I see that rosetta/boinc will randomly choose WU's to process. I wonder why it doesn't sort them by priority via the date of the report deadline? Example would be report deadline for several WU is 08/11, however boinc will randomly choose and process several with later report deadlines(after 08/11). This always results in several WU not being crunched by their deadline.
Thank You

---

shimp

I see that rosetta/boinc will randomly choose WU's to process. I wonder why it doesn't sort them by priority via the date of the report deadline? Example would be report deadline for several WU is 08/11, however boinc will randomly choose and process several with later report deadlines(after 08/11). This always results in several WU not being crunched by their deadline.
Thank You

Boinc Manager handles task scheduling, not Rosetta, and it's infamous for having problems doing so. As such, I think you're not helping the situation by having 1 hour run times and what looks like a 7 day buffer of over 300 tasks waiting to run. Boinc doesn't need any help in messing up its scheduling and so many tasks is almost inviting trouble. It can't be helping you or the tasks either.

Unless there's a pressing reason I'm unaware of I suggest you tweak your run time up to 2 hours for the next couple of weeks and maybe reduce your buffer by a day or so while it settles down, and if that helps maybe get it back up to the default of 3 hours. Juggling 100-ish tasks is bound to be much easier and more successful than juggling 300.

Is there a different problem you're trying to solve with this short run-timelong buffer choice?

---

Actually, the BOINC Manager isn't that bad... ;)

There could be several reasons, why this happens (a single long running model could cause trouble like this), but I guess your cache is just to big.

Under no circumstances you should increase the running time, before reducing the cache. If you already have problems with the deadlines, raising the running time will emphasis this problem (i.e. increasing the dafault running time from 1h to 2h will effectively turn a 7 days cache into a 14 days cache).

I recommend, that you reduce your cache to 2 days, before you increase the running time. And I'm afraid, but you will have to wait, until your cache is really down to a 2 days cache.

---

Actually, the BOINC Manager isn't that bad... ;)

I thought it was fixed, but my main computer has reverted to grabbing one spare task at a time instead of the 2 days worth I've set it to. I have no idea why. I have fewer back-up tasks with 2 days than I did when I had just 0.33. The wierdness seems to have returned. :(

---

Actually, the BOINC Manager isn't that bad... ;)

I thought it was fixed, but my main computer has reverted to grabbing one spare task at a time instead of the 2 days worth I've set it to. I have no idea why. I have fewer back-up tasks with 2 days than I did when I had just 0.33. The wierdness seems to have returned. :(

What are Long Term Debt(LTD) numbers in your client_state.xml file? How about some of the numbers under Boinc Manager, Projects and then properties? What is your Duration Correction Factor?

Actually, the BOINC Manager isn't that bad... ;)

I thought it was fixed, but my main computer has reverted to grabbing one spare task at a time instead of the 2 days worth I've set it to. I have no idea why. I have fewer back-up tasks with 2 days than I did when I had just 0.33. The wierdness seems to have returned. :(

What are Long Term Debt(LTD) numbers in your client_state.xml file? How about some of the numbers under Boinc Manager, Projects and then properties? What is your Duration Correction Factor?

Just got round to looking at this and in the last hour everything's starting to catch up. Part of the reason, I'm sure, is that earlier this evening I set WCG to No New Tasks. But fyi:

Rosetta
<short_term_debt>86400.000000</short_term_debt>
<long_term_debt>0.000000</long_term_debt>

WCG
<short_term_debt>-86400.000000</short_term_debt>
<long_term_debt>-147059.688130</long_term_debt>

---

Actually, the BOINC Manager isn't that bad... ;)

I thought it was fixed, but my main computer has reverted to grabbing one spare task at a time instead of the 2 days worth I've set it to. I have no idea why. I have fewer back-up tasks with 2 days than I did when I had just 0.33. The wierdness seems to have returned. :(

What are Long Term Debt(LTD) numbers in your client_state.xml file? How about some of the numbers under Boinc Manager, Projects and then properties? What is your Duration Correction Factor?

Just got round to looking at this and in the last hour everything's starting to catch up. Part of the reason, I'm sure, is that earlier this evening I set WCG to No New Tasks. But fyi:

Rosetta
<short_term_debt>86400.000000</short_term_debt>
<long_term_debt>0.000000</long_term_debt>

WCG
<short_term_debt>-86400.000000</short_term_debt>
<long_term_debt>-147059.688130</long_term_debt>

You CAN edit those with like Windows Notepad but you MUST stop Boinc AND you must be VERY CAREFUL, if you mess these up it will mean a major Boinc problems!!!!!!!!!!

Basically all you do though is exit from Boinc so it is NOT running, open the file and edit the number to be something like they are but with the number to the left of the decimal point to be a one instead of what it is. If the numbers are already zero leave them alone though. Then save the file and restart Boinc and after just a few seconds you should be getting lots of work again. BUT as stated preciously if you don't do anything it WILL fix itself! What changing the numbers does is get everything back to near normal, as you want it to be, right now. The numbers WILL go back to being funky again in time due to projects being up or down or not having the kind of work you want etc, etc, etc.

What would I want these numbers to show anyway? What does a negative number mean in this context?

---

If you don't wish to tempt fate by directly editing the file, you can use the --set_debts command line argument to boinccmd described here.

---

Rosetta Moderator: Mod.Sense