This definitely is a limit of the Ralph work units

<workunit>
<name>HOMSb7_homDB015_1b72__226_2</name>
<app_name>rosetta_beta</app_name>
<version_num>489</version_num>
<rsc_fpops_est>40000000000000.000000</rsc_fpops_est>
<rsc_fpops_bound>50000000000000000.000000</rsc_fpops_bound>
<rsc_memory_bound>100000000.000000</rsc_memory_bound>
<rsc_disk_bound>200000000.000000</rsc_disk_bound>
<command_line>
...

The workunit is limited to 200000000.000000 bytes. It has nothing to do with my disk-usage parameters. The offending file is probably stdout.txt, whose size is approximatly 1MB/CPU-Minute on my PC. So I expect it to fail after 3.5h.

If you are correct we need to know. Could you set your time parameter to 4 hours and see if that fixes the problem? If not then try 2 hours

The WU failed after 3.25h )see the Result here. The next WU failed after just over 5h. See Result here.
I'll try "Target CPU run time" 4 hours the next time I get a WU.

---

David Kim is aware of this issue. There is a problem with the way BOINC handles the error files which can cause it to write the same error to the file over and over. He has contacted the BOINC development team to have this adjusted. In the meantime he will be reducing the debug settings on the next build of the test application. What for the new release.

But if you can reduce your time setting a little please do so for now. It is supposed to work

My preferences are set to 2 hours and I have had several of these errors with 4.87 and 4.89

I can confirm that

01/03/2006 01:40:25|ralph@home|Scheduler request to https://ralph.bakerlab.org/ralph_cgi/cgi succeeded
01/03/2006 03:29:49|ralph@home|Aborting task HOMSb7_homDB025_1b72__226_7_0: exceeded disk limit: 203574047.000000 > 200000000.000000
01/03/2006 03:29:49|ralph@home|Unrecoverable error for result HOMSb7_homDB025_1b72__226_7_0 (Maximum disk usage exceeded)


is due to an abnormal stdout.txt

Halfway through the WU, it was over 150 MB.

I tried deleting the file on one WU, and then I did not get such an error.
Here is the result of one where I deleted stdout.txt after about 60% of the WU: 12095

I did not babysit it until completion, so I am not sure what the final size was.

Here is one where I did not delete the stdout.txt:
12096

This was with Boinc beta version 5.3.22

I do not remember the exact warning in the stdout.txt file that got repeated, but it was three lines that got repeated several about 100000 lines pr model in the WU. Something about "full atom" requested, but not specied, therefore not computed.


I will try various client versions, both beta and stable to see if I get the same results the next time I receive some work units from Ralph.

---

those WUs were 4.89, if you get work again it should be 4.90 or a even newer version, which does not repeat that lines over and over in the stdout.txt anymore (at least in the ones i was able to look in till now, my fast pc is out of work, my slow one takes its time :=))

I think that the values are being overridden, I updated my prefs to 1 hour then updated BOINC but am still getting 8 hour WU's.

I'm not sure what I was thinking, or not as the case maybe, I actually set my prefs on default then updated, I've now changed them to 2 hours. It seems to be working as they are all around 2 hours, the lowest 1:44:33 the highest 2:02:05.

---