September 2001 Email Thread

Date: Thu, 27 Sep 2001 10:54:44 -0800
To: twf@psfc.mit.edu, mrthompson@lbl.gov, keahey@mcs.anl.gov, crj@cs.utah.edu,
        peng@fusion.gat.com
From: "David P. Schissel" 
Subject: Update on Web on SC01 Demo
Cc: schissel
Status:   


 Kate, Tom, Mary, Chris, Qian -

   I have started updating the workers section on our web site.
Before a general release and in time for tomorrows phone call
I call your attention to

http://www.fusiongrid.org/work/meetings/sc01/sc01.html


  Specifically, I have by hand (ugh...) created an archive of
the email thread on the SC01 demo. One of the emails that Tom
Fredian sent out with a JPG attachment did not make it to all
individuals. You can read this email in the September thread
by going to the email with the JPG image. I have also placed
this image on the SC01 front page.

  We can discuss this image and his idea during our phone call tomorrow.

  Things on my mind to discuss (Chris - you can go first again):
   a) EFIT status at ANL - Qian
   b) Plans for EFIT reading/writing to MDSplus - Qian
   c) Any Vis results or a feeling of value - Chris
   d) Tom's demo scenario and IDL GUI.


  Non-Demo related:
    a) plans to creating long-range working plan - In parallel with
       SC01 demo work?


 The phone call information as set up by Kate:
     The call is setup for Friday, Sept 28 at 11am CST.
     The 8 lines should call into 1-888-677-5720
     Pass code: 48597


  Thanks,
     Dave

**************************************************************************
**************************************************************************

Sender: mrt@lbl.gov
Date: Wed, 26 Sep 2001 10:28:40 -0700
From: Mary Thompson 
Organization: LBNL
X-Accept-Language: en
To: "Thomas W. Fredian" 
CC: "Schissel, Dave" ,
        Kate Keahey , Mary Thompson 
Subject: Re: Minimum requirements for MDSplus/globus-io client
Status:   

That's not so bad. I'm glad to see that the globus libraries are
statically linked.  How are you starting the mdsplus server and how does
the client know what machine the server is running on? I looked at some
of the client scripts, but didn't find a server host name anywhere. 

There is now an empty 2G disk mounted on natasha as /usr/local1. It will
eventually be  mounted as /usr/local once we get the useful binares from
/home/shared/LINUX.local copied there. I made rsi, mdsplus and
fusionDemo directories there if you want to use them. Or that stuff can
stay in /home/natasha.

 
Mary

"Thomas W. Fredian" wrote:
> 
> Hi Mary,
> 
>     I tried to see what the minimum requirements were to use the MDSplus
> client with globus-io. I found that all I needed was to copy the
> following files from natasha to be able to connect from one of our nodes
> to the MDSplus data server on natasha:
> 
> libMdsIpShr.so                        -  MDSplus client library
> ~/.globus/*                              -  my personal certificate
> /etc/grid-security/certificates/*  - CA signing policies and
> certificates
> grid-proxy-init                         - proxy initialization program
> 
> I suspect this would be sufficent to run an MDSplus/globus-io server as
> well.
> 
> -tom

-- 
---------------------------------------------------------------------
Mary R. Thompson				 
Distributed Security Research Group		(510) 486-7408
Lawrence Berkeley National Lab			http://www-itg.lbl.gov/~mrt
----------------------------------------------------------------------

**************************************************************************
**************************************************************************

Date: Wed, 26 Sep 2001 11:50:39 -0400
From: "Thomas W. Fredian" 
X-Accept-Language: en
To: Mary Thompson 
CC: "Schissel, Dave" ,
        Kate Keahey 
Subject: Minimum requirements for MDSplus/globus-io client
Status:   

Hi Mary,

    I tried to see what the minimum requirements were to use the MDSplus
client with globus-io. I found that all I needed was to copy the
following files from natasha to be able to connect from one of our nodes
to the MDSplus data server on natasha:

libMdsIpShr.so                        -  MDSplus client library
~/.globus/*                              -  my personal certificate
/etc/grid-security/certificates/*  - CA signing policies and
certificates
grid-proxy-init                         - proxy initialization program

I suspect this would be sufficent to run an MDSplus/globus-io server as
well.

-tom
**************************************************************************
**************************************************************************

Date: Tue, 25 Sep 2001 11:07:53 -0400
From: "Thomas W. Fredian" 
X-Accept-Language: en
To: "David P. Schissel" 
CC: Mary Thompson 
Subject: Re: scdemo
Status:   

Hi Dave,

    I put together the controller GUI as a prototype/template for running the
demo. It shows how you might globusrun efit and wait for an MDSplus completion
event. Since you guys will be working on efit and the efit viewer can you use
this template and add the bells and whistles to liven it up for the demo? Right
now this stuff is in /home/natasha/demo... on natasha.lbl.gov. I hadn't planned
on doing much more on this except for building/installing the software on the
anl machines as soon as they have stable target machines ready.

    If you need to discuss this with me I am out of the office today but should
be in tomorrow. Keep in mind I leave the office around 4:40PM EST. You always
seem to call shortly after I've headed out to catch my train.

Thanks,
Tom

Mary Thompson wrote:

> Tom,
>   I put two of the files that I got from Dave into
> /usr/natasha/demo/images.
> lhd_control_room is panoramic view of one of the control rooms. It could
> be the initial display. It would need to be cropped to fit into the GUI.
> cmod_plasma6.gif is a picture of a plasma, I think. It could be
> displayed during
> the plasma started -plasma completed part.
>
> ---------------------------------------------------------------------
> Mary R. Thompson                                
> Distributed Security Research Group             (510) 486-7408
> Lawrence Berkeley National Lab                  http://www-itg.lbl.gov/~mrt
> ----------------------------------------------------------------------
**************************************************************************
**************************************************************************


Sender: mrt@lbl.gov
Date: Fri, 21 Sep 2001 16:17:38 -0700
From: Mary Thompson 
Organization: LBNL
X-Accept-Language: en
To: G@PSFC.MIT.EDU, "Thomas W. Fredian" ,
        "David P. Schissel" 
Subject: Re: fusion images
Status:   

Oops, I assumed this was one of the local tokamak control rooms. Do
either of you have such a thing? I would assume that physicists would
recognize these rooms and so pretending (or implying) that this is part
of the National Fusion Collaboratory would make us look foolish.


Martin Greenwald wrote:
> 
> Mary,
>
> for the LHD control room (credit to Tom Fredian)
> 
> "The control room of the LHD (Large Helical Device) an experiment at
> the National Institute for Fusion Studies in Japan.  LHD is the
> largest stellarator, a family of magnetic confinement devices, in
> the world.   U.S. Researchers use the internet to collaborate on
> this experiment."
> 
> - Martin
**************************************************************************
**************************************************************************

Sender: mrt@lbl.gov
Date: Fri, 21 Sep 2001 14:30:29 -0700
From: Mary Thompson 
Organization: LBNL
X-Accept-Language: en
To: Greenwald , "Thomas W. Fredian" ,
        "David P. Schissel" 
Subject: fusion images
Status:   

Martin,
  Thank you for the images. I confused Dave and possibly Tom, by thinkin
they had come from him in a previous mail. They did all come imbedded in
the mail rather than as attachments, but I was able to save them out of
the mail. I think the two that we are mostly likely to use for the
poster are 
lhd_control_room.jpg and crcover.gif. Do you have higher resolution
versions of the crcover one. My graphics consultant wasn't sure that
would be good enough for printing on a poster. (We need 300dpi for an
image about 10"x6"). We also need the captions.

Tom, 
  Since they came from Martin, I probably didn't need to give them to
you, but at least there are copies of the two that seemed most useful
for the demo on /home/natasha/demo/images. The rest are in
~mrt/fusionImages so you don't have to move them across the country
again.


-- 
---------------------------------------------------------------------
Mary R. Thompson				 
Distributed Security Research Group		(510) 486-7408
Lawrence Berkeley National Lab			http://www-itg.lbl.gov/~mrt
----------------------------------------------------------------------
**************************************************************************
**************************************************************************

Date: Fri, 21 Sep 2001 15:54:45 -0400
From: "Thomas W. Fredian" 
X-Accept-Language: en
To: "Schissel, Dave" 
Subject: [Fwd: Prototype demo working]
Status:   

Hi Dave,

	This is a copy of the second attempt to send the message to
nfc-workers.
The image should have only been in the attachment. If this bounced could
you
put the attachment somewhere on the web server and let me know?

Thanks,
Tom

-------- Original Message --------
Subject: Prototype demo working
Date: Fri, 21 Sep 2001 15:48:43 -0400
From: "Thomas W. Fredian" 
To: nfc workers 

Hi All,

    I have a prototype demo working on natasha.lbl.gov. I have a IDL
based gui, controller, with a start pulse button, a status text window
and a
graphics window. When you push the start pulse button you see messages
in the text window such as follows:

Plasma pulse beginning.
Plasma formed in tokomak.
Pulse completed, beginning data acquisition.
Data acquisition complete, beginning analysis.
<<<<< globusrun efit_job on compute server >>>>
globus_gram_client_callback_allow successful
GRAM Job submission successful
GLOBUS_GRAM_CLIENT_JOB_STATE_ACTIVE
GLOBUS_GRAM_CLIENT_JOB_STATE_DONE
Analysis done. Beginning visualization
Visualization done

The first several messages are timed to come out with short delays in
between just to simulate a plasma pulse and data acquisition. The
controller
is then globusrun'ing a job on diesel.lbl.gov which simulates an
analysis job. This job mdsconnects to a data server on natasha.lbl.gov,
fetches
some data from a node in an mdsplus tree, writes data back to the tree,
and issues an EFIT_DONE event. The controller notices the
EFIT_DONE event and mdsconnects to the data server, retrieves the data
and plots the data in the graphics window.  (See attached jpeg).

Currently, this only works if I am running the event_server and
data_server (now running on natasha) because it is configured to use a
host based
certificate which is in a file owned by me. These servers should
eventually be run as root.

The controller gui will have to be redone to do the fancier efit
animation but it should provide a sample to start with.

Seems to work!

Cheers,
Tom
Attachment converted: Macintosh_HD:scdemo.jpg (JPEG/ogle) (0004744E)


**************************************************************************
**************************************************************************

Date: Fri, 21 Sep 2001 12:48:54 -0700 (PDT)
From: owner-nfc-workers@fusion.gat.com
X-Authentication-Warning: apollo.gat.com: majordom set
  sender to owner-nfc-workers@fusion.gat.com using -f
To: owner-nfc-workers@fusion.gat.com
Subject: BOUNCE nfc-workers@fusion.gat.com:     Message too long (>40000 chars)  
Status:   

>From owner-nfc-workers  Fri Sep 21 12:48:52 2001
Received: from PSFC.MIT.EDU (psfc.mit.edu [198.125.176.135])
	by fusion.gat.com (8.9.1/8.9.0) with SMTP id MAA15634
	for ; Fri, 21 Sep 2001 12:48:49 -0700 (PDT)
Received: from psfc.mit.edu ([198.125.176.251]) by PSFC.MIT.EDU with ESMTP
          for nfc-workers@fusion.gat.com; Fri, 21 Sep 2001 15:48:44 -0400
Message-ID: <3BAB999B.C7DA3DAA@psfc.mit.edu>
Date: Fri, 21 Sep 2001 15:48:43 -0400
From: "Thomas W. Fredian" 
X-Mailer: Mozilla 4.7 [en] (WinNT; U)
X-Accept-Language: en
MIME-Version: 1.0
To: nfc workers 
Subject: Prototype demo working
Content-Type: multipart/mixed;
 boundary="------------8EA01425EE2F96A58F3D3DCF"

This is a multi-part message in MIME format.
--------------8EA01425EE2F96A58F3D3DCF
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit

Hi All,

    I have a prototype demo working on natasha.lbl.gov. I have a IDL
based gui, controller, with a start pulse button, a status text window
and a
graphics window. When you push the start pulse button you see messages
in the text window such as follows:

Plasma pulse beginning.
Plasma formed in tokomak.
Pulse completed, beginning data acquisition.
Data acquisition complete, beginning analysis.
<<<<< globusrun efit_job on compute server >>>>
globus_gram_client_callback_allow successful
GRAM Job submission successful
GLOBUS_GRAM_CLIENT_JOB_STATE_ACTIVE
GLOBUS_GRAM_CLIENT_JOB_STATE_DONE
Analysis done. Beginning visualization
Visualization done

The first several messages are timed to come out with short delays in
between just to simulate a plasma pulse and data acquisition. The
controller
is then globusrun'ing a job on diesel.lbl.gov which simulates an
analysis job. This job mdsconnects to a data server on natasha.lbl.gov,
fetches
some data from a node in an mdsplus tree, writes data back to the tree,
and issues an EFIT_DONE event. The controller notices the
EFIT_DONE event and mdsconnects to the data server, retrieves the data
and plots the data in the graphics window.  (See attached jpeg).

Currently, this only works if I am running the event_server and
data_server (now running on natasha) because it is configured to use a
host based
certificate which is in a file owned by me. These servers should
eventually be run as root.

The controller gui will have to be redone to do the fancier efit
animation but it should provide a sample to start with.

Seems to work!

Cheers,
Tom

--------------8EA01425EE2F96A58F3D3DCF
Content-Type: image/jpeg;
 name="scdemo.jpg"
Content-Transfer-Encoding: base64
Content-Disposition: inline;
 filename="scdemo.jpg"



**************************************************************************
**************************************************************************

Date: Thu, 20 Sep 2001 16:33:50 -0700
To: crj@cs.utah.edu
From: Qian Peng 
Subject: sample efit data for vis. part of SC demo
Cc: schissel
Status:   

Chris,

Please get the following two files from anonymous ftp: lithos.gat.com:pub/peng/

	flux.data.gz
	te.data.gz

These are one set of efit data to visualize.  They are in ASCII (I can
provide binary if you prefer that).  What we have in the first file is rho
as a function of X (what we usually call R; X > 0) and Z on plane Y=0, and
then as a function of time (t).  We can think that the plasma is
symmetrical about axis Z for now.  So if you rotate it around Z axis, that
gives you the torus.  Our conventional 2-D graph is rho contour on XZ.  The
second file contains electron temperature (Te) as a function of time and
rho.  We would like to see this information to be presented on the torus
along with the rho values.  To get Te for each (X,Z), you may need
interpolation.  Right now, the time goes from 600 ms to 3900 ms.  To show
plasma evolution more dramatically, we may need to get more data for
earlier times.  But what we have now should be a good starting point.

The data in flux.dat are
nx (dimension of x)	float
x		float(nx)
nz (dimension of z)	float
z		float(nz)
nt (dimension of t)	integer
t		float(nt)
rho		float(nx, nz, nt)

The data in te.dat are
nt (dim. of time) 	integer
t		float(nt)
nrho (dim. of rho)	integer
rho		float(nrho)
Te		float(nt, nrho)

Please let me or Dave know if you have questions about the data.  Thanks,

-Qian

**************************************************************************
**************************************************************************

X-Authentication-Warning: apollo.gat.com: majordom set sender to
  owner-nfc-workers@fusion.gat.com using -f
Date: Thu, 20 Sep 2001 14:59:39 -0700
From: Mary Thompson 
Organization: LBNL
X-Accept-Language: en
To: Kate Keahey 
CC: nfc-workers@fusion.gat.com
Subject: Re: exciting pictures of Fusion work
Sender: owner-nfc-workers@fusion.gat.com
Status:   

I need similar pictures here. Gifs or Tiff in at least 300dpi are the
most useful. Dave had one in his slide of someone standing inside a
tokamak which I thought was cool represenation of  "big science".

Thanks, Mary

Kate Keahey wrote:
> 
> Dear all,
> 
> I will be putting together a poster describing the Globus involvement in
> SciDAC. I want to include some exciting pictures to showcase the Fusion
> work; could somebody point me at such pictures? Many thanks!

===============================================================================

This message was sent to the SciDAC National Fusion Collaboratory (NFC)
workers list nfc-workers.  Visit the Collaboratory at
.

To unsubscribe from this list, please send a message to
majordomo@fusion.gat.com with the following text in the *body* of the
message:  unsubscribe nfc-workers

David P. Schissel:  

**************************************************************************
**************************************************************************

X-Sender: keahey@localhost
Date: Thu, 20 Sep 2001 11:31:59 -0500
To: "David P. Schissel" 
From: Kate Keahey 
Subject: Re: exciting pictures of Fusion work
Status:   

I just got some images rom Martin. I would love to work with your
graphics department as they seem to be producing really good quality
stuff but the problem is that there is a process for getting those
things for the Argonne both which is supposed to make all the posters
look uniform. So it has to be done by a certain company etc. The
examples that I saw from previous years don't look wonderful but this is
what we have to do.

I think it would be really wonderful though if your graphics department
could help with the handout (this would be the thing accompanying the
demo). There are no rules imposed on handouts so we can do whatever we
want and make it good.


At 09:18 AM 9/20/2001 -0800, you wrote:

 Kate -

 Happy to help....We have a nice graphics department with
many images and the like so we should be able to help
put this together. Not sure how to proceed given the
total volume of images we have. Maybe you have some
thoughts on what images you might want?

 -david

This is something additional; I think I posted a message about it
earlier: there is a limit (wall space) on how many posters can be put in
Argonne booth and so it was decided for example to put all the SciDAC
initiatives together. I am in charge of that poster so Fusion will be
featured prominently ;-), but we have to share that poster with 6 other
projects and it all has to be more or less uniform so this is not the
best place to put our "story". The text is going to be about what makes
the project exciting, how we expect to make a difference, what the
impact will be etc. In other words strong motivation stuff and how we
want to go about achieving things. If you would like to have some input
on this I would appreciate it very much.

Kate


**************************************************************************
**************************************************************************

X-Sender: keahey@localhost
Date: Thu, 20 Sep 2001 09:28:56 -0500
To: "David P. Schissel" 
From: Kate Keahey 
Subject: Re: exciting pictures of Fusion work
Status:   

This is something additional; I think I posted a message about it
earlier: there is a limit (wall space) on how many posters can be put in
Argonne booth and so it was decided for example to put all the SciDAC
initiatives together. I am in charge of that poster so Fusion will be
featured prominently ;-), but we have to share that poster with 6 other
projects and it all has to be more or less uniform so this is not the
best place to put our "story". The text is going to be about what makes
the project exciting, how we expect to make a difference, what the
impact will be etc. In other words strong motivation stuff and how we
want to go about achieving things. If you would like to have some input
on this I would appreciate it very much.

At 08:45 PM 9/19/2001 -0800, you wrote:

 Kate -

 Is this part of our demo? Is this the "fusion story" we are trying
to tell? Or, is this something else/additional?

 - david



Dear all,

I will be putting together a poster describing the Globus involvement in
SciDAC. I want to include some exciting pictures to showcase the Fusion
work; could somebody point me at such pictures? Many thanks!

__________________________
Dr. Kate Keahey
Math & Computer Science Div.
Argonne National Laboratory
Argonne, IL 60439, USA
(630) 252-1673


===============================================================================

This message was sent to the SciDAC National Fusion Collaboratory (NFC)
workers list nfc-workers.  Visit the Collaboratory at
.

To unsubscribe from this list, please send a message to
majordomo@fusion.gat.com with the following text in the *body* of the
message:  unsubscribe nfc-workers

David P. Schissel:  


__________________________
Dr. Kate Keahey
Math & Computer Science Div.
Argonne National Laboratory
Argonne, IL 60439, USA
(630) 252-1673

**************************************************************************
**************************************************************************

X-Authentication-Warning: apollo.gat.com: majordom set sender to
owner-nfc-workers@fusion.gat.com using -f X-Sender: keahey@localhost
Date: Wed, 19 Sep 2001 20:08:26 -0500 To: nfc-workers@fusion.gat.com
From: Kate Keahey  Subject: exciting pictures of
Fusion work Sender: owner-nfc-workers@fusion.gat.com Status:

Dear all,

I will be putting together a poster describing the Globus involvement in
SciDAC. I want to include some exciting pictures to showcase the Fusion
work; could somebody point me at such pictures? Many thanks!

__________________________ Dr. Kate Keahey Math & Computer Science Div.
Argonne National Laboratory Argonne, IL 60439, USA (630) 252-1673


===============================================================================

This message was sent to the SciDAC National Fusion Collaboratory (NFC)
workers list nfc-workers.  Visit the Collaboratory at
.

To unsubscribe from this list, please send a message to
majordomo@fusion.gat.com with the following text in the *body* of the
message:  unsubscribe nfc-workers

David P. Schissel: 


**************************************************************************
**************************************************************************



X-Authentication-Warning: apollo.gat.com: majordom set sender to
owner-nfc-workers@fusion.gat.com using -f X-Sender: keahey@localhost
Date: Wed, 19 Sep 2001 19:56:47 -0500 To: nfc-workers@fusion.gat.com
From: Kate Keahey  Subject: telecon Friday next week
Sender: owner-nfc-workers@fusion.gat.com Status:

Here is the information about the next SC01 telecon.

The call is setup for Friday, Sept 28 at 11am CST. The 8 lines should
call into 1-888-677-5720 Pass code: 48597



__________________________ Dr. Kate Keahey Math & Computer Science Div.
Argonne National Laboratory Argonne, IL 60439, USA (630) 252-1673


===============================================================================

This message was sent to the SciDAC National Fusion Collaboratory (NFC)
workers list nfc-workers.  Visit the Collaboratory at
.

To unsubscribe from this list, please send a message to
majordomo@fusion.gat.com with the following text in the *body* of the
message:  unsubscribe nfc-workers

David P. Schissel: 


**************************************************************************
**************************************************************************

Sender: mrt@lbl.gov
Date: Mon, 17 Sep 2001 10:13:11 -0700
From: Mary Thompson 
Organization: LBNL
X-Accept-Language: en
To: "Thomas W. Fredian" 
CC: Kate Keahey , Keith Jackson ,
        "David P. Schissel" 
Subject: Re: Starting the mdsip server
Status:   

Tom,
  My suggestion to use globusrun to start the servers was to provide a
simplier solution than your current one one of ssh`ing in to a globus
machine doing a proxy-init and starting the server by hand. BTW these
servers will only work for the duration of the proxy certificate which I
think is 12hrs, so you can't just start up a server at the beginning of
the week and assume it will be good to go for the whole week. I assume
this hand starting applies to the EFIT and visualization program as well
as the  mdsplus server. Globusrun lets you do a proxy-init on your own
workstation (once every 12 hrs) and then it starts the other servers
running under a proxy credential derived from your credential. This is
the famous Globus "single-signon". 

The stuff about the globus services was just a way to elimate the need
for globus run once things were setup on a server machine.

Thanks for the info about mdsPlus events. I assumed they would work
across a WAN but thought I should check how it was done.

Globusrun should work for you on diesel, clipper, fluffy and slappy all
of whom have a working gatekeeper running. Portnoyc and scott  do not.
Your credential needs to be in the globus grid-mapfile which it is on
our machines, but probably not on pitcairn yet.

Mary

"Thomas W. Fredian" wrote:
> 
> Hi Mary,
> 
>     The MDSplus data server already runs listening to connects from clients. When
> a client connects it forks off a server process for that client. The connect
> completes almost instantaneous. I guess I don't see any great need to globusrun
> anything for the demo. I see the controller as simply a process that sets a
> data_ready MDSplus event. See my description in previous email. Using this
> approach the whole demo could be debugged and run on any systems (including a set
> of non-globus machines). GA could essentially get all of the demo running on
> their local machines. The only different between that implementation and the
> actual demo will be the actual demo will use the globus_io version thus being
> secure.
> 
> Regarding MDSplus events... Events are broadcasted only locally on a single node
> using ipc mechanisms. Remote systems can define environment variables to define
> which system is going to be servicing the events. When the remote system issues
> and event or waits for one, the process automatically connects to a data server
> on the system servicing the events. You can issue an event from the shell using

**************************************************************************
**************************************************************************

Date: Sat, 15 Sep 2001 01:10:30 -0400
From: "Thomas W. Fredian" 
X-Accept-Language: en
To: Mary Thompson 
CC: Kate Keahey , Keith Jackson ,
        "David P. Schissel" 
Subject: Re: Starting the mdsip server
Status:   

Hi Mary,

    The MDSplus data server already runs listening to connects from clients. When
a client connects it forks off a server process for that client. The connect
completes almost instantaneous. I guess I don't see any great need to globusrun
anything for the demo. I see the controller as simply a process that sets a
data_ready MDSplus event. See my description in previous email. Using this
approach the whole demo could be debugged and run on any systems (including a set
of non-globus machines). GA could essentially get all of the demo running on
their local machines. The only different between that implementation and the
actual demo will be the actual demo will use the globus_io version thus being
secure.

Regarding MDSplus events... Events are broadcasted only locally on a single node
using ipc mechanisms. Remote systems can define environment variables to define
which system is going to be servicing the events. When the remote system issues
and event or waits for one, the process automatically connects to a data server
on the system servicing the events. You can issue an event from the shell using a
command such as:

# setevent myevent

or wait for one:

# wfevent myevent

There are also routines which will cause a thread to spin up and run a user
callback function when an event occurs. We've taught IDL how to handle MDSplus
events as well so you can have a GUI recognizing MDSplus events.

At this point in the project our view of the solution are coming from too very
different starting points. I think the fusion folks view the demo as doing pretty
much everything we normally do except we'll be using globus to improve security
whereas you and Kate are looking at how to fit the visualization, efit and data
server into the globus framework. Undoubtedly once I finally read more of the
globus documentation I'll understand that globus provides much better mechanisms
for doing what we have been doing. Unfortunately we don't have a lot of
documentation on MDSplus so you could better understand where I am coming from.

I'll test out this stuff on pitcairn probably on Monday. Thanks everybody for
looking into the problem.

Thanks,
Tom

Mary Thompson wrote:

> There are two ways of starting the servers here and I think there may be
> some confusion.
>
> Plan A - (works for demo)
>   Have controller start MDSplusDataServer, EFIT and the Visualization
> program.  They have to be started at the same time since they are going
> to use MDSplus events and the listening process has to be listening at
> the time the event is broadcast. This can be done by having the
> controller use  3 globusrun commands.
> The reason for the "runme" script is that globusrun does not return
> until the command it is running returns. In this scenario the
> MDSplusDataServer, EFIT and Vis should exit  when they are done. When we
> want a new "shot" the controller will be run and start up everything
> again. Since these guys need to be running simultaneosly we may need a
> little tuning It seemed to take globusrun of /bin/echo about 20 seconds
> to complete. If the DataServer is too fast, it may broadcast its
> completion, before we can get EFIT running. So we may need to start them
> in the reverse order.
>
> BTW, Tom,
>   When the MDSplus documenation talks about "broadcasting" events does
> this actually mean a LAN broadcast, or it is some more general
> notification to processes that it knows are listening. If it actaully a
> local area broadcast, it won't be heard across the WAN.
>
> Plan B - What I think Tom wants for the long term.
> The MDSplusClient library calls the server is such a manner that it gets
> a instance of the server started. This is just like inetd. The Gloubs
> services is the globus way of doing this, that is clever enough to start
> the server with a proxy certiticate. I'm not sure where this is
> documented, but we have one example that Keith did here last year.
>
> Kate Keahey wrote:
> >
> > Mary,
> >
> > Many thanks for explanation of the services process! (and the pointers to
> > security tutorials).
> >
> > I was also assuming that we were going to simply start the processes from
> > the controller (rather than using long-lived servers). The Globus
> > gatekeeper is a long-running server essentially dedicated to the task of
> > startup so if we go this route instead of one server we are going to have
> > two. Out of curiosity: what is the reason that they need to be long-running
> > servers? The Data Server probably takes a long time to initialize and
> > doesn't take that many cycles, but it seems to me that EFIT might as well
> > be started on an "as needed" basis?
> >
> > Also it is not clear to me why it is better for the controller to be a
> > script rather than a program since the changes to the part of API that
> > starts the jobs will be minor (if any) and I don't know if MDSplus has a
> > scripting interface...

**************************************************************************
**************************************************************************

Sender: mrt@lbl.gov
Date: Fri, 14 Sep 2001 12:20:00 -0700
From: Mary Thompson 
Organization: LBNL
X-Accept-Language: en
To: "Thomas W. Fredian" , Von Welch 
CC: Kate Keahey ,
        "Schissel, Dave" ,
        Keith Jackson 
Subject: Re: [Fwd: IDL Evaluation Licenses]
Status:   

Tom,
Pitcarin has been configured to talk to LBL, but its LBL-CA certificate
is out of date.  Sould someeone with root access to pitcairn, please
install the new certificate.

Jason Novotny wrote:
> 
> Hello,
> 
>     Some of you may be experiencing problems using Globus, GSIFTP or
> Myproxy on our Grid machines due to the fact that one of the
> CA certificates in our chain expired yesterday. I've updated the CA cert
> on our machines, however clients will need to update the attached
> CA certificate with the hash f7d75486.0 in their installation in order
> for mutual authentication to work successfully.
> 
>     Please let me know if there are any problems
> 
>     Thanks, Jason
> 
> --
> Jason Novotny               jdnovotny@lbl.gov
> Home: (510) 610-8360        Work: (510) 486-8662
> NERSC Distributed Computing http://www-itg.lbl.gov/Grid
> 
>   ------------------------------------------------------------------------
> -----BEGIN CERTIFICATE-----
> MIICvzCCAiigAwIBAgIBETANBgkqhkiG9w0BAQUFADBbMRkwFwYDVQQKExBET0Ug
> U2NpZW5jZSBHcmlkMSAwHgYDVQQLExdDZXJ0aWZpY2F0ZSBBdXRob3JpdGllczEc
> MBoGA1UEAxMTQ2VydGlmaWNhdGUgTWFuYWdlcjAeFw0wMTAxMzAwMDM0MzlaFw0w
> NjAxMjkwMDM0MzlaMHgxDTALBgNVBAoTBEdyaWQxLjAsBgNVBAoTJUxhd3JlbmNl
> IEJlcmtlbGV5IE5hdGlvbmFsIExhYm9yYXRvcnkxIDAeBgNVBAsTF0NlcnRpZmlj
> YXRlIEF1dGhvcml0aWVzMRUwEwYDVQQDEwxMQk5MLUdyaWQtQ0EwgZ8wDQYJKoZI
> hvcNAQEBBQADgY0AMIGJAoGBAL2t4aX933WXYlofuY+L+16Tdl/KxpAammyfcW8u
> kHHT6RYDjaQdfV1FpNEqfSrRjKNwGGGkrG4XHZWiUO0Di0AlBN04lsRY6jB68l6B
> 5byujfZv+8EeCI2c1ObBLYZYi4lToJf0sm0Hpn3GD7PZBv6BVHLOuwEFDl9z9Dnc
> DFDdAgMBAAGjdjB0MBEGCWCGSAGG+EIBAQQEAwIAhzAOBgNVHQ8BAf8EBAMCAcYw
> HQYDVR0OBBYEFIn+csPVyp+iprpYUIu1SziMQiDxMA8GA1UdEwEB/wQFMAMBAf8w
> HwYDVR0jBBgwFoAUm85P8ry9WHAx1fIyDn6eveJRFOcwDQYJKoZIhvcNAQEFBQAD
> gYEAehijYRFhkM1odxkZ0xUsoI1PjrPCHDRy54sp4oWHl53r5CJdYR+5b33ZDw32
> RKBH9XAp8WJV5zppLtf9T9BAKddfCnmcREvaCOoFjBTWgEPVyVz64ax4TvaPXqRK
> XtPCTyJt4zDYX68qlqmt3VLDnBSIGydTDFZ+QSwd0Tc02R8=
> -----END CERTIFICATE-----

So t
---------------------------------------------------------------------
Mary R. Thompson				 
Distributed Security Research Group		(510) 486-7408
Lawrence Berkeley National Lab			http://www-itg.lbl.gov/~mrt
----------------------------------------------------------------------
**************************************************************************
**************************************************************************

Date: Fri, 14 Sep 2001 14:42:11 -0400
From: "Thomas W. Fredian" 
X-Accept-Language: en
To: Mary Thompson , Kate Keahey 
CC: "Schissel, Dave" 
Subject: Re: [Fwd: IDL Evaluation Licenses]
Status:   

I logged into a machine at anl and built mdsplus on it (pitcairn.mcs.anl.gov,
SunOS). When I try to connect from another process on pitcairn or any processes
on the lbl machines I've built mdsplus on I get an
GLOBUS_IO_ERROR_TYPE_AUTHENTICATION_FAILED error when I try to connect. Is it
likely that this is caused by the ca-signing-policy on pitcairn not accepting
my certificate or something else. The software on pitcairn as well as that on
the lbl machines are using the 1.1.3 version of globus. I have been successful
in connecting between SunOS and linux systems at lbl without any problems.

Thanks,
Tom

Mary Thompson wrote:

> Great. I am planning on bringing natasha.lbl.gov which is 2X400Mh
> pentium linux machine to SC. I want to install redhat 7.1 on it today
> plus the one Globus config file it needs to run globus client side code.
> After that, any one who has an LBL account can use it to submit globus
> jobs.
>
> Mary
> --
> ---------------------------------------------------------------------
> Mary R. Thompson                                
> Distributed Security Research Group             (510) 486-7408
> Lawrence Berkeley National Lab                  http://www-itg.lbl.gov/~mrt
> ----------------------------------------------------------------------

**************************************************************************
**************************************************************************

Sender: mrt@lbl.gov Date: Fri, 14 Sep 2001 10:48:32 -0700 From: Mary
Thompson  Organization: LBNL X-Accept-Language: en
To: "Thomas W. Fredian"  CC: Kate Keahey
, "Schissel, Dave" 
Subject: Re: [Fwd: IDL Evaluation Licenses] Status:

Great. I am planning on bringing natasha.lbl.gov which is 2X400Mh
pentium linux machine to SC. I want to install redhat 7.1 on it today
plus the one Globus config file it needs to run globus client side code.
After that, any one who has an LBL account can use it to submit globus
jobs.

Mary --
---------------------------------------------------------------------
Mary R. Thompson				 Distributed
Security Research Group		(510) 486-7408 Lawrence Berkeley National
Lab			http://www-itg.lbl.gov/~mrt
----------------------------------------------------------------------

**************************************************************************
**************************************************************************

X-Authentication-Warning: apollo.gat.com: majordom set sender to
owner-nfc-workers@fusion.gat.com using -f X-Sender: keahey@localhost
Date: Fri, 14 Sep 2001 12:37:27 -0500 To: nfc-workers@fusion.gat.com
From: Kate Keahey  Subject: SC01 demo: ANL target
systems Sender: owner-nfc-workers@fusion.gat.com Status:

Tom,

We are going to have a "stable" node on our target system for our use on
Monday. Will let you know as soon as it is reconfigured.


__________________________ Dr. Kate Keahey Math & Computer Science Div.
Argonne National Laboratory Argonne, IL 60439, USA (630) 252-1673


===============================================================================

This message was sent to the SciDAC National Fusion Collaboratory (NFC)
workers list nfc-workers.  Visit the Collaboratory at
.

To unsubscribe from this list, please send a message to
majordomo@fusion.gat.com with the following text in the *body* of the
message:  unsubscribe nfc-workers

David P. Schissel: 



**************************************************************************
**************************************************************************


Date: Fri, 14 Sep 2001 13:16:48 -0400
From: "Thomas W. Fredian" 
X-Accept-Language: en
To: Mary Thompson , Kate Keahey ,
        "Schissel, Dave" 
Subject: [Fwd: IDL Evaluation Licenses]
Status:   

We should be all set with IDL licenses for the demo.

-tom

Jim Hill wrote:

> Hi Tom, a message you sent to Donna Carr was forwarded to me.  Donna
> recently decided to leave RSI after 8 years and felt spending time with her
> 3 year old was something she wanted to do.   I am sure Donna took care of
> you and I hope I live up to her standards.   I have been a commercial rep
> for 13 years and this is my second stint at RSI and selling IDL so you will
> only have to break me in on the educational side of "things."   When you are
> ready for the evaluation and have a platform that you will be using let me
> know and I will set you up.  Not a problem.  I use evaluations quick
> frequently with clients to not only showcase their product but IDL also.
>
> All of my information is below.
>
>
> Best regards,
>
> Jim Hill
>
> Jim Hill                              (303)413-3912 Office
> Account Manager               jhill@researchsystems.com
> Northeast Territory              (970)215-5535 Cell
> Research Systems, Inc.   A Kodak Company
> http://www.researchsystems.com
>
> Shelly Schmit                   (303)413-3972
> Maintenance Account Rep   shelly@researchsystems
>
**************************************************************************
**************************************************************************


X-Authentication-Warning: apollo.gat.com: majordom set sender to
owner-nfc-workers@fusion.gat.com using -f Date: Fri, 14 Sep 2001
09:40:01 -0700 From: Mary Thompson  Organization:
LBNL X-Accept-Language: en To: Kate Keahey  CC:
nfc-workers@fusion.gat.com Subject: Re: SC01 demo: more info Sender:
owner-nfc-workers@fusion.gat.com Status:

I wasn't sure how big a deal it is to "install" EFIT on a machine. I
would guess the only installation issue is if a stable version of Globus
is on the machine and since we have 4 "production" Globus nodes here I
thought that might be an immediate solution. --
---------------------------------------------------------------------
Mary R. Thompson				 Distributed
Security Research Group		(510) 486-7408 Lawrence Berkeley National
Lab			http://www-itg.lbl.gov/~mrt
----------------------------------------------------------------------

===============================================================================

This message was sent to the SciDAC National Fusion Collaboratory (NFC)
workers list nfc-workers.  Visit the Collaboratory at
.

To unsubscribe from this list, please send a message to
majordomo@fusion.gat.com with the following text in the *body* of the
message:  unsubscribe nfc-workers

David P. Schissel: 


**************************************************************************
**************************************************************************

X-Authentication-Warning: apollo.gat.com: majordom set sender to
owner-nfc-workers@fusion.gat.com using -f X-Sender: keahey@localhost
Date: Fri, 14 Sep 2001 10:23:36 -0500 To: Mary Thompson
 From: Kate Keahey  Subject: Re:
SC01 demo: more info Cc: nfc-workers@fusion.gat.com Sender:
owner-nfc-workers@fusion.gat.com Status:

Mary,

we have lots of machines here that we can run EFIT on. Tom's point was
that it would be best to install EFIT etc. on target systems i.e.
systems that will actually be used at SC01 for the demo. My reply was
that we might not have such systems in a stable configuration right now
(i.e. for example the os might be upgraded between now and SC) in which
case we might have to reinstall all that software. Running them at LBL
is not a solution as in that case we would *certainly* have to
reinstall.


At 05:30 PM 9/13/2001 -0700, Mary Thompson wrote: We can run EFIT on a
linux grid node at LBL until the Argonne machines are stable.  Does
anyone else need and account here? --
---------------------------------------------------------------------
Mary R. Thompson                                
Distributed Security Research Group             (510) 486-7408 Lawrence
Berkeley National Lab                  http://www-itg.lbl.gov/~mrt
----------------------------------------------------------------------

__________________________ Dr. Kate Keahey Math & Computer Science Div.
Argonne National Laboratory Argonne, IL 60439, USA (630) 252-1673


===============================================================================

This message was sent to the SciDAC National Fusion Collaboratory (NFC)
workers list nfc-workers.  Visit the Collaboratory at
.

To unsubscribe from this list, please send a message to
majordomo@fusion.gat.com with the following text in the *body* of the
message:  unsubscribe nfc-workers

David P. Schissel: 


**************************************************************************
**************************************************************************

X-Authentication-Warning: apollo.gat.com: majordom set sender to
owner-nfc-workers@fusion.gat.com using -f Date: Thu, 13 Sep 2001
17:30:47 -0700 From: Mary Thompson  Organization:
LBNL X-Accept-Language: en To: Kate Keahey  CC:
nfc-workers@fusion.gat.com Subject: Re: SC01 demo: more info Sender:
owner-nfc-workers@fusion.gat.com Status:

We can run EFIT on a linux grid node at LBL until the Argonne machines
are stable.  Does anyone else need and account here? --
---------------------------------------------------------------------
Mary R. Thompson				 Distributed
Security Research Group		(510) 486-7408 Lawrence Berkeley National
Lab			http://www-itg.lbl.gov/~mrt
----------------------------------------------------------------------

===============================================================================

This message was sent to the SciDAC National Fusion Collaboratory (NFC)
workers list nfc-workers.  Visit the Collaboratory at
.

To unsubscribe from this list, please send a message to
majordomo@fusion.gat.com with the following text in the *body* of the
message:  unsubscribe nfc-workers

David P. Schissel: 


**************************************************************************
**************************************************************************

X-Authentication-Warning: apollo.gat.com: majordom set sender to
owner-nfc-workers@fusion.gat.com using -f X-Sender: keahey@localhost
Date: Thu, 13 Sep 2001 17:07:45 -0500 To: nfc-workers@fusion.gat.com
From: Kate Keahey  Subject: SC01 demo: more info
Sender: owner-nfc-workers@fusion.gat.com Status:


Here is some more information about the SC01 demo.

1) Target systems. I am still working on getting one. The stable testbed
that we were planning to contribute to Fusion is not stable (meaning:
people are likely to reinstall the operating system, take down the nodes
without much of a warning, change the configuration, etc). In particular
the configuration of every single node is likely to change between now
and SC01. I am trying to find one stable node where we could get EFIT
and MDSplus installed. Also, the clusters and desksides we (ANL) are
going to take to SC01 are not ready yet and won't be for another 2-3
weeks. So if I manage to find a reasonably stable node for the EFIT
calculation, we can get EFIT and MDSplus there and also (temporarily) do
the "controller" work there. Later, we can do the installation on the
SC01 machines (for both controller and EFIT since in case we loose
outside connectivity at SC we could run the demo on one of the clusters
we will have on the floor). Would this work? I think that realistically
we may have to reinstall things at some point though. As we discussed,
the version of Globus may change for example. How hard is it to
reinstall?

2) Infoviz component. There is a person in our group who is currently
working with the Futures Lab on developing such component. The component
will visualize computation and data transfer across the map of the US.
It is still being designed but chances are it will be something like
this. The moving data will be visualized as a series of dots moving
between two destinations (faster for faster bandwidth), the computing
resources will be say represented by some static information (such as
number of processors, available memory, etc), and the computation itself
will be mapped to some of those nodes. At least this is the current
thinking, this is very much work in progress which means that (1) it is
not there yet (2) we could have some input. Let me know if you have
suggestions/ideas; not all of them can be implemented (at this point)
since this component will be shared across many demos and has to
accommodate the most popular set of functionality, but I will try. The
interface to this component is a bit in a flux right now, but it will
probably be something along the lines of a message periodically sending
information (during a transfer) about how much data was sent. So once we
have something working and also once the interface is better defined we
can think about integrating it into our demo. At this point, it seems
like the esiest way to do that would be to use GridFTP (instead of
Globus io) for data transfer. Doing this would probably entail using the
new version of Globus, etc,so there are some issues here we should
discuss.

3) Backup. In case we loose outside connectivity it would be good to
have a "backup" demo to run on SC01 floor. For example we can run EFIT
on one of the 8-node clusters we are taking to SC. Mary, is LBNL making
similar arrangements (and do you have enough storage)? This is something
that we should probably talk about at next telecon.

4) Interface to controller. It seems that especially if we want to run
the demo on different machines (backup and remotely) the demoee needs a
way to enter some options. In addition, if we do the demo with all the
"bells and whistles" we may want to give the demoee the option of say
doing or not doing the Infoviz component. Also, we may want to regulate
the size of data we will be sending. And choose between parallel and
serial versions of the program. In other words, I still think that it
would be nice to have a GUI for that. I can request one but would have
to act now...



__________________________ Dr. Kate Keahey Math & Computer Science Div.
Argonne National Laboratory Argonne, IL 60439, USA (630) 252-1673


===============================================================================

This message was sent to the SciDAC National Fusion Collaboratory (NFC)
workers list nfc-workers.  Visit the Collaboratory at
.

To unsubscribe from this list, please send a message to
majordomo@fusion.gat.com with the following text in the *body* of the
message:  unsubscribe nfc-workers

David P. Schissel: 


**************************************************************************
**************************************************************************

X-Authentication-Warning: apollo.gat.com: majordom set sender to
owner-nfc-workers@fusion.gat.com using -f X-Sender: keahey@localhost
Date: Wed, 12 Sep 2001 18:16:12 -0500 To: "Thomas W. Fredian"
 From: Kate Keahey  Subject: Re:
SC01 telecon summary Cc: nfc-workers@fusion.gat.com Sender:
owner-nfc-workers@fusion.gat.com Status:

Tom,

Thanks for clarification!

Target systems: I will send you mail tomorrow morning with this
information.

Let's discuss the backup issues later; I tend to think that we will
gather up enough requirements to justify a "GUI" or if not a gui some
other kind of interface where the user will be telling the controller
what to do and on which machines to run.

At 03:14 PM 9/12/2001 -0400, Thomas W. Fredian wrote:

Hi Kate,

This looks good. I have a few minor corrections. In section c) below,
the "Data Server" is not driven by MDSplus events at all. The other
pieces simply connect to the data server and read and/or write data and
issue events and/or wait for events. A flow chart of each piece would be
more or less:

a) Controller

while (1) { optional:  Wait for operator to push go button Issue
raw_data_available event Wait for efit_data_available event Output efit
analysis done message Wait for visualization_done event Output
visualization done message }

b) EFIT computation

while (1) { Wait for raw_data_available event Read raw data from mdsplus
datafile (via MDSplus server) Compute efit data Write efit data to
mdsplus datafile (via MDSplus server) Issue efit_data_available event }

c) Data server

Just running accepting connections and processing requests

d) Scientific Visualization

while (1) { Wait for efit_data_available event Display visualization
Issue visualization done message }

e) Infovis

I presume this is just a fancy way of  displaying the progress being
monitored by the controller.


Regarding "Tasks, timelines and ownership".

It would be best to identify the target systems explicitly so we can get
the demo up and running on the exact systems that will be used for the
demo.

There are other issues regarding the actual execution of the demo. Up to
now I have been logging in to lbl machines and doing a grid-proxy-init
using my Fusion Collaboratory certificate and running the server in one
xterm and then logging into the same or other machine in a separate
window doing the grid-proxy-init and running a client application.
Normally the server would be fired up by inetd when a connection to the
port is received. I have modified the mdsip server to run as a stand
alone process which listens on a port and forks servers as connections
are received. This eliminates the potential problem of doing globus_io
on an already connected socket (via inetd). We could have the MDSplus
server run from inittab with respawn but I'll need some help figuring
out how a server such as this typically does the equivalent of
grid-proxy-init at startup.

We should also have a data server running on the SC floor for backup. It
should be fairly simple to quickly alter the demo to run using a
different data server.

-tom


Kate Keahey wrote:

All,

Here is the summary as I remember it. Please, send amendments,
clarifications, and violent objections.

1) Status (I don't know that we discussed it much, but should have)

Thanks to Tom for integrating Globus into MDSplus.

2) Evolution of demo description

We are going to have 5 components (only 4 components for the basic
version) interacting using events of the MDSplus framework . The MDSplus
framework has been extended to offer the option of using Globus: GSI,
Globus IO, etc. (to the extend required by this demo). The components
are as follows:

a) Controller that sends and receives MDSplus events in order to
coordinate actions of the other components. The controller executes on a
workstation (an SC01 floor deskside). In the current version the
controller automatically orchestrates the interaction of other demo
components (EFIT, data server, etc.), and is invoked by the user (the
demoee) by invoking the controller executable. In other words, invoking
the controller will make the run happen from beginning to the end. This
is for the basic version. In the fancy version we might complicate that
somewhat (see below)

b) EFIT computation: computes EFIT, in the basic version sequentially,
in the fancy version maybe in parallel. EFIT runs on a Linux machine (or
cluster of Linux machines) at Argonne and is driven by MDSplus framework
events generated by the controller, and generates MDSplus events to
write to the data server.

c) Data Server: provides data for EFIT, stores data produced by EFIT,
provides data for visualization, and generally does data server stuff.
Runs at LBNL. The Data Server accepts MDSplus events from the EFIT
computation, and from the controller and responds to them by providing
or storing data.

d) Scientific visualization component: visualizes data output by EFIT.
It runs on the SC floor deskside (could be a part of the controller
process or done separately whatever Tom thinks is better) and is
triggered by MDSplus events generated by the controller. To do its work
it requests data from the data server. The basic version does an
IDL-based visualization; for the fancy version Chris offered to do
something more visually interesting. Thanks to Chris for chiming in and
offering to do that!

e) Infoviz component: visualizes what happens in the demo. It runs on
the SC floor deskside (could be a part of the controller process or done
separately whatever Tom thinks is better) and is driven by MDSplus
events generated by the controller. This would require the controller to
perform monitoring functions as well as initiate work of the components.
In the basic version we will skip this component, in the fancy version
we are planning to have it. Kate is in charge of keeping track of where
we can get it from.

3) Tasks, timelines and ownership

Here is what we agreed to try to do in the next two weeks:

1) Install the necessary software pieces more or less where they are
going to run:

a) Installation

- LBNL: Data Server, MDSplus  (ownership: Tom) - ANL: MDSplus, EFIT (on
one node for now) (ownership: Tom, David) - Send mail to Kate and Mary
to get accounts for whoever needs them to do this.

b) Interaction Code

- Once the installation has been completed Tom will try to write a
controller for the DataServer and EFIT piece  (I somehow expect more
MDSplus to Globus issues especially as starting up those processes
remotely goes; Tom, please feel free to demand from me bits of example
code and such things as necessary).

- My understanding is that the sequential EFIT is already working with
MDSplus, is this right? If it is, nothing to be done there right now.

- With extensive help from Kate (and probably Mary also), Tom will try
to do a first run of the controller, EFIT, and Data Server pieces.

c) Visualization

- David will provide data for Chris

- Chris will play with the data and try to design a cool visualization
for it.

- Kate will try to find out about the infoviz piece.

d) Target. If all of this happens at this point next week we should have
either a basic demo nearly done (in two pieces), or a very specific list
of things that don't quite work yet. Or both.

3) List of possible "extension features" to demo

a) an evolution of the controller (GUI, start button, let the user
"replay" visualization, other functionality we might want) b) infoviz
controller c) parallel version of EFIT d) more data, less data e)
fancier visualization (as proposed by Chris)

4) Other issues that we didn't get to, but perhaps are not the most
urgent for now:

a) backup issues for SC (what happens if we loose outside connectivity?)
b) testbed issues (when and which Globus version to install on sites
other than LBNL and ANL) c) other advertising issues for SC (flyers,
posters, other stuff)


__________________________ Dr. Kate Keahey Math & Computer Science Div.
Argonne National Laboratory Argonne, IL 60439, USA (630) 252-1673

__________________________ Dr. Kate Keahey                   Math &
Computer Science Div. Argonne National Laboratory Argonne, IL 60439, USA
(630) 252-1673

**************************************************************************
**************************************************************************

X-Authentication-Warning: apollo.gat.com: majordom set sender to
  owner-nfc-workers@fusion.gat.com using -f
Date: Wed, 12 Sep 2001 15:33:10 -0700
From: Mary Thompson 
Organization: LBNL
X-Accept-Language: en
To: "Thomas W. Fredian" 
CC: Kate Keahey , nfc-workers@fusion.gat.com
Subject: Re: SC01 telecon summary
Sender: owner-nfc-workers@fusion.gat.com
Status:   

Is there an issue of licenses before we can install IDL on LBL and ANL
machines?

I'll try to come up with the target LBL desktop demo machine ASAP. 

"Thomas W. Fredian" wrote:
> 
> Hi Kate,
> 
>     This looks good. I have a few minor corrections. In section c)
> below, the "Data Server" is not driven by MDSplus events at all. The
> other pieces simply connect to the data server and read and/or write
> data and issue events and/or wait for events. A flow chart of each
> piece would be more or less:
> 
> a) Controller
> 
I think the controller may need to startup the MDSplus data server, the
EFIT and visualization processes - see below for my understanding of how
to do this.
>      while (1)
>      {
>          optional:  Wait for operator to push go button
		show mpeg of tokamak control room for a few seconds to simulate
		producing the data.
>          Issue raw_data_available event
>          Wait for efit_data_available event
>          Output efit analysis done message
>          Wait for visualization_done event
>          Output visualization done message
>      }
As I understand this there are probalby two kind of events here. MDSplus
events and then the two events that the controll outputs which need to
be noticed by the InfoViz software and thus may need to be Globus
events. (Or else we just put together some ad hoc notification system
for this)
> 
> b) EFIT computation
> 
>      while (1)
>      {
>          Wait for raw_data_available event
>           Read raw data from mdsplus datafile (via MDSplus server)
>           Compute efit data
>           Write efit data to mdsplus datafile (via MDSplus server)
>           Issue efit_data_available event
>       }
> 
> c) Data server
> 
>       Just running accepting connections and processing requests
> 
> d) Scientific Visualization
> 
>       while (1)
>       {
>           Wait for efit_data_available event
>            Display visualization
>            Issue visualization done message
>        }
> 
> e) Infovis
> 
>     I presume this is just a fancy way of  displaying the progress
> being monitored by the controller.
> 
> 
> Regarding "Tasks, timelines and ownership".
> 
> It would be best to identify the target systems explicitly so we can
> get the demo up and running on the exact systems that will be used for
> the demo.
> 
> There are other issues regarding the actual execution of the demo. Up
> to now I have been logging in to lbl machines and doing a
> grid-proxy-init using my Fusion Collaboratory certificate and running
> the server in one xterm and then logging into the same or other
> machine in a separate window doing the grid-proxy-init and running a
> client application. Normally the server would be fired up by inetd
> when a connection to the port is received. I have modified the mdsip
> server to run as a stand alone process which listens on a port and
> forks servers as connections are received. This eliminates the
> potential problem of doing globus_io on an already connected socket
> (via inetd). We could have the MDSplus server run from inittab with
> respawn but I'll need some help figuring out how a server such as this
> typically does the equivalent of grid-proxy-init at startup.
> 
    We should use the Globus remote job execution process. This works by
the ser (who will call Tom) doing a grid-proxy-init on his workstation
on the showfloor. Then he starts the dataServer remotely on LBL machine
(PortnoyC) using Globus remote job execution which works via the Globus
gatekeeper on PortnoyC. The Globus Gatekeeer gets Tom's proxy
certificate in the usual course of starting a remote job, does an
additional proxy which it hands off (cert and private key) to the
dataServer. This requires adding the Tom to the grid-map-file on
PortnoyC so that the gatekeeper can execute dataServer under Tom's uid. 

Tom also starts up the EFIT program on the ANL node, so that it can be
waiting
for the MDSplus "data entered" event. This should also be started via
globus remote job execution. The EFIT program will have another
Tom_Proxy certificate which it will use when talking to the dataServer
on PortnoyC.

   The client side is started and run on the showfloor workstation,
using the same proxy cert that Tom has already created. The client talks
globus/io to the server who gets the proxy cert and checks it against
its own map file.

One of the implications of this is that the dataServer must recognize
multiple level proxy certificates which at the moment are just plain
X509 identity certs with one or more _Proxy tacked on the end of the CN.
So the procedure is get the CN, dump one or more _Proxy certs and then
use the resulting DN against your map file. (no magic here)

> We should also have a data server running on the SC floor for backup.
> It should be fairly simple to quickly alter the demo to run using a
> different data server.
> 
> -tom
> 

-- 
---------------------------------------------------------------------
Mary R. Thompson				 
Distributed Security Research Group		(510) 486-7408
Lawrence Berkeley National Lab			http://www-itg.lbl.gov/~mrt
----------------------------------------------------------------------

===============================================================================

This message was sent to the SciDAC National Fusion Collaboratory (NFC)
workers list nfc-workers.  Visit the Collaboratory at
.

To unsubscribe from this list, please send a message to
majordomo@fusion.gat.com with the following text in the *body* of the
message:  unsubscribe nfc-workers

David P. Schissel:  

**************************************************************************
**************************************************************************

X-Authentication-Warning: apollo.gat.com: majordom set sender to
owner-nfc-workers@fusion.gat.com using -f Date: Wed, 12 Sep 2001
16:23:44 -0400 From: "Thomas W. Fredian" 
X-Accept-Language: en To: Kate Keahey  CC:
nfc-workers@fusion.gat.com Subject: Re: SC01 telecon summary Sender:
owner-nfc-workers@fusion.gat.com Status:

Hi Kate,

A reminder... We will need IDL installed on the SC floor deskside to run
the IDL-based visualization. In theory we could also implement the
"Controller" using IDL.

Thanks, Tom

Kate Keahey wrote:

All,

Here is the summary as I remember it. Please, send amendments,
clarifications, and violent objections.

1) Status (I don't know that we discussed it much, but should have)

Thanks to Tom for integrating Globus into MDSplus.

2) Evolution of demo description

We are going to have 5 components (only 4 components for the basic
version) interacting using events of the MDSplus framework . The MDSplus
framework has been extended to offer the option of using Globus: GSI,
Globus IO, etc. (to the extend required by this demo). The components
are as follows:

a) Controller that sends and receives MDSplus events in order to
coordinate actions of the other components. The controller executes on a
workstation (an SC01 floor deskside). In the current version the
controller automatically orchestrates the interaction of other demo
components (EFIT, data server, etc.), and is invoked by the user (the
demoee) by invoking the controller executable. In other words, invoking
the controller will make the run happen from beginning to the end. This
is for the basic version. In the fancy version we might complicate that
somewhat (see below)

b) EFIT computation: computes EFIT, in the basic version sequentially,
in the fancy version maybe in parallel. EFIT runs on a Linux machine (or
cluster of Linux machines) at Argonne and is driven by MDSplus framework
events generated by the controller, and generates MDSplus events to
write to the data server.

c) Data Server: provides data for EFIT, stores data produced by EFIT,
provides data for visualization, and generally does data server stuff.
Runs at LBNL. The Data Server accepts MDSplus events from the EFIT
computation, and from the controller and responds to them by providing
or storing data.

d) Scientific visualization component: visualizes data output by EFIT.
It runs on the SC floor deskside (could be a part of the controller
process or done separately whatever Tom thinks is better) and is
triggered by MDSplus events generated by the controller. To do its work
it requests data from the data server. The basic version does an
IDL-based visualization; for the fancy version Chris offered to do
something more visually interesting. Thanks to Chris for chiming in and
offering to do that!

e) Infoviz component: visualizes what happens in the demo. It runs on
the SC floor deskside (could be a part of the controller process or done
separately whatever Tom thinks is better) and is driven by MDSplus
events generated by the controller. This would require the controller to
perform monitoring functions as well as initiate work of the components.
In the basic version we will skip this component, in the fancy version
we are planning to have it. Kate is in charge of keeping track of where
we can get it from.

3) Tasks, timelines and ownership

Here is what we agreed to try to do in the next two weeks:

1) Install the necessary software pieces more or less where they are
going to run:

a) Installation

- LBNL: Data Server, MDSplus  (ownership: Tom) - ANL: MDSplus, EFIT (on
one node for now) (ownership: Tom, David) - Send mail to Kate and Mary
to get accounts for whoever needs them to do this.

b) Interaction Code

- Once the installation has been completed Tom will try to write a
controller for the DataServer and EFIT piece  (I somehow expect more
MDSplus to Globus issues especially as starting up those processes
remotely goes; Tom, please feel free to demand from me bits of example
code and such things as necessary).

- My understanding is that the sequential EFIT is already working with
MDSplus, is this right? If it is, nothing to be done there right now.

- With extensive help from Kate (and probably Mary also), Tom will try
to do a first run of the controller, EFIT, and Data Server pieces.

c) Visualization

- David will provide data for Chris

- Chris will play with the data and try to design a cool visualization
for it.

- Kate will try to find out about the infoviz piece.

d) Target. If all of this happens at this point next week we should have
either a basic demo nearly done (in two pieces), or a very specific list
of things that don't quite work yet. Or both.

3) List of possible "extension features" to demo

a) an evolution of the controller (GUI, start button, let the user
"replay" visualization, other functionality we might want) b) infoviz
controller c) parallel version of EFIT d) more data, less data e)
fancier visualization (as proposed by Chris)

4) Other issues that we didn't get to, but perhaps are not the most
urgent for now:

a) backup issues for SC (what happens if we loose outside connectivity?)
b) testbed issues (when and which Globus version to install on sites
other than LBNL and ANL) c) other advertising issues for SC (flyers,
posters, other stuff)


__________________________ Dr. Kate Keahey Math & Computer Science Div.
Argonne National Laboratory Argonne, IL 60439, USA (630) 252-1673

**************************************************************************
**************************************************************************

X-Authentication-Warning: apollo.gat.com: majordom set sender to
owner-nfc-workers@fusion.gat.com using -f Date: Wed, 12 Sep 2001
15:14:44 -0400 From: "Thomas W. Fredian" 
X-Accept-Language: en To: Kate Keahey  CC:
nfc-workers@fusion.gat.com Subject: Re: SC01 telecon summary Sender:
owner-nfc-workers@fusion.gat.com Status:

Hi Kate,

This looks good. I have a few minor corrections. In section c) below,
the "Data Server" is not driven by MDSplus events at all. The other
pieces simply connect to the data server and read and/or write data and
issue events and/or wait for events. A flow chart of each piece would be
more or less:

a) Controller

while (1) { optional:  Wait for operator to push go button Issue
raw_data_available event Wait for efit_data_available event Output efit
analysis done message Wait for visualization_done event Output
visualization done message }

b) EFIT computation

while (1) { Wait for raw_data_available event Read raw data from mdsplus
datafile (via MDSplus server) Compute efit data Write efit data to
mdsplus datafile (via MDSplus server) Issue efit_data_available event }

c) Data server

Just running accepting connections and processing requests

d) Scientific Visualization

while (1) { Wait for efit_data_available event Display visualization
Issue visualization done message }

e) Infovis

I presume this is just a fancy way of  displaying the progress being
monitored by the controller.


Regarding "Tasks, timelines and ownership".

It would be best to identify the target systems explicitly so we can get
the demo up and running on the exact systems that will be used for the
demo.

There are other issues regarding the actual execution of the demo. Up to
now I have been logging in to lbl machines and doing a grid-proxy-init
using my Fusion Collaboratory certificate and running the server in one
xterm and then logging into the same or other machine in a separate
window doing the grid-proxy-init and running a client application.
Normally the server would be fired up by inetd when a connection to the
port is received. I have modified the mdsip server to run as a stand
alone process which listens on a port and forks servers as connections
are received. This eliminates the potential problem of doing globus_io
on an already connected socket (via inetd). We could have the MDSplus
server run from inittab with respawn but I'll need some help figuring
out how a server such as this typically does the equivalent of
grid-proxy-init at startup.

We should also have a data server running on the SC floor for backup. It
should be fairly simple to quickly alter the demo to run using a
different data server.

-tom


Kate Keahey wrote:

All,

Here is the summary as I remember it. Please, send amendments,
clarifications, and violent objections.

1) Status (I don't know that we discussed it much, but should have)

Thanks to Tom for integrating Globus into MDSplus.

2) Evolution of demo description

We are going to have 5 components (only 4 components for the basic
version) interacting using events of the MDSplus framework . The MDSplus
framework has been extended to offer the option of using Globus: GSI,
Globus IO, etc. (to the extend required by this demo). The components
are as follows:

a) Controller that sends and receives MDSplus events in order to
coordinate actions of the other components. The controller executes on a
workstation (an SC01 floor deskside). In the current version the
controller automatically orchestrates the interaction of other demo
components (EFIT, data server, etc.), and is invoked by the user (the
demoee) by invoking the controller executable. In other words, invoking
the controller will make the run happen from beginning to the end. This
is for the basic version. In the fancy version we might complicate that
somewhat (see below)

b) EFIT computation: computes EFIT, in the basic version sequentially,
in the fancy version maybe in parallel. EFIT runs on a Linux machine (or
cluster of Linux machines) at Argonne and is driven by MDSplus framework
events generated by the controller, and generates MDSplus events to
write to the data server.

c) Data Server: provides data for EFIT, stores data produced by EFIT,
provides data for visualization, and generally does data server stuff.
Runs at LBNL. The Data Server accepts MDSplus events from the EFIT
computation, and from the controller and responds to them by providing
or storing data.

d) Scientific visualization component: visualizes data output by EFIT.
It runs on the SC floor deskside (could be a part of the controller
process or done separately whatever Tom thinks is better) and is
triggered by MDSplus events generated by the controller. To do its work
it requests data from the data server. The basic version does an
IDL-based visualization; for the fancy version Chris offered to do
something more visually interesting. Thanks to Chris for chiming in and
offering to do that!

e) Infoviz component: visualizes what happens in the demo. It runs on
the SC floor deskside (could be a part of the controller process or done
separately whatever Tom thinks is better) and is driven by MDSplus
events generated by the controller. This would require the controller to
perform monitoring functions as well as initiate work of the components.
In the basic version we will skip this component, in the fancy version
we are planning to have it. Kate is in charge of keeping track of where
we can get it from.

3) Tasks, timelines and ownership

Here is what we agreed to try to do in the next two weeks:

1) Install the necessary software pieces more or less where they are
going to run:

a) Installation

- LBNL: Data Server, MDSplus  (ownership: Tom) - ANL: MDSplus, EFIT (on
one node for now) (ownership: Tom, David) - Send mail to Kate and Mary
to get accounts for whoever needs them to do this.

b) Interaction Code

- Once the installation has been completed Tom will try to write a
controller for the DataServer and EFIT piece  (I somehow expect more
MDSplus to Globus issues especially as starting up those processes
remotely goes; Tom, please feel free to demand from me bits of example
code and such things as necessary).

- My understanding is that the sequential EFIT is already working with
MDSplus, is this right? If it is, nothing to be done there right now.

- With extensive help from Kate (and probably Mary also), Tom will try
to do a first run of the controller, EFIT, and Data Server pieces.

c) Visualization

- David will provide data for Chris

- Chris will play with the data and try to design a cool visualization
for it.

- Kate will try to find out about the infoviz piece.

d) Target. If all of this happens at this point next week we should have
either a basic demo nearly done (in two pieces), or a very specific list
of things that don't quite work yet. Or both.

3) List of possible "extension features" to demo

a) an evolution of the controller (GUI, start button, let the user
"replay" visualization, other functionality we might want) b) infoviz
controller c) parallel version of EFIT d) more data, less data e)
fancier visualization (as proposed by Chris)

4) Other issues that we didn't get to, but perhaps are not the most
urgent for now:

a) backup issues for SC (what happens if we loose outside connectivity?)
b) testbed issues (when and which Globus version to install on sites
other than LBNL and ANL) c) other advertising issues for SC (flyers,
posters, other stuff)


__________________________ Dr. Kate Keahey Math & Computer Science Div.
Argonne National Laboratory Argonne, IL 60439, USA (630) 252-1673


**************************************************************************
**************************************************************************

X-Authentication-Warning: apollo.gat.com: majordom set sender to
owner-nfc-workers@fusion.gat.com using -f X-Sender: keahey@localhost
Date: Wed, 12 Sep 2001 13:23:22 -0500 To: nfc-workers@fusion.gat.com
From: Kate Keahey  Subject: SC01 telecon summary
Sender: owner-nfc-workers@fusion.gat.com Status:

All,

Here is the summary as I remember it. Please, send amendments,
clarifications, and violent objections.

1) Status (I don't know that we discussed it much, but should have)

Thanks to Tom for integrating Globus into MDSplus.

2) Evolution of demo description

We are going to have 5 components (only 4 components for the basic
version) interacting using events of the MDSplus framework . The MDSplus
framework has been extended to offer the option of using Globus: GSI,
Globus IO, etc. (to the extend required by this demo). The components
are as follows:

a) Controller that sends and receives MDSplus events in order to
coordinate actions of the other components. The controller executes on a
workstation (an SC01 floor deskside). In the current version the
controller automatically orchestrates the interaction of other demo
components (EFIT, data server, etc.), and is invoked by the user (the
demoee) by invoking the controller executable. In other words, invoking
the controller will make the run happen from beginning to the end. This
is for the basic version. In the fancy version we might complicate that
somewhat (see below)

b) EFIT computation: computes EFIT, in the basic version sequentially,
in the fancy version maybe in parallel. EFIT runs on a Linux machine (or
cluster of Linux machines) at Argonne and is driven by MDSplus framework
events generated by the controller, and generates MDSplus events to
write to the data server.

c) Data Server: provides data for EFIT, stores data produced by EFIT,
provides data for visualization, and generally does data server stuff.
Runs at LBNL. The Data Server accepts MDSplus events from the EFIT
computation, and from the controller and responds to them by providing
or storing data.

d) Scientific visualization component: visualizes data output by EFIT.
It runs on the SC floor deskside (could be a part of the controller
process or done separately whatever Tom thinks is better) and is
triggered by MDSplus events generated by the controller. To do its work
it requests data from the data server. The basic version does an
IDL-based visualization; for the fancy version Chris offered to do
something more visually interesting. Thanks to Chris for chiming in and
offering to do that!

e) Infoviz component: visualizes what happens in the demo. It runs on
the SC floor deskside (could be a part of the controller process or done
separately whatever Tom thinks is better) and is driven by MDSplus
events generated by the controller. This would require the controller to
perform monitoring functions as well as initiate work of the components.
In the basic version we will skip this component, in the fancy version
we are planning to have it. Kate is in charge of keeping track of where
we can get it from.

3) Tasks, timelines and ownership

Here is what we agreed to try to do in the next two weeks:

1) Install the necessary software pieces more or less where they are
going to run:

a) Installation

- LBNL: Data Server, MDSplus  (ownership: Tom) - ANL: MDSplus, EFIT (on
one node for now) (ownership: Tom, David) - Send mail to Kate and Mary
to get accounts for whoever needs them to do this.

b) Interaction Code

- Once the installation has been completed Tom will try to write a
controller for the DataServer and EFIT piece  (I somehow expect more
MDSplus to Globus issues especially as starting up those processes
remotely goes; Tom, please feel free to demand from me bits of example
code and such things as necessary).

- My understanding is that the sequential EFIT is already working with
MDSplus, is this right? If it is, nothing to be done there right now.

- With extensive help from Kate (and probably Mary also), Tom will try
to do a first run of the controller, EFIT, and Data Server pieces.

c) Visualization

- David will provide data for Chris

- Chris will play with the data and try to design a cool visualization
for it.

- Kate will try to find out about the infoviz piece.

d) Target. If all of this happens at this point next week we should have
either a basic demo nearly done (in two pieces), or a very specific list
of things that don't quite work yet. Or both.

3) List of possible "extension features" to demo

a) an evolution of the controller (GUI, start button, let the user
"replay" visualization, other functionality we might want) b) infoviz
controller c) parallel version of EFIT d) more data, less data e)
fancier visualization (as proposed by Chris)

4) Other issues that we didn't get to, but perhaps are not the most
urgent for now:

a) backup issues for SC (what happens if we loose outside connectivity?)
b) testbed issues (when and which Globus version to install on sites
other than LBNL and ANL) c) other advertising issues for SC (flyers,
posters, other stuff)


__________________________ Dr. Kate Keahey                   Math &
Computer Science Div. Argonne National Laboratory Argonne, IL 60439, USA
(630) 252-1673

**************************************************************************
**************************************************************************

X-Sender: keahey@localhost
Date: Tue, 11 Sep 2001 14:19:47 -0500
To: "David P. Schissel" 
From: Kate Keahey 
Subject: Re: SC01 Fusion demo
Status:   

OK. I am somewhat concerned that the project will keep being perceived
as two different projects. I am not sure if this is good or bad. In
general things will be good as long as we avoid last minute pre-SC01
panic.


At 03:45 PM 9/10/2001 -0800, you wrote:

 Kate -

 The feeling i got was that this demo was not going to have a  new
Visualization component. We were not planning something on a grander scale.
So, this demo, was just the security/remote show with a visualization
one to follow at a later date and time.

  But, let us discuss this Wednesday morning during our phone call.

 - david


What I was trying to ask is whether for the SC01 demo we should plan to
go with the IDL-based visualization that we were agreed on at the
August/Chicago meeting or should we plan for something grander? Or were
there perhaps other visualization ideas that would affect the
remote/security demo in some way? Or are there going to be two separate
demos for this project and we should plan on coming up with a unifying
story in the future?


At 03:20 PM 9/10/2001 -0800, David P. Schissel wrote:

  Chris -

 My memory of the SF discussion was that our data was simple enough
that there was no need for a fancy visualization effort. This fact
then raised your concern that our demo would be a bit on the boring
side. From there we moved into the discussion of how to tell a
compelling story, to motivate the demonstration, to try to make
it interesting. I need to do some work here - be a story teller.

  What we were going to do to show the data was a simple IDL based
animation of the plasma (something we can do today). We were
going to try and combine that with a map of the US showing how
the data and executions were traveling between LBNL, ANL, and
the Denver. This was something that someone at the meeting (Rick?)
thought already existed.

 That is my recollection and what my notes tell me.

 - dave


Hi Kate,

Yes, we did discuss this at HPDC.  I think we were going to try and do
something simple (since we aren't going to have much in the way of new
research done by then).
We discussed several possibilities, but I don't think we converged on a
specific demo yet.

David - do you remember any specifics?

Cheers,

Chris

Kate Keahey writes:
 > Chris,
 >
 > I seem to recall (and I may be wrong it was rather early in the day ;-) you
 > telling me at HPDC that you were planning a visualization demo for the
 > Fusion project for SC01. Did I get that right, and if I did what exactly
 > were you planning and how is it going to be related to the "remote
 > computation" demo? Many thanks!
 >
 >
 > __________________________
 > Dr. Kate Keahey
 > Math & Computer Science Div.
 > Argonne National Laboratory
 > Argonne, IL 60439, USA
 > (630) 252-1673

__________________________
Dr. Kate Keahey
Math & Computer Science Div.
Argonne National Laboratory
Argonne, IL 60439, USA
(630) 252-1673


__________________________
Dr. Kate Keahey
Math & Computer Science Div.
Argonne National Laboratory
Argonne, IL 60439, USA
(630) 252-1673

**************************************************************************
**************************************************************************

Date: Mon, 10 Sep 2001 15:57:51 -0800
To: Chris Johnson 
From: "David P. Schissel" 
Subject: Re: SC01 Fusion demo
Cc: keahey@mcs.anl.gov, schissel
Status:   


  Chris -

 The data we have is so simple. We make contour plots of
constant magnetic flux. See the PDF file
http://Fusion.gat.com/comp/analysis/publications/Schissel-softvgs.pdf
and look at page 18.

 The plot on the right hand side is a plot of the magnetic equilibrium
of the plasma that is calculated by the code EFIT (which is the code
we are demonstrating at SC2001). In time, the shape
of the plasma evolves. So the animation would be the growing evolution
of this contour plot.

 Now, what could be done is you take this information, and spin it
into the shape of the torus (we can supply the major radius). We
might then color code the field lines to temperature, and then the
animation would be a 3dish type thing of the plasma shape and
temperature evolution.

  You can get an idea of what I am talking about by looking at the
PDF file
http://Fusion.gat.com/comp/analysis/publications/schussma.pdf

 This is Greg Schussman's (UC Davis) IEEE Vis 2000 paper. Color
Plate 1 gives you a feeling of the torus shape.
Color Plate 3 gives you an idea of the view in 3D. Note, that
this paper was done for a different problem even though it looks
at magnetic field lines. Magnetic Islands are a special case that
the EFIT code does not calculate.

  Does this make any sense via email?

 -David


Hi,

What kind of plasma data do you have?  Do you have large-scale
time-dependent scalar (and/or vector) data sets?  If so, we might be able
to use our real-time ray tracer and make some nice images (either live or
record them).

Cheers,

Chris

David P. Schissel writes:
 >
 >    Chris -
 >
 >   My memory of the SF discussion was that our data was simple enough
 > that there was no need for a fancy visualization effort. This fact
 > then raised your concern that our demo would be a bit on the boring
 > side. From there we moved into the discussion of how to tell a
 > compelling story, to motivate the demonstration, to try to make
 > it interesting. I need to do some work here - be a story teller.
 >
 >    What we were going to do to show the data was a simple IDL based
 > animation of the plasma (something we can do today). We were
 > going to try and combine that with a map of the US showing how
 > the data and executions were traveling between LBNL, ANL, and
 > the Denver. This was something that someone at the meeting (Rick?)
 > thought already existed.
 >
 >   That is my recollection and what my notes tell me.
 >
 >   - dave
 >
 >
 >
 > >Hi Kate,
 > >
 > >Yes, we did discuss this at HPDC.  I think we were going to try and do
 > >something simple (since we aren't going to have much in the way of new
 > >research done by then).
 > >
 > >We discussed several possibilities, but I don't think we converged on a
 > >specific demo yet.
 > >
 > >David - do you remember any specifics?
 > >
 > >Cheers,
 > >
 > >Chris
 > >
 > >Kate Keahey writes:
 > >  > Chris,
 > >  >
 > >  > I seem to recall (and I may be wrong it was rather early in the day ;-) you
 > >  > telling me at HPDC that you were planning a visualization demo for the
 > >  > Fusion project for SC01. Did I get that right, and if I did what exactly
 > >  > were you planning and how is it going to be related to the "remote
 > >  > computation" demo? Many thanks!
 > >  >
 > >  >
 > >  > __________________________
 > >  > Dr. Kate Keahey
 > >  > Math & Computer Science Div.
 > >  > Argonne National Laboratory
 > >  > Argonne, IL 60439, USA
 > >  > (630) 252-1673
 >

**************************************************************************
**************************************************************************

X-Sender: keahey@localhost Date: Mon, 10 Sep 2001 17:41:08 -0500 To:
"David P. Schissel" , Chris Johnson
 From: Kate Keahey  Subject: Re:
SC01 Fusion demo Cc: schissel@fusion.gat.com Status:

What I was trying to ask is whether for the SC01 demo we should plan to
go with the IDL-based visualization that we were agreed on at the
August/Chicago meeting or should we plan for something grander? Or were
there perhaps other visualization ideas that would affect the
remote/security demo in some way? Or are there going to be two separate
demos for this project and we should plan on coming up with a unifying
story in the future?


At 03:20 PM 9/10/2001 -0800, David P. Schissel wrote:

Chris -

My memory of the SF discussion was that our data was simple enough that
there was no need for a fancy visualization effort. This fact then
raised your concern that our demo would be a bit on the boring side.
From there we moved into the discussion of how to tell a compelling
story, to motivate the demonstration, to try to make it interesting. I
need to do some work here - be a story teller.

What we were going to do to show the data was a simple IDL based
animation of the plasma (something we can do today). We were going to
try and combine that with a map of the US showing how the data and
executions were traveling between LBNL, ANL, and the Denver. This was
something that someone at the meeting (Rick?) thought already existed.

That is my recollection and what my notes tell me.

- dave



Hi Kate,

Yes, we did discuss this at HPDC.  I think we were going to try and do
something simple (since we aren't going to have much in the way of new
research done by then). We discussed several possibilities, but I don't
think we converged on a specific demo yet.

David - do you remember any specifics?

Cheers,

Chris

Kate Keahey writes: > Chris, > > I seem to recall (and I may be wrong it
was rather early in the day ;-) you > telling me at HPDC that you were
planning a visualization demo for the > Fusion project for SC01. Did I
get that right, and if I did what exactly > were you planning and how is
it going to be related to the "remote > computation" demo? Many thanks!
> > > __________________________ > Dr. Kate Keahey > Math & Computer
Science Div. > Argonne National Laboratory > Argonne, IL 60439, USA >
(630) 252-1673


__________________________ Dr. Kate Keahey Math & Computer Science Div.
Argonne National Laboratory Argonne, IL 60439, USA (630) 252-1673

**************************************************************************
**************************************************************************

X-Authentication-Warning: apollo.gat.com: majordom set sender to
owner-nfc-workers@fusion.gat.com using -f X-Sender: keahey@localhost
Date: Mon, 10 Sep 2001 14:39:36 -0500 To: nfc-workers@fusion.gat.com
From: Kate Keahey  Subject: telecon on Wednesday
Sender: owner-nfc-workers@fusion.gat.com Status:

All,

I have set up a telecon to discuss SC01 demo issues for Wednesday, Sept
12, at 11:00am CST. We have 8 lines for the duration of one hour. If you
want to participate but have not sent me mail, please do so, so that I
can make sure that this is enough.

The 8 lines should call into: 1-800-857-6556. The pass code is 21465.
Talk to you then!


__________________________ Dr. Kate Keahey Math & Computer Science Div.
Argonne National Laboratory Argonne, IL 60439, USA (630) 252-1673


===============================================================================

This message was sent to the SciDAC National Fusion Collaboratory (NFC)
workers list nfc-workers.  Visit the Collaboratory at
.

To unsubscribe from this list, please send a message to
majordomo@fusion.gat.com with the following text in the *body* of the
message:  unsubscribe nfc-workers

David P. Schissel: 



**************************************************************************
**************************************************************************

X-Authentication-Warning: apollo.gat.com: majordom set sender to
owner-nfc-workers@fusion.gat.com using -f Date: Sat, 08 Sep 2001
17:48:42 -0700 From: Mary Thompson  Organization:
LBNL X-Accept-Language: en To: Kate Keahey  CC:
nfc-workers@fusion.gat.com Subject: Re: SC01 demo tasks Sender:
owner-nfc-workers@fusion.gat.com Status:

Kate, 9am Wedensday works for me for a telcon.

Dave, LBL's target size for posters is about 32X48" in a portrait
aspect. I suspect the printer they used last year had a 32" width. So
this is an approximate measure. We will have room for a generic Fusion
Collab poster (as long as it mentions LBNL, or course). If you do one on
heavy paper that can  be mailed in a tube, I would assume that I could
just "borrow" it and you could use it at APS or other places. We are
also panning a DOE Science Grid poster where the top part will be about
the 4 nodes of the Grid and at the bottom we will have some eye-catching
Cosmology and Fusion graphics with the message "Science on the Grid". We
also have some space behind the demo monitors and are thinking about
having something 11x17" that describe the actual demo. So the
"compelling story" could  go there.

Mary --
---------------------------------------------------------------------
Mary R. Thompson				 Distributed
Security Research Group		(510) 486-7408 Lawrence Berkeley National
Lab			http://www-itg.lbl.gov/~mrt
----------------------------------------------------------------------

===============================================================================

This message was sent to the SciDAC National Fusion Collaboratory (NFC)
workers list nfc-workers.

To unsubscribe from this list, please send a message to
majordomo@fusion.gat.com with the following text in the *body* of the
message:  unsubscribe nfc-workers

David P. Schissel: 


**************************************************************************
**************************************************************************

X-Authentication-Warning: apollo.gat.com: majordom set sender to
owner-nfc-workers@fusion.gat.com using -f X-Sender: keahey@localhost
Date: Fri, 07 Sep 2001 20:26:22 -0500 To: nfc-workers@fusion.gat.com
From: Kate Keahey  Subject: SC01 demo tasks Sender:
owner-nfc-workers@fusion.gat.com Status:

I propose that we consider the demo in stages; the first stage would be
very basic, enough to get the demo to run and make a story around it.
Once we are done with that, we can add more complex features such as:
EFIT executing in parallel, or reading/writing more data, and other
features that would make the application more realistic.

Below is a list of some tasks to accomplish for the basic version (based
on results of the August meeting as described in David's document).
During the telecon I would like to fix a list of features and ownership
for each task, and come up with some deadlines for their completion so
we can track our progress. I think it would probably be easiest if I try
to write a "demo skeleton" with interactions of some placeholder
"components" that we will then be able to replace by the EFIT, the Data
Server and the GUI (but I am open to other suggestions). Based on this
assumption we have the following tasks.

1) Globus interface to the MDSplus DataServer. Tom has been working on
it quite intensively for a while now. I am not sure I understand to what
extent we want to tackle security issues for the demo, what is realistic
and so forth...

2) Visualization component: this I understand still is the basic
IDL-based visualization that we were talking about at the meeting. Are
there any other plans? Who has the ownership of this piece?

3) A basic "demo framework"; a scaled-down 3 component interactive
scenario containing more or less all the functionality in the picture
from our document. The idea is that this would be something we could
integrate all the pieces into. I am working on it.

4) GUI. Who has the ownership of this piece?

5) Integration of MDSplus into the "demo framework".

6) Integration of EFIT into the framework.

7) Integration of GUI into the framework.

8) Testing, extensions, etc.

So far it looks like Wednesday 11 CST (9 Pacific, 12 Eastern) would be
good bet, is this OK for everybody interested? Please let me know if
this is bad for you, I would like to schedule the telecon Monday
morning.




__________________________ Dr. Kate Keahey Math & Computer Science Div.
Argonne National Laboratory Argonne, IL 60439, USA (630) 252-1673


===============================================================================

This message was sent to the SciDAC National Fusion Collaboratory (NFC)
workers list nfc-workers.

To unsubscribe from this list, please send a message to
majordomo@fusion.gat.com with the following text in the *body* of the
message:  unsubscribe nfc-workers

David P. Schissel: 



**************************************************************************
**************************************************************************


X-Authentication-Warning: apollo.gat.com: majordom set sender to
owner-nfc-workers@fusion.gat.com using -f X-Sender: keahey@localhost
Date: Fri, 07 Sep 2001 10:12:21 -0500 To: nfc-workers@fusion.gat.com
From: Kate Keahey  Subject: SCO1 demo telecon
Sender: owner-nfc-workers@fusion.gat.com Status:

All,

I would like to call a telecon for next week for anybody
working/interested in the SC01 demo. The purpose of the telecon is to
fill in some more detail into the description David sent out earlier and
to identify some concrete tasks and deadlines that we have to hit for
the demo, and discuss their dependencies. Tom has been making a lot of
progress globizing MDSplus, but there is still a lot of work that needs
to be done on other components. I will send out a list of tasks later
today so that we have something to disagree with ;-).

Time: if you would like to participate, please send me a note by say COB
today with your time constraints. I am good all next week except Monday.


__________________________ Dr. Kate Keahey Math & Computer Science Div.
Argonne National Laboratory Argonne, IL 60439, USA (630) 252-1673


===============================================================================

This message was sent to the SciDAC National Fusion Collaboratory (NFC)
workers list nfc-workers.

To unsubscribe from this list, please send a message to
majordomo@fusion.gat.com with the following text in the *body* of the
message:  unsubscribe nfc-workers

David P. Schissel: 


**************************************************************************
**************************************************************************

about the fusion grid | fusiongrid research Last modified 09/25/01. Comments? webmaster