5.59 Science application often locks up

Message boards : Number crunching : 5.59 Science application often locks up

To post messages, you must log in.

AuthorMessage
Prototype

Send message
Joined: 7 Feb 06
Posts: 1
Credit: 89,406
RAC: 0
Message 40126 - Posted: 1 May 2007, 2:16:33 UTC

Having major problems with R@H units locking up completely, not just % freezing like some others have reported but the science application itself will often lock up completely and the core will go unused .

The watchdog (or whatever its called) never restarts or terminates the application, I've left the frozen applications for over 8 hours to see if something would happen but nothing ever did.

If I restart BOINC the frozen application will usually continue around the area it froze.

Incidently the time where most of the freezing seems to occur is somewhere around 2 hours 40 min to 2 hours 55 min mark.

Machine is core 2 on XP.

ID: 40126 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Krupp

Send message
Joined: 8 Mar 06
Posts: 3
Credit: 4,088
RAC: 0
Message 40214 - Posted: 2 May 2007, 13:47:32 UTC

The same thing have happened several times to me aswell. Not sure about the time scale though. Luckily the fan speed drops considerably when this happens so I usually notice it right away.
ID: 40214 · 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 40222 - Posted: 2 May 2007, 16:39:07 UTC

So on the advanced view of the BOINC Manager, on the tasks tab, what is shown for the task in the status column? Is it waiting for memory? waiting to run? or Running?

And when this occurs, am I correct that the CPU time shown in the graphic does not increase?

There have been problems in the passed where BOINC seems to lose contact with the crunching thread. And this prevents the watchdog from detecting the problem. But I had thought that was resolved now on the newer versions of BOINC. But it looks like you are already using BOINC 5.8.16.

How frequently are you seeing this happen?
When you restart, do the tasks complete normally?
Rosetta Moderator: Mod.Sense
ID: 40222 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Krupp

Send message
Joined: 8 Mar 06
Posts: 3
Credit: 4,088
RAC: 0
Message 40263 - Posted: 3 May 2007, 9:24:13 UTC - in response to Message 40222.  

So on the advanced view of the BOINC Manager, on the tasks tab, what is shown for the task in the status column? Is it waiting for memory? waiting to run? or Running?

And when this occurs, am I correct that the CPU time shown in the graphic does not increase?

There have been problems in the passed where BOINC seems to lose contact with the crunching thread. And this prevents the watchdog from detecting the problem. But I had thought that was resolved now on the newer versions of BOINC. But it looks like you are already using BOINC 5.8.16.

How frequently are you seeing this happen?
When you restart, do the tasks complete normally?


The BOINC manager says it is running. I am almost, but not 100%, sure that the CPU time does not increase. Is has happened 3 times in the past two or three weeks now. Although I am not sure that only Rosetta tasks have freezed, maybe this is a BOINC problem. I'm still using BOINC 5.8.15, I'll update it and see if it gets better.

Also, the tasks continues normally after I restart BOINC.
ID: 40263 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote

Message boards : Number crunching : 5.59 Science application often locks up



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