Well we're out of work again. According to latest Server Status there are about 275,000 out in the field and nothing to send.

Any idea when the next batch of work will be created and if so what type of wu?

I managed to pickup a resend of a HADAM3P_EU but thats all I have managed to get.

Thanks,
MarkJ

I've made a News item about this.

---

Backups: Here

Thanks for the reply Les.

Do we know if they will be waiting on all 275,000 wu to finish then before they produce another set of wu? Presumably they will have an idea how they are progressing, especially the Rapit ones seeing as they have a shorter deadline, to decide if another set are required or not.

There are 2 lots of info on the RAPIT experiment:
RAPID-RAPIT: What is the risk of thermohaline circulation collapse?
Welcome to the RAPIT experiment!

Information about the 3 Regional models are scattered a bit over the 2 boards.

No other information is available or expected.

---

Backups: Here

Dear MarkJ

Most of those 275,000 WU are likely lost units that died in ways that server never heard about (like a hard disk crash or someone uninstalling boinc with a running WU) and the server has just not given up on them yet.

Its going to be a long time before most of those WUs call home.

---

I seem to recall a script the project admins ran that detected idle workunits. If there is a workunit with "In Progress" tasks, it checks that at least one trickle has been received in the last 6 weeks or so. If not, they could cause those tasks to abort and resend.

I think running such a query this weekend would be insightful.

Right now, there's about a thousand hadcm3n available for download.
That changes every hour.
I'm only guessing , but I believe that the project admins are not using the BOINC infrastructure all that much -- not for timeout, reissue, verified -- what I guess is that when they get enough results for the models that interest the researchers they issue another batch with parameters that will refine those results. Or, with hadcm3n when there are enough results from one stage to issue wu for the next stage -- or reissue the lost ones.
The BOINC infrastructure is not real well suited to the climateprediction project, but it's what's available, so, as has been posted here many times, some of the the BOINC concepts just don't apply. CPDN has such __long-running__ models.

That 275000 models thing is a BOINC artifact -- don't give it moment's consideration

Right now, the longest hadcm3n models take a month or so.

Me , a geezer, years back, ran a few that took 4 months.

I believe that the tech crew are doing a good job issuing high-priority wu as needed by the science people. It's just that it takes a long time to know when wu aren't going to finish.

Keep on crunching.

Eric

---

It seems there has been some work generated since this message thread started.

I managed to pickup a bunch of regional models on 1 machine and there are also 126 HADCM3N's ready to go. Once I knock over the regional work units i'll see if I can grab a couple of them before they all disappear.