Looking over some of the work my computers were assigned yesterday when the project came back online. I find many groups of work that show "In Progress" on Seti's web site but are missing from my computers. Certainly they are "ghost work units".

Perhaps my memory is not as it should be but, I seem to remember a time where on Seti's web site the work would show as "downloading" and then switch to "In Progress" when the work was successfully downloaded and present on the client computer. Is my memory failing me? Again??

---

Boinc....Boinc....Boinc....Boinc....

I believe you are thinking of the BOINC Manager. I don't remember the webpage ever showing downloading/in progress. I think the webpage just says in progress until you report it.

---

PROUD MEMBER OF Team Starfire World BOINC

I seem to have "ghost WUs" also. The SETI site shows 8 WUs sent to my laptop (CPUID 4677975) now, but only one shows in the Task tab of the BOINC Manager (version 6.10.56).

I noticed "ghosts" a couple of days ago when my cache of about 30 disappeared and I was concerned then, but just let things alone. I had set my cache days from 10 down to 0 (zero) in preparation for my return to my home, and expected my laptop to be empty soon. I was surprised that the "ghosts" were in my machine! Yes, the WUs were here and are being processed, but I can't see them locally, only through the SETI website. I also set BOINC to NOT ALLOW NEW TASKS to prove the WUs were being processed from my computer.

My other projects also have "ghosts" now, so it is NOT unique to SETI, but seems to be a BOINC issue.

I could not find a thread that addresses this issue directly, though. Has another noticed this issue?

---

Darrell,
On your tasks tab in the BOINC Manager do you have a button marked "Show all tasks" ? If you do, click that. It should change to "Show active tasks" and you will see all your work units in the manager.


In case anyone wonders why I think it's the "Show all tasks" Button, Darrell gave a clue here..

Yes, the WUs were here and are being processed, but I can't see them locally, only through the SETI website.

---

PROUD MEMBER OF Team Starfire World BOINC

Ghost work units occur when the servers at Berkeley are overloaded with traffic. It's almost a DOS attack. (denial of service) Your client computer sends out a request for work. Berkeley hears the the request and responds by assigning your computer work and sending out the notice to your computer to download it. Meanwhile the notice of your new work never arrives at your computer and your request dies out with no response received.

This is most likely to occur during periods of high activity when Berkeley has been down for some time and then comes back online. Like the currently scheduled 3 day science fair where Berkeley goes down for 3 days and then everyone is trying to report their work and get more work, at the same time.

---

Boinc....Boinc....Boinc....Boinc....

BINGO! I have never (knowingly) clicked on the "Show active tasks" buttons, but your answer was dead on. Thanks!

---

I know I have got ghost units as my account shows over 100 and there was only 80 in my active task list, most of mine was due to the 3 day outage when they increase the Wus from 20 to the new figure before the outage I get http errors or not connect so sorry wingmen about this

---

Looks like I've picked up over 50 ghosts this time. This is getting a bit tiresome but I don't know what might be done to prevent them.

---

PROUD MEMBER OF Team Starfire World BOINC

Suspend network activity of your BOINC if the SETI@home server have high (maxed out) load.. :o)


I guess my host have 300 - 500 ghost WUs.
IIRC, this will damage the daily WU quota if they reach the deadline, right?

---

Suspend network activity of your BOINC if the SETI@home server have high (maxed out) load.. :o)


I guess my host have 300 - 500 ghost WUs.
IIRC, this will damage the daily WU quota if they reach the deadline, right?

I do not believe so; I have had some hit the deadline and not affect the WU quota.

I now seem to have a whole lot of WU's on one of my machines that I was supposed to have received on July 5th & 6th ( which I never appear to have actually gotten ) that have now started to time out. Quite annoying...

Is there any way to report ghost units manullay, in order that the WU can be sent to another user instead of waiting for the DL?

---

Is there any way to report ghost units manullay, in order that the WU can be sent to another user instead of waiting for the DL?

Unfortunately, IMO the only way to do this is to detach from the project and reatach again. That'll mark all WUs for the specific host as Client detached and trigger the reissue. Of course, it is important to run down your existing (real) cache before detaching.

---

Who the hell is General Failure and why is he reading my harddisk?¿

Greetings,

I think I have had the same ghosts.
20+ wu reached their time out on 20 July.
My client showed all work completed.
( I like to run down all work and reboot once a week.)

Significant??

• All the wu were assigned at the same time- 6 Jul 2010 5:53:57 UTC
  
• In each case, both initial hosts timed out.
  

Here is a link WU stats

So. Is this, possibly, the results of a db crash?

Thanks in advance,
Jay

I too had a large number of "ghost" units.. that I supposedly received July 5th.
I vaguely recall some pending downloads "vanishing" about then.. Possibly involving a re-boot of the local computer.

In any case.. they are starting to hit timeouts.. and coming in as errors.

And life goes on.

---

Janice

Greetings,

I think I have had the same ghosts.
20+ wu reached their time out on 20 July.
My client showed all work completed.
( I like to run down all work and reboot once a week.)

Significant??

• All the wu were assigned at the same time- 6 Jul 2010 5:53:57 UTC
  
• In each case, both initial hosts timed out.
  

Here is a link WU stats

So. Is this, possibly, the results of a db crash?

Thanks in advance,
Jay

Read Joe Segur's post in similar thread.

---

Who the hell is General Failure and why is he reading my harddisk?¿

Same here, about 35 WUs from July 5 timed out.

Read Joe Segur's post in similar thread.

here

Thanks Zeus!
Thanks Joe!

Is there any way to report ghost units manullay, in order that the WU can be sent to another user instead of waiting for the DL?

Unfortunately, IMO the only way to do this is to detach from the project and reatach again. That'll mark all WUs for the specific host as Client detached and trigger the reissue. Of course, it is important to run down your existing (real) cache before detaching.

Zeus,
I did a detach over 20 hours ago.
((Yes, I had worked off ALL WU and reporting before detaching.))
The GHOST WUs are still in the system.
Example
http://setiathome.berkeley.edu/workunit.php?wuid=629999178 -
with the Computer ID of 5234833.

I **assume** that a detach will notify SAH of WU that are physically on your machine.
It doesn't look like a detach will 'release' the GHOST WU for reassignment.

I'm going to look through the forums - to see if there is an 'early release' program' for the ghost WU.

Have a great weekend,
Jay

Is there any way to report ghost units manullay, in order that the WU can be sent to another user instead of waiting for the DL?

Unfortunately, IMO the only way to do this is to detach from the project and reatach again. That'll mark all WUs for the specific host as Client detached and trigger the reissue. Of course, it is important to run down your existing (real) cache before detaching.

Zeus,
I did a detach over 20 hours ago.
((Yes, I had worked off ALL WU and reporting before detaching.))
The GHOST WUs are still in the system.
Example
http://setiathome.berkeley.edu/workunit.php?wuid=629999178 -
with the Computer ID of 5234833.

I **assume** that a detach will notify SAH of WU that are physically on your machine.
It doesn't look like a detach will 'release' the GHOST WU for reassignment.

I'm going to look through the forums - to see if there is an 'early release' program' for the ghost WU.

Have a great weekend,
Jay

You missed the reatach in the post you quoted - that's essential (see the other thread).