these are the only running threads. where's the rest of my CPU being used?

top - 14:10:34 up 46 days, 23:53, 7 users, load average: 2.00, 2.00, 1.95
Tasks: 99 total, 2 running, 97 sleeping, 0 stopped, 0 zombie
Cpu(s): 30.2% us, 64.8% sy, 4.7% ni, 0.0% id, 0.0% wa, 0.0% hi, 0.3% si
Mem: 510628k total, 502340k used, 8288k free, 87060k buffers
Swap: 498004k total, 156868k used, 341136k free, 170928k cached

PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
29259 root 39 19 16856 14m 2612 R 4.7 3.0 0:03.31 setiathome_4.07
29260 kc 16 0 10372 1260 9.9m R 0.0 0.2 0:00.03 top

30 % User, 64,8 % sys. Looks like your system is doing something with higher priority ?
What did the rest from top say ?
What OS do you have ?

---

30 % User, 64,8 % sys. Looks like your system is doing something with higher priority ?
What did the rest from top say ?
What OS do you have ?

What did "vmstat 5" or "iostat 5" say ?

---

30 % User, 64,8 % sys. Looks like your system is doing something with higher priority ?
What did the rest from top say ?
What OS do you have ?

gentoo on amd64

these are the only running threads (top -i). the rest are sleeping. if i renice seti to 0 i can get cpu up to 50% but i cant find where the other 50% is going. unless it's just idle, in which case seti sould be taking priority

vmstat output

procs -----------memory---------- ---swap-- -----io---- --system-- ----cpu----
r b swpd free buff cache si so bi bo in cs us sy id wa
2 0 156868 4292 87060 170936 0 0 4 11 3 6 75 23 2 0
2 0 156868 4164 87060 170936 0 0 0 0 1063 202 87 13 0 0
2 0 156868 10520 87024 164112 0 0 0 0 1062 193 86 14 0 0
2 0 156868 10520 87024 164112 0 0 0 0 1065 203 85 15 0 0
3 0 156868 10520 87024 164112 0 0 0 0 1067 206 87 13 0 0
2 0 156868 10520 87024 164112 0 0 0 0 1063 192 87 13 0 0
2 0 156868 10512 87024 164112 0 0 0 31 1082 247 87 13 0 0
2 0 156868 10512 87024 164112 0 0 0 9 1068 195 86 14 0 0


edit:
fixed-width font would be useful
renice'ing to -1 gets 87% which vmstat shows as user. 13% is still system but i cant tell what that is allocated to from top

30 % User, 64,8 % sys. Looks like your system is doing something with higher priority ?
What did the rest from top say ?
What OS do you have ?

gentoo on amd64

these are the only running threads (top -i). the rest are sleeping. if i renice seti to 0 i can get cpu up to 50% but i cant find where the other 50% is going. unless it's just idle, in which case seti sould be taking priority

How much "sys" shows top, if you renice seti to 0 ? - Look for IO (iostat 5 and vmstat 5). IO is a posible reason for high "sys" on top !

---

30 % User, 64,8 % sys. Looks like your system is doing something with higher priority ?
What did the rest from top say ?
What OS do you have ?

gentoo on amd64

these are the only running threads (top -i). the rest are sleeping. if i renice seti to 0 i can get cpu up to 50% but i cant find where the other 50% is going. unless it's just idle, in which case seti sould be taking priority

How much "sys" shows top, if you renice seti to 0 ? - Look for IO (iostat 5 and vmstat 5). IO is a posible reason for high "sys" on top !

35% system if i renice to 0
seems i have to install iostat.. just a sec

edit for iostat:
avg-cpu: %user %nice %sys %iowait %idle
64.67 0.00 35.33 0.00 0.00

Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn
sda 15.77 0.00 198.00 0 992

Your system is doing something work, but i am not shure what exactly is the problem.
Can you please do a "iostat 5" and "netstat -in" and let it generate a little bit output.
What does top (without -i) mean ? - Sorry have no linux here, so I am a little confused about the different outputs i get normaly...
Your swap looks high, do you have a memory leak ?

A reboot could solve it, but will not eliminate cause of the problem.

---

top shows all the system threads in order of CPU%. only those two threads are using anything
top -i shows only running threads, it hides everything with no cpu usage.
there are only 3 connections in netstat, 2 ssh connections and a web connection to another server on the lan


netstat -in:
Kernel Interface table
Iface MTU Met RX-OK RX-ERR RX-DRP RX-OVR TX-OK TX-ERR TX-DRP TX-OVR Flg
eth0 1500 017356332 0 0 016479431 0 0 0 BMNRU
lo 16436 0 379094 0 0 0 379094 0 0 0 LRU


iostat 5:
Linux 2.6.9-gentoo-r14 (halo) 07/29/05

avg-cpu: %user %nice %sys %iowait %idle
9.22 65.31 23.03 0.00 2.43

Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn
sda 0.93 8.88 21.54 36094844 87505040

avg-cpu: %user %nice %sys %iowait %idle
65.20 0.00 34.80 0.00 0.00

Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn
sda 0.60 0.00 6.40 0 32

avg-cpu: %user %nice %sys %iowait %idle
65.20 0.00 34.80 0.00 0.00

Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn
sda 0.00 0.00 0.00 0 0

UPDATE:

it may be a memory leak.
i didn't think i'd be able to do a reboot until later because im not the only user, but it did get rebooted and so far everything looks normal.

since its gentoo ill try to go about a world update and keep an eye open to see if it happens again

Dotsch, thanks for the assistance

UPDATE:

it may be a memory leak.
i didn't think i'd be able to do a reboot until later because im not the only user, but it did get rebooted and so far everything looks normal.

since its gentoo ill try to go about a world update and keep an eye open to see if it happens again

Dotsch, thanks for the assistance

You're welcome !

A good way to detect a memory leak is to look at the memory util for example with "ps -auxwww" or with top. There is a option in top where you can sort by memory util. Let the system running some time, and you will find the process with the high util.

Btw. I have seen, that you have a binary for NetBSD on alpha, do you what to share it with the other users ? - I am collecting some thridy party links in the moment, and will report them to Berkley to add it on the website (boinc.berkeley.edu/download_other.php).

One other thing, is it posible to let FreeBSD binaries running on other BSDs, like OpenBSD or NetBSD, without a lost in performance ?

Lars

---

UPDATE:

it may be a memory leak.
i didn't think i'd be able to do a reboot until later because im not the only user, but it did get rebooted and so far everything looks normal.

since its gentoo ill try to go about a world update and keep an eye open to see if it happens again

Dotsch, thanks for the assistance

A good way to detect a memory leak is to look at the memory util for example with "ps -auxwww" or with top. There is a option in top where you can sort by memory util. Let the system running some time, and you will find the process with the high util.

My favorite tool to detect performance problems is HP Glance Plus. It is avaiable for HPUX, Solaris and Linux for a free 60 day trial version :
http://openview.hp.com/products/gplus/download.html

For long time performance and bottleneck detection is HP performance agent and performance manager my favorite tool : http://openview.hp.com/products/ovperf/download.html. Its available on the same OSes.

Lars

---

Btw. I have seen, that you have a binary for NetBSD on alpha, do you what to share it with the other users ? - I am collecting some thridy party links in the moment, and will report them to Berkley to add it on the website (boinc.berkeley.edu/download_other.php).

One other thing, is it posible to let FreeBSD binaries running on other BSDs, like OpenBSD or NetBSD, without a lost in performance ?

Lars

i don't know if it works. you might notice that computer has 0 credit. i've let it run boinc for 4+ weeks without any finishing, so either the alpha is too slow, the client is too slow, or the client doesn't work at all. i'd delete the host, but i like seeing things up there that aren't x86/win/linux. i run BeOS too, but it can't compile boinc so its stuck with seti classic.

i do recall someone on the boards here actually making optimized alpha/BSD clients. it would have taken me months to test different optimizations. ive never tried anything cross-platform, but i know the NetBSD package manager works on all of them.

top - 14:10:34 up 46 days, 23:53, 7 users, load average: 2.00, 2.00, 1.95
Cpu(s): 30.2% us, 64.8% sy, 4.7% ni, 0.0% id, 0.0% wa, 0.0% hi, 0.3% si
Mem: 510628k total, 502340k used, 8288k free, 87060k buffers
Swap: 498004k total, 156868k used, 341136k free, 170928k cached

A constant load of 2 when running 1 process mostly indicates 1 process going rampant. 150MB used Swap and 65% system use might show, that this process is blocking a large portion of memory, so other processes (like seti) have to constantly swap their memory (65% system might be 65% swap-daemon).

I'd look around for sleeping processes and restart them one-by-one (probably starting with boinc itself, already had such problems when switching between multiple projects within boinc), and look if the load gets down.
To get rid of the extra I/O you might even consider disabling swap space.
Your system could cope without it, according to your data. But don't leave it that way, or the system might lock up, when the leak filled up all of the memory.