Hi there.
Hopefully someone might have a quick answer for me. I'm running three projects with equal resource share on my computer - one of them being from climateprediction.net.
All was working fine until recently I noticed that the climateprediction job is stuck at 24.986%. It says that it is running, and the elapsed time increases and the time remaining counter counts down. However, the Task Manager program indicates that the job is not being crunched.
If I suspend the climateprediction job, the other job waiting to run (world community grid) kicks in and starts crunching away as it should. The task manager indicates that the cpu is 100% busy for that job.
Switch back to the climateprediction job, and BOINC states it is running, but again, task manager states it isn't.
I just upgraded BOINC manager a few days ago to version 6.12.34.
Could this have been a problem?
Any help would be greatly appreciated.
Thanks
-Tom-

Hi Tom

This type of model is, unfortunately, very touchy about being interrupted in any way at the point at which it zips up the data to return it to the project. This occurs every 25%, so it looks like you've lost that model.
25% for the model, can equate to a range of percentages in the BOINC manager, which is only approximate.
They shouldn't be interrupted between late November and about the 5th of December. (Well after the next check point.)
So about the only thing to do is to cut your losses and Abort it.

The currently available models, the PNW regional models are a lot more robust.

---

Backups: Here

I have what is apparently a similar problem. My climateprediction task says it's running, but I doubt it. The graphics display says it's 99.79% done (7 December), but the BOINC manager says 100% and running. It isn't doing anything. I've another climateprediction task that was waiting to run, so I switched to that one, and then back, but to no avail. So I've just continued running the new task, which seems to be doing what it should. Shall I abort the finished or nearly-finished task?

---

Shall I abort the finished or nearly-finished task?

Unfortunately, that's what you'll have to do. You may well find that just stopping and restarting BOINC (or the machine) will cause that model to crash. Very frustrating so close to the end ...

BOINC manager says 100%

When BOINC loses contact with a work unit, in this case a climate model, it's programmed to assume that the wu has finished, and then displays 100% completion.
So, as Iain says, Abort it.
And as I said earlier in this thread, be careful about letting anything interrupt the running model anywhere near it's "zip points".

---

Backups: Here