New rescheduler

Message boards : Number crunching : New rescheduler
Message board moderation

To post messages, you must log in.

Previous · 1 · 2 · 3 · 4 · 5 · 6 · 7 . . . 11 · Next

AuthorMessage
Profile Tim Norton
Volunteer tester
Avatar

Send message
Joined: 2 Jun 99
Posts: 835
Credit: 33,540,164
RAC: 0
United Kingdom
Message 1019533 - Posted: 25 Jul 2010, 19:59:15 UTC
Last modified: 25 Jul 2010, 19:59:46 UTC

Fred

the problem that hansr posted earlier in the week is still there in ver 1.3

what appears to be happening is that after a reschedule boinctray.exe is running (can be seen in task manager) - this is not normally running and appears to stop the boinc client from running/restarting

if its killed in task manager then boinc mamager and the client run ok

have not tried it in 1.4 but the change log implies its not a bug fix and have nothing to reschedule yet

this behaviour only appears in 1.2 and 1.3

happened on two win7 64bit pc's

not a show stopper just a pain but easy enough the manually work around

great app by the way
Tim

ID: 1019533 · Report as offensive
Richard Haselgrove Project Donor
Volunteer tester

Send message
Joined: 4 Jul 99
Posts: 14650
Credit: 200,643,578
RAC: 874
United Kingdom
Message 1019555 - Posted: 25 Jul 2010, 21:56:00 UTC - in response to Message 1019533.  

In theory, boinctray.exe is purely there to monitor mouse and keyboard activity, so that BOINC itself can be signalled to stop and start according to 'user activity' preferences.

If boinctray.exe running continuously, while boinc.exe is stopped and restarted by an external process, causes problems, then it would be worth investigating further and reporting back to the BOINC developers.

I've never had any problems with previous reschedulers, but I run XP with BOINC in service mode as a rule - this problem may be more related to user mode (where running according to user activity preferences is perhaps more likely).
ID: 1019555 · Report as offensive
Profile S@NL - eFMer - efmer.com/boinc
Volunteer tester
Avatar

Send message
Joined: 7 Jun 99
Posts: 512
Credit: 148,746,305
RAC: 0
United States
Message 1019651 - Posted: 26 Jul 2010, 7:34:37 UTC - in response to Message 1019533.  

Fred

the problem that hansr posted earlier in the week is still there in ver 1.3

what appears to be happening is that after a reschedule boinctray.exe is running (can be seen in task manager) - this is not normally running and appears to stop the boinc client from running/restarting

if its killed in task manager then boinc mamager and the client run ok

have not tried it in 1.4 but the change log implies its not a bug fix and have nothing to reschedule yet

this behaviour only appears in 1.2 and 1.3

happened on two win7 64bit pc's

not a show stopper just a pain but easy enough the manually work around

great app by the way

Maybe I build in too may security features.
The problem is the BOINC client sets a mutex to prevent it from starting twice.
It seems the mutex is not cleared properly at exit.
Looks like, when another program has stated the BOINC client, the mutex is not cleared. It only clears after the other program, that started the client is closed as well.

There are probably reschedulers that don't check this. But safety first.

In the next version I will make a setting in the expert tab to disable this extra security feature.
TThrottle Control your temperatures. BoincTasks The best way to view BOINC. Anza Borrego Desert hiking.
ID: 1019651 · Report as offensive
Profile [DPC] hansR Project Donor
Volunteer tester
Avatar

Send message
Joined: 14 Jul 00
Posts: 47
Credit: 235,829,569
RAC: 8
Netherlands
Message 1019662 - Posted: 26 Jul 2010, 9:11:57 UTC - in response to Message 1019651.  
Last modified: 26 Jul 2010, 9:12:18 UTC

Found the rescheduler log directory :-)

Log of another problem last friday (V 0.9):

23 July 2010 - 22:03:49 Shutting down BOINC client
23 July 2010 - 22:03:52 Shutdown of BOINC client completed
23 July 2010 - 22:03:53 Out of range: rsc_fpops_bound < 500000000000000000.000000, total: 5
23 July 2010 - 22:03:53 Found: CPU: 0, VLAR: 0, VHAR: 0
23 July 2010 - 22:03:53 Found: GPU: 5, VLAR: 0, VHAR: 0
23 July 2010 - 22:03:53 Rescheduling CPU version: 603 ,Gpu version: 608 planclass: cuda
23 July 2010 - 22:03:53 ERROR: unexpected empty CPU: or GPU: 608 Plan:
23 July 2010 - 22:03:53 ERROR: Move completed with an error.

SETI was low on workunits, CPU clients were all running astropulse jobs.

Now running V 1.4, no problems sofar ;-)
ID: 1019662 · Report as offensive
Terror Australis
Volunteer tester

Send message
Joined: 14 Feb 04
Posts: 1817
Credit: 262,693,308
RAC: 44
Australia
Message 1019668 - Posted: 26 Jul 2010, 9:34:37 UTC

Had the same error over the weekend, V0.6

T.A.
ID: 1019668 · Report as offensive
Profile S@NL - eFMer - efmer.com/boinc
Volunteer tester
Avatar

Send message
Joined: 7 Jun 99
Posts: 512
Credit: 148,746,305
RAC: 0
United States
Message 1019669 - Posted: 26 Jul 2010, 9:38:54 UTC - in response to Message 1019668.  

Had the same error over the weekend, V0.6

T.A.

Was fixed in 1.1
TThrottle Control your temperatures. BoincTasks The best way to view BOINC. Anza Borrego Desert hiking.
ID: 1019669 · Report as offensive
Profile S@NL - eFMer - efmer.com/boinc
Volunteer tester
Avatar

Send message
Joined: 7 Jun 99
Posts: 512
Credit: 148,746,305
RAC: 0
United States
Message 1019671 - Posted: 26 Jul 2010, 9:45:29 UTC - in response to Message 1019533.  

Fred

the problem that hansr posted earlier in the week is still there in ver 1.3

what appears to be happening is that after a reschedule boinctray.exe is running (can be seen in task manager) - this is not normally running and appears to stop the boinc client from running/restarting

if its killed in task manager then boinc mamager and the client run ok

have not tried it in 1.4 but the change log implies its not a bug fix and have nothing to reschedule yet

this behaviour only appears in 1.2 and 1.3

happened on two win7 64bit pc's

not a show stopper just a pain but easy enough the manually work around

great app by the way

What may help is Exit the BOINC Manager and let everything stop.
Reschedule and restart the Manager.
This way the Manager is no longer the one that started the client.
After I did something similar, closing BoincTasks and shut down the client, everything worked fine on Win 7 64.

TThrottle Control your temperatures. BoincTasks The best way to view BOINC. Anza Borrego Desert hiking.
ID: 1019671 · Report as offensive
Profile BilBg
Volunteer tester
Avatar

Send message
Joined: 27 May 07
Posts: 3720
Credit: 9,385,827
RAC: 0
Bulgaria
Message 1019677 - Posted: 26 Jul 2010, 10:57:03 UTC
Last modified: 26 Jul 2010, 11:22:02 UTC



Some minor changes needed in About (BOINC Monitor -> BOINC Rescheduler):





I will be more confident if in the Expert tab there is text similar to:
"
Leave the above boxes blank to use the default folders.

BOINC Data folder used now by BOINC Rescheduler is: C:\..........
TEMP folder used now by BOINC Rescheduler is: C:\...............
"

Or:
"
BOINC Rescheduler uses now the following folders:

BOINC Data: C:\..........
TEMP files: C:\..........
"


Maybe also add buttons to open these folders (for easy access)?






 


- ALF - "Find out what you don't do well ..... then don't do it!" :)
 
ID: 1019677 · Report as offensive
Profile S@NL - eFMer - efmer.com/boinc
Volunteer tester
Avatar

Send message
Joined: 7 Jun 99
Posts: 512
Credit: 148,746,305
RAC: 0
United States
Message 1019678 - Posted: 26 Jul 2010, 11:00:09 UTC - in response to Message 1019555.  

In theory, boinctray.exe is purely there to monitor mouse and keyboard activity, so that BOINC itself can be signalled to stop and start according to 'user activity' preferences.

If boinctray.exe running continuously, while boinc.exe is stopped and restarted by an external process, causes problems, then it would be worth investigating further and reporting back to the BOINC developers.

I've never had any problems with previous reschedulers, but I run XP with BOINC in service mode as a rule - this problem may be more related to user mode (where running according to user activity preferences is perhaps more likely).

Probably the stopping of the BOINC client as a service works.
The --quit method doesn't

The problem is that the boinc.exe client is started by someone, mostly the BOINC Manager.
So boinc.exe is a child of the Manager. Even as the --quit closes boinc.exe it's still linked to the Manager.
So you need to close the Manager in order to clear the Mutex.
But this may be OS dependent I have seen no problems on XP and plenty on WIN 7.

The best workaround is exit the Manager with the option to close everything down.
Than reschedule. At that point the rescheduler, that does release the link between the Rescheduler and boinc.exe, by closing with the --detach command.
Now everything works.

TThrottle Control your temperatures. BoincTasks The best way to view BOINC. Anza Borrego Desert hiking.
ID: 1019678 · Report as offensive
Profile S@NL - eFMer - efmer.com/boinc
Volunteer tester
Avatar

Send message
Joined: 7 Jun 99
Posts: 512
Credit: 148,746,305
RAC: 0
United States
Message 1019698 - Posted: 26 Jul 2010, 13:33:17 UTC - in response to Message 1019677.  



1) Some minor changes needed in About (BOINC Monitor -> BOINC Rescheduler):


I will be more confident if in the Expert tab there is text similar to:
"
2) Leave the above boxes blank to use the default folders.

BOINC Data folder used now by BOINC Rescheduler is: C:\..........
TEMP folder used now by BOINC Rescheduler is: C:\...............
"

Or:
"
BOINC Rescheduler uses now the following folders:

BOINC Data: C:\..........
TEMP files: C:\..........
"


3) Maybe also add buttons to open these folders (for easy access)?



1) Copied to much, corrected
2) Changed, good idea.
3) [...] now opens the default folder if blank.
TThrottle Control your temperatures. BoincTasks The best way to view BOINC. Anza Borrego Desert hiking.
ID: 1019698 · Report as offensive
Profile S@NL - eFMer - efmer.com/boinc
Volunteer tester
Avatar

Send message
Joined: 7 Jun 99
Posts: 512
Credit: 148,746,305
RAC: 0
United States
Message 1019701 - Posted: 26 Jul 2010, 13:56:05 UTC - in response to Message 1019698.  

V 1.5

Add: Settings Dialog: A check to not restart the BOINC client after rescheduling.
Add: Settings Dialog: A button to start the BOINC client.
Add: Expert Dialog: A check not to lock the BOINC client while rescheduling (mutex).
Add: Expert Dialog: Text for the used folders, when the boxes are blank.
Changed: Expert Dialog: [..] now starts in the default folder.
TThrottle Control your temperatures. BoincTasks The best way to view BOINC. Anza Borrego Desert hiking.
ID: 1019701 · Report as offensive
Profile BilBg
Volunteer tester
Avatar

Send message
Joined: 27 May 07
Posts: 3720
Credit: 9,385,827
RAC: 0
Bulgaria
Message 1019708 - Posted: 26 Jul 2010, 14:47:11 UTC - in response to Message 1019701.  


OK :)

Now I can easy look at the Log & Capture files
- click [...] button
- right-click the directory
- choose Open






 


- ALF - "Find out what you don't do well ..... then don't do it!" :)
 
ID: 1019708 · Report as offensive
Profile Tim Norton
Volunteer tester
Avatar

Send message
Joined: 2 Jun 99
Posts: 835
Credit: 33,540,164
RAC: 0
United Kingdom
Message 1019736 - Posted: 26 Jul 2010, 16:35:12 UTC - in response to Message 1019671.  

Fred

the problem that hansr posted earlier in the week is still there in ver 1.3

what appears to be happening is that after a reschedule boinctray.exe is running (can be seen in task manager) - this is not normally running and appears to stop the boinc client from running/restarting

if its killed in task manager then boinc mamager and the client run ok

have not tried it in 1.4 but the change log implies its not a bug fix and have nothing to reschedule yet

this behaviour only appears in 1.2 and 1.3

happened on two win7 64bit pc's

not a show stopper just a pain but easy enough the manually work around

great app by the way

What may help is Exit the BOINC Manager and let everything stop.
Reschedule and restart the Manager.
This way the Manager is no longer the one that started the client.
After I did something similar, closing BoincTasks and shut down the client, everything worked fine on Win 7 64.

Fred

i always shut the Boinc manager down completely before i run the reschedule as you describe above

but i usually have your reschedulear running to do a test run first before i close the boinc manager - so i close BM, wait 5 secs or so for the running wu to clear from task manager etc then do the reschedule

may be i should exit out of reschedule at the same time and then run it up again to do a proper run etc - will test to see with the latest version (1.5) and let you know etc

Tim

ID: 1019736 · Report as offensive
Profile S@NL - eFMer - efmer.com/boinc
Volunteer tester
Avatar

Send message
Joined: 7 Jun 99
Posts: 512
Credit: 148,746,305
RAC: 0
United States
Message 1019741 - Posted: 26 Jul 2010, 16:49:00 UTC - in response to Message 1019736.  


Fred

i always shut the Boinc manager down completely before i run the reschedule as you describe above

but i usually have your reschedulear running to do a test run first before i close the boinc manager - so i close BM, wait 5 secs or so for the running wu to clear from task manager etc then do the reschedule

may be i should exit out of reschedule at the same time and then run it up again to do a proper run etc - will test to see with the latest version (1.5) and let you know etc

You can do a test with everything running, that's no problem.
Normally you only have to close down the manager once, because when the Manager is restarted and the BOINC client is already running it does nothing except connecting.
The next version of BoincTasks will be rescheduler friendly, so that shouldn't give any problems.
TThrottle Control your temperatures. BoincTasks The best way to view BOINC. Anza Borrego Desert hiking.
ID: 1019741 · Report as offensive
Profile Sutaru Tsureku
Volunteer tester

Send message
Joined: 6 Apr 07
Posts: 7105
Credit: 147,663,825
RAC: 5
Germany
Message 1019965 - Posted: 27 Jul 2010, 5:16:51 UTC
Last modified: 27 Jul 2010, 5:18:33 UTC

Fred, a big 'Thanks a lot!' for this nice tool!

Sorry if it was mentioned already..
Your tool work also with 609 and cuda23 entries in the app_info.xml file?

To now I have 608 and cuda in the app_info.xml file, but get .vlar WUs for GPU.
I guess maybe it's because of the 608 and cuda entries.

If I add 609 and cuda23 to my current app_info.xml file, I wouldn't trash my DLed WU cache, or?
And then all new DLed WUs will be named as 609_cuda23, right?
And your tool will work also with both 608_cuda and 609_cuda23 entries in the app_info.xml file?

At SETI@home Beta Test I have (only) 609 and cuda23 entries in my app_info.xml file and my BOINC don't get .vlar WUs for GPU.
ID: 1019965 · Report as offensive
Profile BilBg
Volunteer tester
Avatar

Send message
Joined: 27 May 07
Posts: 3720
Credit: 9,385,827
RAC: 0
Bulgaria
Message 1020072 - Posted: 27 Jul 2010, 12:10:49 UTC - in response to Message 1019701.  


Fred,
Thank you for the effort to make this program (and all the other of yours BOINC "relatives" :) )!


Can you allow in Simulation mode that new client_state.xml file be created
(in Temp (capture) directory) even if error occurs?

The reason -
I have no CUDA capable GPU but want to try/use the feature to increase the
rsc_fpops_bound (and maybe DCF)

but get the Error:

27 July 2010 - 14:35:34 Out of range: rsc_fpops_bound < 500000000000000000.000000, total: 188
27 July 2010 - 14:35:34 Found: CPU: 187, VLAR: 18, VHAR: 2
27 July 2010 - 14:35:34 Found: GPU: 0, VLAR: 0, VHAR: 0
27 July 2010 - 14:35:34 Rescheduling needed

27 July 2010 - 14:35:37 WARNING: Simulation mode selected
27 July 2010 - 14:35:37 Out of range: rsc_fpops_bound < 500000000000000000.000000, total: 188
27 July 2010 - 14:35:37 Found: CPU: 187, VLAR: 18, VHAR: 2
27 July 2010 - 14:35:37 Found: GPU: 0, VLAR: 0, VHAR: 0
27 July 2010 - 14:35:37 No version for Cpu or Gpu


I will try this with BOINC completely exited
and compare (by Total Commander) the old (in the BOINC Data directory)
and the new (in Temp (capture) directory) client_state.xml files
to see what changes are made before manual copy.


app_info.xml:
<app_info> 
    <app>
        <name>setiathome_enhanced</name>
    </app>
    <file_info>
        <name>AK_v8b_win_SSE3_AMD.exe</name>
        <executable/>
    </file_info>
    <app_version>
        <app_name>setiathome_enhanced</app_name>
        <version_num>603</version_num>
        <file_ref>
           <file_name>AK_v8b_win_SSE3_AMD.exe</file_name>
            <main_program/>
        </file_ref>
    </app_version>
    <app>
        <name>astropulse_v505</name>
    </app>
    <file_info>
        <name>ap_5.05r409_SSE.exe</name>
        <executable/>
    </file_info>
    <app_version>
        <app_name>astropulse_v505</app_name>
        <version_num>505</version_num>
        <file_ref>
            <file_name>ap_5.05r409_SSE.exe</file_name>
            <main_program/>
        </file_ref>
    </app_version>    
</app_info> 


 


- ALF - "Find out what you don't do well ..... then don't do it!" :)
 
ID: 1020072 · Report as offensive
Profile S@NL - eFMer - efmer.com/boinc
Volunteer tester
Avatar

Send message
Joined: 7 Jun 99
Posts: 512
Credit: 148,746,305
RAC: 0
United States
Message 1020075 - Posted: 27 Jul 2010, 12:18:17 UTC - in response to Message 1020072.  
Last modified: 27 Jul 2010, 12:18:46 UTC


Fred,
Thank you for the effort to make this program (and all the other of yours BOINC "relatives" :) )!


Can you allow in Simulation mode that new client_state.xml file be created
(in Temp (capture) directory) even if error occurs?

The reason -
I have no CUDA capable GPU but want to try/use the feature to increase the
rsc_fpops_bound (and maybe DCF)

You should provide a valid app_info.xml with an GPU application.
There are plenty of examples around.
Or just take the lunatics installer.
This will probably work even when there is no GPU around.
But I'm not 100% sure if the client_state will include the GPU if it doesn't find one.
TThrottle Control your temperatures. BoincTasks The best way to view BOINC. Anza Borrego Desert hiking.
ID: 1020075 · Report as offensive
Profile BilBg
Volunteer tester
Avatar

Send message
Joined: 27 May 07
Posts: 3720
Credit: 9,385,827
RAC: 0
Bulgaria
Message 1020079 - Posted: 27 Jul 2010, 12:34:05 UTC - in response to Message 1020075.  
Last modified: 27 Jul 2010, 12:45:10 UTC



So you find it no easy to make some switch so this tool will
just change the rsc_fpops_bound and DCF
and not try to reschedule WUs (change ver. and plan class)?

Of course I can do this in Notepad but it will be cleaner and easier using your tool :)


 


- ALF - "Find out what you don't do well ..... then don't do it!" :)
 
ID: 1020079 · Report as offensive
Profile hiamps
Volunteer tester
Avatar

Send message
Joined: 23 May 99
Posts: 4292
Credit: 72,971,319
RAC: 0
United States
Message 1020225 - Posted: 28 Jul 2010, 1:34:36 UTC
Last modified: 28 Jul 2010, 1:36:38 UTC

Odd, I just ran the 1.5 and now I have completion times for some CPU units at 17 hours and some are around 15 minutes. Wonder why it didn't do them all?

EDIT: Maybe I missed something, now they are at 1 hour 17 minutes and 15 minutes...LOL
Official Abuser of Boinc Buttons...
And no good credit hound!
ID: 1020225 · Report as offensive
Profile Sutaru Tsureku
Volunteer tester

Send message
Joined: 6 Apr 07
Posts: 7105
Credit: 147,663,825
RAC: 5
Germany
Message 1020246 - Posted: 28 Jul 2010, 3:17:38 UTC

I hope I'm not in Fred's ignore list.. :o)

Maybe someone have knowledge/answers to the questions in Message 1019965 ?

ID: 1020246 · Report as offensive
Previous · 1 · 2 · 3 · 4 · 5 · 6 · 7 . . . 11 · Next

Message boards : Number crunching : New rescheduler


 
©2024 University of California
 
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.