Among other things, I am running work unit sulphur_hduv_000811111_0, also known as
Result ID 1547834 and Workunit 1019600. This work unit has recently started on my slow machine (two 550 MHz Pentium IIIs with 512MBytes RAM), and trickles very slowly because there are two other climate-prediction work units running there with shorter deadlines.

I manually suspended one of those other work units until I got a trickle from this one (and then resumed it the one I suspended). Nevertheless, I get:

Result ID 1547834, Computer 164427, Sent 19 Dec 2005 13:40:07 UTC, Time reported 1 Dec 2006 19:00:07 UTC, Server state Over, Outcome Unknown, Client state New

That implies that I either finished this work unit (definately not) or timed out (surely not, since it is not due until next December).

The system seems to have accepted my trickle, so I am puzzled that it reports this way.

---

Do not worry about the server state. Strange things sometimes happen in that area, yet the results are still viewable by the project scientists.

But running sulphur on a 550 MHz PC is going to be stretching the limits to be sure. Running 60 percent slower than slab per timestep, with two more phases...good luck.

Carl/Tolu use the server state to control what gets sent out.
e.g. At one stage, all finished results were set to \"Inactive\". I forget why now, but I suddenly had lots of slab models like this.
The latest is setting it to \"Over\" to prevent re-sending the model.
Carl has posted <a href=\"http://climateapps2.oucs.ox.ac.uk/cpdnboinc/forum_thread.php?id=3746#18635\"> here</a> on \"Over\".

You\'ll know your model has finished, good or bad, when the other two columns change.

Do not worry about the server state. Strange things sometimes happen in that area, yet the results are still viewable by the project scientists.

OK. Sure is misleading, though.

But running sulphur on a 550 MHz PC is going to be stretching the limits to be sure. Running 60 percent slower than slab per timestep, with two more phases...good luck.

I do not think that would be a problem, but it is not going to make much progress for a while. The deadline is Friday, December 01, 2006, so I probably have enough time to do it. But there are two slab models ahead of it, and only two processors, so it will not get much time until one of them completes. Those two are due October 16, and have a bit over 2000 estimated hours each to go. But by my calculations, they really have about 1425 hours to go. At 168 hours per week, that is less than 10 weeks until the sulfur 4.22 starts getting full time at a processor. The machine runs almost 24/7, down only when I run Windows on it for an hour or so a week.

---