Out of Work

[UBT - Chris Suddick]

I seem to be having a bit of a problem with Rosetta. I've been doing it for about 5 weeks now. About 2 weeks ago the work stopped coming in and my cache ran completely dry. I tried the Update button several times but to no avail so I Reset the project but that did no good. Fanally, I detached and reattached and that seemed to work OK.

Today it has done it again. Completely dry. This is good news for the BBC CCE which shares my Linux box 50-50 with Rosetta but not so good for Rosetta.

I'll give it a few hours and see if it cures itself but I guess I'm going to have to detach and reattach again.

Does anyone else see this?

Chris.

[UBT - Halifax-lad]

It is to do with the CPDN WU you have got going in there I am willing to bet that you have this in your BOINC logs

Using earliest-deadline-first scheduling because computer is overcommitted.

If you want to be sure of what the problem is you could always copy these two files from BOINC and e-mail them to me

stderrdae.txt & stdoutdae.txt both in the main BOINC directory usually some of the last files in there (my e-mail is on the forum just press the e-mail button)

[UBT - Chris Suddick]

Thanks for the help David which is possibly right. Unfortunately, the two files you mentioned haven't been written to since Dec 2nd 2005 which is probably the day I installed Linux on that PC and got it working. So, I'm not going to get any help from them.

Are there any Linux gurus who might know why the stderrdae.txt and stdoutdae.txt files might not be being written to? (Sorry about ending a sentence with a preposition - I'll go and shoot myself)

Chris.

[UBT - Halifax-lad]

OK that's fine, there is a add-on for BOINC that works with Windows but I can't fine one that would work with Linux

The Windows one is BOINC Debt Viewer so if anyone knows of a Linux one please let us know in this thread.

Now for an explanation of what is happening on your BOINC

As you have CPDN & Rosetta running on BOINC 50:50 the following is what BOINC does

Each project runs up a Short Term Debt (STG) & Long Term Debt (LTD), so BOINC plays a juggling act to keep both projects going and to honor the resource share & time worked on each project.

So on your computer CPDN has now run up a LTD and BOINC is letting it have its turn of processing, leave it running and Rosetta will soon build up a LTD, then BOINC will switch again.

[Temujin]

Are there any Linux gurus who might know why the stderrdae.txt and stdoutdae.txt files might not be being written to?

don't know about the guru bit but on my work linux box the file is boinc.log

[UBT - Chris Suddick]

There's no boinc.log on my Linux system that I can find. Dave had the right file names but they are just not being appended to. I had a look at the messages tab in boincmgr and Dave was spot on. There are several messages over the last few days about being overcommitted so I think I'll leave it be for a few days and let it sort itself out.

This explains an oddity I noticed the other day. I usually trickle about every 3 days on the BBC but the last one took over five days despite the seconds/timestep being lower. Seemed a bit strange at the time. Also, explains why my Rosetta RAC has been climbing over the last day or two.

Chris.

[Temujin]

There's no boinc.log on my Linux system that I can find.

How are you starting boinc?
maybe you're redirecting output to /dev/null  or some other file somewhere ??

[UBT - Halifax-lad]

There's no boinc.log on my Linux system that I can find. Dave had the right file names but they are just not being appended to. I had a look at the messages tab in boincmgr and Dave was spot on. There are several messages over the last few days about being overcommitted so I think I'll leave it be for a few days and let it sort itself out.

This explains an oddity I noticed the other day. I usually trickle about every 3 days on the BBC but the last one took over five days despite the seconds/timestep being lower. Seemed a bit strange at the time. Also, explains why my Rosetta RAC has been climbing over the last day or two.

Chris.

Oh to be right again  :lol:  :lol:

[UBT - Chris Suddick]

How are you starting boinc?
maybe you're redirecting output to /dev/null  or some other file somewhere ??

There are a couple of files run_manager and run_client, both are very similar with a single line:

cd "/root/BOINC" && exec ./boincmgr $@

and

cd "/root/BOINC" && exec ./boinc $@

I'm not sure where these came from, either from the Boinc installation or I created them. If the latter then I must have got it from a Google search or similar because I couldn't have worked it out for myself :scratch:

Chris.

[Temujin]

There are a couple of files run_manager and run_client, both are very similar with a single line:

they're the standard files that come with boinc for linux, I don't use them myself but would've thought they should produce the 2 files Dave mentioned earlier.
Oh well

[UBT - JohnR]

I have the same problem on my windows box that runs Climateprediction
and other projects.  Rosetta is the only other project that gives me the same problem as you.

If in Tasks I suspend CPND and then go into Projects and also suspend CPND
and then do an update to Rosetta, it then downloads work (last time I tried this).

You can then resume CPND.  Rosetta will not get new work you have to fool it
every time that other tasks have more than a few days work in the buffer.

[UBT - Halifax-lad]

It is best to leave BOINC to it and let it do what it does best, micromanaging that way only messed up STD & LTD.

Failing that get rid of CPDN as this is the one that causes any troubles