|
I am so new to the BOINC environment I am not sure if I am simply doing something wrong. Please let me desribe what I am trying to do and then perhaps someone can tell me if I am approaching it in the right ways.
I have five home machines on a LAN
Two machines are "fast" using WinXPPro
Three machines are "slow" using Win2KPro
I want the fast machines to run BOINC with both S&H and CP
I want the slow machines to run BOINC with just S&H
I want to run modified preferenes on all fast machines, so I "Add separate General preferences for work," set "Leave applications in memory while preempted?" to Yes and call these "Work" machines. I do this on both S&H and CP project sites.
I want to run default preferences on all slow machines, so I leave "Leave applications in memory while preempted?" set to No and call these "Home" machines. I do this on the S&H project site.
I set the fast machines' "Location" to Work.
I set the slow machines' "Location" to Home.
I select "Update" on all projects on all machines, fast and slow.
I did this all last night, but this morning I noticed the following in my message log:
climateprediction.net - 2004-09-19 07:31:40 - Restarting result 2p90_100147310_0 using hadsm3 version 4.03
SETI@home - 2004-09-19 07:31:40 - Pausing result 26ap04aa.25756.18000.153414.58_2 (removed from memory)
This told me my "Leave applications in memory while preempted?" setting had not stuck.
When I went back to my S&H and CP web sites, I found all Work locations have been reset to Home.
So I started over, reset the site "Locations" and again "Updated" all of my projects.
On my fast machines, I receive the following messages:
SETI@home - 2004-09-19 10:57:06 - Sending request to scheduler: http://setiboinc.ssl.berkeley.edu/sah_cgi/cgi
SETI@home - 2004-09-19 10:57:09 - Scheduler RPC to http://setiboinc.ssl.berkeley.edu/sah_cgi/cgi succeeded
climateprediction.net - 2004-09-19 10:58:28 - Sending request to scheduler: http://climateapps2.oucs.ox.ac.uk/cpdnboinc_cgi/cgi
climateprediction.net - 2004-09-19 10:58:31 - Scheduler RPC to http://climateapps2.oucs.ox.ac.uk/cpdnboinc_cgi/cgi succeeded
Note that among these responses I do not receive any "General preferences have been updated" messages, I receive only the two lines (each) above.
When I then go back to my S&H and CP web sites, I find all Work locations have again been reset to Home.
So I said whathehell, removed all Work definitions, set all Default definitions "Leave applications in memory while preempted?" to Yes and set all machine "Locations" to Home.
I again selected "Update" on all projects on all machines, fast and slow.
On my fast machines, I received the following messages:
SETI@home - 2004-09-19 11:08:39 - Sending request to scheduler: http://setiboinc.ssl.berkeley.edu/sah_cgi/cgi
SETI@home - 2004-09-19 11:08:42 - Scheduler RPC to http://setiboinc.ssl.berkeley.edu/sah_cgi/cgi succeeded
SETI@home - 2004-09-19 11:08:42 - General preferences have been updated
--- - 2004-09-19 11:08:42 - General prefs: from SETI@home (last modified 2004-09-19 11:08:20)
--- - 2004-09-19 11:08:42 - General prefs: no separate prefs for home; using your defaults
climateprediction.net - 2004-09-19 11:10:08 - Sending request to scheduler: http://climateapps2.oucs.ox.ac.uk/cpdnboinc_cgi/cgi
climateprediction.net - 2004-09-19 11:10:11 - Scheduler RPC to http://climateapps2.oucs.ox.ac.uk/cpdnboinc_cgi/cgi succeeded
climateprediction.net - 2004-09-19 11:10:11 - General preferences have been updated
--- - 2004-09-19 11:10:11 - General prefs: from climateprediction.net (last modified 2004-09-19 11:09:59)
--- - 2004-09-19 11:10:11 - General prefs: no separate prefs for home; using your defaults
Soon followed by:
climateprediction.net - 2004-09-19 11:31:41 - Restarting result 2p90_100147310_0 using hadsm3 version 4.03
SETI@home - 2004-09-19 11:31:41 - Pausing result 26ap04aa.25756.18000.153414.58_2 (left in memory)
The "(left in memory)" makes me believe the "Leave applications in memory while preempted?" parameter has been successfully change and I am now swapping instead of removing.
The price I am paying for this ad hoc "solution" is that all of my machines, even the slow ones running S&H only, will be swapping. This would seem to be unnecessary overhead, something an already slow machine doesn't need any more of.
If you have stuck with me this far (I almost lost myself a few times) please entertain these questions:
1. Is what I attempted to do reasonble (enabling memory swapping for multiple projects, disabling it for single projects)?
2. Did I attempt to do it in a reasonable way (establish a Work "Location" for swapping and a Home "Location" for not swapping)?
3. Were the steps I took to accomplish #2 correct (should they have worked)?
4. If either #1 or #2 are incorrect, can you please help me do the right thing the right way?
Thanks for your patience....
Dick Flanagan
| 
