►
From YouTube: Docker Containers & Dark Matter: Overview Of the Spin Container Platform with Highlights from LZ
Description
Presented by Cory Snavely, Quentin Riffard, & Tyler Anderson
May 27, 2020
Spin is a container-based platform at NERSC designed for deploying science gateways, workflow managers, databases, API endpoints, and other network services to support scientific projects. Spin leverages the portability, modularity, and speed of Docker containers to allow NERSC users to quickly deploy pre-built software images or design their own. The underlying Rancher orchestration system provides a secure, managed infrastructure with access to NERSC systems, storage, and networks.
One project making use of Spin as part of its engagement with the Superfacility project is the LZ Dark Matter Experiment, which is preparing to operate a 10-ton, liquid-xenon-based detector a mile underground at the Sanford Underground Research Facility (SURF) in South Dakota. The collaboration of some 250 scientists and 37research institutions is busily readying the detector and associated software and data systems.
Services that will run in Spin to support the LZExperiment range from databases to data transfer monitoring and have been exercised during mock data challenges. In this demonstration, NERSC staff will give an overview of the Spin platform and show how a simple service is created in a few seconds. LZ staff will then describe the science of dark matter detection and give an overview of their work in Spin so far, focusing on the Event Viewer, a science gateway that allows researchers to examine significant detector events.
A
Right,
hi,
everyone
and
welcome
to
the
fourth
in
our
series
of
super
facility
demos.
Super
facility
project
is
designed
to
help
support
experimental
science
at
berkeley
lab
and
within
the
doe
complex,
and
it
is
coordinating
work
and
effort
in
crd
and
ess
and
at
nurse
in
supporting
these
workflows
and
supporting
these
very
data-centric
workloads.
A
And
so
we've
been
running
as
a
project
now
for
something
like
a
year
and
a
half
and
we've
organized
this
demo
series
to
to
share
and
to
show
off
a
little
bit
some
of
the
work
that's
gone
in
gone
into
the
super
facility
project
and
some
of
the
tools
and
some
of
the
capabilities
that
we've
been
developing
to
support
this
workload.
So
today's
topic
is
spin.
A
B
Okay,
so
you're
familiar
with
nurse
and
one
of
the
characteristics
of
of
nurse's
work
workload
and
the
number
of
different
projects
and
users
that
we
support.
Is
this
diversity
and
a
part
of
the
part
of
what
comes
with
that?
That
makes
it
fun
to
work
at
nurse?
Is
the
the
number
of
different
kinds
of
experiments
and
workflows
and
research
that
we
help
to
support
part
of
what
comes
along
with
that
diversity
of
workload?
Is
the
need
for
services
and
other
kind
of
and
other
kind
of
science
gateways
and
other
resources?
B
B
So
this
is
sort
of
the
impetus
behind
creating
spin
is
receiving
questions
from
our
users
about.
How
can
I
run
some
of
my
services
in
conjunction
with
my
computational
projects
that
access
my
data
or
can
tie
into
the
hpc
networks
and
can
scale
up
to
you
know
commissioning
a
detector
or
operating
in
real
time
at
full
volume
and
by
the
way
I
need
to
use
custom
software
in
these
services,
and
I
need
to
have
things
stay
up
outside
of
just
when
jobs
are
running.
B
That's
what
spin
was
designed
to
do
and
it's
a
platform
where
users
can
deploy
their
own
science
gateways
or
these
types
of
tools
using
docker
containers
and
docker
is
a
really
nice
fit
for
this
use
case,
because
it
lets
people
who
are
developing
software
on
these
projects,
kind
of
control
what's
inside
the
docker
container,
but
deploy
it
to
a
managed
infrastructure,
that's
hard
to
do
on
any
traditional
sort
of
platform,
even
a
cloud
kind
of
platform.
That's
based
on
virtual
machines.
B
You
can
easily
end
up
with
if
you're
a
researcher,
you
can
easily
end
up
with
far
too
much
stuff
to
manage.
You
don't
want
to
manage
virtual
machines.
Essentially,
you
want
to
try
and
manage
just
just
the
software
and
docker
helps,
helps
make
that
possible.
So
in
spin
you
can
use
public
or
custom
software
images
and
we'll
show
that
you
can
access
the
file
systems
and
network
here
at
nurse.
B
B
People
are
actively
working
with
so,
if
you're
familiar
with
docker
at
all,
you're,
probably
familiar
with
this
methodology
of
build
ship
and
run
that's
very
compatible
with
how
people
use
bands
of
building
images.
If
you
need
custom
software,
you
can
do
that
on
your
laptop.
You
can
ship
them
into
what's
a
registry
which
is
a
docker
image,
repository
for
version
control
and
safe
keeping.
B
B
So
how
does
that
actually
work?
How
do
you
actually
interact
with
spin?
So
with
our
with
the
new
version
of
rancher
that
we've
just
announced,
you
can
use
a
user
interface.
That's
what
the
the
demo
I'm
about
to
do
will
show,
or
you
can
use
yaml
declarations,
and
there
are
some
kind
of
toy
examples
for
creating
a
docker
image
and
also
for
describing
how
how
your
application
is
constructed
and
when
you
apply
you
either
build
with
a
ui
or
apply
those
yaml
rules
to
to
design
a
system
like
this.
B
C
B
B
So
you
can
compose
these
things
together
and
then
the
rancher
orchestration
takes
care
of
number
one
plumbing
a
private
overlay
network
between
all
of
them,
so
they
can
communicate
with
each
other,
but
they're
isolated
to
themselves,
and
it
also
takes
care
of
running
those
containers
on
the
nodes
within
then
you
don't
have
to
worry
about
where
your
node
is
running.
Where
your
container
is
running,
you
don't
have
to
track
that.
Rancher
takes
care
of
all
that
for
you
and
also
make
sure
the
containers
stay
running
if
they
stop
restarts
them,
reschedules
them.
B
B
B
B
B
So
you
can
see.
This
is
the
fun
part.
In
a
number
of
seconds.
I
went
from
just
giving
a
declaration
to
a
running
database
so
now
I'm
going
to
deploy
the
python
application.
On
top
of
that,
I'm
going
to
call
that
app
because
it's
an
application-
and
this
is
going
to
use
a
custom
image.
That's
part
of
our
spin
up
training
workshop,
and
this
is
the
this-
contains
the
application
that
shows
the
shows
the
cosmology
images
I'm
going
to
keep
that
app
workload
in
the
same
demo
workspace.
B
This
will
reveal
this
application
was
actually
created
by
roland
thomas
who's,
part
of
the
spin
working
group.
It's
pulling
out
of
his
directory.
That's
going
to
mount
inside
serv
static
inside
the
container
and
then
last
I'm
going
to
apply
some
more
security
settings
to
reduce
the
privilege
in
this.
In
this
container,
this
container
actually
doesn't
need
any
special
privileges
at
all.
B
And
the
host
name:
that's
going
to
be
that
we'll
use
to
access.
It
is
kind
of
a
predictable
name
based
on
the
name
of
the
ingress
and
the
name
space
and
the
development
spin
instance.
That's
going
to
support
80
on
that
host
is
going
to
connect
to
the
app
workload
port
5000,
and
then
I'm
going
to
do
one
little
trick
here
to
to
set
the
dns
time
to
live
on
the
dynamic
record.
That's
created.
B
B
D
So,
let's
start
with
some
some
background.
So
in
the
universe
we
have
a
few
evidences
of
the
existence
of
dark
matter.
One
of
these
is
the
rotation
curves.
So
what
you
can
see
here
on
this
on
this
animation
is
the
simulation
of
the
rotation
of
stars
in
a
galaxy
with
a
halo
of
dark
matters,
running
the
galaxy
and
without
the
dark
matter
in
in
in
the
galaxy.
So
what
you
can
see
here
is
without
dark
matter.
D
While
when
we
have
dark
matter,
what
you
can
see
here
is
that
the
rotation
velocity
of
the
stars
is
more
homogeneous
at
large
radius
and
guess
this
is
what
we
are
observing.
So
this
is
the
measurements
here,
the
dots
that
you
can
see
here-
and
this
is
telling
us
that
we
have
dark
matter
in
in
the
galaxy,
because
this
is
the
only
explanation
that
we
can
add
on
top
of
it,
to
explain
why
we
are
observing
the
planet
too.
Instead
of
of
decreasing.
D
D
If
we
just
take
the
sum
of
the
mata
component,
the
dark
matter
represents
83
percent
of
the
matter,
which
is
the
dark
matter,
is
dominating
the
tomato.
So
we
don't
know
anything
about
this
and
in
this
puzzle
of
the
dark
matter,
so
we
have
here.
I
really
like
this
this
this
smaller
diagram,
because
it
shows
that
this
is
just
a
part
of
the
what
we
call
the
visible
matter.
D
So
this
is
an
interesting
diagram
because
we
can
read
it
in
few
directions,
so
we
could
first
start
to
try
to
produce
dark
matter,
so
we
just
take
two
standard
model
particles
like
protons
and
we
abuse
them
and
we
produce
dark
matter.
So
this
is
what
lhc
people
are
typically
doing,
and
there
are
some
elite
searches
about
dark
matter,
as
we
have
dark
matter
everywhere
in
the
universe.
D
D
We
have
a
skin
and
also
an
outside
detector,
to
avoid
to
help
to
the
discrimination
of
the
different
backgrounds.
But
I
don't
want
you
to
go
too
much
in
the
details
so
right
now,
the
lcd
detector
is
on
the
on
the
mine
at
the
surf
labor
in
south
dakota,
and
I
I'm
sorry,
but
I
completely
missed
you
to
put
pictures
of
them
of
the
detector
right
here.
D
So
if
we
just
try
to
think
about
now,
what
is
the
expected
signal
in
lz
is
oops.
I
just
wanted
to
start
the
video.
Yes,
we
have
a
wimp
or
dark
matter
battery
that
arrives,
it
interacts
and
then
here
we
had
the
first
flash
of
flight
in
the
detector,
and
this
is
going
to
create
a
small
signal
here
and
then
it
creates
some
source
of
electrons
and
because
we
have
an
electric
field.
D
Those
electrons
are
drifting
toward
the
the
liquid
to
reach
here,
the
the
allot
grid,
and
then
let
me
close
on
at
the
very
end
of
the
video,
oh,
my
gosh,
okay,
for
some
reason
I
can
do
that.
But
when
the
electrons
arrive
on
the
anode
grid,
they
will
be
amplified
by
higher
electric
field
and
then
they
will
emit
a
lot
of
flight
and
this
light
will
be
detected
on
the
on
the
pmt.
So
we
are
going
to
have
the
first
signal.
D
That
is
our
the
signal
that
we
are
going
to
call
s1
and
then
s2
later
on,
that
is
going
to
be
this
energy
ionizations.
You
know,
let's
say-
and
this
signal
will
also
give
us
the
capability
based
on
its
position
and
the
different,
the
different
light
types
of
among
the
different
pnts.
We
are
going
to
be
able
to
reconstruct
also
the
position
in
the
x
y
plane
of
the
events
and
thanks
to
the
drift
time,
because
we
know
the
electron
drift
velocity
here,
we
can
reconstruct
the
z
position.
D
D
D
D
And
but
thanks
to
the
new
tools
that
nursekee
is
setting
up,
the
lc
collaboration
is
not,
let's
say
limited
in
cpu
usage
and
data
storage.
We
are
using
other
services
provided
by
miners
and
spin
is
a
very
good
services
that
we
are
using
a
lot.
We
are
using
it
for
the
data
transfer
because
our
software's
space
is
living
on
on
spin.
We
have
the
monitoring
the
software
monitoring
for
spade.
D
That
is
also
living
on
skin
the
job
system,
our
job
submission
engine,
that's
not
even
on
skin,
which
is
in
b
squared
and
also
it's
monitoring.
So
you
have
here
two
screenshots
of
the
of
the
monitors
we
have
the
the
game
viewer
that
tyler
is
going
to
introduce
in
in
few
minutes
our
prime
software
for
the
offline
data
quality,
monitor
and
also
some
database
mirrors
that
are
also
moving
on
on
spin.
D
E
E
Everyone,
this
is
tyler
anderson,
I'm
a
graduate
student
at
stanford,
studying
physics,
and
I
work
at
slack
as
part
of
the
lz
collaboration,
and
today
I'm
going
to
be
showing
off
the
lc
offline
event,
viewer,
which
is
one
of
the
more
visually
interesting
services
that
lz
has
up
on
spin
right
now.
We
use
it
to
look
at
raw
data
so
like
waveforms
and
pmt
heat
maps
and
it's
kind
of
the
one
service
that
I've
been
maintaining
and
working
on.
E
E
So
you
could
see
that
once
I
opened
this
file,
a
bunch
of
pretty
colors
popped
up
and
that's
this
first
event
in
the
file
being
loaded.
You
can
see
here
there
are
74
events
in
this
this
file,
so
I'll
give
you
a
tour
around
the
page
here
in
the
top
left.
You
have
kind
of
just
metadata
about
this
first
event
below
you
have
some
waveforms
and
then
to
the
right.
E
E
The
other
two
correspond
to
waveforms
from
your
outer
detector
parts,
which
are
mostly
used
as
veto,
so
I'll
mostly
focus
on
just
the
tpc
and
in
the
pmt's.
That's
this
top
circle
here
and
the
bottom
circle
here
are
the
top
and
bottom
arrays
of
the
tpc,
which
are
a
little
more
clear.
If
I
go
to
this
tpc
tab,
and
then
you
can
only
see
those,
so
you
have
the
top
pmt
array
and
then
the
bottom
pmtree
and
these
waveforms
are
summed
up
over
all
of
the
individual
pmt
waveforms,
at
least
on
this
tab.
E
E
E
E
You
create
a
bunch
of
new
photons
so
to
show
you
this
s2,
I
can
actually
use
a
new
feature.
So
I'll
say
this
is
pulse
20
and
if
I
come
up
to
selection
and
just
put
in
20
it'll
zoom
to
that
s2
until
you,
some
nice
information
about
it,
and
you
can
see
that
it's
a
very
wide
pulse
and
there's
a
lot
of
there's
a
lot
more
photons.
Of
course
they
come
from
this
folks
and
that's
very
obvious.
E
Well,
you
can
see
that
we
can
get
an
idea
of
where
the
event
happened,
using
the
idea
that
pmt's,
just
above
the
interaction,
collect
the
most
light.
So
just
looking
at
this
heat
map,
you
can
see.
Oh
this
event
probably
happened
at
this
location
in
the
detector
kind
of
near
the
edge
of
the
detector
or
the
tvc.
E
Then
you're
probably
curious
what
all
this
other
stuff
is
below.
So
if
we
zoom
in
on
that,
it's
mostly
just
single
electrons
that
are
kind
of
lagging
behind
and
I'll
use.
This
opportunity
opportunity
to
just
show
off
you
know
this
other
feature
where
you
can
just
do
a
custom
time
range
or
also
show
off
that
the
legend
here
is
clickable.
E
So
if
you
wanted
to
only
see
you
know
these
single
electrons,
which
are
labeled
in
red,
you
can
click
everything
else,
and
then
you
just
have
these
red
dots,
where
your
single
electrons
are
so
it's
kind
of
nice.
If
you're
looking
for
specific
type
of
pulse
and
finally
over
here,
you
can
also
customize
these.
This
heat
map
color
range.
E
Are
summed
up
over
all
of
the
individual
waveforms
from
the
pmt's,
but
if
you
wanted
to
look
at
one
team
to
use
waveform
instead,
you
can
do
that
in
this
tab.
So
if
I
wanted
to
look
at
just
the
waveform
coming
from
pmt
181,
which
is
this
guy
here,
I
can
enter
that
in
and
hit
plot
and
you'll
just
see
that
specific
waveform.
So
this
is
kind
of
its
contribution
to
the
s2
pulse.
E
E
So
that's
about
all
of
the
features
I
want
to
show
off.
The
last
thing
is
just
noting
that
you
can
move
to
different
events
within
the
file,
which
is
important.
So
if
I
wanted
to
go
to
the
second
event
in
the
file
and
when
it
does
it
just
you
know,
changes
reloads,
all
the
way,
forms
and
all
of
this
metadata.
E
So
that's
it.
Let
me
just
reiterate
that
the
that
part
of
this
is
made
possible
with
polka,
which
is
a
nice
interactive
tool,
and
users
can
use
this.
All
the
lz
users
can
use
this
pretty
easily
because
it's
hosted
on
spin
and
spin
as
very
close
to
or
it
lets
them
use
all
the
data
they'd
normally
be
working
with
anyway.
D
B
So
again
I
want
to
thank
quentin
and
tyler
for
going
into
depth
on
the
on
the
project
and
kind
of
and
and
showing
the
tools,
because
it's
it's
very
interesting
for
us
to
see
how
how
you
all
are
able
to
kind
of
apply
these
tools
for
in
such
creative
ways,
and
that's
so
I'm
seeing
that
detail
for
the
first
time
myself
too.
So
it's
just
it's
fascinating!
B
B
That's
just
free
form
that
we
staff
to
kind
of
get
get
your
project
launched
if
you're
an
existing
user,
we're
opening
access,
transfer,
v2
through
self-guided
exercise,
and
then
we
have
office
hours
as
well
to
help
out
with
your
implementations
there.
So
and
if
you're
curious,
you
can
look
at,
you
can
look
more
at
our
webpage.
B
D
So
at
the
moment
we
don't
have
automatic
job
submission
to
to
be
squared.
We
are
just
doing
it
by
hand
because
we
don't
have
the
data
stream
coming
from
from
yet,
but
what
we
are
going
to
do
is
when
the
data
file
will
be
written
here.
At
most,
we
will
have
a
hook
that
will
tell
to
b
square
to
put
this
file
on
the
queue
for
processing.