Dynamic cpu_run_time_pref possible?

Message boards : Number crunching : Dynamic cpu_run_time_pref possible?

To post messages, you must log in.

AuthorMessage
Nothing But Idle Time

Send message
Joined: 28 Sep 05
Posts: 209
Credit: 139,545
RAC: 0
Message 37883 - Posted: 16 Mar 2007, 15:01:48 UTC

I know that one can change the cpu_run_time_pref on the project server at any time and communicate that to the local Rosetta app by way of the project's update button.

Where does this parameter reside on the local host and how is it communicated to the science app? Can it be communicated, say, at a checkpoint?

Which Rosetta tasks will notice the change: Running, Waiting to Run, or only tasks D/L since the update?

Can cpu_run_time_pref be modified locally and communicated to the science app without updating the Rosetta server? With the newer boinc 5.8 we now have override ability and I would like the same ability with the cpu_run_time_pref here at Rosetta.

Those of you who run Einstein know that it had many problems in the last few months and was down for 1-2 days at a time. I wonder if Rosetta should experience the same thing and I happen to have 2 or 3 Rosetta tasks onboard with 4-hour run time pref, could I dynamically and locally increase the cpu_run_time_pref and force Rosetta tasks to continue creating decoys well beyond my normal run time pref while the servers are being repaired.



ID: 37883 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile anders n

Send message
Joined: 19 Sep 05
Posts: 403
Credit: 537,991
RAC: 0
Message 37886 - Posted: 16 Mar 2007, 15:52:03 UTC - in response to Message 37883.  

I know that one can change the cpu_run_time_pref on the project server at any time and communicate that to the local Rosetta app by way of the project's update button.

Where does this parameter reside on the local host and how is it communicated to the science app? Can it be communicated, say, at a checkpoint?

Which Rosetta tasks will notice the change: Running, Waiting to Run, or only tasks D/L since the update?

Can cpu_run_time_pref be modified locally and communicated to the science app without updating the Rosetta server? With the newer boinc 5.8 we now have override ability and I would like the same ability with the cpu_run_time_pref here at Rosetta.

Those of you who run Einstein know that it had many problems in the last few months and was down for 1-2 days at a time. I wonder if Rosetta should experience the same thing and I happen to have 2 or 3 Rosetta tasks onboard with 4-hour run time pref, could I dynamically and locally increase the cpu_run_time_pref and force Rosetta tasks to continue creating decoys well beyond my normal run time pref while the servers are being repaired.


I don't know the whole story but All tasks notice the change.
If you lower the time a wu will finish at the next checkpoint.
If you set the time higher the wu will continue.

Yes you can change the time , but don't ask me where :)

Anders n

ID: 37886 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Mod.Sense
Volunteer moderator

Send message
Joined: 22 Aug 06
Posts: 4018
Credit: 0
RAC: 0
Message 37888 - Posted: 16 Mar 2007, 17:04:40 UTC

The preference is called
<cpu_run_time>xxxxx</cpu_run_time>
it is in seconds and located in the project's xml file. By default that would be here on Windows:
C:Program FilesBOINCaccount_boinc.bakerlab.org_rosetta.xml

Only modify it when BOINC is ended. Changes will take effect whenever a WU is restarted. So, in your scenario, if Rosetta's servers went down, you could end BOINC, revise the file, restart BOINC and the tasks would crunch through to the new preference. 86400 seconds is the 24hr maximum.

Note that since BOINC was used to the history of your old runtime preference, it will be a little confused about % complete and estimated time to completion as these run. But this will not be a problem for the science application.

Then, since in the above scenario, you did not change the preference on the Rosetta preferences webpage, your preference would be set back to your original value the next time update to project completes. ...and again BOINC will be confused about how long to expect the thing to run.

Another way to prepare for such a server outage would be to keep a larger cache of work on-hand. This is done in the General Preferences via the setting to "connect to network about every ... days".
Rosetta Moderator: Mod.Sense
ID: 37888 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
FluffyChicken
Avatar

Send message
Joined: 1 Nov 05
Posts: 1260
Credit: 369,635
RAC: 0
Message 37889 - Posted: 16 Mar 2007, 17:05:11 UTC

Setting local preference's are only the basic 'global preferences' that you can set via the Simple GUI in BOINC, though you can set a larger number if you don't mind editing the global_prefs_override.xml yourself.

There is no provision to alter project preferences via this and there are none in future versions (yet, though I have ask and it has been asked a few times).


There is a file you can edit.. sorry have to rush but it's either the rosetta one in boinc or in the rosetta project folder or the one in the 'slot'
though you have to restart 'rosetta' for it to get picked up and it will change back the next time it comunicates with the server. I know it works since i have done it before a while back...

So you would need to make a program to monitor it (and at the same time monitor why you want do this in the first place) etc...



Team mauisun.org
ID: 37889 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote

Message boards : Number crunching : Dynamic cpu_run_time_pref possible?



©2025 University of Washington
https://www.bakerlab.org