I am using BOINC for Rosetta@Home. After a reboot, I will see two tasks running in BOINC Manager, and my Windows Task Manager shows 50% of CPU time dedecated to each task.

Within a few hours (less than 4), the progress on both tasks will have stopped, but BOINC Manager will still indicate that both tasks are still "Running." CPU usage will be 99% Idle Process, and the two minirosetta_2.11_windows_intelx86.exe processes will be at 0% CPU usage. It makes no difference what else is running. It happened twice during the night when nothing else was running, and it happened twice yesterday when I was using the machine for other things. The "Messages" show nothing at the time that the processing ceases.

This is getting to be a real nuisance. Nothing works to restart the tasks except rebooting my machine or aborting the tasks (then they seem to be reassigned, and the whole process starts all over again).

I am running XP-SP3 (all updates current) on a Pentium D with 2 Gigs RAM and a GTX 295 GPU. I have 24 Gigabytes free on my HDDs. I have seen posts from others about this same problem, and responses from Site Moderators to the effect that the problem can't be reproduced. What is causing this, and how can it be fixed?

deesy

I have seen posts from others about this same problem, and responses from Site Moderators to the effect that the problem can't be reproduced. What is causing this, and how can it be fixed?

If it cannot be reproduced it's difficult to figure what's causing it and get a cure.

Make a file (with Notepad) called cc_config.xml in your BOINC Data directory (default C:\Documents and Settings\All Users\Application Data\BOINC\ and a hidden directory, but just navigate to it)

In it add the following lines:

<cc_config>
<log_flags>
<cpu_sched_debug>1</cpu_sched_debug>
<app_msg_receive>1</app_msg_receive>
<app_msg_send>1</app_msg_send>
<task_debug>1</task_debug>
<mem_usage_debug>1</mem_usage_debug>
</log_flags>
<options>
<max_stdout_file_size>20971520</max_stdout_file_size>
</options>
</cc_config>

Make sure the file's extension is .xml, not .txt
Save the file in ANSI format, not UTF-8 or something else.

Then open BOINC Manager->Advanced view->Advanced->Read config file.

Run with that until it happens again and post the corresponding log of it (not all, just about the last 3 minutes before it to 3 minutes after it). The <max_stdout_file_size> option I added in gives you a 20MB stdoutdae.txt file, so it won't fill up overnight and overwrite the possible hang moment that easily.

Do know that these flags will give a lot of data. You may want to increase the stdoutdae.txt file even further, if you do so do know the value is in bytes. So 20MB = 20 * 1024 * 1024

---

Okay, Jord. I'll do that.

FYI, I used BOINC Manager to suspend the two tasks that were "Running, High Priority," and the two tasks that were waiting to start, so that I could run a complete AV scan on my system (Windows Live OneCare). It found no malware, so I restarted the suspended tasks. The two that had been running at high priority were reported by BOINC Manager to have resumed running at high priority, and the other two were reported as waiting to start. Unfortunately, neither of the tasks that were reported as running were really doing so (unless BOINC has found a way to run without using the CPU). The Windows Task Manager reported the System Idle Process as accounting for 99% of CPU usage, and my CPU fan was very quiet.

deesy

I wrote the XML file and put in the correct folder, then read the config file per your directions. One of the two tasks has now ceased running, although the BOINC Manager reports that it is running at high priority. I have no idea when the task ceased running, so I have no way of finding a 6-minute section of a log. BTW, which log should I be looking at? What, specifically, should I be looking for? The script appears to be wotking because I now see a constant stream of messages in the Advanced view Messages window.

I will leave the machine processing the remaining one task for the rest of the night, but I am pretty confident that it will be doing nothing by morning.

deesy

BTW, which log should I be looking at?

It's the one that Jord mentioned a couple of times: stdoutdae.txt, and it's in the same place as the cc_config.xml that you've just created.

As of this morning, the first task is still stuck at 62.114%, even though BOINC Manager says that it is "Running, high priority." The second really is running.

I have two files named "stdoutdae," one has a ".txt" extension and the other has a ".old" extension. They have the exact same time/date stamp of 2/29/2010 7:29 AM (even though I did not suspend the tasks until about 7:56 AM, and my system clock is accurate. stdoutdae.old is 2,048 KB in length, while stdoutdae.txt is 756 KB in length.

Entries in stdoutdae.old have times from approximately 6:59 AM to approximately 7:42 AM today. stdoutdae.txt appears to still be receiving entries. The first entry in it was at 6:59 AM, and the last entry was at 8:11 AM, about five minutes ago.

I assume that neither of these files will contain useful information. I can save them and send them if you wish, but I would like to reboot my machine and get the task that is stuck at 62% running again. It is taking me several days to complete tasks because of the hang ups.

Please advise.

deesy

I can save them and send them if you wish, but I would like to reboot my machine and get the task that is stuck at 62% running again.

Send them to me. I will send you my email address in a private message.

Please zip or otherwise compress them before emailing.
And you can normally reboot. These files will be kept by BOINC.

---