The SETI@home project suddenly vanished from the Projects tab of BOINC! I'm using BOINC version 5.2.13. This happened on my Windows XP Pro computer at work earlier today, and just happened now on my Windows XP laptop at home too! (Same version of BOINC on both computers.) I didn't ask to delete it, detach from it, or do anything else. Suddenly <poof!> I went from having 5 projects to only having 4 problems. I had to look for a moment. Rosetta, Einstein, LCH, and Predictor are all there. Yep. Only SETI is gone. Eeek!

Is this an issue with the transition to the new version of SETI that uses finer resolution -- and takes longer to crunch? I see no error or other indication in the Messages tab.

Same thing happened to someone at uFluids. I've lost every project because of a lost heartbeat error. But I'm not sure if that's what is causing this problem.

---

me@rescam.org

Me to, but the work usnits are still being worked on. I tried reattaching but the client says that I'm already subscribed.

Update: It's still getting new WUs.

I exited BOINC (the server application) and restarted it via the client. Now the project and application values for all work units of the Work tab are empty! Some of the units are NOT SETI, so it seems to be getting worse! Gulp!

I have the same problem; SETI vanished form the BOINC-manager.
The trouble started with the bad abeta-WU's from predictor. I can't stop or abort them. I also can't re-attach to SETI. And I got WU's without name.

---

OK, here's another observation:

The master+*.xml files for projects other than SETI contain a lot (~16KB) of information. The master_setiathome.berkeley.edu.xml on my host only contains:

<scheduler>http://setiboinc.ssl.berkeley.edu/sah_cgi/cgi</scheduler>

I agree that it occured coincidentally with the bad abeta* WUs of predictor at home but am not knowlegable to connect the two. It would open a can of security worms if it were true: one project nuking another project.

I have the same problem; SETI vanished form the BOINC-manager.
The trouble started with the bad abeta-WU's from predictor. I can't stop or abort them. I also can't re-attach to SETI. And I got WU's without name.

Okay! I was now able to reset the predictor@home project on my computer and I'm not accepting new WU's from them (at least for the moment).
The nameles WU's (4) are all from SETI. My computer still crunches them. I'll try uploading when they're finished.

---

I had the same problem, but after I detached Predictor@Home Seti was back in list.

---

I had the same problem, but after I detached Predictor@Home Seti was back in list.

Thanks for your help! I did the same and SETI was back in the list. The WU's got their name back. I then reattached to predictor@home but am now accepting no new WU's. So far all is back in order.

---

Since the BOINC Client would not allow me to abort a remaining Predictor abeta work unit which had not yet started to run, I did a Detach for Predictor followed by an Attach. That wiped away the abeta work units.

For the "fun" of it, I deleted all files with SETI in the file name, then restarted the BOINC server. No change/improvement.

Then I tried the Detach/Attach process with SETI and, after a short time, SETI@home showed up again in the Projects tab. AND, the the Project and Application names columns of the Work tab are back again.

I wonder if a simple Detach/Attach would have fixed SETI? If I had to D/A Predictor first, etc.?

I was a bit tired when I created my previous post. Forget the bit about Detaching from SETI. Obviously I can't do that if it doesn't show up as a project! My bad.

I'm at work now. I detached from Predictor to remove all abeta work units, then Attached to it, then immediately set Predictor to "No new work".

In less than a minute, SETI@home reappeared on the Projects tab, and the Project and Application columns of the Work tab were repopulated.

Does this process work for everybody else? Sounds pretty much like what another poster had tried.

- Bill

---

I've had similar problem few weeks ago with web application BoincGUI-PHP (web interface replacement of Boinc Manager). I attached one machine to Pirates@home, but could not see the project or its tasks among other projects or running / transferred tasks. One week later it got much worse - only first two projects in order showed up (Seti Beta and Pirates) and only Seti Beta's units could be seen).

The problem was that the Boinc RPC data stream contained few blank lines and the BoincPHP-GUI parse loop terminated as if end of file were reached. I found out that the files account_pirates.spy-hill.net.xml and account_einstein.phys.uwm.edu.xml contained empty lines and these got exactly copied to the RPC data stream. The rest of the data stream (containing info about other projects and tasks) was ignored and disappeared.

I could only solve this problem by modifying the php application. Maybe something similar (any art of corruption) happened to some of your Boinc files and BCC can't recognize or correctly parse them - what is then possibly solved with resetting and/or reattaching to the projects.

Peter

[edit]After reading other threads - yes, this problem could be caused by ill-formed Predictor WUs.[/edit]

Same thing here. Detached preditor and got Seti back. Reattached Preditor and lost Seti again so detached again, that was yesterday. Same thing today so do I keep unattached from predictor for a while?

Same thing happened to someone at uFluids. I've lost every project because of a lost heartbeat error. But I'm not sure if that's what is causing this problem.

What IS a lost heartbeat?

---

From the Predictor boards:
"there's a missing </gui_url> for <name>Abeta news</name> in sched_reply_predictor.scripps.edu.xml that causing display trouble in Boinc Gui for predictor and following project.
Of course, problem disappear when changing account_predictor1.scripps.edu.xml and sched_reply_predictor.scripps.edu.xml by adding this missing </gui_url> until next update."
(http://predictor.scripps.edu/forum_thread.php?id=2335)

It's indeed the cause of the problem, but since those files are fetched each time Predictor has to contact the server, it'll reappear until it's fixed server-side.

---

I verified that the workaround of adding </gui_url> to *both* files in the appropriate place does solve the problem. seti@home reappears when the client is restarted.

This also would prompt a client fix to shield other projects from XML coding errors no?

I am having the same problem.

I'm sorry but I'm not techno-savvy. There appear to be several solutions here but I honestly don't understand them. Could someone spell out the solution in the most basic terms, please? And from the beginning.

---

I am having the same problem.

I'm sorry but I'm not techno-savvy. There appear to be several solutions here but I honestly don't understand them. Could someone spell out the solution in the most basic terms, please? And from the beginning.

The simple solution is to detach from the Predictor project until someone wakes up and fixes the bug on their end. It's simply an ommission of a closing XML tag in two files that's causing the UI failure in the BOINC client, it's cosmetic for the most part.

The other solution requires editing of the xml files but that's only temporaty as the next time the client updates from the Predictor site, it will undo the fix.

Just detach from Predictor until they wake up and fix the files.

---

What IS a lost heartbeat?

No hearbeat explanation, in effect the process stopped responding.

---

me@rescam.org