Up to top level
AO15   AO16   AO17   AO18   Backgrounds   Calibration   Conference   Data   Docs   EPICMOS   EPICpn   Feedback   Gallery   Misc   OM   Pending   PhD_Theses   Publications   RGS   RadMonitor   SAS_Hardware   SAS_WS   SASv16.0   SASv16.0_Installation   SASv16.1   SASv16.1_Installation   SASv17.0   SASv17.0_Installation   SciSim   Simulators_other   Suggestions   Trash   Visibility   XMM-bouncing   XMM-news   XRPS   XSA   esas   incoming  

Logged in as guest

Viewing XRPS/9176
Full headers

From: dks@pha.jhu.edu
Subject: Get rid of XRPS
Compose reply
Download message
Move To:
3 replies: 1 2 3
2 followups: 1 2

Private message: yes  no

Notes:

Notification:


Date: Wed, 17 Sep 2003 17:33:41 GMT
From: dks@pha.jhu.edu
To: xmmhelp@xmm.vilspa.esa.es
CC: dks@pha.jhu.edu
Subject: Get rid of XRPS
Full_Name: Dave Strickland
Submission from: (NULL) (128.220.233.91)


I've just wasted more than 2 hours trying to log into XRPS while its been timing
out or 
giving me typically useless Microsoft error pages (the person who suggested 
you use Microsoft deserves to be fired). XRPS is a total disaster, and has been
from AO1. XRPS is totally unacceptable in terms of performance, ease of use, 
robustness, etc. etc.  Get rid of it.


Reply 1

Resend
From: Matteo Guainazzi <xmmhelp@xmm.vilspa.esa.es>
To: dks@pha.jhu.edu
Subject: Re: Get rid of XRPS (PR#9176)
Date: Thu Sep 18 07:42:17 2003
Dear Dr. Strickland,

 we greatly appreciate your constructuve suggestions on the XMM-Newton proposal
submission process. We will take them in careful consideration for future
XMM-Newton Announcement of Opportunity.

 Regards,

 Matteo Guainazzi

P.S.: if you would provide us with a slightly more technical description of the
problems you have encountered, we may still be in the condition to provide you
with some more practical help.


Followup 1

Compose reply
Download message
Subject: Re: Get rid of XRPS (PR#9176)
From: Dave Strickland <dks@proteus.pha.jhu.edu>
To: Matteo Guainazzi <xmmhelp@xmm.vilspa.esa.es>
Cc: Dave Strickland <dks@pha.jhu.edu>
Date: 18 Sep 2003 10:15:08 -0400
On Thu, 2003-09-18 at 03:42, Matteo Guainazzi wrote:
> Dear Dr. Strickland,
> 
>  we greatly appreciate your constructuve suggestions on the XMM-Newton
proposal
> submission process. We will take them in careful consideration for future
> XMM-Newton Announcement of Opportunity.

Getting rid of XRPS was, in fact and in all honesty, a constructive
suggestion. Reusing the same system that caused so many problems in AO1
and AO2 with only minor optimizations ("Network and firewall approaches
have been adapted, hardware performance has been enhanced, and extensive
efficiencies have been achieved in software implementation.") without
removing the real bloated performance hog (XRPS itself) aint going to do
much. That said, the more technical description of problems you wanted
follows below.

>  Regards,
> 
>  Matteo Guainazzi
> 
> P.S.: if you would provide us with a slightly more technical description of
the
> problems you have encountered, we may still be in the condition to provide
you
> with some more practical help.

I go to the XRPS main page. I click the "Load Proposal" button. I
wait... some minutes later the browser reports a time out. I try again.

I try again. Some thing happens. 

Repeat ad nauseum for several hours.

On one or two occasions instead of a time out I get a very uninformative
Microsoft error page.

I tried this using 3 different browsers - Mozilla and Konqueror under
Linux, and Netscape 4.7 on an ancient Sun I manged to find lying around.
Exact same symptoms with all of them. I'd even followed the browser
cache etc etc setup info the XRPS page suggests.

Does that help?

There were ~950 observations accepted, and lets assume that there are
typically 2-3 observations per accepted proposal. Thats approx 400
people who need to do phase II. You've divied them into 3 bins, say 120
people in this weeks bin. Lets assume they all tried connected the same
time I did yesterday. A server that can handle 120 concurrent
connections? Its a joke, even with the java bloatware considered - and
wasn't XRPS outsourced to a commercial software vendor? And you probably
can't fix the code yourselves now either. Contrast the performance to
the Chandra proposal and observation planning system which works fine
and uses simple non-commercial tools. Sure, they have more money, but
they accurately assessed their needs and asked for what was in reality
needed. The fundamental problems are at the upper management level, the
same complaints that have followed XMM around since launch.
-- 
----------------------------------------------------------------------
Dave Strickland	                | E-mail: dks@pha.jhu.edu
Dept. of Physics & Astronomy,   | WWW: http://proteus.pha.jhu.edu/~dks/
The Johns Hopkins University,   | Tel: 410-516-2881
Baltimore, MD 21218-2686, USA   | Fax: 410-516-5096
----------------------------------------------------------------------



Reply 2

Resend
From: Matteo Guainazzi <xmmhelp@xmm.vilspa.esa.es>
To: dks@pha.jhu.edu
Subject: Re: Get rid of XRPS (PR#9176)
Date: Fri Sep 19 07:50:22 2003
Dear Dave,

 thanks for your detailed description of the indeed annoying problem you have
encountered using the XRPS. I have conveyed it to our Software Support Team. I
will come back to you as soon as their analysis is available.

 Of course I don't dare hoping to change your robust and consolidated opinion on
the XMM-Newton project, neither I am being paid for that. As you surely now, an
XMM-Newton User Group was set up with the exact scope to convey complaints and
suggestions to the project. The conclusions of this board are public
(http://xmm.vilspa.esa.es/external/xmm_user_support/usersgroup/index.shtml), and
slightly disagree with your general sharp judgement. I warmly encourage you to
make use of this official board in the future. Anyhow, simply for the due
respect to the naked facts, to which all of us are committed, I feel compelled
to correct some statements in your last e-mail, which do not reflect the reality
as it is.

1. each XMM-Newton observer has had at least 6 weeks (not 1) to complete her/his
2nd Phase submission. This changes your estimate of the workload associated with
the proposal submission by a significant factor, not taking into account that 20
users in a week bin means, on the average, 3 users per day

2. indeed, about 70% of the AO3 proposals had already been submitted at the end
of last week (now they are probably something more). Although some quirks have
been discovered in the course of the process (cf.
http://xmmrps.vilspa.esa.es/wphrp_reports.asp), the process has been in general
rather smooth (I regret to stress that you are the first user who is asking SOC
to fire somebody in relation to the XRPS). The next meeting of the XMM-Newton
Users' Group is scheduled for September 22-23. It could be a good chance for
unsatisfied users to convey their complaints, and turn down our view of the
process, if too positive

3. the XRPS code is under our full (SOC) control, and therefore can be fixed in
whatever moment

4. indeed, the AO3 XRPS is **not** the same system as the AO1 or (above all)
AO2, even if the layout was kept as similar as possible to the previous
Announcements to help the final users. Rightful criticisms on those system do
not automatically apply to the AO3 one, although of course new may apply

 Said that, I hope to come back with a practical solution for your troubles as
soon as possible.

 Regards, Matteo


Reply 3

Resend
From: Matteo Guainazzi <xmmhelp@xmm.vilspa.esa.es>
To: dks@pha.jhu.edu
Subject: Re: Get rid of XRPS (PR#9176)
Date: Fri Sep 19 16:12:06 2003
Dear david,

 our Software Support Team has given a deeper look at the problem you have
encountered in connecting with the XRPS. It is indeed puzzling. The time out
problems you refer to happen not only when you load "dynamical" URLs, but even
with simple "static" pages. In other word: the reason for the slow connection
seems to have less to do with XRPS, the proposal database or the Java Virtual
Machine, than with the speed of the Internet connectivity to VILSPA, as pages
accessed prior to XRPS launch seems to be problematic for you to load as well.
This is confirmed indirectly by the fact that you encounter the same problems
with a large spectrum of different platforms.

 What about your connection speed to a "normal" XMM-Newton web page (e.g.:
"http://xmm.vilspa.esa.es")? Have you noticed any time-dependency in the speed
of your network connection to VILSPA/Spain?

 I am sorry if my explanation is so far of little practical use for you.
However, from our analysis there is little which looks attributable to the XRPS
as such. We are collecting elements to find a reasonable clue. XRPS has been
already used by ~80% of XMM-Newton AO3 PIs during the Phase II, and has coped so
far with a maximum load of 20 simultaneous users without any appreciable loss of
performance, as far as we know. 

 Regards, Matteo


 

 


Followup 2

Compose reply
Download message
Subject: Re: Get rid of XRPS (PR#9176)
From: Dave Strickland <dks@proteus.pha.jhu.edu>
To: Matteo Guainazzi <xmmhelp@xmm.vilspa.esa.es>
Date: 19 Sep 2003 12:19:50 -0400
Thanks for looking into it - today I can actually log on and confirm
that XRPS is indeed friendlier than its earlier incarnations.

At the time I didn't notice any appreciable time lags while surfing the
main xmm pages and xmm handbook.

Have a pleasent weekend.
Dave

On Fri, 2003-09-19 at 12:12, Matteo Guainazzi wrote:
> Dear david,
> 
>  our Software Support Team has given a deeper look at the problem you have
> encountered in connecting with the XRPS. It is indeed puzzling. The time
out
> problems you refer to happen not only when you load "dynamical" URLs, but
even
> with simple "static" pages. In other word: the reason for the slow
connection
> seems to have less to do with XRPS, the proposal database or the Java
Virtual
> Machine, than with the speed of the Internet connectivity to VILSPA, as
pages
> accessed prior to XRPS launch seems to be problematic for you to load as
well.
> This is confirmed indirectly by the fact that you encounter the same
problems
> with a large spectrum of different platforms.
> 
>  What about your connection speed to a "normal" XMM-Newton web page (e.g.:
> "http://xmm.vilspa.esa.es")? Have you noticed any time-dependency in the
speed
> of your network connection to VILSPA/Spain?
> 
>  I am sorry if my explanation is so far of little practical use for you.
> However, from our analysis there is little which looks attributable to the
XRPS
> as such. We are collecting elements to find a reasonable clue. XRPS has
been
> already used by ~80% of XMM-Newton AO3 PIs during the Phase II, and has
coped so
> far with a maximum load of 20 simultaneous users without any appreciable
loss of
> performance, as far as we know. 
> 
>  Regards, Matteo
> 
> 
>  
-- 
Dave Strickland <dks@proteus.pha.jhu.edu>


Up to top level
AO15   AO16   AO17   AO18   Backgrounds   Calibration   Conference   Data   Docs   EPICMOS   EPICpn   Feedback   Gallery   Misc   OM   Pending   PhD_Theses   Publications   RGS   RadMonitor   SAS_Hardware   SAS_WS   SASv16.0   SASv16.0_Installation   SASv16.1   SASv16.1_Installation   SASv17.0   SASv17.0_Installation   SciSim   Simulators_other   Suggestions   Trash   Visibility   XMM-bouncing   XMM-news   XRPS   XSA   esas   incoming  

Logged in as guest


Please make your (short) question the subject of your request!


Web interface using JitterBug ... back to the XMM home page