WUs with a deadline in November and December? It's early October!

WTF do I want these for? I'm processing WUs at an average speed of 34m 36s. Even my slowest system averages 3h 5m a WU.
They all run 24/7 with a 97 to 99 percent BOINC dedication, and only one project - SETI.
I have a 3.3 day work buffer and an always-on connection.

Being given WUs with 30 to 60 day deadlines is just plain daft. These should be reserved for 386sx systems running just 2 or 3 hours a week.

Worse, they probably are going to those systems too and I'll be waiting till December for them to complete and get my credit. No wonder my pending credit is so high and my RAC is lagging.

So I'll just carry on aborting anything more than 10 days off...

---

WUs with a deadline in November and December? It's early October!

WTF do I want these for? I'm processing WUs at an average speed of 34m 36s. Even my slowest system averages 3h 5m a WU.
They all run 24/7 with a 97 to 99 percent BOINC dedication, and only one project - SETI.
I have a 3.3 day work buffer and an always-on connection.

Being given WUs with 30 to 60 day deadlines is just plain daft. These should be reserved for 386sx systems running just 2 or 3 hours a week.

Worse, they probably are going to those systems too and I'll be waiting till December for them to complete and get my credit. No wonder my pending credit is so high and my RAC is lagging.

So I'll just carry on aborting anything more than 10 days off...

. . . i believe that Dr. Korpela has said that he was goin' to 'fix' that problem (long deadlines) just recently /

So I'll just carry on aborting anything more than 10 days off...

Aborting them just makes someone else crunch them instead.

We all know that the deadlines (for all angle ranges) is higher than it needs to be, but that's not the real problem.

The real problem is: people are afraid of EDF.

When we had shorter deadlines, there was some urgency, and BOINC turned those tight deadline WUs around fast to avoid missing deadlines -- by using EDF.

But people complained.

So, now the deadlines are nice and long and everyone complains about it.

If you have your "connect every" interval at 10 days (the maximum) then pretty much everything is coming back in about 20 days unless things go badly wrong.

Doesn't matter what the CPU might be.

---

If you don't want them I'll take them! I'm in no hurry... :)

---

PROUD MEMBER OF Team Starfire World BOINC

So I'll just carry on aborting anything more than 10 days off...

Ned is right, aborting work manually is just not right. Boinc will take care of it if it needs to be aborted.

EDF is your friend!

WUs with a deadline in November and December? It's early October!

WTF do I want these for? I'm processing WUs at an average speed of 34m 36s. Even my slowest system averages 3h 5m a WU.
They all run 24/7 with a 97 to 99 percent BOINC dedication, and only one project - SETI.
I have a 3.3 day work buffer and an always-on connection.

Being given WUs with 30 to 60 day deadlines is just plain daft. These should be reserved for 386sx systems running just 2 or 3 hours a week.

Worse, they probably are going to those systems too and I'll be waiting till December for them to complete and get my credit. No wonder my pending credit is so high and my RAC is lagging.

So I'll just carry on aborting anything more than 10 days off...

I think the consensus sentiment is: why do you care? The only reason anyone should care about such deadlines is that the long deadlines put extra burden on the servers, to keep the information lying around for months, in the worst case. We 'clients' don't really care as we will get credit in the end (unless seti shuts down and then we are all screwed).

On the other hand, long deadlines enable the slow-end of our client distribution to make a contribution, as you noted. Instead of worrying about deadlines, we should worry about minimizing our queues. If everyone had minimum queues the issuance of credit would be a lot faster and the server storage burden would be minimized.

There was a poster on these boards once who had a good philosophy you should consider: Just chill out, pop a beer, and stop trying to out boinc boinc.

So I'll just carry on aborting anything more than 10 days off...

Interesting snippet from MikeSW17's profile:

3. Ignore the silly a**es that try to micro-mange BOINC and then get upset at the result of their meddling -- left alone, BOINC does it all just right.

Mike, please follow your own advice. Don't micro-manage, and that means don't abort work that doesn't need aborting. At the very least, it means someone else has to waste Berkeley's, and their own, bandwidth downloading it. There are also database update costs, and if the re-issue goes to a slower computer, medium-to-long term storage costs. As we have seen from the weekend just past, the SETI end-to-end workflow system is balanced on a knife-edge: anything which destabilises that balance for selfish reasons is antisocial.

I do wish that people would remember that they cannot earn credit at SETI by their own unaided efforts. It requires a three-way collaboration between Berkeley, the rest of the SETI user community (represented by your wingman), and yourself. Please factor that three-way collaboration into your decision-making processes - if you damage any part if it, you will suffer the consequences too.

Ned is right, aborting work manually is just not right. Boinc will take care of it if it needs to be aborted.

EDF is your friend!

What is EDF?. Could someone explain please.

PJ

---

What is EDF?. Could someone explain please.

Earliest Deadline First

Ned is right, aborting work manually is just not right. Boinc will take care of it if it needs to be aborted.

EDF is your friend!

What is EDF?. Could someone explain please.

PJ

It just means that if SETI/BOINC sees a deadline that is too close it will grab that WU and crunch it first so that it will get in on time. If the deadlene is really close it may even suspend a WU it is already crunching and work on the close one. It is really no big deal but it keeps things running a bit smoother.

---

PROUD MEMBER OF Team Starfire World BOINC

It just means that if SETI/BOINC sees a deadline that is too close it will grab that WU and crunch it first so that it will get in on time. If the deadlene is really close it may even suspend a WU it is already crunching and work on the close one. It is really no big deal but it keeps things running a bit smoother.

Thanks Perryjay.

---

So I'll just carry on aborting anything more than 10 days off...

Interesting snippet from MikeSW17's profile:

3. Ignore the silly a**es that try to micro-mange BOINC and then get upset at the result of their meddling -- left alone, BOINC does it all just right.

Mike, please follow your own advice. Don't micro-manage, and that means don't abort work that doesn't need aborting. At the very least, it means someone else has to waste Berkeley's, and their own, bandwidth downloading it. There are also database update costs, and if the re-issue goes to a slower computer, medium-to-long term storage costs. As we have seen from the weekend just past, the SETI end-to-end workflow system is balanced on a knife-edge: anything which destabilises that balance for selfish reasons is antisocial.

I do wish that people would remember that they cannot earn credit at SETI by their own unaided efforts. It requires a three-way collaboration between Berkeley, the rest of the SETI user community (represented by your wingman), and yourself. Please factor that three-way collaboration into your decision-making processes - if you damage any part if it, you will suffer the consequences too.

You're absolutely right Richard, don't micro-manage.

This is the first time in about 4 months that I have even bothered to look at any of my BOINC systems. I even missed passing 2,000,0000 (my Sig supplier isn't updating stats) credits for days!
BOINC just gets on with the job.

The only reason I'm looking now is I've just resurected a core2duo machine that died a few months ago and wondered why its credit and RAC was growing so slowly. Now I know - silly deadlines.
If I want to get this machine up to speed quickly and be able to forget about it, I don't need to be paired-up with screen-saver run only machines turning in a task every fortnight.

---

So I'll just carry on aborting anything more than 10 days off...

Interesting snippet from MikeSW17's profile:

3. Ignore the silly a**es that try to micro-mange BOINC and then get upset at the result of their meddling -- left alone, BOINC does it all just right.

Mike, please follow your own advice. Don't micro-manage, and that means don't abort work that doesn't need aborting. At the very least, it means someone else has to waste Berkeley's, and their own, bandwidth downloading it. There are also database update costs, and if the re-issue goes to a slower computer, medium-to-long term storage costs. As we have seen from the weekend just past, the SETI end-to-end workflow system is balanced on a knife-edge: anything which destabilises that balance for selfish reasons is antisocial.

I do wish that people would remember that they cannot earn credit at SETI by their own unaided efforts. It requires a three-way collaboration between Berkeley, the rest of the SETI user community (represented by your wingman), and yourself. Please factor that three-way collaboration into your decision-making processes - if you damage any part if it, you will suffer the consequences too.

You're absolutely right Richard, don't micro-manage.

This is the first time in about 4 months that I have even bothered to look at any of my BOINC systems. I even missed passing 2,000,0000 (my Sig supplier isn't updating stats) credits for days!
BOINC just gets on with the job.

The only reason I'm looking now is I've just resurected a core2duo machine that died a few months ago and wondered why its credit and RAC was growing so slowly. Now I know - silly deadlines.
If I want to get this machine up to speed quickly and be able to forget about it, I don't need to be paired-up with screen-saver run only machines turning in a task every fortnight.

Just an opinion, as is yours........
You might wanna watch who you are calling a silly a**.
Some of us like to micro manage our rigs, and do NOT complain when things do not work out to our expectations. Some of us do it for the sport, and to see how the system reacts to different inputs. I happen to try a lot of different settings to see how things work out.
I have done a lot of testing for Simon and Crunch3r, and currently Francois, and would be a bit offended if someone considered me a silly a** for doing so.
One of the reasons that the stock app is crunching so nicely these days is that a few of us silly a**es have taken the time to test and tweak and coddle the system to see what works best.
Blanket statements like that are rarely the wisest.
And if I mistook your comment for being antagonistic, I apologize to you.

---

"Time is simply the mechanism that keeps everything from happening all at once."

The only reason I'm looking now is I've just resurected a core2duo machine that died a few months ago and wondered why its credit and RAC was growing so slowly. Now I know - silly deadlines.
If I want to get this machine up to speed quickly and be able to forget about it, I don't need to be paired-up with screen-saver run only machines turning in a task every fortnight.

You are assuming that most machines just barely beat the deadlines.

I'm waiting for three work units now, the machines crunching them average about 15 days.

That's a long way from two months.

The rest seem to validate and get credit almost instantly.

I'm currently carrying four days of work, and connecting once/day to report/pick up work.

Edit: I just reported three, and got credit instantly. The other three machines average 18 hour turnaround times.

---

WTF do I want these for? WUs with a deadline in November and December? It's early October! So I'll just carry on aborting anything more than 10 days off...

WTF would you rather have NO work?

I will agree the deadlines are a bit long, maybe there is a reason for this that I am not aware of?

Here is 1 example of a WU that may take 6+ Months to complete the quorum.

WU: http://setiathome.berkeley.edu/workunit.php?wuid=147070191 , issued on August 13th to me and fellow cruncher. My 1st partner did not return the WU by October 4th so it has been reissued to another cruncher. It looks as though my 2nd partner is not returning WU’s either. November 27th is his deadline. If this WU is reissued to 3rd person and they actually complete and return it, could be 5-6 Months. If they don’t, could be longer…maybe 8+ months? At some point there are going to be freak occurrences where someone could go a year (maybe even longer) before a WU’s quorum is met with these long deadlines.

WTF, aborting WU’s with a deadline of 10 days or less. Right now the earliest deadline in my entire cache of 130 WU’s is October 30th. WTF, should I follow your lead and abort them all?