►
From YouTube: NUG Monthly Webinar, May 2023
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
All
right,
well,
hi,
everyone
I
hope
everyone
is
doing
well,
this
last
Thursday
of
May
going
into
the
holiday
weekend.
Thank
you
all
for
gathering
with
us
today
for
our
monthly
meeting
again.
My
name
is
Charles
Lively
myself
and
Lippy
we're
the
science
engagement
engineers
and
happy
to
be
working
with
all
of
you
to
help
further
engagement
and
Community
amongst
our
users,
and
so
we'll
get
started
with
our
plan
for
today.
A
Again,
of
course,
we
want
everyone
to
interact
as
much
as
possible.
Give
us
your
thought,
thoughts
or
any
questions
throughout
the
meeting
today,
as
well
as
make
sure
you
participate
on
the
slack
with
any
questions,
tips
or
even
just
look.
What
I
learned
for,
as
you
continue
throughout
the
research
Journey
for.
E
A
Right
all
right
so
as
I
was
saying
yeah,
please
speak
up
like
if
I'm
not
showing
slides
when
I
should
be
again
today
we'll
go
ahead
and
go
through
with
our
usual
win
of
the
month,
any
win
of
the
months
that
you
may
have
what
you
might
have
learned
updates
on
Quarry
retirement.
It
is
happening,
announcements
and
call
for
participation
and
then
just
a
little
discussion
on
using
slurm
and
sharing
tips
and
tricks
and
any
any
tips
that
you
might
have
as
well.
F
C
Okay,
so
for
today's
sort
of
win
of
the
month,
we
were
thinking
we'd
like
to
actually
have
everybody
share
and
chat.
First,
sometimes
it's
a
little
scary
to
like
unmute.
So
if
you
have
a
recent
win,
which
just
means
I
mean
it
just
is
a
win
for
you,
so
it
doesn't
have
to
be
like
you
know
it.
You
know.
C
Maybe
it's
a
paper
accepted,
or
maybe
you
just
figured
something
out
that
you
hadn't
figured
out
yet
we'd
love
to
hear
about
it,
and
so,
if
people
have
anything
to
share,
please
feel
free
to
drop
it
in
the
chat.
Give
people
a
couple
minutes
just
for
some
context.
We
like
to
share
this
win
of
the
month
because
it
helps
us
know
like
kind
of
what
are
people
doing
on
the
system.
What
are
people
accomplishing
and
then
we
also
like
to
highlight
things.
C
We
have
a
couple
different
ways
that
that
we
highlight
science,
that
people
are
doing
at
nurse
but
also
we'd
like
to
move
into
highlighting
technical
accomplishments
that
are
being
done
so
I
have
my
chat
open
if
anyone
has
had
a
chance
to
think
of
anything.
You're
also
welcome
to
unmute
and
say
something,
but
please
feel
free
to
share
in
the
chat.
C
C
Well,
if
anyone
thinks
of
any
wins
feel
free
to
share,
we
do
have
a
couple
different
ways
that
you
can
share
information
with
us
in
terms
of,
if
you'd
like
us
to
maybe
highlight
an
achievement,
and
then
you
know,
if
there's
also
these
awards
that
are
given
I
think
every
year,
and
so
you
can
find
out
some
more
information
about
those
as
well.
C
If
you
decided
that
you've
done
something
really
outstanding
and
you'd
like
us
to
know
about
it,
cool,
okay,
well,
let's
move
on
then
maybe
then
maybe
there
will
be
more
for
the
next
topic
by
the
way
I
wanted
to
share
this
I
think
this
is
something
everybody
should
know
about,
especially
I
think
relevant
to
Scientific
Computing,
because
most
of
us,
some
of
us
are
not
from
Computing
backgrounds.
So
this
is
XKCD
a
fantastic
Cartoon.
C
If
you
don't
know
about
it,
and
basically
I'm,
not
gonna
read
all
of
this,
but
the
the
sense
here
is
that
you
know
people
sometimes
make
fun
of
people
or
judge
people
for
not
knowing
something.
We
definitely
get
this
a
lot.
I
think
for
scientific
competing
I.
Remember
when
I
was
a
grad
student.
C
Just
starting
off
at
nurse
I
was
really
worried
and
scared
to
ask
questions,
because
I
was
like
someone's
gonna
judge
me
for
for
the
fact
that
I
don't
know
this
already,
but
I
I
kind
of
like
this
reframing
of,
like
you
know,
on
any
given
day,
there's
probably
about
10
000
people
who
don't
know
something
that
are
going
to
learn
it
that
day,
that's
kind
of
the
thesis
of
this
cartoon
and
instead
of
judging
them
or
you
know,
making
fun
of
them.
C
You
could
actually
be
the
person
to
tell
them
about
it
and
you
know
enjoy
their
reaction
to
learning
that,
especially
if
it's
something
really
fun
and
interesting.
So
that's
sort
of
the
reframing
I
think
for
the
today.
I
learned,
you
know
it's
it's
not
about
you
know
it
doesn't
have
to
be
something
super
high
level.
It
could
be
something
that
you.
B
C
Maybe
you're
worried
is
really
basic
or
that
someone
might
judge
you
for,
but
you
know
that's
not
the
case.
We'd
really
like
to
see
our
users
realize
that
scientific
Computing
is
really
hard
and
you
come
from
a
different
background.
You're,
not
necessarily
a
computer
scientist.
So
there
might
be
something
that
you
learned
about
using
terminal
about
python,
about
C
plus
plus
whatever
it
is
that
you're
doing
that.
You
didn't
know
before
that.
Someone
else
probably
doesn't
know
either.
C
F
C
Person
I've
run
some
some
simulations
that
were
written
in
like
portion,
fantasy
plus,
but
I've
never
had
to
actually
compile
them
or
do
anything,
and
so
she
was
telling
me
about
the
correct
words
to
use
to
describe
like
you,
don't
execute
a
make
file
you
run
make
and
then
the
make
file
like
does
whatever
it
needs
to
do,
and
so
this
is
just
like
some
kind
of
vernacular
understanding
of
the
fact
that
you
know
what
is
a
make
file
like
what
is
make.
C
What
are
these
you
know
kind
of
terminology,
so
that's
something
that
I
literally
learned
yesterday.
C
So
this
is
another
chance.
If
anyone
has
anything
they've
learned
recently,
please
feel
free
to
share
it
in
the
chat
or
just
on
mute
and
and
share
it.
D
Well,
I'll
tell
you
one
that
I
learned
so
Charles
and
I.
We
were
working
on
the
main
topic
of
today's
presentation,
so
I
won't
give
away
all
of
that,
but
we
just
learned
so
much
from
just
talking
to
a
system
administrator
and
just
kind
of
watching
him
do
his
thing
for
an
hour,
like
my
mind,
was
blown
within
the
first
10
minutes.
There
were
commands,
I
didn't
know,
existed
in
Linux,
but
that
exists
and
that
they
use
to
really
to
really
maximize
what
they
get
out
of.
A
That
and
also
even
like
when
we
met
with
Chris
and
he
was
showing
us
the
Nvidia,
wrapper
libraries
for
the
power
power,
aware
metrics
that
are
included
in
the
library.
That
was
something
that
I
learned,
that
was
kind
of
I,
guess
new
for
Pearl
Motor
on
you
know
going
forward
where
we
could
use
the
counters
to
get
the
expected
power
consumption
at
a
per
node
level
and
then
for
CPUs
as
well.
A
Well,
I
knew
we
could
do
it,
but
the
way
he
showed
us
was
a
new
way
for
me.
So.
G
This
is
coaching.
This
is
not
me
running
anything,
but
you
know:
I
subscribed
this
Fortune
user
group
and
ask.
H
G
Then
in
the
stock
Channel
I
think
about
the
week
ago,
I
don't
know
him,
but
Robert
shared
his
interesting
experience
using
Fortran.
He
was
I,
think
running
Ultron
called
with
the
double
precision
and
looking
at
some
you
know
six
or
seven
decimal
place.
You
know
value
as
testing.
He
did
very
very
minor
edit,
like
we
will
print
command
for
debugging,
for
example,
or
remove
one,
if
condition
to
print
something
for
debugging.
G
Climate
science-
sometimes
we
produce
an
example
by
explore
the
sensitivity
of
the
atmosphere
of
climate,
to
the
initial
condition.
Do
we
add
machine,
friction
level,
cultivation
in
initial
condition,
but
at
least
temperature
distribution
after
running
simulation
for
6
months
and
10
years
or
100
years,
they
just
produce
very
different
climate.
G
If
you
just
look
at
five
years
that
this
is
somewhere
so
I
do
learn
a
little
bit
last
year
when
I'm
writing
a
paper
about
coming
model
and
then
go
through
some
high
performance,
Computing
textbook
first
time
to
understand
the
paper
written
by
nice
stuff,
but
and
then
both
the
rules
of
operations,
then
just
a
little
subtle
change.
How
was
sort
of
very
intriguing
for
me
and
by
the
way
I
I
it's
four
times.
Also
it's
really
nice
just
joining
then
I
can
learn
bunch
of
stuff
like
never
be
able
to
run
myself.
C
Oh
that's
great.
Thank
you.
Thank
you
for
sharing
and
a
good
plug
for
the
Fortran
users
group,
the
fun
group.
If
you're
interested,
please
feel
free
to
join
great
thanks,
koichi,
okay,
any
to
anyone
else.
Last
call
feel
free
again,
if
you
think
of
something,
please
feel
free
to
share
it
in
the
chat.
It
could
really
help
somebody
else
too.
C
C
Okay,
so
yeah.
This
is
important.
We
are
going
to
be
turning
Corey
off
and
retiring
Corey
on
May
31st
at
noon,
so
you'll
still
have
access
to
login
on
the
login
nodes
and
access
to
Corey's
scratch
until
Wednesday
June
7th,
but
at
noon
on
Wednesday
all
the
cues
will
be
turned
off.
Any
jobs
that
are
left
will
be
deleted
and
no
jobs
will
be
running
on
Corey
anymore,
and
so,
if
you
have
any
data
files,
anything
that
you
need
to
move,
you
really
need
to
do
it.
C
Try
to
get
everything
off
of
there,
especially
in
case
you
have
any
issues.
If
something
doesn't
transfer
correctly,
then
you'll
have
some
time
to
submit
a
ticket
or
get
help.
If
you
need
any
help.
So
there's
a
lot
of
resources
on
how
to
move
things
off
of
Quarry
and
to
Pearl
Mudder
there's
you
can
use
Globus.
C
You
can
use
the
community
file
system
as
a
sort
of
in
between
query
scratch
and
promoter
scratch,
and
if
you
have
data
that
you
want
to
use
the
hpss
tape
archive
for
that's
also
a
possibility
for
for
data-
that's
not
used,
maybe
as
often
and
there's
information
in
the
documentation
about
all
of
that.
Please
make
sure
you
put
any
Irreplaceable
data
in
CFS
or
on
hpss,
because
scratch
storage
is
not
permanent.
C
C
There's
going
to
be
two
coming
up,
one
on
Friday
and
one
on
Tuesday:
that's
10
a.m,
to
noon,
both
on
Friday
and
Tuesday,
Pacific
and
so
I
think
the
documentation
is
being
kind
of
changed
so
that
there's
a
lot
more
information
about
either
getting
like
everything
onto
promoter
or
running
on
Paul
Mudder,
because
Corey
is
being
retired
absolutely
on
May,
31st,
okay,
any
questions
so
far.
Please
feel
free
to
unmute
and
ask
questions
or
raise
your
hand
if
you
have
any
questions
or
put
them
in
chat.
C
Okay,
so
I
just
want
to.
We
just
want
to
share
a
couple
imminently
approaching
seminars,
events
and
trainings
coming
up
on
May
30th
is
a
advanced
SQL
training,
and
so
this
is
there
if
you,
if
you're,
not
familiar
with
Siegel,
there's
I
believe
some
resources
on
maybe
how
to
prepare
I
think
it's
maybe
about
four
hours
of
content
that
you
can
look
at
beforehand
in
case
you
want
to
catch
up
and
participate
in
that
training.
So
that's
going
to
be
on
May
30th.
C
There
is
an
introduction
to
nurse
resources,
training
on
June,
8th
again
the
information
for
that
will
be
on
the
nurse
website
in
the
training
category.
C
C
Great
okay
awesome,
so
this
might
be
good
for
students.
If
you
have
students
that
are
joining
for
the
summer,
please
ask
them
to
register
for
this,
because
it's
meant
to
be
for
like
summer
interns
I
think
that's
sort
of
the
target
audience
is
that
correct.
Okay,.
C
Okay,
great
okay,
fantastic
good,
and
to
that
also,
you
can
find
information
in
the
weekly
emails
all
this
information,
as
always
in
the
weekly
emails,
but
it's
also
on
the
nurse
website.
Okay,
next.
C
Okay,
so
there
is
a
call
for
submissions
for
the
usrsc,
so
U.S
research,
software
engineer,
Association
conference.
This
conference
will
be
in
October,
I
believe
in
Chicago,
and
so
the
the
poster
abstracts
are
due
June
19th-
and
this
is
a
very
good
reminder
for
me,
because
I
also
am
submitting
a
couple
things
for
that,
so
I
need
to
get
get
those
in
as
well.
So
don't
forget:
June,
19th,
there's
also
an
RSE
science,
Workshop
I,
think
that
must
be
part
of
this.
C
The
usrsc
conference
itself
within
that
so
there's
also
an
abstract
submission
for
that
June,
30th
and
then
as
part
of
SC.
There
is
a
workshop
on:
let's
see
what
it
says:
sixth
annual
parallel
applications,
Workshop
alternatives
to
MPI
plus
x,
which
I
guess
they're,
calling
paw
ATM
I
love
that
the
acronym
and
the
submissions
will
close
July
24th.
So
you
have
a
little
bit
of
time.
C
C
Okay,
I!
Think!
That's
it!
For
this,
let's
see:
okay,
okay,
so
this
is
a
webinar
through
ECP
the
ideas
ECP
webinar,
which
is
on
open,
ssf,
best
practices,
fetch
program,
I,
actually
don't
know
what
open
ssf
is.
If
anyone
knows
what
that
is,
please
feel
free
to
unmute
and
share
or
put
it
in
the
chat
but
and
looks
like
the
link.
The
link
is
probably
in
the
in
the
weekly
email.
C
So
if
you're,
if
you're
interested
in
finding
out
more
check
out
the
weekly
email
and
they'd,
probably
have
a
description
about
what
what
is
open
ssf,
if
you
don't
know
already
again
the
fun
group,
The
410
users
group
is
still
running
their
llvm
flying
survey.
So
you
can
get
this
link
these.
These
two
should
still
probably
be
in
the
weekly
email
or
I.
I.
Actually
can't
remember
if
they're
still
in
there,
yeah.
C
They
were
okay
good,
so
if
you
want
to
participate
in
that
survey
or
if
you
want
to
participate
in
the
nurse
message
of
the
day
survey,
so
the
message
of
the
day
is
kind
of
like
our
status
page,
a
nurse
on
nurse.gov
and
we're
thinking
about
maybe
making
it
you
know
changing
in
a
little
bit,
maybe
improving
it
a
little
bit.
So
if
you
have
any
thoughts
about
that,
please
feel
free
to
check
out
that
survey.
A
Did
anyone
have
any
events
that
maybe
we
might
have
missed
that
you
wanted
to
bring
to
our
attention
and
or
maybe
you
could
post
a
message
in
our
chat
about
that,
as
well
as
on
the
slack.
C
To
say
weekly
email
at
nurse.gov,
maybe
it's
a
Lincoln.
C
That
would
be
great
feel
free
to
use
that
if
you
have
an
event
that
you
want
to,
let
the
nurse
user
Community
know
about,
we
will
add
it
to
the
weekly
email.
Okay,
thanks
all
right
go
for
it.
Charles.
A
Okay,
all
right,
so
we
will
move
on
to
our
topic
of
the
day
and
basically,
it's
going
to
be
just
kind
of
just
a
refresher
and
some
pretty
cool,
quarries
and
options
that
you
can
do
on
slurm.
So
let
this
be
in
engaging
and
interactive
feel
free
to
chime
in
share
your
thoughts
share,
how
you
might
use
any
of
the
the
flags
or
commands
as
well,
and
so
you
know
the
basics
slurm
by
what
it
stands
for.
Is
it's
a
simple
Linux
utility
for
resource
management?
A
That
name
comes
out
pretty
pretty
good
for
what
it
applies
for
and
we
you
know,
we
use
slurm
for
managing
our
resources,
and
you
know
hybrid
large-scale
environments,
because
it
helps
us
to.
You
know,
make
sure
that
we
are
getting
the
most
out
of
the
functionality
of
our
jobs
that
we're
submitting,
that
we
are
able
to
schedule
and
allocate
efficiently
and
that
we
have
a
management
software
that
is
reliable
and
scalable
and
provides
good
insight
into
our
jobs
and
the
projects
that
we're
working
on.
A
So
really
it's
it's
very
important
that
we
are
able
to
utilize
the
software
in
an
efficient
manner
and
learn
different
ways
to
use
it
to
fit
our
needs
and
what
this
topic
of
the
day
it'll
evolve
and
continue
on.
A
And
so,
of
course,
we
have
our.
You
know
our
commonly
used
for
submitting
our
batch
shops
and
running
those
shops,
and
this
is
just
kind
of
like
a
refresher
overview
highlight
when
we're
using
s
batch
and
sron,
and
we
have
our
different
ways
in
which
we
can
allocate
and
launch
tasks
using
our
different
in
task
Flags
or
tasks
per
node,
CPUs,
of
course,
per
node,
as
well
as
different
flags
for
using
for
quality
of
service
or
if
we
need
to
control
whether
we're
going
to
kill
a
job
appropriately
or
not.
A
So
these
are
just
a
couple
of
the
the
high
level
flags
that,
of
course,
we
all
use
for
anyone
in
particular
do
are.
There
are
any
of
these
commands
or
any
of
these
flags
do.
You
feel,
are
most
useful
for
the
type
of
work
that
you
do
for
allocating
and
assigning
tasks
at
a
per
node
or
per
CPU
level.
A
A
I
D
So
so
Roman
one
of
those
is
the
first
one.
Is
the
total
number
of
tasks
total.
D
My
job
okay,
then
they
would
essentially
be
the
same
for
your
purses.
But
if
you
needed
a
thousand
nodes
right,
then
you
would
you
would
do
minus
n
or
minus
minus
n
tasks
equals
the
number
of
tests
that
is
in
total
across
all
1000
nodes,
whereas
the
minus
minus
tasks
or
the
minus
minus
and
test
per
node
is
the
total
number
of
tests
on
one
individual
node
then
times
a
thousand?
If
that
makes
sense
so.
I
Okay.
The
second
question,
for
example,
I,
need
to
run
some
tasks
somewhere
on
the
cluster
and
I
need
number
of
CPUs
for
it
for
Simplicity.
Let
the
see
the
deck
I
need,
like
20
CPUs,
for
my
task,
it's
multi-thread
or
whatever
I
do
in
it's
my
business.
What
parameter
for
us
by
should
I,
provide
to
submit
this
job
well,
I
want
to
run
one
particular
task
from
my
opinion.
It's
one
task
from
slum
point
of
view.
I
D
Okay,
so
that
it's
going
to
be
minus
m
one
or
you
know
the
first
one
on
there
you're
going
to
set
that
to
be
one
and
then,
if
it's
going
to
do.
If,
if
it's
going
to
use
multiple
CPUs
in
that
one
task,
then
this
minus
lowercase
C
on
the
third
line.
That's
going
to
be
what
you're
going
to
use.
D
D
We
also
have
the
JavaScript
generator,
and
that
is
something
I
would
recommend
using
because
it'll
give
you
all
the
right
values.
You
know
you
just
input
what
you
want
and
then
it'll
give
you
all
the
right
values.
This
minus
C1
can
be
especially
confusing
for
people,
and
so
that's
why
we
have
that
tool
available
because
it
always
gets
it
right
and
your
human
brain
may
not
always
get
it
right.
A
And
Annette
just
put
it
into
the
link
to
the
generator
into
the
chat.
Do
you
is
there
any
do
you
know
if
there's
any
variance
in
the
way
that
you
execute
your
applications?
Where
is
it
if
it's
MPI,
have
you
done
any
experiments
to
see
how
limiting
the
number
of
CPUs
per
node
would
affect
performance,
or
is
that
something
you've
looked
into.
A
Whether
you're,
considering
the
Bashan
versus
a
number
of
CPUs
per
task,.
I
Are
you
asking
me?
Yes?
Yes,
sorry:
okay,
yeah,
I'm,
sorry,
it's
okay,
yeah!
What
I
have
discovered
I'm,
not
sure
that
specific
or
per
motor,
because
on
Corey
it
was
slightly
different
policy.
As
far
as
I
understand
I
had
if
I
don't
need
the
whole
mode
for
my
task,
I
need
to
specify
quality
of
service,
shared
I,
believe
otherwise.
I
I'll
get
the
whole
note
and
will
be
judged
for
a
whole
128
CPUs,
no
matter
if
I
requested
only
like
clink
for
my
task
yeah,
if
you
use
the
whole
node
and
yeah,
but
my
impression
was
like
on
Quarry
I
kind
of
can
just
specify
that
I
need
specific
number,
of
course,
for
my
task
and
don't
care
about
anything
else.
I
will
be
charged
only
for
CPU
hours,
I
actually
used
on
per
motor.
I
It
seems
like
it's
slightly
different,
so
I
explicitly
have
to
say
that
there
will
be
one
task
which
will
require,
like
whatever
20,
CPUs
and
quality
of
service
should
be
shared,
because
if
I
ask
for
normal
for
regular
I,
don't
remember
exactly
as
available
the
whole
mode
will
be
allocated
for
me,
and
I
will
burn
a
lot
of
CPU
hours
for
nothing.
Yeah.
D
That's
that's
true,
Roman.
Basically,
the
reason
we
the
way
that
we
charge
is
essentially
like
what
resources
is
a
user
does
user
have
access
to
that?
Other
users
can't
have
access
to
right.
So,
even
if
you
were
just
using
a
fraction
of
a
node,
but
you
had,
you
had
exclusive
use
of
that
node
we
have
to
charge
for
the
whole
node
yeah.
I
I
understand
your
point,
but
my
point
of
view.
You
know
I
ask
it
specifically
for
number
of
course,
I
didn't
ask
for
a
whole
note:
I
didn't
provide
minus
10
Capital
to
get
mold
in
exclusive
mode.
I
asked
for
one
task
and
like
whatever
20
10
CPUs,
why
you
are
giving
me
the
whole
Northern
exclusive
mode.
I
didn't
ask
for
it,
so
this
is
kind
of
not
obvious
yep.
D
I
yeah
I
understand
your
point
of
view,
but
so
anyway,
so.
I
A
I
I
can
clarify
a
little
bit.
Our
software
we
are
using
is
actually
another
kind
of
work
load
manager,
we're
just
asking
for
resources
and
split
it
to
the
computational
slots
by
ourselves.
So
we
submitted
some
job,
which
then
will
execute
our
comments
independently
from
from
slide.
So
we
are
asking,
for
example,
give
us
the
whole
node
for
12
hours
and
then
kind
of
use
all
128
cores
doing
our
stuff
internally.
So
we
are
not.
B
G
Oh
yeah
I
have
a
separate
question.
I
just
just
came
to
my
mind
recently.
When
do
we
use
S
run
and
went
to
not
so
when
I
started,
the
job
is
a
running
simulations
or
processing
data.
G
When
I
run
any
code
I
compiled,
which
is
either
MPI
or
Oppo.
Mp
I
use
a
strong
command
within
a
tax
script,
because
mainly
because
that's
what
I
get
the
minor
job
description
generator,
but
when
I'm
process
data
with
some
other
applications
like
I,
have
some
pies,
including
python,
and
a
couple
of
other
data
processing
applications
I
tend
to
listen
to
use,
Google
parallel
and
develop
like
command
into
the
parallel
and
when
I
went
to
a
good
part
of
documentation.
Page
on
mask
some
example:
even
not
those
each
processors
in
pi.
G
Some
cases
is
a
surround
to
launch
those
raft,
shell
scripts
so
and
I
when
I
come
to
think
of
it,
even
though
those
tools
I
use
data
process
they're
between
like
C
and
compiled
by
somebody
and
then
sits
there.
So
it's
also
compiled
code,
even
though
I
don't
think
that
I
use
those
compiled
binary
with
open
Pure
API.
G
So
my
understanding
is,
if
I
don't
use
an
API
or
3MP
it's
just
serial
code.
I
just
run
the
binary
executable
by
typing.
It's
firing
that
if
using
open
vmpi
was
using
SRAM
command
and
then
name
of
the
binary
other
options.
But
is
that
a
case,
or
is
that
other
occasions
this
reading
these
little
power
of
documentation
in
detail,
kind
of
making
a
little
bit
wonder
about
what
is
really
is
wrong
is
doing.
D
That's
a
great
question,
so
let
me
try
to
answer
it
so
generally:
yes,
if
you
are
using
MPI
or
openmp,
you
should
definitely
use
S
run
if
you're
not
doing
that,
then
it
becomes
a
little
bit
optional.
D
D
You
know
one
instance
of
it
on
each
node,
let's
say,
but
maybe
with
different
inputs,
you're
still
you're
going
to
want
to
use
S
run
in
order
to
do
that,
because
otherwise,
because
what
what
happens
is
when
you,
when
you
get
a
job
and
it
starts
you're
kind
of
it's,
it
works
the
same
way
with
interactive
and
with
with
a
batch
job,
it's
easier
to
think
interactive
right,
so
you're
dumped
onto
one
of
the
nodes
of
your
interactive
job,
okay
and.
H
D
G
B
G
If
I
don't
use
extra
I
only
using
the
particular
node
I
learned
it
that's
right:
okay,
okay,
thank
you!
So
sometimes
I
use
multiple
nodes
for
my
data
process
through
parallel,
in
some
case,
probably
better
to
use
S1
I
tried
because
using
S
run
and
not
these
are
just
parallel.
I
didn't
see
much
difference
in
in
the
performance
like
time
it
takes
so
I.
Did
it
last
much
so
yeah
paying
attention
too
much,
but.
B
D
A
Okay,
and
so
we'll
continue
on
with
a
few
more
examples
of
how
to
make
use
of
a
couple
of
other
commands,
and
so
of
course,
we
have
SQ,
which
we
can
use
for
getting
appropriate
information
about
jobs
that
are
in
our
scheduled
queue.
And
then
we
also
have
sqs,
which
is
kind
of
wrapped
into
sq1
on
promoter
to
get
the
different
information.
A
Sq
is
going
to
focus
more
on
monitoring
for
job
status,
real-time
updates
and
for
filtering
and
sorting,
and
we
can
also
use
flags
and
sqs
for
getting
Q
status,
fair
share
usage
and
other
resource
utilization
using
slurm
for
any
of
the
jobs
that
we've
submitted.
A
So
we
can
use
those
and
also
use
S
accounts
for
getting
our
resources
that
have
been
executed,
running
or
terminated
based
off
of
our
individual
tasks
that
we
have
submitted.
And
so
we
have
a
number
of
different
flags
that
you've,
of
course,
are
familiar
with
or
button
them
might
not
use.
As
much.
Of
course,
we
have
by
filtering
by
user,
with
the
Hue
or
user
job
ID
filtering
by
a
Time
range,
where
you
can
indicate
a
start
and
then
time
using
dash
s
or
start
time
or
E
and
end
time.
A
And
then
you
have
different
formatting
and
output
options
that
you
can
do
as
well,
and
so
you
can
make
use
of
these
flags
and
some
of
the
output
options
to
display
specific
fields
that
you
might
need
need.
And
so
you
can.
You
can
limit
the
fields
that
you
want
to
display
based
off
of
the
job.
Id
user
start
in
time,
the
state,
the
exit
code,
as
well
as
any
resource
usage
usage,
metrics
that
you
can
utilize.
A
And
so
let's
a
few
examples
in
which
you
can
you
know
kind
of
make
use
with
the
the
idea
of
formatting
and
limiting
start
and
endpoints.
So
if
you
wanted
to
specify
a
Time
range
with
your
with
your
command,
you
can
do
a
start
and
specify
whatever
the
date
is
and
the
format
of
year
month
day,
and
then
you
can
do
a
end
date
as
well.
A
With
that
same
format
and
that'll
get
you
specify
the
the
status
for
jobs
within
that
range,
and
then
you
can
also,
of
course,
format
it
like
we
previously
mentioned
and
using
the
formatting
flag,
we
can
specify
those
specific
fields
that
we
want
to
gain
access
to.
So
if
we
just
stash
Dash
format,
we
can
list
whatever
fields
we
want
from
accounts.
Job
IDs
submit
start
and
elapsed
Etc,
and
we
can
further
modify
those
as
well.
A
We
could
use
our
Dash
X
flag
for
providing
a
summary
of
each
job
instead
of
multiple
lines
that
continue
on
and
other
flags
that
we
can
use.
We
can
use
our
state
flag,
which
can
be
used
to
filter
only
on
jobs
that
have
successfully
completed
versus
those
that
had
other
problems
are
ended
and
all
of
different
flags
and
options.
We
can
use
and
format
to
custom,
customize
or
output
for
desired
results
that
we
could
get.
A
So
some
other
examples
using
formats
we
can
specify
specific
options
for
width
with
the
US
field.
If
we
wanted
it
to
start
with,
with
the
30
or
versus
an
aligned
version
with
a
30,
we
can
use
the
dash
P
flag
for
providing
output.
That's
Dylan
dilemma
delimited
using
the.
A
What
is
that
line
symbol
called?
Is
it
just
a
bar
line?
Pipe
pipe!
Is
that
okay
for
the
pipe
we
can
specify
based
off
of
duplicates,
and
we
can
also
filter
based
off
of
different
node
options
that
we
have
available,
and
so,
if
we
we
have
a
few
examples
in
which
example
command
in
which
we
could
do
it
for
S
account.
A
So
we
can
use
different
slurm
flag
options
to
get
more
insights
and
statistics
and
for
identifying
the
patterns
that
that
might
persist
in
the
jobs
that
we've
executed,
either
from
our
accounts
or
even
on
different
nodes.
A
Now
one
question
I
do
have
perhaps
maybe,
if
Rebecca
do
you
know
how
long
the
history
of
different
jobs
would
remain
on
the
system.
D
It's
an
excellent
question:
I
think
it
remained
on
Corey
I.
Think,
like
everything
from
the
beginning
is,
is
actually
available,
but
we
can
only
query
one
month
of
data
at
a
time
just
to
avoid
putting
too
much
of
a
strain
on
the
slurm
database.
A
Oh
wow,
okay,
cool
I
I
thought
it
might
be
like
a
year
or
two
or
something
like
that.
But
wow.
That's
that's
pretty
cool
that
it's
all
seven
years.
That's
a
lot
of
data
in
does.
Does
anyone
have
any
tips
or
questions
on
how
that
how
they've
been
able
to
use
some
of
these
flags
or,
if,
if
you
have
or
have
not
used
them
before
in
the
past,.
A
A
So
we'll
just
continue
on,
but
feel
free
to
chime
in
and
give
your
thought
your
thoughts
kalichi
is
trying
it
now
good
good,
so
other
useful
commands
that
a
lot
of
a
lot
of
engineers
in
our
group
and
Consultants
have
used.
A
Also
is
the
S
control
that
a
lot
of
admins
will
use
in
slurm,
and
we
can
use
that
to
get
detail.
We
can
do
things
like
show
information
related
to
a
specific
reservation
and
how
it's
been
used.
A
We
can
use
it
to
show
information
for
specific
types
that
have
been
used,
such
as
whether
it's
a
debug
reservation,
and
then
we
can
also
include
appropriate,
Flags
or
whatnot
to
get
information
pertaining
to,
starting
in
time,
duration,
the
nodes
used
and
a
lot
of
other
information,
and
so
we
can
use
S
control
show
for
showing
different
aspects
about
some
of
the
submissions
and
the
jobs
that
have
been
executed
on
the
system.
G
G
D
D
So
that's
one
that
has
one
thing.
Another
thing
is
sometimes
people
want
to
have
a
reservation
for
for
various
different
purposes.
D
You
know
they
may
they
may
need
to
run
some
really
big
jobs
like
for
you
know
for
Gordon,
Bell
prize
sort
of
things
or
other
reasons,
and
sometimes
people
need
a
reservation
because
they
have
a
job
that
they
want
to
debug
and
they
want
to
just
be
able
to
debug
it
in
real
time.
Oh.
D
A
Yeah
but
when
Micah
shared
a
good
tip
in
using
S
control
for
even
adjusting
different
attributes
after
the
job
has
already
started.
D
B
D
And
that's
really
so.
There
are
other
s.
Control
update,
commands
that
don't
work
as
well,
but
the
one
that
does
work
is
is.
F
I
always
had
the
problem
later,
looking
up,
what
exactly
the
job
script
for
a
specific
job,
ID
looked
like
so
I
found
that
and
so
I
created
the
script
which
basically
submits
the
ticket
and
then
immediately
queries
the
job
ID,
because
as
control
also
allows
you
to
find
out
what
the
script
for
a
specific
job
was,
and
this
is
combined
to
then
write
it
in
some
directory
of
your
choosing.
And
then
you
have
basically
all
your
scripts
saved
with
the
specific
ID
they
were
submitted.
A
Good
good,
thank
you
for
sharing
good
deal,
okay
and
continuing
on
other
useful
commands
that
we
have.
Of
course,
we
have
S
info,
which
we
can
use
for
getting
a
better
understanding.
A
So
our
Magics
are
available
in
Jupiter,
awesome,
understanding
about
specific
job
types
and
within
the
queue
and
resource
availability.
So
we
can
use
the
S
info
and
the
dash
o
flag
for
outputting
the
format
specifically
for
different
partitions,
based
on
the
availability,
the
nodes
and
node
status
that
are
going
to
be
used,
and
so
we
can
use
just
the
formatting
options
to
help
the
the
information
be
displayed
in
a
logical
manner
where
we
can
break
down
different
aspects
about
the
different.
A
The
availability
of
the
different
types
of
nodes
that
are
up
there
and
again.
This
can
be
used
to
help
identify
which
nodes
are
available
in
use
idle
and
basically
can
serve
as
kind
of
a
tool
for
helping
you
to
see
what
resources
are
available
for
you
to
use
right
now
for
submitting
a
job
as
well.
A
So
other
okay,
another
and
we
have
I,
don't
want
to
hold
anyone
I'll
quickly
go
through
a
few
examples
on
aliases,
but
these
will
also
be
available
and
you
can
use
the
Alias
to
for
shorthand
to
get
specific
information
and
format
it
in
different
ways,
and
so
Alias
aliases
can
be.
A
You
know,
time
saving
they
can
help
you
customize
things
and
as
well
as
you
know,
provide
consistency,
and
so
we
have
a
couple
of
different
examples
and
how
you
can
use
aliases
and
slurm
you'll
just
use
the
keyword,
Alias
and
then
you'll
name
it,
and
then
you
will
specify
the
type
of
command
you
want
that
Alias
to
be
based
off
of
so
here.
A
We
have
two
examples
where
you
can
create
an
alias
for
for
getting
specific
information
and
a
dedicated
format,
and
so
this
is
going
to
be
a
s
info
nodes
Alias
that
is
based
off
of
the
S
info
command
and
some
of
the
different
options
that
you
can
use
for.
Formatting
and
basically
the
output
for
this
as
simple
nodes
Alias
will
show
basically
the
partitions
allocated
nodes,
node,
State
features
and
node
names
that
have
been
utilized
and
other
types
of
aliases.
There
are
I
believe
four
or
five
different
examples.
A
I
will
end
with
this
one,
which
is
just
one
that
uses
a
it's
a
check.
Job
Alias
for
using
the
S
account
flag,
and
you
can
use
that
for
getting
information
on
from
the
F
account
command
using
the
dasho
flag
and
the
information
that
will
be
gathered
here
is
based
off
a
job,
ID
user
job
name,
qos
Etc,
and
so
we
can
do
that
and
we
can
use
the
dash
J
flag
to
specify
the
job
ID
that
we
want
to
check
as
well.
A
G
Oh
yeah,
my
child
was
I,
tried
to
kind
of
type
as
I
see
the
screens,
but
it
would
be
nice
if,
like
this,
is
how
many
documentation
for
the
data
I
use
and
copy
paste.
So
I
asked
in
the
chat
by
you.
Are
you
going
to
put
these
informations
sometime
in
the
documentation.
A
Yeah
and
what
we're
gonna
do
well,
of
course,
this
PowerPoint
will
be
available,
but
we're
working
on
updating
and
revamping
documentation
and
what
will
definitely
work
on
is
a
a
useful
section
on
tips
and
tricks
and
helpful
slurm
commands
and
tricks
that
you
can
use
as
well.
A
Okay,
good
well
I,
do
not
we're
a
little
bit
over.
A
So
thank
you
all
for
attending
this
month's
user
meeting
upcoming
topics
that
we
have
in
the
pipeline,
Jupiter
Hobbit
nurse,
as
well
as
other
tips
and
tricks,
as
well
as
security,
and
how
it's
handled
at
nurse
and
accessing
our
systems,
tips
and
tricks
is
something
that
we're
looking
to
evolve
and
get
feedback
from
everyone
moving
forward,
so
we'll
be
sending
out
a
few
surveys
for
ideas
on
possible
tips
and
tricks
that
you
think
would
be
beneficial
for
all
users,
so
feel
free
to
provide
your
thoughts
on
that,
so
that
we
can
use
that
for
future
sessions
as
well.
A
If
you
have
any
other
topics,
you
know,
please
do
feel
free
to
put
that
in
and
nominate
it
as
well
as
any
science
highlight
submissions
as
well,
and
with
that
that
is
the
end
of
our
user
meeting
for
this
month.
Did
anyone?
Thank
you
all
for
attending.
Did
anyone
have
any
last
minute
questions
or
thoughts
or
feedback.
D
D
Probably
is
still
the
limit.
That's
probably
a
limitation
on
on
just
the
number
I
mean
that
that
sounds
like
a
power
of
two
sort
of
thing.
D
I
A
Well,
if
not,
thank
you
all
for
taking
time
out
of
your
day
to
attend
this
month's
meeting.
I
hope
that
everyone
has
a
good
end
of
the
week
as
well
as
a
good
holiday
weekend
and
be
safe
and
remember
to
come
to
any
of
the
Quarry
to
promoter
office
hours.
If
you
have
any
other
questions
as
we
work
on
retiring
Quarry
as
well,
but
other
than
that
hope,
you
guys
have
a
great
one,
and
we
will
see
you
next
month,
if
not
before
that.