I just had the strangest thing happen. When checking the status of my various projects a few minutes ago, I noticed that CPDN wasn\'t updating the WORK DONE counter in BOINC. I pressed the UPDATE button and the CPDN client promptly reported my task as complete and deleted it. I was actually only less than 1% complete as it was a new task. I had previously noticed that the end-date had been changed to 2010 but thought little of that since CPDN apparently doesn\'t enforce end-dates anyway.

Anybody got a clue what happened, because I sure don\'t.

Cheers, all.

bob graham in boise, idaho, usa

---

Hello, Bob.

The model is presumably this one: 7855805 - in which case it has, unfortunately, crashed. When the update button was pressed, BOINC simply reported the task that had crashed previously.

The error message at the bottom of the dump for that model is not one I\'ve seen before, \"Model crashed: READHIST: End of file in READ from history file for namelist NLIHISTO tmp/pipe_dummy 2048\". But it was probably due to closing down the machine while the model was running. The PC is not the speediest, and stopping BOINC manually is advisable in order to ensure a smooth closedown. In fact, at a timestep interval of 12 seconds, the model would take 19 months to complete running full-time.

The model before that one, 6972023, completed successfully. It was a \'slab\' model, which is considerably shorter. I would suggest you change the climateprediction.net preferences in your account (here) to select only slab models - i.e. HADSM3. That way you should have a run of completed models.

Iain