Bug#778278: mate-settings-daemon: very bad condition (99% cpu load and more) after high system load

[treaki]

Package: mate-settings-daemon
Version: 1.8.1-1~bpo70+1
Severity: important

hi,

that problem befall/occured after a porblem with wgets mirror mode, i have discribed that in Bug#778277, thees problem lead into a process (wget) that used up all memory and swap it could resoulded in very high system load and after that kernel killed that process. ive got some messages about kernel modules which had problems that way but nothing about what i have discovered after that.

shortly after my system recovered from this odd situation the numlock led on my keyboard started to go on and of rapidly in a very fast and but not stedy frequency. first i took another look into dmesg but i couldnt find anything new there. after that i saw that my cpu load were up by nearly 100%. I started my system process manager (htop) and found out that the process called mate-settings-daemon were using up that cpu cycles. I took out my phone started the camara software and documented that in a video. After that i killed that thread of mate-settings-daemon using htop but after some time my system cpu io and after that my system cpu user value began to raise again and my numlock lead did the same stuff. another look into htop revealed that another thread of that mate-settings-daemon was doing the same shit. this time i typed in 

$ killall mate-settings-daemon 

into another terminal emulator...

and guess what, afer some time the same shit happened again... now i got a little bit more rude:

$ killall -v mate-settings-daemon 
mate-settings-daemon(6023) mit Signal 15 beendet
$ killall -v mate-settings-daemon 
mate-settings-daemon(6053) mit Signal 15 beendet
$ killall -v mate-settings-daemon 
mate-settings-daemon(6080) mit Signal 15 beendet
$ killall -v mate-settings-daemon 
mate-settings-daemon(6080) mit Signal 15 beendet
$ killall -v mate-settings-daemon 
mate-settings-daemon(6080) mit Signal 15 beendet
$ killall -v mate-settings-daemon 
mate-settings-daemon(6080) mit Signal 15 beendet
$ killall -v mate-settings-daemon 
^[[Amate-settings-daemon(6080) mit Signal 15 beendet
$ killall -v mate-settings-daemon 

and after i waited some time the problem with the numlock led didnt happened again. But i still got some problems with high cpu load. after another look into htop i found out that the process dconf-service had a high persentage of that load. i tryed to kill, first using htop after that using killall, but it resisted. After that i got again a bit more rude and used the sigkill (9) signal and killed mate-settings-daemon again after that.:

$ killall -v dconf-service 
dconf-service(4353) mit Signal 15 beendet
$ killall -v dconf-service 
dconf-service(4353) mit Signal 15 beendet
$ killall -v dconf-service 
dconf-service(4353) mit Signal 15 beendet
$ killall -v dconf-service 
dconf-service(4353) mit Signal 15 beendet
$ killall -v dconf-service 
dconf-service(4353) mit Signal 15 beendet
$ killall -v -9 dconf-service 
dconf-service(4353) mit Signal 9 beendet
$ killall -v mate-settings-daemon 
mate-settings-daemon(6100) mit Signal 15 beendet


 i guess i will now have a uncomplete mate session and the need to log off and on again :( (not as bad as rebooting the whole system but still not good, need to invent an uptime counter for the mate session uptume to show off...).


i am sorry that i havent got the necessary knowledge to debug that, that had been a better solution then just killing that thread, but i hope you will be able to reproduce that problem with that information. 


I guess the videos taken dosnt contain more relevant information, just some funny reactions to this situation, but if you thing different i am willing to share the relevant frames and samples of that video. Just ask...

i hope this information are hapefull and you will be able to put yourself into my position and reproduce that situation to debugg that pice of software. And if you like to smile you will have a little smile to ;). please keep me up and running on this issue to let me smile to. there is nothing better then funny behavior of data processing systems (EDP-Systems) and that situations around that and all that stuff...

thanks a lot and keep up the good work!!

greetings treaki

-- System Information:
Debian Release: 7.7
  APT prefers stable-updates
  APT policy: (500, 'stable-updates'), (500, 'stable')
Architecture: amd64 (x86_64)
Foreign Architectures: i386

Kernel: Linux 3.2.0-4-amd64 (SMP w/2 CPU cores)
Locale: LANG=de_DE.UTF-8, LC_CTYPE=de_DE.UTF-8 (charmap=UTF-8)
Shell: /bin/sh linked to /bin/dash

Versions of packages mate-settings-daemon depends on:
ii  mate-settings-daemon-gstreamer  1.8.1-1~bpo70+1

mate-settings-daemon recommends no packages.

mate-settings-daemon suggests no packages.

-- no debconf information