think
the
way
that's
worked
now
is
like
each
different
SIG's
are
owning
different
parts
of
the
concerns
that
were
raised,
but
I
think
the
working
group
still
might
exist
as
a
way
to
like
make
sure
everybody
is
actually
following
through
and
delivering
all
the
things
they
said
they
might
deliver.
I,
don't.
H
Know
yeah
the
way
Brian
grant
kind
of
weirded
thing
is
in
the
discussion
on
the
working
groups
thread
was
that
it's
okay
for
working
groups
to
kind
of
form,
small
task
forces
that
go
and
contribute
improvements
to
a
wide
range
of
tools,
but
those
improvements
have
to
ultimately
be
owned
by
the
sub
project
or
the
stage
or
whatever
owns
those
tools.
These
are
kind
of
like
short-term
resource
needs
across
a
whole
bunch
of
different
stages
or
sub
projects
to
kind
of
be
tackled
by
working
group,
but
yeah
you're.
E
Bio
work,
yeah,
we're
not
planning
on
I
mean
the
dislike.
The
ownership
of
things
like
mini
EQ
or
you
know,
tools
like
scaffolding.
Zaft
are
not
like
CNCs
projects,
but
you
know
they.
We
we
want
to
make
sure
that
we're
we're
building
these
in
a
way.
So
if
we
build
out
the
multi
node
support,
we
want
to
make
sure
that
we're
building
in
a
way
that
you
know
could
address
some
of
the
sig
testing
needs.
If
possible,
you
know
it
doesn't
really.
You
know,
I
mean
I
guess.
E
One
other
thing
to
address
is
that
I
think
a
lot
of
people
are
scared
of
more
meetings.
I
mean
these
meetings
are
already
going
on.
So
it's
it's
not
they're,
just
informal
in
the
sense
that
they're,
not
an
official
working
group
meeting-
and
you
know-
is
that
right
or
wrong
I
mean
it
was
born
out
of
necessity.
Sure.
A
I
think
the
concern
I'm
trying
to
raise
is
the
flat
hierarchy
we
have
of
some
30-something
different
groups
is
really
not
good
for
human
discoverability,
so
I'm
trying
to
think
of
it
in
terms
of
a
hierarchy.
That
makes
it
easier
to
kind
of
understand
where
I
should
go.
If
I'm
interested
in
this
particular
topic,
as
opposed
to
searching
in
amongst
thirty-one
instead
of
thirty
groups
or
something
yeah.
E
H
You're,
absolutely
right,
though,
there
is
asleep
at
some
point.
There
is
a
flat
list
of
meeting
if
they're
overlaid
on
to
everyone's
weekly
schedule.
So
whether
it's
organized
is
like
there
aren't
really
it
there's
no
notion
of
sub
ciggies
and
even
working
groups
are
supposed
to
be
sponsored
by
a
state
but
they're
not
actually
ever
nested
I
would.
A
Say
sub
projects
are
kind
of
what
I'm
thinking
of
what
I
think
read
when
you
say
sub
sick,
that's
what
I
think
of
a
group
of
people
who
are
interested
in
delivering
on
some
piece
of
code
that
accomplishes
some
discrete
objective,
but
so
I
kind
of
want
to
hear
from
like
Eric
or
Steve
or
or
other
folks
longer
live
with
the
same.
Like
do
you
see
other
areas
of
overlap
here?
My
impression
is
more
of
the
overlap
of
this
sega's
with
contributor
experience.
I
Seeing
head
nodding
so
yeah
I
would
definitely
say
that
the
you
know,
local
testing
right
is
definitely
something
that
Quinton
has
been
working
on
and
I
feel
I
could
be.
You
know
useful
if
Quinton
and
the
other
people
working
on
the
nested
Esther
nested
docker
solutions
are,
you
know
talking
to
the
people
working
on
the
meeting,
cube
efforts,
I,
think
that
would
be
useful
to
make
sure
that
we're
developing
and
going
in
a
good
way
there,
depending
on
how
developer
tools
you
know
work
out.
I
Scaffolds
kind
of
you
know
like
prow
sort
of
has
ideas
of
like
a
build,
workflow
right
and
so
I
could
potentially
see
overlap
there
and
cubic
I.
Don't
know
exactly
you
know
about
the
whole
bootstrapping
interface.
That's
potentially
is
similar
to
cube
tests
or
whatever
so
I
feel
I.
Don't
I
don't
feel
strongly
about.
You
know
what
the
hierarchy
is,
but
I
do
feel.
C
Okay,
yeah
I
guess
this.
My
only
comment
would
be
like
on
the
overlap
on
the
build
flow
side
of
things.
I
guess
we
have
a
pretty
sophisticated
like
set
of
ideas
about
that
and
OpenShift,
which
I
think
we're
expecting
to
make
at
least
voice
some
of
those
opinions
or
lessons
learned
in
the
working
group
and
so
I
guess
I'm,
not
sure
it's
a
sick
testing
involvement
as
much
as
just
like
yeah
I,
wouldn't
expect
that
to
be
like
a
super
huge
involvement
from
the
city.
I
Yeah
I
guess
my
only
other
thing
is:
you
know
I
sort
of
feel
like
in
an
ideal
world
right.
Just
as
Tim
was
talking
about
how
you
know
and
I
like
I
feel
like
there,
there
should
be.
It
is
a
Venn
diagram
and
I
feel
like
there
should
be
like
it
would
get
really
good
developer
tools
that
apply
to
people
in
making
kubernetes
applications.
Since
a
lot
of
the
things
about
kubernetes
are
you
know,
kubernetes
deployments
themselves.
A
Cool
so
I
think
we
look
forward
to
hearing
sort
of
what
what
you
learn
in
chopping
us
around
with
with
other
SIG's
like
like
I,
said
I
feel
like
there's
a
lot
of
interest
in
collaborating
for
sure,
and
some
of
these
efforts
aren't
are
great
and
we'd
love
to
push
that
forward.
It's
just,
oh,
my
god,
so
many
meetings,
so
whatever
we
can
do
to
help
you
accomplish
what
you're
trying
to
accomplish
in
the
way
that
it's
the
most
sense.
You
know
looking
forward
to
moving
forward
on
that
I
work
for
you,
man,
yeah,.
E
Yeah,
if
I
guess,
if
you
want
a
comment
on
the
thread,
just
what
we
talked
about
or
I
could
I
could
post
a
summary
just
to
loop
in
the
non-stick
testing
people,
that's
kind
of
what
I've
been
doing.
Just
as
we,
we
don't
run
these
differences
I'm
just
to
kind
of
let
communities
devs
know
exactly
what
we
talked
about
sure.
A
F
E
So
we
we've
been
having
this
this
local
dev
meeting.
It's
mainly
been
with
the
mini
chef
folks,
but
we've
been
trying
to
as
part
of
this
proposal,
move
that
into
the
developer
tools
working
group,
but
we're
we're
still
having
those
meetings
now
so
I
could
loop
even
on
those
anywhere
which
is
part
of
it
almost
discoverability.
E
A
That
thing
for
what
it's
worth
sends
definitely
like
a
sub-project
to
me
and,
and
it
would
be
a
matter
of
figuring
out
which
sync
should
like.
Oh
that's
some
project,
but
it's
a
project
that
owns
its
own
destiny
right,
okay,
so
we're
kind
of
at
the
30
minute
mark,
which
is
traditionally
when
we've
ended
these,
but
still
have
to
kind
of
substance
of
things
on
the
net
flix
going
to
keep
on
going.
One
of
them
was
a
discussion
that
Christoph,
Blocher
and
Erik
Vita
have
about
like
what?
What
does
lgt
I
mean?
A
I
I
Obviously,
the
largest
by
far
the
user
of
prowl
is
the
kubernetes
community,
like
we
use
it
internally
at
Google
for
some
stuff,
I
think
you
know,
Red
Hat
uses
it
for
some
stuff
as
well,
and
so
we
should
sort
of
you
know,
try
and
D
couple
the
what
do.
What
is
best
for
the
kubernetes
community,
how
they
want
to
use
prowl
and
what
is
sort
of
you
know?
How
do
we
want
to
prowl
to
be
used
for
other
people,
and
so
I
don't
know
Christophe?
J
First
hi
everybody
I,
haven't
really
attended
this
meeting
much
but
I
I'm
here,
I
am
probably
sitting
around
flap
and
top
of
that
kind
of
stuff.
I
also
want
to
just
introduce
myself,
is
hi
I'm
your
tech,
new
technical
lead
over
at
the
career
experience,
so
you
might
be
seeing
more
of
me
here
as
I
kind
of
come
as
an
ambassador,
so
on
this
particular
topic,
I
completely
agree
with
what
you're
saying
Eric
that
there
that
we
should
kind
of
just
just
distinctly
separate.
J
What
is
the
best
for
prowl
like
what
defaults
we
want
to
set
is
the
best
for
prowl
as
a
tool
in
itself
and
what
settings
we
choose
for
the
kubernetes
community
when
we
as
we
you
know,
enable
prowl
on
kubernetes
repo
I,
have
I
care,
definitely
more
about
the
how
we
enable
and
use
it
for
the
kubernetes
repos
itself.
I
have
opinions
on
the
other,
but
I
I
definitely
care
more
about
the.
How
we
use
it,
but
yeah
I,
think
that
might
be
a
good
like
the
topic
might
be
a
good
thing
for
a
breakout.
I
Ok,
maybe
I'll,
try
and
find
some
invites
schedule
that
we've
had
success
with
those
in
the
past,
so
maybe
later
this
week,
I'll
try
and
find
some.
Let
me
know
on
slack
if
you
are
interested,
you
know
anybody
interested
and
I'll
also
toss
out.
The
meeting
invites
to
the
mailing
list
generally,
but
try
and
optimize
people
who
say
they
want
to
attend
finding
time
that
works
for
all
of
those.
I
A
H
H
There's
a
couple:
it's
running
through
all
the
different
you
know:
can
you
reach
kubernetes,
not
defaulting
on
those
and
at
some
point
it
is
looking
under
service
tech,
cluster
dot
logo
which
doesn't
exist
if
you're,
if
you've
overridden
the
cluster
domain?
Okay,
we
really
I'm
just
yeah.
It
is
a
conformance
test,
so
I
mean
it's
it's
a
networking
test,
so
I
should
I
go
bug
that
working
about
this.