Error message cant finish WU

Message boards : Number crunching : Error message cant finish WU

To post messages, you must log in.

AuthorMessage
malmal

Send message
Joined: 29 Nov 08
Posts: 12
Credit: 728,356
RAC: 0
Message 59268 - Posted: 4 Feb 2009, 2:12:22 UTC
Last modified: 4 Feb 2009, 2:15:27 UTC

My S2895 is boincing away , two WUs at a time.
The messages part says it is asking for more work and the server says no- claims my cpus wont finish in time. cpu is at 100% with boinc at 17%. Task manager shows boinc as the only prog running.

It is online all the time. The OS is XP X64 with boinc and antivirus as the only two programs installed.
The rac is a little less than my other dual cpu [K8N-DL], and is slowly overtaking it.
ID: 59268 · 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 59269 - Posted: 4 Feb 2009, 4:12:40 UTC

BOINC is looking back historically at your usage on your machine and sees that either your machine is often powered off, or BOINC is not running. That is where it is getting the 17% from.

If you now intend to leave the machine on, it will adapt over the coming several days and realize you will need more work to keep you busy. Until then, it is going to be conservative and try to assure you don't get so much work that you miss the deadlines.

It also factors in time for other projects, if you have attached to more then just Rosetta, and it's trying to assure each gets their fair share in the days ahead. (where "fair" is defined by the resource shares you have configured)
Rosetta Moderator: Mod.Sense
ID: 59269 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
malmal

Send message
Joined: 29 Nov 08
Posts: 12
Credit: 728,356
RAC: 0
Message 59270 - Posted: 4 Feb 2009, 4:27:58 UTC - in response to Message 59269.  

Thanks for the reply. Rosetta is the only one. Machine reboot for xp updates only.
Time and date changed on their own though [once]. Keeping an eye on it. :)
ID: 59270 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
mikey
Avatar

Send message
Joined: 5 Jan 06
Posts: 1895
Credit: 9,188,754
RAC: 3,501
Message 59276 - Posted: 4 Feb 2009, 10:07:21 UTC - in response to Message 59268.  

My S2895 is boincing away , two WUs at a time.
The messages part says it is asking for more work and the server says no- claims my cpus wont finish in time. cpu is at 100% with boinc at 17%. Task manager shows boinc as the only prog running.

It is online all the time. The OS is XP X64 with boinc and antivirus as the only two programs installed.
The rac is a little less than my other dual cpu [K8N-DL], and is slowly overtaking it.


How long does Boinc say a unit will take to complete?
ID: 59276 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
malmal

Send message
Joined: 29 Nov 08
Posts: 12
Credit: 728,356
RAC: 0
Message 59308 - Posted: 5 Feb 2009, 0:23:46 UTC - in response to Message 59276.  
Last modified: 5 Feb 2009, 0:34:39 UTC

I thought anybody could see my computers.
Here is the link:https://boinc.bakerlab.org/rosetta/show_host_detail.php?hostid=967950
ID: 59308 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
malmal

Send message
Joined: 29 Nov 08
Posts: 12
Credit: 728,356
RAC: 0
Message 59309 - Posted: 5 Feb 2009, 0:41:35 UTC - in response to Message 59276.  
Last modified: 5 Feb 2009, 0:48:27 UTC

How long does Boinc say a unit will take to complete?[/quote]
#1 WU 86% 57 minutes
#2 WU 38% 57 hours 11 minutes
but #2 went from 70 hours to 57 hours in about 10 minutes!
edit while typing this it went from 57 hours to 45 hours 36 minutes.
ID: 59309 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
malmal

Send message
Joined: 29 Nov 08
Posts: 12
Credit: 728,356
RAC: 0
Message 59311 - Posted: 5 Feb 2009, 3:08:47 UTC - in response to Message 59309.  
Last modified: 5 Feb 2009, 3:09:36 UTC

How long does Boinc say a unit will take to complete?

#1 WU 86% 57 minutes
#2 WU 38% 57 hours 11 minutes
but #2 went from 70 hours to 57 hours in about 10 minutes!
edit while typing this it went from 57 hours to 45 hours 36 minutes.[/quote]
One more
#1 WU cpu time=1:57 65% time to complete=10:34
#2 WU cpu time= :52 24% time to complete=103:32
I left the seconds out.
ID: 59311 · 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 59336 - Posted: 5 Feb 2009, 14:29:49 UTC

malmal the target (it's not perfect, but generally close) runtime is configurable in your Rosetta preferences on the website. It defaults to 3 hours. While the numbers are messed up, they are just for your viewing pleasure anyway, and do not effect the science being done.
Rosetta Moderator: Mod.Sense
ID: 59336 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
malmal

Send message
Joined: 29 Nov 08
Posts: 12
Credit: 728,356
RAC: 0
Message 59372 - Posted: 6 Feb 2009, 1:42:42 UTC - in response to Message 59350.  
Last modified: 6 Feb 2009, 2:02:56 UTC

Looking at his results, his runtime preference is the default of 3 hours. So when initially his time to completion is 70 hours, that will influence how much work he is getting (as I understand the mechanism). Reading the messages of users with the same problem (on other projects) the R(esult) D(uration) C(orrection) F(actor) will not correct itself. Maybe a better course of action would be to downgrade to 6.2.19 if you don't need the CUDA capabilities in BOINC.

So I could go to an earlier x64 boinc? v6.4.5 now
@ mod sense. My message page is all red with server denials.
I tried uninstalling boinc, but when I reinstall it uses my old settings.
Is there a sure fire way of deleting all boinc stuff on my comp?
My thinking is Boinc will behave itself then. My other comps are crunching steady [ one other with x64 boinc] All on Rosetta only. :)
edit cpu 2hours 33 min 79% 2 hours 25 min left
#2 cpu 2 min .5% 198 hours 3 min left [ just started]
ID: 59372 · 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 59373 - Posted: 6 Feb 2009, 2:45:45 UTC
Last modified: 6 Feb 2009, 2:47:05 UTC

malmal, apparently the bugs in that version of the BOINC client will end up causing a similar problem again. But, if you look at the messages tab when BOINC first starts, it will show you the path to the BOINC data directory. Get to a point where you have all the completed work reported back, so you don't lose it. Then uninstall BOINC. Then delete that data directory. Then reinstall BOINC.

But the "downgrade" idea should avoid the problem reoccurring. You can download the older BOINC versions here

[edit] now I look at your task list and it looks like you just had a large pile of work come down. You must have found the other thread about manually adjusting the DCF?
Rosetta Moderator: Mod.Sense
ID: 59373 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
malmal

Send message
Joined: 29 Nov 08
Posts: 12
Credit: 728,356
RAC: 0
Message 59375 - Posted: 6 Feb 2009, 3:00:23 UTC - in response to Message 59373.  
Last modified: 6 Feb 2009, 3:19:16 UTC

malmal, apparently the bugs in that version of the BOINC client will end up causing a similar problem again. But, if you look at the messages tab when BOINC first starts, it will show you the path to the BOINC data directory. Get to a point where you have all the completed work reported back, so you don't lose it. Then uninstall BOINC. Then delete that data directory. Then reinstall BOINC.

But the "downgrade" idea should avoid the problem reoccurring. You can download the older BOINC versions here

[edit] now I look at your task list and it looks like you just had a large pile of work come down. You must have found the other thread about manually adjusting the DCF?

My task list shows 2 wu
#1cpu 1hour 26% 79 hours left
#2cpu 36min 17% 112 hours left
I got v6.2.19 x64. I'll install as you say. :)
Found out something else. I go to stats page. My graph starts at 194,000 and is a vertical line down to half way ,then all the way straight across ,then vertical down to the bottom.[178,000]
The dates on the bottom grid usually have the last few days. now it reads:
04 feb 09, nov 46, aug 84, may 22, feb60, dec97,.......oct86
reload tommorrow morning.
ID: 59375 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
malmal

Send message
Joined: 29 Nov 08
Posts: 12
Credit: 728,356
RAC: 0
Message 59405 - Posted: 7 Feb 2009, 1:30:48 UTC - in response to Message 59379.  

All is well in BoincWorld. I deleted 6.4.5 x64 and all the files, then installed 6.2.19 x64 . Everything is working like my other comps. full page of pending work ,stats graph normal and not one red message.
Now I worry about merging the two S2895's .:)
ID: 59405 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
malmal

Send message
Joined: 29 Nov 08
Posts: 12
Credit: 728,356
RAC: 0
Message 59407 - Posted: 7 Feb 2009, 1:41:19 UTC - in response to Message 59379.  
Last modified: 7 Feb 2009, 2:11:29 UTC

Are you talking about the tasks running? In that case you might want to take a look at your preferences, you have to enter 4 cpu's there (for the computer we're talking about). Looking at your tasklist you have close to 40 tasks "ready to start". Lack of work can't be a real problem I'd say.
[/quote]

I think your idea about time referencing in v 6.4.5 sounds plausible, but S2885 has v 6.4.5 and has no problems.
ID: 59407 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
malmal

Send message
Joined: 29 Nov 08
Posts: 12
Credit: 728,356
RAC: 0
Message 59620 - Posted: 17 Feb 2009, 1:04:04 UTC

My second x64 computer started acting up with erroneous dates and times this weekend. So I downgraded from v 6.4.5 to v6.2.19. Now everything is stable again.
ID: 59620 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote

Message boards : Number crunching : Error message cant finish WU



©2024 University of Washington
https://www.bakerlab.org