So while checking my account I noticed a block of tasks that were shown as in progress somewhat lower in the list after a bunch of completed tasks.

Looking at their names and looking at the list of tasks in my BOINC Manager I do not find them. Where did they go? How do I clean them from my online account so wingmen will not have to wait till they expire and wonder what I am doing with them since I do not have them.

---

You can't do anything other than just be patient at this point. Just wait it out.

-Dave

---

Are you running Boinc 6.10.x?, can't tell as your computers are hidden,
If so, try pressing the Button 'Show All Tasks'

Claggy

So while checking my account I noticed a block of tasks that were shown as in progress somewhat lower in the list after a bunch of completed tasks.

Looking at their names and looking at the list of tasks in my BOINC Manager I do not find them. Where did they go? How do I clean them from my online account so wingmen will not have to wait till they expire and wonder what I am doing with them since I do not have them.

Ghost Tasks - Tasks assigned to you but due to an error your BOINC client software never got the message.

---

SETI@home classic workunits: 93,865 CPU time: 863,447 hours
Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[

HAL 9000 writes
"Ghost Tasks - Tasks assigned to you but due to an error your BOINC client software never got the message."

Ok. So this is a bug or some such in the BOINC system? Seems there should be a way to clear them early so they can be properly assigned to someone so they are being processed.

I am using BOINC version 6.10.18 btw.

---

So this is a bug or some such in the BOINC system? Seems there should be a way to clear them early so they can be properly assigned to someone so they are being processed.

"Earlier" is a relative expression;-) Seti has long deadlines, so ghosts get more press time here than in other projects that might have 5 day deadlines. Also, the boinc client/server interaction is designed to not require human intervention ... you set it and forget it.

So this is not a bug per se, rather, it's a reasonable design compromise to contend and scale up when dealing with millions of separate active contracts. The back end seti server really only knows that it gave you a WU of a certain deadline, and eventually a result set either comes back (including the possibility of an abort) or it does not. To track any other client/server progress is unnecessary to the end goal, as Time (if the form of patience first followed by time outs when necessary) correct the problems.

But a method still exists to achieve what you ask, but to me it sounds pretty painful and involved. I've never tried this, but I've read here if you set your project to "No New Tasks," and let your work buffer run out, and then detach from the project, rejoin, and then detach again ... that any WU's the backend system thought were still assigned to you from that project would be NULLED and added back into the general pool. <-- But lets see if anyone confirms those steps before trying them out.

...
But a method still exists to achieve what you ask, but to me it sounds pretty painful and involved. I've never tried this, but I've read here if you set your project to "No New Tasks," and let your work buffer run out, and then detach from the project, rejoin, and then detach again ... that any WU's the backend system thought were still assigned to you from that project would be NULLED and added back into the general pool. <-- But lets see if anyone confirms those steps before trying them out.

Good except for the second detach, should be attach.

If you are running optimized apps, you also need to re-install them. You may also want to back up your local copy of statistics_setiathome.berkeley.edu.xml if you want to keep the stats in BOINC Manager.

I would agree that if you have 1 or 2 ghosts then it is not worth the effort. If you have dozens, or hundreds then it may by worth while.

HAL 9000 writes
"Ghost Tasks - Tasks assigned to you but due to an error your BOINC client software never got the message."

Ok. So this is a bug or some such in the BOINC system? Seems there should be a way to clear them early so they can be properly assigned to someone so they are being processed.

I am using BOINC version 6.10.18 btw.

As other people have already posted, S@h processes millions of tasks per week. A few hundred or thousand get lost, but they eventually time out and get re-sent. That is what many of your long-term pendings are. Sometimes hosts download thousands of tasks and just dissapear!

BOINC does have a feature to check that hosts do have the tasks they are supposed to have, but it puts a lot of strain on the database, so it is currently turned off for SETI, and possibly will never be turned on again.

The project staff appear to be satisfied with the "status quo". It is not perfect, but it works well enough most of the time.

The data that SETI is processing was recorded months or years ago, and has taken hundreds of years to come from distant stars. Another couple of weeks is not going to make a lot of difference, apart from some peoples pending credit.

As other people have already posted, S@h processes millions of tasks per week. A few hundred or thousand get lost, but they eventually time out and get re-sent. That is what many of your long-term pendings are. Sometimes hosts download thousands of tasks and just dissapear!

Actually, none of them get lost.

As you point out, BOINC has to handle the case of disappearing systems. They have no idea what the owner is likely to do with a computer, including formatting the drive or throwing it away.

The case of work that is assigned, but never shows up in the assigned client gets handled by the same logic that handles the disappearing host.

But all of those errors get handled eventually, and the work eventually reaches two hosts that both complete the work.

... and the extra load of detecting those early is expensive.

---