In perusing WUs, I’m surprised how many instances there are where a person has downloaded ONE (1) WU, and then is never heard from again. I know this doesn’t stop, but only delays, WU completion due to WU being reissued.

What I wonder is: Has there ever been any effort to CONTACT these one-WU folks? Maybe see if they had some problem or need help, clarification, or encouragement?

Maybe if first WU is not returned by completion date …
(1) the server side sends a simple canned e-mail encouraging them & pointing them to forums or someplace
(2) the client side pops up a message suggesting <something> to remind & encourage them to continue participating


Since some projects are having longer and longer WUs, has there been any discussion of making the FIRST WU a new client gets be a TEST-WU which just ensures client can go thru process?

---

Quite a few people need more than one workunit to discover what it's all about, what boinc and the workunit do and what the person has to do and not do. The high user attrition rate probably has many causes, including the general incomprehensibility of many boinc manager messages for non-geek newbies. Some projects such as cpdn and BBC have a boinc manager button that goes straight to the project forum, but the button for many other projects takes members to the project website which is less helpful for people who don't know what they're doing.

But does a boinc manager project forum button bring all the members with crashing workunits onto the forums for help? Very rarely. Many would probably love to post but are afraid to expose themselves as incompetent.

At cpdn the admins and moderators have recently sent quite a few emails or private messages (which generate an email notification) to members whose username is obscene, the founders of teams with obscene names, members with a specific OS who need advice, and members who download their daily quota of workunits for months (hundreds or thousands of climate models) without ever actually crunching them.

Some of these people have taken action and some have made contact. I couldn't say yet what the response rate is but I don't think it's very high. This isn't surprising as we've targeted particular problem groups. But the contact strategy is still probably worth while as long as the privacy of members is not invaded with repeated messages from a project.

But in the case of people who just try one or two workunits and then decide that running boinc projects isn't what they want to do with their computer or spare time, this is their choice and I don't think we should be proactively contacting them. Proactive contact should be limited to members who appear to be trying to run workunits but failing over an extended period, and those who cause a problem for the project.

When WUs time out or crash I do think it would be useful if the boinc manager messages included something like 'Visit the ***** project forum for help with diagnosing workunit crashes/completing workunits before the deadline'. I'm planning a Trac ticket about BoincSpeak in the boinc manager messages in the hope that some of the lingo can be made more accessible and comprehensible to the general public.

In perusing WUs, I’m surprised how many instances there are where a person has downloaded ONE (1) WU, and then is never heard from again. I know this doesn’t stop, but only delays, WU completion due to WU being reissued.

What I wonder is: Has there ever been any effort to CONTACT these one-WU folks? Maybe see if they had some problem or need help, clarification, or encouragement?

Maybe if first WU is not returned by completion date …
(1) the server side sends a simple canned e-mail encouraging them & pointing them to forums or someplace
(2) the client side pops up a message suggesting <something> to remind & encourage them to continue participating


Since some projects are having longer and longer WUs, has there been any discussion of making the FIRST WU a new client gets be a TEST-WU which just ensures client can go thru process?

Strange, I thought Boinc was a GNU free project, based on volunteers free to choose what they want to do, to help one or many scientists, universities or companies and not the others because X or Y.
So, we're free to work on one project, but if we choose to quit, here is the newletters, spam or else.

May be is it just a hardware problem ?
Crash disk, restore, not enough memory or GHz, human error, ...

As Helfin, some wants to help on CPDN, but receiving the first WU and seeing an estimated CPU time of 5000 hours may cause a bad uninstall of Boinc, because, the CPU is not enough powerful, because the person don't want to let its computer running night and day because it enforced climatic reheating, or computer is too noisy to sleep, or else ...
Some project need too many ressources, and there's 2 solutions, quit or buy. I think it's easier to quit (PacMan or the latest 3D game?) ... for now.

Please, no mails, newletters or else, just freedom (and may be shorter WUs ?).

[quote]In perusing WUs, I’m surprised how many instances there are where a person has downloaded ONE (1) WU, and then is never heard from again. I know this doesn’t stop, but only delays, WU completion due to WU being reissued.
[\\quote]

All I can give you is my experiences.

I first connected to SETI because it gets a lot of publicity. When I connected up however, it was down after crunching just 1 or 2 WU's because of a server outage. Then when it was finally restored (weeks later) the work units kept failing to download successfully because the checksum failed. I finally tracked this down to my firewall settings. Then Seti work units kept bombing. Apparently they aren't set up for quad processors or something. There was a lot of cycle waste. I felt like my contribution was taken for granted too so I moved on to some other projects.

One of these was CPDN. Every time I connect to a new project I get flooded with work units so I learned how to control work manually and tune the parameters to stop the bad behavior. Imagine bening new to BOINC getting flooded with CPDN wu's. More than 1 is too many of those monsters. I've crunched for grid.org since 2000 but this was a trip.

CPDN was the second project I connected to. I about fell out of my chair when I saw the hours. Thought it was a mistake! No its for real. So the machine I'm doing crunching on is a dual boot machine. I can't run a WU on both OS's because of the way BOINC identifies a machine. If I stay on one of the OS's for too long I'll miss the deadline. Now I run a special resident program as a service along with BOINC to tightly control how much work is downloaded for any particular project. Trying to get BOINC modified for decent task loading behavior is an effort in futility I think. I moved CPDN to another single cpu machine machine I have that's up 24/7. I experienced several wu bombs at different percentages. I have yet to finish one CPDN wu successfully but there is one wu cranking now over half done. No I'm not going to do special backups to restart a CPDN that crashes. If that happens because I took a stop or something they need to look at the dump.

Finally I found a few projects (QMC and Einstein) that run a decent length work unit and are stable. uFluid is cool because its wu's are very short but they don't feed work all the time. That's OK.

I'll run Superlink when it gets back on line and that's enough tasks for my quad. I try hard to return back every wu I pull but because of the uneven behavior of BOINC I've had to scrap a few. My apologies to those projects that I wasted a little time on because I had to kill a wu.

Some folks won't put up with all this.

---

Ty < finally.. thinks he's got it fixed now