I run offline, and connect only infrequently. I have a fair amount of 28-day-deadline WUs still queued on my system. I was surprised the last time I connected, when 7-day-deadline WUs arrived and started to 'preempt' the work already queued up. I wanted to "clear up" the old work before accepting yet more 'new' work. So I specified to BOINC Manager "no more work from rosetta" (it's the _only_ BOINC project on that system).

But when I connected this time, MORE 7-day-deadline WUs were downloaded -- __despite__ me having specified "no more work". WHY is the rosetta project not giving me control over what work my system does ?
.

But when I connected this time, MORE 7-day-deadline WUs were downloaded -- __despite__ me having specified "no more work". WHY is the rosetta project not giving me control over what work my system does ?
.

Are you sure your really specifying not to get any more work from Rosetta.

When the Tab shows No New Work you will actually get more work & when the Tab shows Allow New Work is when you won't get any more work from Rosetta.

It really can be a little confusing until you get used to which way the Tab should be ... IMO

To see if it is set

-BoincManager
-Projects Tab -> Rosetta line
-> at the end should be status, it will say "Won't get new work"

---

Team mauisun.org

Mikus
I have had Rosetta do the same when set to NNW, this is the only program that dues not obey the NNW command.

Have you tried setting the connect time in your preferences to a shorter time like .5 days when you have work left and you don't want new work? You have to update that from another program or Rosetta will request more before it updates your connect time.

Ray

---

Pizza@Home Rays Place Rays place Forums

I have had Rosetta do the same when set to NNW, this is the only program that dues not obey the NNW command. Ray

It does work, I have never gotten new work from the Rosetta Project when I didn't request any ...

...(it's the _only_ BOINC project on that system).

But when I connected this time, MORE 7-day-deadline WUs were downloaded -- __despite__ me having specified "no more work". WHY is the rosetta project not giving me control over what work my system does ?
.

It's the boinc client, that decides if a computer needs new work, the Rosetta application is not involved. And (IIRC) if there are no other projects to pull from (or all other projects have no WUs) then work is fetched from NNW-projects.

Norbert

---

After selecting "No new work" it may be necessary to click on "Update" to insure that BOINC re-checks your Preferences... That worked for me when I quit crunching here because of the W/U's humongous size... All puters just run out of work...

And (IIRC) if there are no other projects to pull from (or all other projects have no WUs) then work is fetched from NNW-projects.

hi Norbert,

There are three different things, and it is easy to get them confused.

NWF: no work fetch if the computer seems overcommitted.

LTD: The client will akso not fetch work if the project has a large negative long term debt

NNW: The user has specified No New Work

You are right that NWF and LTD are ignored if there is a vacant cpu and no other project has work.

However, NNW should still be respected by the client. This is how a user would finish gracefully if they want to leave that project or want to leave BOINC altogether.

River~~

Since before the situation I first posted about, in (Linux) BoincManager on the 'Projects' panel the only line (it's for rosetta@home) has had at the left of its 'Status' field (and still has): "Won't get new work".

An update: I connected today (first connect after posting) to report accumulated results, and new work was __not__ downloaded. However (as advised here) I had earlier changed (on the web) the connect time in my preferences to a short value, and immediately upon allowing network activity at the client I clicked on 'Update' for the project. So I don't yet know whether the 'not_downloading' of new work was due to the "Won't get new work" status, or due to my existing queue still holding more 4-week WUs than can be processed in the shorter connect interval that is now in effect. At least now all the 1-week WUs that pre-empted the existing queue have been reported.

I'll watch when I next connect (once my 4-week WUs are completed) to see whether the project does/does_not download new WUs despite the rosetta "Won't get new work" status.


---------
p.s. Note that 10 days ago I had a situation where the project _kept__ downloading more and more WUs to my system, DESPITE claiming that work fetch was suspended because the "computer is overcommitted" and DESPITE switching to earliest-deadline-first scheduling because of all the WUs it was downloading. (Perhaps confusion was caused by the arrival of long-duration WUs when the client's internally calculated "average WU duration" value was still short.) Anyway, that's how come the queue of 4-week WUs at my system got to be so long.
.

It does work, I have never gotten new work from the Rosetta Project when I didn't request any ...

Today it WORKED for me the way it is supposed to.

I finally worked off the oversupply of 28-day WUs that were downloaded to my system on Jan 24. With zero WUs in either 'running' or 'ready' state, when I did an 'Update' all it did was report the uploaded WUs -- it did *not* download any new work as long as 'No new work' was shown in the 'Projects' panel. When I "un-set" the 'No new work' state for rosetta, it immediately started downloading 7-day WUs.

mikus

--------
p.s. Right now I have specified (via the website) in my preferences: 'connect every 8 days'. Right after I clicked on 'Update', the client put out the message: "Using earliest-deadline-first scheduling because computer is overcommitted'. This is BEFORE the first download of a WU could complete !!

All I can think of is that it reacted stupidly to being allowed to download eight day's work, when what it was downloading all had to finish in seven days. Stopping the download after getting seven day's work would have been smarter.
.

Mikus
When you got the first 7 day deafline units did it go into EDF part way through crunching the 28 day units?

It should have done the first downloaded first, which if you had a lot of the 28 day units could have put you into EDF as the new ones got closer to the deadlines.

I imagine that happened to some who we don't here from over here on the boards, mine came close but not quite.

I sure wish that they cut if back to 14 days rather than 7 days. I was going to get an older system that I am not using loaded up with them (when they had the 28 day units) and bring it to a storage unit to run on there power. Than every week to 2 weeks pull the HD to bring home and turn them in and get new ones. But the 7 day units are not good for that.

Cheers
Ray

---

Pizza@Home Rays Place Rays place Forums