►
From YouTube: NUG Monthly Meeting June 16, 2022
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
Record
that,
actually
that's
a
that's
a
good
reminder.
So
zoom
has
this
very
handy,
live
transcript,
which
can
be
really
useful
when,
when
people
speaking
such
as
me,
don't
have
a
local
accent
which
makes
us
a
little
harder
to
understand.
I
suspect-
and
yes,
it's
just
kind
of
a
good
thing
to
have
so
yeah,
so
so
we'll
run
through
win
of
the
month.
A
Today
I
learned
some
announcements
and
then
for
our
topic
of
the
day,
annette
is
here
and
will
tell
us
a
bit
about
sharing
and
managing
and
publishing
data
at
nurse,
then
we'll
kind
of
have
a
quick
look
at
some
things
that
are
coming
up
and
also
a
quick
look
at
some
numbers
from
last
month.
A
A
A
It
might
be
a
scientific
achievement
and
some
of
those
can
be
candidates
for
a
science
highlight
so
nurse
sends
publishers
and
sends
to
department
of
energy
every
every
month
or
recorder.
A
collection
of
highlights
of
scientific
work
has
been
done
using
nurse
resources
and
also
we
have
an
annual
high
impact,
scientific
achievement,
award
and
innovative
use
of
high
performance
computing
award,
and
in
fact
I
think
each
of
those
has
two
categories
for
early
early
career
researchers,
as
well
as
for
more
established
researchers.
B
I
want
to
give
a
shout
out
to
rebecca
hartman
baker
for
the
crash
course
in
super
computing.
That
was
a
superb
training.
A
That's
good
to
hear
so
that
was,
I
think,
tuesday,
that
that
was
just
just
this
week.
So
yeah,
it's
good
to
good,
to
hear
that
you
found
it
valuable.
I.
B
Learned
about
by
omp
that
day,
which
is
awesome,
I
think,
bringing
openmp
to
python.
A
That
is
interesting.
Actually,
I
haven't
haven't
stumbled
across
that
one
myself,
I'm
curious:
how
does
it?
How
does
it
do
it
in
terms
of
the
actual
openmp,
because
python
is
famously
challenging
around
multi-threaded
applications
because
of
its
global
global
interpreter,
lock.
A
So
yeah,
I
think,
rebecca's
actually
not
on
at
the
moment,
but
yeah
pass
that
on.
A
Let's
move
across
to
the
kind
of
the
flip
side
of
that
coin,
which
which
is
a
today
I
learned
this
is
an
opportunity
to
talk
about
things
that
what
necessarily
wins
something
that
surprised
you
that
might
benefit
others
to
know
about,
and
then
this
can
be
something
that
didn't
work
like
discovering.
What
doesn't
work
is
quite
important
and
valuable.
A
Dead
ends
difficulties,
and
you
know
if
it's
a
if
it's
a
still
ongoing
difficulty.
It
might
be
that
somebody
here
has
a
has
an
idea
of
what
you
can
try
next.
It
can
also
be
a
tip
that
you've
learned
recently.
So
in
a
way,
what
will
was
speaking
about
just
now
with
being
able
to
use
pi
pi
omp
to
get
openmp
from
python
programs
is
a
pretty
good
tip.
A
Already
got
something
to
call
out
on
that
front.
A
See
I
actually
stumbled
across
something
recently
there's
more
kind
of
hpc
adjacent.
I
was
using
it
for
a
hpc
related
project,
but
it's
not
directly
hpc,
which
is
you
may
have
heard
of
mermaid
diagrams.
A
So
mermaid
is
a
kind
of
a
markdown
like
you
know,
markup
type
language
that's
fairly
easy
to
write
and
I
think
he
uses
something
like
graphics
in
the
background.
But
so
you
can
write
kind
of
you
know,
text
descriptions
of
a
diagram
like
that.
You
have
a
flowchart
or
a
class
diagram
or
an
architecture
diagram
that
your
network
diagram
and
we'll
generate
a
graph
of
it.
A
The
particular
example
I
was
actually
trying
to
generate
in
this
case
was:
it
was
a
gantt
chart
which
is
one
of
its
things
that
it
does.
But
what
I
learned
is
that
you
can
use
these
diagrams
and
it's
got
a
clause
called
click,
and
so
you
can
make
elements
clickable
and
if
you
open
that
in
a
markdown
capable
notebook-
or
at
least
certain
markdown,
capable
notebooks
or
cdf,
was
the
one
I
was
playing
with,
you
can
actually
make
a
a
kind
of
a
you
know,
a
graphic
or
a
flowchart,
or
an
architecture.
A
Diagram
click
on
the
elements
of
it
and
jump
to
essentially
a
wiki
page
that
you
set
up
about
that
which
I
thought
was
a
really
neat
way
of.
You
know
navigating
the
documentation
and
architecture
of
a
project.
A
Any
other
tips
or
lessons
that
people
might
want
to
pass
on.
D
A
little
bit
quiet
this
week.
A
There
there
were
a
few
maintenances
actually
in
the
last
in
the
last
few
weeks,
if
you
haven't,
haven't
looked
too
closely
at
this
month's
ones
yet,
but
for
for
last
month's
numbers,
I'm
going
to
look
at
previous
months
on
so
yeah,
we
did
have
a
few
things
sort
of
coming
up
and
down.
A
I
think
what
we're
yeah,
something
we're,
we're,
clearly
learning
more
and
more
about
how
to
run
helmet,
how
to
how
to
set
up
and
run
palmata
effectively,
and
these
maintenances
are
sort
of
you
know,
building
out
more
of
the
you
know
the
the
the
software
stack
in
the
systems
better.
A
We
have
kind
of
a
bunch
of
announcements
and
and
various
calls
for
participation
in
things.
Some
were
in
the
most
recent
weekly
emails.
We've
got
a
few
that
we
can
call
out
here
and
yeah
there'll
be
a
chance
in
a
moment
to
add
to
them
first,
and-
and
I
think
this
also
starts
to
fall
a
little
bit
into
the
the
the
width
of
the
muffler.
I
thought
this
was
a
great
progress
and
I
see
creatures
here
that
sort
of
celebrate.
It
was
the
one
who
really
kicked
this
off.
A
So
no
now
has
a
special
interest
group,
particularly
for
wharf
users.
So
wrf
is
weather,
research
and
forecasting
application.
It's
a
you
know
a
fairly
important
application.
We
have
a
quite
a
number
of
users
that
nurse
use
it.
A
It's
also,
you
know
famously
tricky
to
you,
know,
set
up
and
get
running
right,
and
this
is
setting
up
a
seat,
for
it
is
part
of
an
effort
to
enable
nurse
users
to
be
able
to
to
collaborate
more
in
ways
that
are
you
kind
of
generally
helping
other
nurse
users
in
a
way
and
yeah
expand
the
the
the
ability
to
make
good
use
of
nurse
resources
beyond
just
setting
tickets.
A
You
find
your
domain
expertise,
that's
relevant
to
particular
things,
and
so
the
sig
is,
if
it
in
that
direction
and
it
had
a
kickoff,
and
so
it's
coming
along
quite
nicely
and
there's
a
little
bit
more
detail
in
the
weekly
email
announcing
that,
but
on
the
nug
slack.
If
you
look
in
the
browse
channels,
there's
a
wrf
user
that
has
some
more
information.
A
Another
announcement,
just
in
the
last
few
weeks,
that's
also
in
the
nurse
in
the
weekly
email,
is
that
we
have
federated
id,
has
rolled
out
a
little
bit
further.
So
it's
not
all
institutions
yet,
but
it's
to
most
doe
labs.
A
So,
particularly
if
you're
at
a
doe
lab
you'll
see
when
you
go
into
the
single
sign-on
for
nurse
a
screen
that
looks
a
little
bit
like
like
this
one
over
here,
where
you
can
choose
your
institution
and
log
in
with
your
login
from
your
home
institution
kind
of
as
an
alternative
to
logging
in
at
nurse.
Basically,
it
works
for
most.
A
You
know
web
services
yeah,
so
it's
not
for
the
ssh
login,
but
for
things
like
jupyter
and
servicenow,
and
and
so
on,
like
that,
so
it's
quite
handy,
especially
if
you're
in
multiple
institutions,
one
less
password
to
remember
you-
can
use
your
institutional
log
on
it's.
I
think
a
little
more
secure
because
of
that
as
well.
A
We
have
you
may
have
seen
a
couple
of
announcements
in
emails
about
the
nurse
machine
learning
users
survey.
Please
fill
this
out
if
you
haven't,
and
but
you
will
particularly
the
work
that
you
do
uses
machine
learning.
This
is
to
help
us
to
optimize
perlmutter
and
future
systems
for
ml
capability
and
performance.
A
A
number
of
calls
for
participation
out
at
the
moment
the
smoky
mountains
data
challenge,
which
is
hosted
by
oakridge.
If
I
remember
correctly,
is
now
accepting
or
has
a
cfp
out
the
george
michael
memorial
hpc
fellowship
nominations
are
open
and
actually
falling
due
fairly
soon
at
the
end
of
this
month
and
that
one's
accepting
self
nominations.
So
this
is
for
phd
students
doing
hpc
related
research.
So
if
you
are
a
phd
student
or
have
a
phd
student
who
this
year
might
be
applicable
to
go,
take
a
look.
A
You
know
it
comes
with
a
certain
amount
of
money,
as
well
as
recognition
in
acm
and
expenses
for
sc
22,
as
well
as
there's
an
award
involved,
picking
up
and
publicly
recognized
on
the
stage
there.
The
third
international
symposium
on
checkpointing
for
supercomputing
is
also
looking
for
papers.
Participation
at
the
moment
that,
if
I
remember
earlier
that
one
is
held
in
conjunction
with
supercomputing,
so
in
november
a
few
training
events
coming
up.
There's
a
up
workshop
starting
next
week.
A
There's
a
work
tutorial
with
an
nvidia
on
profiling,
deep
learning
applications
using
nvidia
insight.
I
think
the
ones
with
video
at
the
end
of
this
month
there's
also
a
tutorial
on
coordinating
dynamic,
ensembles
of
computations
with
lib
ensemble,
since
it's
kind
of
a
you
know,
ensemble
not
quite
the
same
as
a
workflow.
If
I
understand
it
correctly
manager,
but
you
know
a
way
of
managing
ensembles
of
simulations.
A
So
that's
the
ones
that
we
know
about.
Are
there
any
others
that
people
would
like
to
announce.
A
E
I
don't
know:
what's
going
on
with
yours,
pretty
strange
all
right,
so
I'm
just
gonna
talk
about
data
sharing
at
nursk
in
all
its
various
forms
and
give
people
some
pointers
about
how
to
do
it.
E
Some
of
you
will
know
this
pretty
well
already,
but
I
think
it
bears
pointing
out
a
few
tips
that
relate
to
them
I'll
talk
about
internal
data
sharing
within
nurse
either
within
a
project
or
outside
a
project,
and
then
I'll
talk
a
bit
about
external
data
sharing
and
sharing
through
science
gateways
or
just
transferring
data
so
starting
out
with
notes
on
permissions,
so
people
are
probably
pretty
well
acquainted
with
the
user
group
and
other
permissions
that
you
see
in
a
file
directory
when
you
do
ls
minus
l.
E
E
Yes,
and
in
fact
your
your
project
director,
you
should
have
that
automatically
enabled
to
start
with,
but
you
may
need
to
kind
of
add
that
to
directories
below
that,
and
so
others
on
the
system.
E
Of
course,
you
know
one
way
to
to
share
if
somebody
isn't
already
in
a
group
that
you
have
in
common,
you
could
make
something
just
available
to
others,
but
of
course
you
need
to
keep
in
mind
that
when
you're
doing
that,
you're
you're
sharing
it
with
everyone
on
the
system,
not
just
the
people
that
you're
thinking
of
as
your
collaborators.
E
So,
of
course
you
know,
the
unix
permissions
are
kind
of
a
combination
of
who
the
users
and
groups
are
that
are
actually
the
owners
of
the
file
and
then
the
modes
that
are
set
for
each
of
them
and
those
are
the
little
rwx
and
friends
that
we
see
in
lsl,
minus
l
and
just
to
know.
You
know
when
you're
setting
these
things
on
a
directory.
The
behavior
is
just
a
little
bit
different
from
when
you're
putting
it
on
a
regular
file,
whereas
the
the
plus
x
isn't
just
making
it
executable.
E
It
means
it's
it's
making.
The
directory
executable,
which
just
means
that
you
can
traverse
it
and
because
that
is
actually
the
function
of
a
directory
and
then
plus
r
means
that
you
can
run
ls
on
it.
So
you
can
actually
see
the
content
while
you're
working
through
it.
E
So
at
nurse
we
like
to
to
try
to
remind
people
to
remember
the
principle
of
least
privilege
when
you're
sharing
things,
you
want
to
make
sure
that
you're
you're
sharing
with
people
who
really
need
to
be
able
to
see
it,
and
so
it's
always
good
to
just
to
make
sure
that
anything
that
needs
to
be
kept
private
is
get
private.
And
so,
if
you
do
set
up
specific
directories
that
you
have
more
open
permissions
than
the
default,
it's
a
good
idea
to
set
up
an
area
where
you
have
your
own
private
stuff.
E
That
has
the
permissions
where
the
group
and
other
do
not
have
read,
write
and
execute
and
do
not
have
any
of
those
so
that
you
can.
You
just
know
that
that's
locked
down
and
then
you
set
up
separate
directories
that
have
the
sharing
permissions
that
you
want
and
remember.
When
you're
doing
this,
you
know
it's
kind
of
easy
to
to
just
open
up
a
directory
and
open
up.
A
So-
and
that
seems
like
a
fairly
neat
tip,
does
that
mean
that
if
I
want
to
share
a
specific
file
or
a
specific
subdirectory,
but
not
but
that's
sort
of
deeper
in
my
home
directory,
but
not
the
things
behind
it?
I
can
use
just
plus
x,
give
them
the
full
path
to
the
file
and
they
can
get
to
that
but
yeah,
but
it
doesn't
give
them
access
to
all
the
directories
underneath
it.
E
Yeah
exactly
yeah,
and
then
with
that
comes
up
a
lot
in
web
sharing,
which
I'll
talk
about
later,
where
people
will
need
to
share
a
specific
directory
so
that
the
web
server
can
get
to
it,
and
the
web
server
will
need
to
be
able
to
traverse
those.
But
it
doesn't
need
to
be
able
to
look
at
what's
inside.
It.
E
Okay
and
the
other
thing
is
we
just
mentioned
from
that
previous
question-
the
set
gid
on
a
directory,
so
this
preserves
the
group
ownership
so
that
when,
when
somebody
creates
a
new
file
or
directory
inside
a
directory
that
has
this
bit
set
that
ends
up
having
the
correct
group,
it
shares
the
same
group
name
so
that
helps
keep
the
shared
directories
accessible.
E
Of
course,
one
thing
that
happens
over
time
is
a
lot
of
times.
People
will
forget
to
set
that
look,
maybe
create
a
new
directory,
and
maybe
it's
a
couple
layers
below
where
you
had
that
s
bit
set.
E
So
you
might
need
to
do
something
to
just
kind
of
bring
things
along
and
catch
everything
up
to
having
the
permissions
that
you
want.
The
group
to
have
and
nurse
has
this
handy
little
script
called
make
dur
group
readable,
and
you
can
give
it
a
group
name,
and
you
can
give
it
a
list
of
directories
and
it
will
go
through
and
it
will
set
the
right
group
name.
It
will
set
the
right
permissions
on
there
and
make
it
all
group
readable,
and
it
will
set
the
s
bit
as
well.
E
E
And
then
just
to
mention
also
another
piece
of
the
equation
is
the
umask,
which
is
basically
the
settings
for
when
you
create
a
new
filer
directory
this,
in
conjunction
with
the
system
defaults
which
are
kind
of
like
you
can
think
of
it
as
the
mask
is
sort
of
like
subtraction
from
what
the
the
defaults
are.
E
The
defaults
at
nurse
are
going
to
be
basically
for
full
permissions
for
a
directory
and
just
read
and
write
permissions
for
a
file,
so
that
can
be
changed,
but
it's
usually
you're
not
going
to
want
to
try
to
worry
too
much
about
that,
because
it's
sort
of
complicated
with
computing,
it's
compute
nodes,
but
we
do
have
some
extra
documentation
about
how
to
handle
this
as
well.
E
In
fact,
we
have
a
lot
of
documentation
about.
You
know:
expel
permissions
on
the
docs.nurse.gov
website,
so
it's
a
good
place
to
go
for
more
info
about
that
all
right.
Let's
look
at
sharing
within
nurse
within
a
specific
project,
so
obviously
the
permissions
are
something
that
I
just
mentioned.
E
E
Some
groups
find
that
they
need
to
set
up
an
additional
project
directory
for
sharing,
and
this
is
something
that
can
be
done
on
request,
and
you
know
basically,
these
end
up
being
a
project
directory
on
the
community
file
system
and
once
those
are
in
place,
a
pi
can
make
changes
on
their
own
to
the
setup
there.
So
the
pi
toolbox
will
allow
a
pi
to
go
in
and
reset
the
ownership
and
the
permissions
on
specific
files
can
be
any
file
in
their
project.
E
It
doesn't
have
to
be
one
that
they
own,
so
this
is
a
little
extraordinary
because
usually
in
a
unix
system,
you
can
only
change
what's
what's
owned
by
you,
but
we
have
this
special
facility
for
pis
to
go
in
and
make
those
kind
of
changes.
A
E
Alright,
so
for
sharing
outside
of
a
project
again,
you
can
use
permissions
to
do
that,
but
be
you
know,
be
careful
about
how
much
you're
opening
up,
because
remember
not
everybody
on
the
system
is
necessarily
the
ones
you're
thinking
of
when
you
set
these
permissions,
but
there's
also
a
very
nice
facility
called
give
and
take-
and
these
are
just
two
commands
that
allow
you
to
give
a
file
to
somebody
else
or
to
take
it
from
somebody
else.
You
can
only
take
what's
been
given
to
you.
E
So
if
somebody
gives
a
file
to
you,
they
just
specify
the
recipient
and
the
file,
and
then
you
go
in
and
to
take
it.
You
have
to
know
who's,
giving
it
to
you
and
you
can
use
the
minus
a
which
will
just
grab
everything
that
that
person
is
trying
to
give
you.
You
don't
have
to
remember
the
file
names
or
you
can
use
the
more
lengthy
version
of
it,
and
you
can
tell
it
a
specific
location
in
which
you
want
your
copy
of
that
to
be
landed.
E
So
that
is
sharing
within
nurse
now,
let's
look
at
outside
of
nurse,
so
starting
out
looking
at
file
transfers
pretty
much
anytime.
Anyone
is
trying
to
share
data
by
transferring
it
to
somebody
else.
We're
gonna
recommend
that
they
try
globus
first,
because
globus
is
a
really
really
well
put
together
system,
it's
very
performant.
E
E
Another
thing
that's
great
is
the
the
data
transfer
nodes.
We
use
those
particularly
because
they
are
set
up
to
have
the
right
configuration
for
moving
data,
and
so
you
know
there
they
mount
all
of
our
file
systems,
they're
optimized,
for
allowing
data
to
move
efficiently.
So
please
take
advantage
of
those.
E
Let's
see,
tools
for
file
transfer,
you're,
probably
pretty
familiar
with
scp
the
secure
copy
command
just
using
ssh
to
move
a
file.
You
may
be
less
familiar
with
bbcp,
which
is
just
the
I
think
that
stands
for
babar
copy,
that's
built
by
the
high
energy
physics
committee.
I
mean
community
and
it
works
like
sap,
but
it
it's
more
robust
and
built
to
work
with
large
files.
E
And
finally,
we
don't
really
generally
recommend
it,
but
we
do
have
an
ftp
upload
service
that
occasionally
ends
up
being
kind
of
the
last
the
last
way
to
to
move
files
like.
If,
if
you
have
a
collaborator,
who
is
not
a
nurse
user
and
doesn't
have
a
way
to
use
globus,
then
that
may
be
the
way
to
go.
We
are
working
on
a
replacement
for
that
and
we
also
just
like
people
to
try
to
find
alternatives
to
using
that.
E
Let's
see
so.
I
have
in
here
also
a
link
to
information
about
globus
and
I'll
talk
about
globus
a
little
more
here,
one
aspect
of
it
that
they
call
guest
collections
there.
It
used
to
be
called
globus
sharing
for
those
who
are
familiar
with
that,
but
what
it
does
is
it
allows
people
to
share,
share
data
with
unauthenticated
users,
so
you
can
basically
make
it
public
using
globus,
and
that
means
that
somebody
can
can
use
globus
and
doesn't
have
to
have
a
nurse
login
to
get
at
your
data.
E
E
So
the
next
opportunity
is
science
gateways,
so
science
gateways
can
be
much
more
than
simply
sharing
data
files,
like
you
know,
they're,
basically,
scientific
resources
for
collaboration,
and
that
can
be
all
kinds
of
computer
cooperative
work.
It
could
be
sharing
data,
but
could
also
be
sharing
tools
or
sharing
services.
E
Whatever
you
want
to
build
to
share
with
collaborators
there's
a
group
called
sciencegateways.org
that
is
basically
geared
toward
helping
people,
build
science
gateways
and
understand
how
they
can
be
useful,
and
if
you
check
them
out,
though
they
have
a
lot
of
useful
information,
although
they're
kind
of
geared
toward
the
nsf
end
of
things
but
yeah,
they
have
a
lot
of
info
about
how
to
go
about
building
them
so
nurse
how
to
go
about
building
them.
E
E
So,
as
you
know,
we
support
it,
but
we're
you're
not
going
to
get
the
kind
of
level
of
support
that
you
might
get
from.
You
know
say
like
amazon
or
or
maybe
a
google
site
could
even
be
a
really
good
alternative
as
well.
If
you
want
a
higher
level
of
support,
the
other
alternative
is
build.
A
service
in
our
cloud-based
system
called
spin.
E
So
there
you
can
use
any
server
side
framework
that
you
want.
You
can
give
it
in
a
custom
domain
name.
If
you
happen
to
own
one,
you
can
connect
directly
to
nurse
resources
like
databases
in
the
community
file
system
and
again
you
can
still
use
javascript
apis
and,
like
the
super
facility
api.
So
all
these
things
are
still
available,
but
this
system
gives
you
a
much
more
robust
infrastructure.
With
the
past
couple
of
years,
they've
had
99
uptime,
so
that's
been
pretty
darn
reliable,
but
you
are
sort
of
self-administering
the
service.
E
You
have
to
build
the
images
that
that
will
be
used
through
that
system
and
ship
it
as
containers.
E
All
right
so
sharing
a
www
directory,
so
you
basically
just
set
it
up
inside
a
project
directory
in
the
community
file
system.
As
long
as
you
name
it
www
and
you
give
it
the
path
between
the
the
top
level
project
directory
and
your
www
directory
make
sure
those
those
directories
are
all
executable.
Then
the
web
server
at
portal.nurse.gov
or
portalofs.nurse.gov
will
be
able
to
reach
that
directory
and
it
will
serve
those
files
as
any
regular
website.
E
So
I
mean
this
means
that
whatever
runs
is
kind
of
ephemeral
in
the
sense
that
you
don't
have
state
in
between
deployments
of
it.
So
that
has
a
lot
of
pluses
in
in
terms
of
scaling
and
being
able
to
just
kind
of
restart
a
system
anytime.
You
want
it's
a
very
flexible
system,
but
you
just
have
to
be
able
to
think
of
things
as
an
ephemeral
service,
but
you
can
can
build
any
x86
image,
build
it
on
a
laptop
with
docker
and
you
can
ship
it
over.
E
You
can
use
the
nurse
registry
or
just
docker
hub
itself,
and
then
you
can
run
these
things
through
the
rancher
graphical
user
interface,
which
makes
it
really
pretty
straightforward
to
hook
it
up
with
the
networking
that
you
want
and
have
it
make
that
service
available,
so
spin
is
meant
for
apps.
It's
not
for
computing,
but
it
gives
you
access
to
databases.
E
E
There's
lots
of
documentation
like
in
the
nurse.gov
website
gives
you
a
lot
of
info
about
spin
and,
if
you're
interested
in
giving
spin
a
try,
we
have
a
series
of
workshops.
We've
been
doing
these
for
a
while
and
they
are
continuing
through
june
august.
We
have
some
in
october
and
november
and
there'll
be
more
scheduled
in
the
future
as
well.
So
keep
an
eye
on
that.
If
you're
at
all
interested.
E
So
that
is
it:
if
anybody
has
any
questions,
I'd
be
happy
to
answer
them.
Yeah.
C
I
have
a
question
one
of
the
projects
I'm
going
to
be
working
on,
I'm
I'm
generating
a
lot
of
simulation
data
using
gosx
and
I'm
going
to
be
keeping
it
at
certain
time
points
but
there'll
be
internal
time
points
that
are
also
generated,
and
I
wanted
to
know
if
there's
a
way
to
like
kind
of
archive
those
because
they'll
be
pretty
large
and
you
know
I
can
come
back
and
get
more
data.
If
I
need
it,
I
don't
want
to
fill
up
cfs
with
you
know,
a
petabyte
of
data.
Is
there
someplace?
E
C
D
D
C
To
resample
it
basically,
so
I
I
it's
going
to
take
a
lot
of
compute
hours
to
generate
this
data.
I
don't
want
to
throw
out
all
the
points
in
between
just
because
I
don't
think
I
need
them
now,
but
you
know
you
know,
but
I
want
to
be
able
to
get
back
to
them.
If
I
need
them,
but
I
don't
want
to,
you
know,
use
high
performance,
you
know
file
systems
that
are
designed
for,
like
you
know,
gigabyte
access
on
the
fly.
E
Right
yeah,
that's
the
archive!
That's
okay!
Definitely
a
place
to
consider.
I
don't
know
you
know
how
much
data
you
have.
If
that's
going
over
your
limit
in
hpss.
What
I
would
do
is
maybe
file
a
ticket
on
servicenow
and
kind
of
explain
your
predicament.
You
might
be
able
to
get
more
space
if
you
need
it.
Okay,
but
yeah.
C
My
my
ercap
allocation
is
still
in
review
right
now,
but
I'm
just
trying
to
figure
out
like
how
do
I
work
that
trade-off
between
recomputing
and
just
storing
the
data.
E
Yeah
yeah
yeah,
so
hpss
is
really
optimized
for
storage.
You
need
to
format
your
data
in
some
specific
ways
to
put
it
in
there
correctly,
so
you're
not
loading
it
with
a
done,
but
with
tons
of
small
files,
because
that
would
be
anything.
Oh.
A
E
F
Okay,
thanks
yeah,
that's
too
bad,
but
it'd
be
really
helpful
for
us
to
have
yeah.
I
didn't
know
about
this
global
guest,
global's
guest,
but
that's
actually
really
useful
for
me
and
my
colleagues
because,
okay,
I
have
another
related
question
that,
because
when
we
publish
papers,
we
sometimes
have
to
put
the
data
somewhere
publicly
accessible
for
in
case,
for
example,
readers
want
to
reproduce
or
actually
these
days,
there's
a
really
journals
that
focused
on
data
like,
for
example.
F
This
is
the
scientific
data
channel,
that's
under
nature,
yeah
and
then,
but
to
publish
paper
or
publish
data.
We
have
to
have
stored
data
in
data
center
repositories
that
meet
some
criteria,
that
all
general
screen
terms
have
some
same
criteria
like
long-term
persistence
and
in
public
access,
and
so
my
related
question
is
that
is
it?
F
Is
it
possible
for
nas
to
expand
a
little
bit
on
its
like
a
data
related
service,
so
that
really
to
consider
those
criteria
set
up
by
several
high-profile
data
journals
and
then
even
better?
If
we
have
some
framework
to
have
some
doi
on
the
data
on
us,
because
I
am
just
about
to
publish
paper,
something
half
of
the
scope
is
to
publish
the
data,
but
then
I
have
to
move
data
to
somewhere.
I
can
get
doi,
for
example,
and
then,
but
because
all
the
data
is
produced
at
nask.
F
Last
quality
has
really
long
stable.
You
know
archive
of
hpss,
so
I
think
it's
really
nice
for
everyone
to
have
asked
to
have
a
bit
more
capability
or
play
a
role
as
data
repository.
Then
you
know
we
can
it's
not
discuss
even
more
high
profile
paypal
data
publication
from
that.
So
what?
What
do
you
think
about
those
possibilities.
E
Yeah,
this
is
something
we're
actually
looking
into
so
currently,
and
I
would
strongly
suggest
you
do
this.
Is
you
can
get
a
doi
from
anywhere
and
point
it
at
directories
that
are
a
nurse
and
the
beautiful
thing
about
a
doi
is
if
you
do
end
up
having
to
move
it
elsewhere,
then
you
can
do
that,
because
you
can
just
change
where
the
doi
is
pointing.
So.
D
E
F
E
E
Will
retain
your
data
for
a
year?
You
know
we
promise
you
at
least
that
much,
but
it
would.
I
think,
a
lot
of
people
are
feeling
that
it
would
be
nice
to
have
more
than
that
for
something
that's
publicly
shared,
and
especially,
if
you
have
it,
you
know
published
through
a
doi
it'd,
be
nice
to
keep
that
working
without
you
having
to
to
change
it.
So
we're
looking
into
that.
C
F
Yeah,
oh
yeah,
so
if
actually
I
searched
for
how
to
get
the
doi
myself,
but
it's
so
cool
to
me.
It's
very
complicated.
So
if
it's
not
that
complicated,
straightforward
it'd
be
very
helpful.
If
we
have
some
ask
documentation,
have
some
page
to
explain
some
general
procedure
or
point
to
some
resources,
because
again
many
of
my
colleagues
are
generating
huge
amounts
of
data
at
mask.
So
it
would
be
nice
if
you
can
get
doi
the
data
at
mask
or,
like
you
say,
or
anywhere.
E
E
That's
pretty
easy,
pretty
fast,
but
generally
I
mean
you
want
the
doi
to
be
minted
through
a
place
that
you
have
a
long-lasting
relationship
with,
so
so
you
can
get
to
it
to
make
corrections
right,
but
yeah
I
mean
we
definitely
could
be
providing
some
extra
information
about
that,
so
that
people
can
understand
what
their
options
are.
A
So
there's
another
question
in
the
chat,
and
I
also
see
that
stephen
has
his
hand
up
in
the
chat
is,
is
filezilla
supported
for
connecting
with
the
dvds.
E
So
filezilla
can
use
our
ftp
upload
service
it'll
only
allow
for
upload
and
not
download,
and
it's
kind
of
funny,
because
once
you
put
something
up
there,
you
sort
of,
though,
because
of
the
security
features
on
it,
you
don't
see
what's
in
there.
So
it's
it's
a.
D
G
Hi
thanks
for
this
info,
I
have
a
tip
a
request
and
a
question,
so
the
tip
is
first,
I
found
globus
to
be
the
most
effective
way
of
moving
files,
even
internally
to
nursk,
so
like
moving
between
scratch
and
cfs.
Globus
is
way
faster
than
copy
or
rsync
or
something
so
that's
great.
G
So
the
request
is
I'm
under
the
impression
that
nursk
has
bought
into
globus
at
a
level
that
gets
them
like
priority
requests
on
features,
and
so,
if
nurse
could
help
get
globus
to
support
syn
links,
that
would
be
great.
We
would
use
globus
a
lot
more
if
it
actually
supported
siblings,
fully
in
the
way
that
and
also
file
metadata.
So
if
globus
was
more
of
a
drop-in
replacement
for
rsync,
we
would
use
it
more
so
request
a
nurse
on
that
and
then
the
question
is
within
our
collaboration.
G
But
to
do
that,
we
have
to
grant
write
permission
which
to
the
group,
which
also
means
that
users
can
accidentally
delete
each
other's
files,
and
we
haven't
found
a
good
solution
for
how
to
manage
that
where
we
like,
create
the
ability
for
someone
to
add
something,
but
not
step
on
someone
else
and
accidentally
remove
someone
else.
Do
you
have
a
best
practice
suggestion
for
that.
C
Do
there's
a
in
unix
there's
a
thing
called
a
sticky
bit,
which
is
the
t
on
on
chmod.
So
if
you
give
the
full
permission
with
the
t
that
it
gives
the
ability
of
people
to
create
things,
but
they
can't
remove
them
unless
they
owned
it.
C
C
A
That's
a
handy
tip
and
I
think
I've
used
that
one
before
oh
today,
you
learned
another.
C
A
Today
I
learned
so
so.
Stephen's
comment
was
interesting.
That
was
one
of
the
questions
I
was
going
to
ask
because
I
remember
a
little
while
ago
it
used
to
be
the
case
that
actually
moving
data
within
nurse
via
globus
was
faster
than
most
other
methods,
and
was
that
still
true-
and
it
sounds
like
the
answer-
is
yes.
G
E
A
Okay
in
another
couple
of
comments,
I
noticed
that
and
it
gave
a
pointer
to
the
unix
file
permissions
page
in
nurse
docs,
and
I
think
historically,
that
has
has
been
our
most
visited
page.
I
think
google
or
something
sends
users
there
when
they
ask
for
file
permissions.
So
it's
actually
a
really
popular
page.
A
The
other
thing
that
was
somewhere
between
a
question
and
a
comment,
whereas
we
we
also
have
for
most
of
our
file
systems
acls,
and
so
you
can
do
quite
fine-grained
permissions
management
with
set
accola
and
set
feckle
and
get
faculty.
I
I
don't
know
if,
and
it
wants
to
say
anything
about
that
in
terms
of
sort
of
recommendations.
It
is
a.
It
is
a
little
more
complicated
to
use.
E
Yeah,
that's
definitely
an
option,
I
think
for
most
groups,
that's
not
really
necessary,
but
yeah.
Sometimes
it
gives
you
a
little
more
granularity
and
sometimes
that's
actually
what
you
need
so
yeah.
A
A
And
thanks
people
for
insightful
questions
as
well.
A
And
finish
up
with
a
couple
of
last
items,
which
is
a
quick
look
at
last
month's
numbers,
there's
a
very
limited
set
of
numbers.
It's
really
just
the
outages,
it's
interested
in
in
other
numbers
still
coming
around
to
to
how
to
actually
do
that.
So
we
did
have
a
few
outages
in
may.
Corey
had
two
two
incidents
on
top
of
it
scheduled
maintenance
and
pelmeter
had
a
few
scheduled
maintenances
and
a
few
unscheduled
incidences,
which,
unfortunately,
were
mostly
short.
A
And
finally,
a
quick
look
at
what's
coming
up,
so
we
have
penciled
in
for
next
month
a
talk
from
bjorn
from
nursk,
about
interactive
hpc,
workflows
for
experimental
user
facilities
and
also
kind
of
in
our
in
our
coming
up
list,
but
not
actually
scheduled.
Yet
is
a
bit
about
hpss
interfaces.
A
Also,
we
are
very
interested
in
hearing
more
about
inter
talks.
You
know
from.
D
A
Users
about
about
the
work
that
you're
doing
you
know
nurse
users
are
doing
really
really
interesting,
work
and
yeah.
I
think
it'd
be
great
to
see
a
little
of
what
some
of
our
users
are
doing.
So
if
you've
got
a
lightning
talk
or
you
would
like
to
nominate
somebody
that
you
know
has
has
something
that
they
would
like
to
present.
But
it
is
an
interesting
thing
for
furnace
users
to
see.
Please
let
us
know
these
slides.
A
C
Do
you
know
when
the
the
gpus
are
going
to
start
actually
allocating
against
our
allocation.
C
A
No,
so
I
think
the
charging
will
be
all
at
once
when
it
when
what
happens
for
for
both
types
of
nodes,
but
so
so
the
cpu
nodes
were
still
kind
of
in
the
process
of
integrating
those,
and
you
know
bringing
the
system
up
so
so
they
are
there
and
we
encourage
you
to
to
try
them
out
and
let
us
know
about
any
issues
that
you
find,
but
also
remember
that
it's
it's
not
a
production
system,
yet
that's
still
being
brought
up.
So
there
might
be
some.
A
D
A
A
So
there's
now
a
link
in
the
zoom
chat.
So
if
you
have
a
topic
that
you're
interested
in
or
would
like
to
present
or
yeah
something
to
to
nominate,
please
let
us
know
about
advice,
form.
A
Thanks
all
for
joining
us
today,
the
slides
and
recording
will
be
on
the
website
kind
of
soon
and
thanks
again
annette
for
a
great
overview
of
sharing
and
we'll
see
you
all
in
the
coming
months.