UID PID PPID CPU PRI NI VSZ RSS WCHAN STAT TT TIME COMMAND
31 356 348 0 12 19 235812 66128 - RN ?? 83:18.32 setiath
31 357 348 0 12 19 106632 67536 - RN ?? 80:20.12 setiath
0 3839 356 0 0 19 0 0 - ZN ?? 0:00.00 (ps)
0 3840 356 0 0 19 0 0 - ZN ?? 0:00.00 (ps)
0 3844 356 0 0 19 0 0 - ZN ?? 0:00.00 (ps)
0 3845 356 0 0 19 0 0 - ZN ?? 0:00.00 (ps)
0 3850 356 0 0 19 0 0 - ZN ?? 0:00.00 (ps)
0 3851 356 0 0 19 0 0 - ZN ?? 0:00.00 (ps)
0 3855 356 0 0 19 0 0 - ZN ?? 0:00.00 (ps)
0 3856 356 0 0 19 0 0 - ZN ?? 0:00.00 (ps)
0 3860 356 0 0 19 0 0 - ZN ?? 0:00.00 (ps)
0 3861 356 0 0 19 0 0 - ZN ?? 0:00.00 (ps)
0 3913 356 0 0 19 0 0 - ZN ?? 0:00.00 (ps)
0 3914 356 0 0 19 0 0 - ZN ?? 0:00.00 (ps)

OK, I just noticed that the parent process of all the zombies is the first of the two SETI@home applications. I will confer with others on the SETI@home team to see if they have any ideas.

---

Charlie Fenton
BOINC / SETI@home Macintosh & Windows Programmer

OK, I just noticed that the parent process of all the zombies is the first of the two SETI@home applications. I will confer with others on the SETI@home team to see if they have any ideas.

I've spent quite a bit of time looking into this, running various tests on both an Intel iMac and a dual G4 PowerPC Mac.

I've found that each task always spawns exactly one zombie. This is true for Rosetta and Einstein, so it is not specific to SETI@home. It happens with S@h 5.13 as well as S@h 5.18. It happens in 5.10.0 and 5.9.12, so it is nothing new. And it happens in OS 10.4.4, so it is not new to OS 10.4.9.

I'm guessing the zombies may be an artifact of the way we launch project applications under the Mac's sandbox security system; we launch the switcher application which in turn launches the project application under a different user and group. But in general I don't think they normally cause any problems.

On both my Macs, the zombie disappears when its parent is removed from memory, and I've not seen any task spawn more than one zombie.

Clearly, you are seeing different behavior, since your output from ps shows that one SETI@home application has spawned many zombies. Do you have your preferences set to leave applications in memory when they are suspended? If so, you might try setting that option off. Please let me know if this helps.

i will continue checking to see if I observe the problem that you describe on my Macs.

Cheers,
--Charlie

---

Charlie Fenton
BOINC / SETI@home Macintosh & Windows Programmer

OK, I just noticed that the parent process of all the zombies is the first of the two SETI@home applications. I will confer with others on the SETI@home team to see if they have any ideas.

I've spent quite a bit of time looking into this, running various tests on both an Intel iMac and a dual G4 PowerPC Mac.

I've found that each task always spawns exactly one zombie. This is true for Rosetta and Einstein, so it is not specific to SETI@home. It happens with S@h 5.13 as well as S@h 5.18. It happens in 5.10.0 and 5.9.12, so it is nothing new. And it happens in OS 10.4.4, so it is not new to OS 10.4.9.

I'm guessing the zombies may be an artifact of the way we launch project applications under the Mac's sandbox security system; we launch the switcher application which in turn launches the project application under a different user and group. But in general I don't think they normally cause any problems.

On both my Macs, the zombie disappears when its parent is removed from memory, and I've not seen any task spawn more than one zombie.

Clearly, you are seeing different behavior, since your output from ps shows that one SETI@home application has spawned many zombies. Do you have your preferences set to leave applications in memory when they are suspended? If so, you might try setting that option off. Please let me know if this helps.

i will continue checking to see if I observe the problem that you describe on my Macs.

Cheers,
--Charlie

Last night I suspended all WUs except the two active ones. I then waited for that first instance to
complete. This time (falling back to BOINC 5.8.17) when that initial instance completed all the
extra 'ps' processes went away. (note there were only the 2, one per process under this version
of BOINC until the 2 hour mark when several others had appeared.) At any rate, I then unsuspended
a WU and went to bed expecting 'bad things' to happen overnight. That has not occurred.

My plan is to watch things today and if I don't hear anything to the contrary, try going to the new
BOINC later on today and see if the disaster of yesterday repeats itself.

Thanks for you time and effort.

By the way, I tried to check some things out and it appears the new version of SETI had been
sent/gathered to my machine just a few hours before I downloaded the new BOINC. That is why
I thought the BOINC download and SETI upgrade were related. Now I see that they were
distinct, unrelated events that just had happened close together.

OK, I just noticed that the parent process of all the zombies is the first of the two SETI@home applications. I will confer with others on the SETI@home team to see if they have any ideas.

I've spent quite a bit of time looking into this, running various tests on both an Intel iMac and a dual G4 PowerPC Mac.

I've found that each task always spawns exactly one zombie. This is true for Rosetta and Einstein, so it is not specific to SETI@home. It happens with S@h 5.13 as well as S@h 5.18. It happens in 5.10.0 and 5.9.12, so it is nothing new. And it happens in OS 10.4.4, so it is not new to OS 10.4.9.

I'm guessing the zombies may be an artifact of the way we launch project applications under the Mac's sandbox security system; we launch the switcher application which in turn launches the project application under a different user and group. But in general I don't think they normally cause any problems.

On both my Macs, the zombie disappears when its parent is removed from memory, and I've not seen any task spawn more than one zombie.

Clearly, you are seeing different behavior, since your output from ps shows that one SETI@home application has spawned many zombies. Do you have your preferences set to leave applications in memory when they are suspended? If so, you might try setting that option off. Please let me know if this helps.

i will continue checking to see if I observe the problem that you describe on my Macs.

Cheers,
--Charlie

I did have the option to leave suspended processes in memory set to yes. I now have set it
to no. I'll see how things go later today when I again try the new BOINC.

OK, I just noticed that the parent process of all the zombies is the first of the two SETI@home applications. I will confer with others on the SETI@home team to see if they have any ideas.

I've spent quite a bit of time looking into this, running various tests on both an Intel iMac and a dual G4 PowerPC Mac.

I've found that each task always spawns exactly one zombie. This is true for Rosetta and Einstein, so it is not specific to SETI@home. It happens with S@h 5.13 as well as S@h 5.18. It happens in 5.10.0 and 5.9.12, so it is nothing new. And it happens in OS 10.4.4, so it is not new to OS 10.4.9.

I'm guessing the zombies may be an artifact of the way we launch project applications under the Mac's sandbox security system; we launch the switcher application which in turn launches the project application under a different user and group. But in general I don't think they normally cause any problems.

On both my Macs, the zombie disappears when its parent is removed from memory, and I've not seen any task spawn more than one zombie.

Clearly, you are seeing different behavior, since your output from ps shows that one SETI@home application has spawned many zombies. Do you have your preferences set to leave applications in memory when they are suspended? If so, you might try setting that option off. Please let me know if this helps.

i will continue checking to see if I observe the problem that you describe on my Macs.

Cheers,
--Charlie

I did have the option to leave suspended processes in memory set to yes. I now have set it
to no. I'll see how things go later today when I again try the new BOINC.

Well it has been nearly 24 since I re-installed the new BOINC and many 'ps' zombie problem
has not recurred. Not sure what happened or what fixed it but I am glad it is no longer happening.

Thanks to all who offered help/suggestions.

When I upgraded to this official 5.10 build I lost both SOCKS and HTTP proxy support on a Vista Business 32bit and XP SP2 system. Is there any way to fix this?

When I upgraded to this official 5.10 build I lost both SOCKS and HTTP proxy support on a Vista Business 32bit and XP SP2 system. Is there any way to fix this?

Same here.

Also 5.10.7 (which was an official build up to recently) does not preserve width of window if you open BOINC Manager after a reboot in Windows XP SP2 - 5.8.16 does not have this problem.

To explain further - in advanced view, I normally set the width of the BOINC Manager window to be wide enough to see all columns. However, 5.10.7 resets the window width to something approximating the width of the Simple View. It then retains the dimensions I have set for the window as long as the machine is on, but reverts to the annoying narrow window after I restart the machine.

I am reporting this here on the grounds that 5.10.7 was briefly in the wild, even though it has now been put back in the alpha cage!

---

Also 5.10.7 (which was an official build up to recently) does not preserve width of window if you open BOINC Manager after a reboot in Windows XP SP2 - 5.8.16 does not have this problem.

To explain further - in advanced view, I normally set the width of the BOINC Manager window to be wide enough to see all columns. However, 5.10.7 resets the window width to something approximating the width of the Simple View. It then retains the dimensions I have set for the window as long as the machine is on, but reverts to the annoying narrow window after I restart the machine.

I am reporting this here on the grounds that 5.10.7 was briefly in the wild, even though it has now been put back in the alpha cage!

Actually this has been around for quite a while. I remember a post here about it a few weeks back. I think it was John Keck who made it.

---

Kathryn :o)

Also 5.10.7 (which was an official build up to recently) does not preserve width of window if you open BOINC Manager after a reboot in Windows XP SP2 - 5.8.16 does not have this problem.

To explain further - in advanced view, I normally set the width of the BOINC Manager window to be wide enough to see all columns. However, 5.10.7 resets the window width to something approximating the width of the Simple View. It then retains the dimensions I have set for the window as long as the machine is on, but reverts to the annoying narrow window after I restart the machine.

I am reporting this here on the grounds that 5.10.7 was briefly in the wild, even though it has now been put back in the alpha cage!

Actually this has been around for quite a while. I remember a post here about it a few weeks back. I think it was John Keck who made it.

Yep this is a bug that will not go away. Usually if you exit the manager via file->exit the settings will be remembered but not always.

---

BOINC WIKI

BOINCing since 2002/12/8

5.8.16 is the best new version again, since 5.10.x showed problems.

---

Usually if you exit the manager via file->exit the settings will be remembered but not always.

Works for me on both my Windows XP machines unless I power the machine down :-)

---

After installing Boinc 5.10.7 I noticed that Boinc was 'hanging' often. This requires 'killing' the process, closing boinc and restarting everything. So I installed Boinc 5.10.17 and have now found that it does the same thing.

I am running on Mac OS 10.4.10 with 'rossetta, Rectilinear and WCG' being the jobs.

I was running with Climate Prediction but found that Boinc would calling for more and more of the 282 hour jobs and then eventually assigning them to all eight cores even though I have only 8% of the resource assigned. Assuming the it was a Climate Prediction problem I am no longer running that job.

Any suggestions about the 'hangs' would be helpful.

---

The problem of getting unwanted workunits (eg unwanted climate models) isn't because of cpdn; it's because of how boinc works. If you suspend any workunit(s) in the Tasks tab, boinc finds itself without work and immediately looks for more.

You can avoid this by suspending workunits in the Activity menu - if you want to suspend all of them. You can also, in the Projects tab, select a project and click the No new tasks button.

I'm afraid that someone else will need to help you with the problem of boinc hanging.

---

After installing Boinc 5.10.7 I noticed that Boinc was 'hanging' often. This requires 'killing' the process, closing boinc and restarting everything. So I installed Boinc 5.10.17 and have now found that it does the same thing.

I am running on Mac OS 10.4.10 with 'rossetta, Rectilinear and WCG' being the jobs.

I was running with Climate Prediction but found that Boinc would calling for more and more of the 282 hour jobs and then eventually assigning them to all eight cores even though I have only 8% of the resource assigned. Assuming the it was a Climate Prediction problem I am no longer running that job.

Any suggestions about the 'hangs' would be helpful.

Do the older versions (5.8 or 5.4) give you the same hangs? Any error messages in the system logs (sorry, I don't know where they would be). Can you register for the alpha list and report there?

---

Kathryn :o)

After installing Boinc 5.10.7 I noticed that Boinc was 'hanging' often. This requires 'killing' the process, closing boinc and restarting everything. So I installed Boinc 5.10.17 and have now found that it does the same thing.

I am running on Mac OS 10.4.10 with 'rossetta, Rectilinear and WCG' being the jobs.

I was running with Climate Prediction but found that Boinc would calling for more and more of the 282 hour jobs and then eventually assigning them to all eight cores even though I have only 8% of the resource assigned. Assuming the it was a Climate Prediction problem I am no longer running that job.

Any suggestions about the 'hangs' would be helpful.

Do the older versions (5.8 or 5.4) give you the same hangs? Any error messages in the system logs (sorry, I don't know where they would be). Can you register for the alpha list and report there?

The older versions did not get into the 'hang' state. I have searched all the logs and there is no record of the Boinc failure or of me cancelling the job. Seems strange but I can't find anything in the logs. I have registered as you have suggested. Thanks for the quick response.

---

After installing Boinc 5.10.7 I noticed that Boinc was 'hanging' often. This requires 'killing' the process, closing boinc and restarting everything. So I installed Boinc 5.10.17 and have now found that it does the same thing.

I am running on Mac OS 10.4.10 with 'rossetta, Rectilinear and WCG' being the jobs.

I was running with Climate Prediction but found that Boinc would calling for more and more of the 282 hour jobs and then eventually assigning them to all eight cores even though I have only 8% of the resource assigned. Assuming the it was a Climate Prediction problem I am no longer running that job.

Any suggestions about the 'hangs' would be helpful.

Do the older versions (5.8 or 5.4) give you the same hangs? Any error messages in the system logs (sorry, I don't know where they would be). Can you register for the alpha list and report there?

The older versions did not get into the 'hang' state. I have searched all the logs and there is no record of the Boinc failure or of me cancelling the job. Seems strange but I can't find anything in the logs. I have registered as you have suggested. Thanks for the quick response.

Well here is a few hours after registering for Boinc Alpha and I have now unregistered. During the night I got six emails from Boinc of various people scrapping with each other about a problem I know nothing about. The final straw was an email saying that an Administrator had moderated me and no explanation given. I have no idea what all that stuff is about so that is the end of trying to be helpful. If Boinc keeps hanging I'll just stop doing any processing for any of these projects and take the hassle out of my life.

Thanks for your help any ways. Good Bye.

---

Well here is a few hours after registering for Boinc Alpha and I have now unregistered. During the night I got six emails from Boinc of various people scrapping with each other about a problem I know nothing about.

You do know what the meaning of an email list is? It's not a list that everyone sends emails to which are then read by one or two people with access to that list. It's read by everyone who is registered to that list.

So your problem comes in at everyone else who is registered and some may eventually answer your problem. On the same list there are also other problems by other people, whose emails you will all get. No need to be answering them. Although you could learn of them.
If you want to, that is.

The final straw was an email saying that an Administrator had moderated me and no explanation given.

That was an email by KSMarksPsych (admin and moderator of the forums here - just as I am) which showed that she had deleted your duplicate post in this thread. Earlier you had made one post with only the quotes of the posts before that, but without any input from yourself. The post under that did have input of you, so Kathryn deleted the earlier one. And you (as original poster of the post), I and Kathryn (as moderators) get a notification of that. Quite simple.

Nothing to be upset about.

---