(2013-03-18, 21:24)philhu Wrote: (2013-03-18, 20:37)popcornmix Wrote: A reboot is usually provoked by too high an overclock, or an insufficient power supply.
Can you disable overclock and confirm it still fails. Measuring voltage could identify a power supply problem:
http://elinux.org/R-Pi_Troubleshooting#T...r_problems
If both those are okay, then it may be an out of memory issue. Enabling swap is one solution. That's a distribution question. Try here:
http://forum.stmlabs.com/forumdisplay.php?fid=7
Yes, I'll cut the clocks to standard for a few days, see if it stabilizes. Thanks
Hi, I lowered my clocks and these memory or whatever problems still happen. I really think it is memory or something with memory leaks because, for grins, I installed a swapfile which seems to have made the problems happen less often, but over a few days, my system just gets slower and slower until I reboot or it reboots. (I added a 256m swapfile on a USB installation).
I also posted another message about swapfile and the rPI doing nothing at all but running, for days, and my watching the space used in the swapfile increasing slowly over time by ssh'ing in and doing a 'free' command.
My clocks are standard, my power supply is more robust than required. It does seems to be ememory-related as the rPi does not have alot of extra memory to play with.
Right now, after 2 days of doing nothing on this rPi, this is my free count:
Code:
root@raspbmc:/usr/local/bin# free
total used free shared buffers cached
Mem: 384468 345796 38672 0 38608 74804
-/+ buffers/cache: 232384 152084
Swap: 255996 332 255664
Tomorrow, it will be around 505 and day after 740 or so. It seems to go exponential after that, 1000,2200,3800, then it usually dies or typing anything or bringing up a xbmc menu hangs or reboots at that point.
The question being, if it is not designed for swapfiles, why is it using any?
One interesting point. On another rPi, I cut down my CIFS mounts from 5 to 1 with links. I was pointing at different movie,tv,stub, etc directories as different mounts. Now I mount the drive and then use ln links to point to the correct place on it. That seems to have stabilized it quite a bit. Not completely but alot. Maybe disk buffer manipulation?
I do not know the internal code, but as a developer in Linux, have seen my share of memory/disk issues and this just seems exactly to be one.