Here's how zombies were explained to me...

Eric Myers wrote:

What is a zombie? It helps to know about parent and child processes. One process (the parent) can launch another process (the child). The terminal shell launches another process to run whatever command you type. Or think of BOINC, where the core client launches the app as a child process.

Normally the parent process waits for the child to finish, and does some approprite cleanup. And normally if a parent dies the child dies too, but you can change a setting so that this is not the case (eg. launching the boinc client as a daemon in the background).

So a zombie process is a child who kept running even after the parent died, and then the child died as well. The child is inherited by the init process (pid 1 always). This new parent will clean up the dead child, but not right away (gotta make sure it's really dead), and so the child process hanging around until everything is cleaned up is called a zombie.

---

Kathryn :o)

I'm still seeing this problem... It seems to be limited to seti@home. I suspect that this process isn't correctly respecting signals from the boinc client.

*PING*

I'm still seeing this. It's actually a real annoyance that forces me to have to shut down boinc (and then often forget to restart it). I'm running 6.2.18 now.

Manager Reads: "Suspended - user activity"
Top reads: "797 setiathome 80.6"

80.6% of my processor is NOT suspended.

I used to have a script that would duty cycle setiathome to control the amount of CPU usage. When the control was introduced I didn't need it anymore. Now it appears that the feature is broken and I will have to revive the script.

As a recap, in the manager I have set under Advanced - Preferences:
Switch between applications every: 60.00 minutes
On multiprocessor systems use at most: 100% of processors (I only have 1 core)
Use at most 50.00%

On my account I have set Computing Preferences:
Switch between applications every (recommended: 60 minutes) 60 minutes
On multiprocessors, use at most 1 processors Enforced by version 5.10 and earlier
On multiprocessors, use at most 100 % of the processors Enforced by version 6.1+
Use at most (Can be used to reduce CPU heat) 50 percent of CPU time Enforced by version 5.6+

What am I missing? Why does this no longer work?

So more likely something is going funky with pointers. I wonder if it is related to my laptop going to sleep?

That's what caught my attention when rereading your posts. Do you use

Suspend work if no mouse/keyboard activity in last --- minutes 
(Needed to enter low-power mode on some computers) 
Enforced by version 5.10.14+

with a value a bit lower than the time until your pc goes into hibernation? If not, I can imagine that there "something is going funky with pointers".

And are you aware of the "throttling" mechanism of BOINC? Some seconds computing with 100% cpu alternating with some seconds of 0%.

Gruß,
Gundolf

---

Computer sind nicht alles im Leben. (Kleiner Scherz)

I am experiencing 100% CPU usage on 6.2.19, running on XP SP3. When I open task manager there are two instances of astropulse_4.35_windows_intelx86.exe running, under username boinc_project, each one using between 40 and 49% of my CPU usage. terminating these processes returns CPU usage to a normal level.

nate

I am experiencing 100% CPU usage on 6.2.19, running on XP SP3. When I open task manager there are two instances of astropulse_4.35_windows_intelx86.exe running, under username boinc_project, each one using between 40 and 49% of my CPU usage.

That's exactly as it should be.

terminating these processes returns CPU usage to a normal level.

That's the wrong action taken. If you terminate the processes via task manager, they can't be done properly and eventually end up without credit for you.

Gruß,
Gundolf

[edit] If you want to reduce cpu usage, try

On multiprocessors, use at most 50 % of the processors

in your Computing preferences under Processor usage. [/edit]

---

Computer sind nicht alles im Leben. (Kleiner Scherz)

I will agree that this is seti's problem, and I'll go an annoy them for a solution. Sadly killing the process is the only solution I have available to me. Once the task goes sideways there is little that can be done to save it (the manager losses all control). As such if I want my computer to be functional, I have to shut it down. The task starts back up, though I have no way of knowing if the results data is corrupt. I would strongly suspect that it is, but the task won't respond to a 'kill -HUP' which would actually give it time to shut down.

I will give your suggestion a whirl... but it *shouldn't* have any effect if they actually mean the option to control what they say it controls.

What is this 'Task Manager' of which you speak :P ("ps aux | grep setiathome | cut -d' ' -f2 | xargs kill" for the win!). Oh windows is there anything you can't corrupt.

Actually running OSX...

Ok, my laptop hasn't sleep cycled in several days. It appears that the process runs normally, respecting the run limits for a period of time, and then starts to disobey the limits. I don't have an exact time frame, but maybe if I'm not doing anything I'll figure it out.

I have tried setting the preference to not leave the app in memory, but that has no effect.

I can't find any direct place on the setiathome message board to report this problem. It does appear to be limited to setiathome. I wonder if it is caused by the fact that I am now running more than one task.

all right, I have it set up on the debugging info.

I'm getting a lot of noise as the manager duty cycles the process in a running/suspended loop (I'm getting a message per transaction).

I think it will uncover what happens when the system breaks though.

Ok, currently on my computer setiathome is running. Last messages in boinc manager:

"
Tue Nov 4 22:38:41 2008|SETI@home|[task_debug] task_state=EXECUTING for 14oc08ad.25488.11524.3.8.19_0 from unsuspend
Tue Nov 4 22:38:42 2008||Suspending computation - user is active
Tue Nov 4 22:38:42 2008||[app_msg_send] sent <suspend/> to 14oc08ad.25488.11524.3.8.19_0
Tue Nov 4 22:38:42 2008|SETI@home|[task_debug] task_state=SUSPENDED for 14oc08ad.25488.11524.3.8.19_0 from suspend
Tue Nov 4 22:38:42 2008||Suspending network activity - user is active
Tue Nov 4 22:38:42 2008||[app_msg_receive] got msg from slot 0: <current_cpu_time>5.014766e+04</current_cpu_time><checkpoint_cpu_time>2.830884e+04</checkpoint_cpu_time><fraction_done>8.186549e-01</fraction_done>