Hi,

I am running seti@home (currently stalled Dec-9-05 with growing pains, Climate Prediction and Rosetta.

I have a dual core processor which runs two separate WUs at the same time. Seti is out of the picture at the moment as none of my completed WUs are being uplaoded to the server and one DL from the seti server has been stuck "downloading" all morning. Rosetta has been downloading regularly until today and now the last completed WU has left my machine but no new Rosetta WU has been downloaded.

Seti, Climate & Rosetta each have 33% assigned resources but for the last several hours the only project running is Climate as it takes me aproximately one month to complete their WUs and I still have two weeks of crunching left to do on this current WU.

I clicked on Rosetta in the projects field and clicked on update but no WU came down from this manual tickle.

What can I do to get new Rosetta WUs?

Thanks,

Gary

---

Hi,

I am running seti@home (currently stalled Dec-9-05 with growing pains, Climate Prediction and Rosetta.

I have a dual core processor which runs two separate WUs at the same time. Seti is out of the picture at the moment as none of my completed WUs are being uplaoded to the server and one DL from the seti server has been stuck "downloading" all morning. Rosetta has been downloading regularly until today and now the last completed WU has left my machine but no new Rosetta WU has been downloaded.

Seti, Climate & Rosetta each have 33% assigned resources but for the last several hours the only project running is Climate as it takes me aproximately one month to complete their WUs and I still have two weeks of crunching left to do on this current WU.

I clicked on Rosetta in the projects field and clicked on update but no WU came down from this manual tickle.

What can I do to get new Rosetta WUs?

Thanks,

Gary

Suspend cpdn short ;)

I have no problems with my P4 HT here running all the time one cpdn and 1 rosetta Wu and getting new work from rosetta :p

---

What I would do is this:
- In the Projects tab, choose "No new work" for Seti.
- In the Work tab, Abort the stucked SETI-download.

Then you should be able to get some more Rosetta-wu's.

Remember to choose "Allow new work" in the Projects tab when Seti is up and running again...

Good luck!

---

Greetings from Norway.
Crunching is fun!

Without the info from the Messages tab, we're all guessing at what the problem (if any) may be. If your computer is overcommitted on CPDN, or a Rosetta update says "not requesting new work" - in which case it's simply "in debt" to SETI, or, or, or...

Try suspending SETI and then updating Rosetta. That's the best guess, given their current problems, and one in "downloading" status - and if that doesn't get Rosetta work, you're probably better off _not_ forcing any to download. The scheduler is pretty smart, and you won't be idle until that CPDN WU finishes.

"No new work" and abort the download for SETI would be my second choice. Don't leave it in NNW for very long though, suspend is better if they aren't up within a day.

DON'T hit "reset" on anything. It won't help, and WILL hurt.

---

Hello all,

Seti's un-completed download seems to be stuck. Selecting it and entering "abort" is futile, it persists. Stopped & restarted BOINC but it's there like a bad penny.

I do now have Seti as "no new work" I earlier had seti selected to "no new work" but selected "allow new work" to see if the Seti server logjam was fixed. It was during this time the persistant seti DL appeared.

Rosetta did download a new WU after restarting BOINC but there must have been 3-4 hours without BOINC DLing a new Rosetta WU and I had earlier restated BOINC several times to induce it to DL a new WU with no success.

There are no Rosetta WU's in cache, just the one being crunched at this moment. I do have two CDPN wu's onboard but have suspended one of them. When this WU is completed in another 30 days or so, I'll resume the suspended one.

This computer is not overextended with CPDN; when only that one WU was available to run & there were no seti or Rosetta WU's to crunch, both processor cores fluctuated around 50% usage. When two WU's are being crunched at the same time there is a 100% usage of both cores.

(Nice to have dual cores & do two wu's at the same time)

So as it stands I don't know why the new Rosetta wu downloaded but at this moment there is none in cache.

Is there a method of storing a cache of several Rosetta WUs?

Thanks,

Gary

---

Rosetta is Downloading New WU's, I just Downloaded several Days worth to all my Computers ... :)

Is there a method of storing a cache of several Rosetta WUs?

Yes... If your cache is currently at the default 0.1 days, I would increase it to 0.5 or so. If that's not enough, try 1.0, and so on until you're content. DON'T go above 3 or 4 unless you really know how it's going to affect things... The cache is set on your General Preferences section under Your Account on the website. Be aware that it affects _all_ projects equally, this is not separate by project.

You still never said you suspended SETI. "No new work" will not remove SETI from the scheduler calculations, only Suspend will do this. You said aborting the downloading result didn't work - did you do this in the "Transfers" tab? I've never seen that fail...

And a reminder - you can't "trick" BOINC into downloading work for a project if the scheduler says it doesn't need any. Restarting, Updating, none of that is going to work. The ONLY thing that will work is changing what the scheduler is looking at. In this case, the problem is SETI, so you must suspend SETI to have it give Rosetta more work. Trust the scheduler... except when SETI is lying to it by saying "this work will be there any second now", it works in your best interests.

---

Without the info from the Messages tab, we're all guessing at what the problem (if any) may be. If your computer is overcommitted on CPDN, or a Rosetta update says "not requesting new work" - in which case it's simply "in debt" to SETI, or, or, or...

Try suspending SETI and then updating Rosetta. That's the best guess, given their current problems, and one in "downloading" status - and if that doesn't get Rosetta work, you're probably better off _not_ forcing any to download. The scheduler is pretty smart, and you won't be idle until that CPDN WU finishes.

"No new work" and abort the download for SETI would be my second choice. Don't leave it in NNW for very long though, suspend is better if they aren't up within a day.

DON'T hit "reset" on anything. It won't help, and WILL hurt.

Slight thread hijack here. Just trying to get the operation of LTD sorted out.

Boincview is nice in that it shows LTD, something I have not figured how to do with Boincmgr. That being said, some questions for you:

1. The sum of all LTD values for a given host will be zero. True or false? I believe true.

2. As a project does work its LTD decreases, and the LTD of all other projects on that host increases. True or false? I believe true.

3. A project will only fetch work if it's LTD is above some value (it seems to be in the order of -3,000 or so for my systems here). True or false? I believe true.

Please correct any misunderstanding I have about this.

It should be noted that I run Rosetta 99% and Predictor 1%, and the general way things appear to work is that Predictor will get some work, do it, and leave things where Rosetta has a huge +ve LTD (often anywhere from 30,000 to 60,000), and Predictor has a corresponding -ve LTD. Always the same number, just negative. Then as Rosetta does work, it's +ve LTD gets smaller (slowly but surely), as Predictor's -ve LTD gets smaller in magnitude (i.e. an absolute increase). Then when Predictor hits about -3,000 or so, it grabs more work, goes into EDF mode, clears the Predictor WU's and restores the status quo.

---

1. The sum of all LTD values for a given host will be zero. True or false? I believe true.

2. As a project does work its LTD decreases, and the LTD of all other projects on that host increases. True or false? I believe true.

3. A project will only fetch work if it's LTD is above some value (it seems to be in the order of -3,000 or so for my systems here). True or false? I believe true.

Almost perfect! 1) with more than two projects, the sum may only be "close" to zero.

3) the "some value" is your cache size setting.

:-)

---

1. The sum of all LTD values for a given host will be zero. True or false? I believe true.

2. As a project does work its LTD decreases, and the LTD of all other projects on that host increases. True or false? I believe true.

3. A project will only fetch work if it's LTD is above some value (it seems to be in the order of -3,000 or so for my systems here). True or false? I believe true.

Almost perfect! 1) with more than two projects, the sum may only be "close" to zero.

3) the "some value" is your cache size setting.

:-)

1 is true to a pretty good accuracy (i.e. last digit in the floating point number).

2 is true.

3 the "some value" is the negative of switch applications every X, not the cache size setting. If everything is running round robin, all of the LTDs will be between + and - switch applications every X.

---

BOINC WIKI

3 the "some value" is the negative of switch applications every X, not the cache size setting.

DUH! Obviously my brain was not engaged when I typed that... Thanks John!

While you're here... look at "Something wrong with Server-Side-Scheduler" thread, please?

---