This is the second time that a WU crashes with a 0xfffffffb error, this time at about 97% completion (Result ID <A HREF="http://climateapps2.oucs.ox.ac.uk/cpdnboinc/result.php?resultid=516911">516911</A>). My question is: is all the work that has been done over the last 2½ months just lost?

It appears as if BOINC did anyway upload the result, according to the message log below, but is it of any value?

2005-03-25 19:23:13 [climateprediction.net] Sending request to scheduler: http://climateapps2.oucs.ox.ac.uk/cpdnboinc_cgi/cgi
2005-03-25 19:23:14 [climateprediction.net] Scheduler RPC to http://climateapps2.oucs.ox.ac.uk/cpdnboinc_cgi/cgi succeeded
2005-03-25 19:26:48 [climateprediction.net] Sending request to scheduler: http://climateapps2.oucs.ox.ac.uk/cpdnboinc_cgi/cgi
2005-03-25 19:26:49 [climateprediction.net] Scheduler RPC to http://climateapps2.oucs.ox.ac.uk/cpdnboinc_cgi/cgi succeeded
2005-03-25 20:24:15 [climateprediction.net] Unrecoverable error for result 45m5_000215845_0 ( - exit code -5 (0xfffffffb))
2005-03-25 20:24:15 [climateprediction.net] Deferring communication with project for 59 seconds
2005-03-25 20:24:15 [climateprediction.net] Computation for result 45m5_000215845 finished
2005-03-25 20:24:16 [climateprediction.net] Started upload of 45m5_000215845_0_1.zip
2005-03-25 20:24:16 [climateprediction.net] Started upload of 45m5_000215845_0_2.zip
2005-03-25 20:24:37 [climateprediction.net] Finished upload of 45m5_000215845_0_1.zip
2005-03-25 20:24:37 [climateprediction.net] Throughput 54545 bytes/sec
2005-03-25 20:24:37 [climateprediction.net] Started upload of 45m5_000215845_0_3.zip
2005-03-25 20:24:44 [climateprediction.net] Finished upload of 45m5_000215845_0_2.zip
2005-03-25 20:24:44 [climateprediction.net] Throughput 50688 bytes/sec
2005-03-25 20:24:44 [climateprediction.net] Started upload of 45m5_000215845_0_4.zip
2005-03-25 20:25:05 [climateprediction.net] Finished upload of 45m5_000215845_0_3.zip
2005-03-25 20:25:05 [climateprediction.net] Throughput 55701 bytes/sec
2005-03-25 20:25:05 [climateprediction.net] Started upload of 45m5_000215845_0_5.zip
2005-03-25 20:25:12 [climateprediction.net] Finished upload of 45m5_000215845_0_5.zip
2005-03-25 20:25:12 [climateprediction.net] Throughput 48179 bytes/sec
2005-03-25 20:25:13 [climateprediction.net] Finished upload of 45m5_000215845_0_4.zip
2005-03-25 20:25:13 [climateprediction.net] Throughput 55298 bytes/sec

Unfortunate time for an error. But yes it is definitely of value. It has uploaded what it can. The team has posted that a phase or 2 is useful even without completing the 3 phases. Also they will get around to looking at what is causing errors eventually. Since you reached 97%, there is more information to go on and it is less likely to be due to an unstable computer. When they do start looking at crashed model, ones that reached over 95% may well be one of the best places to start.


There is some discussion of the -5 error in this thread:

http://climateapps2.oucs.ox.ac.uk/cpdnboinc/forum_thread.php?id=1103

But since you got so far through your model, I would not worry too much about it.

Thanks for your explanation. Can I assume from the thread you referenced that the -5 error is fixed now in the 4.10 core application?

&gt; Thanks for your explanation. Can I assume from the thread you referenced that
&gt; the -5 error is fixed now in the 4.10 core application?
&gt;
&gt;
It is not impossible that the 4.12 version has alleviated the problem but I wouldn't bet on it. A lot of AMD processors run no problem a few seem to always hit this problem. No idea why. But that shouldn't worry you if you are only using Intel. Some intel processors get the problem with these stability testing the computer often eventually finds the problem. If you get another one (or had encountered it much earlier than 97%) then I would be recommending some stability tests. There is enough people successfully running the project to say that it is more likely to be a stability problem with the computer or a software conflict or something like that rather than a fatal bug in the code that needs a fix.