I received a similar message today.

</stderr_txt>
<message>
upload failure: <file_xfer_error>
<file_name>E225098_640_S.320.C44H28N4.MUTYYGLCJGKFND-UHFFFAOYSA-N.4_s1_14_0_1</file_name>
<error_code>-131</error_code>
</file_xfer_error>

Hi all.

I think this is a BOINC issue rather than something my side, and I will raise it with the IBM guys. I do note that these are some of the largest molecules we put out, but since there are less jobs than before, I would guess that this was not the issue. Will keep everyone in the loop.

Your Harvard CEP Team

Well, I couldn't be left out of the party, now could I?

</stderr_txt>
<message>
upload failure: <file_xfer_error>
<file_name>E225106_934_S.326.C42H30N4S1.XSTKVABCJZSPJF-UHFFFAOYSA-N.5_s1_14_0_1</file_name>
<error_code>-131 (file size too big)</error_code>
</file_xfer_error>

CJSL

Crunching for a brighter future...

I think it's a work unit build up, and a bad exspection of results. Have to remember we're dealing with quantum mechanics calculations and they went to one basic molecule/electron configuration.

It's my guess they rushed the work when creating the units by giving a base for all that, maybe a ring one or two +/- legs/arms hanging off, and other atoms exchanged and about. A kind of Mr. Potato head. Well it looked good in theory and Bata may have been just to see if the basic idea worked. Then the real world hit and the electron clouds became unexpected for the validation program at the time. That's because each electron structure was laid over a basic pattern, but each different molecule was far from that exspection and the fast build idea went boom. Remember it was only about a month+ when they said one person changed how they decide what molecules they test, then it went to Bata stage.

So they went into that (validation problem) seeking a fast fix about a week ago. Found they had a deeper problem and need to rebuild the units; and they just recently said that.

Also remember they started off about a week ago saying they had everything they needed to correct the problem, then the whole project fell apart.

That's my best guess, wonder if I'm right. It's the only current theory that fits what went on. Anyone got anything better?? I blame Harvard.

The file size too big is a BOINC configuration thing. It has happened on the MCM project a few weeks ago (see thread) and a configuration change fixed the problem for new workunits and resends.

I don't know what the currently configured max result size is for this project, I know it does have some fairly large result sets which have returned successfully. So unless the CEP scientists or IBM guys notice something wrong with these WU, other than the larger result size , then it needs someone like uplinger to update the configuration to allow the larger files.