On Fri, Aug 08, 2008 at 10:44:14AM -0400, Rob Mahurin wrote: > On Aug 8, 2008, at 12:40 AM, Dale wrote: >> 2008/8/6 Dale <quail.linux@gmail.com>: >>> Hi all, >>> >>> I have had this problem for a little while now, when I come out of >>> suspend / hibernation on on my Acer Aspire 5601AWLMi laptop I have >>> very high system load. I am wondering if any one else has or noticed >>> this problem. I have been informed it is a kernel problem and I am >>> at >>> a lost how to debug it. >>> >>> If am one can help with this problem it be very appreciated >> [.....] >> >> After doing some research and testing I still have not been able to >> stop the initial system load when coming out of suspend / hibernate. >> The commands 'ps aux' and 'top' are to showing me what is causing the >> system load. > > from the proc(5) man page: > >> /proc/loadavg >> The load average numbers give the number of jobs in the >> run queue (state R) or waiting for disk I/O (state D) >> averaged over 1, 5, and 15 minutes. They are the same >> as the load average numbers given by uptime(1) and other >> programs. > > While your machine is suspended, isn't every process waiting? So if > your machine has 150 processes running and a normal load of 0, and the > suspend/resume occupies the processor for 0.1 minutes = six seconds, the > one-minute load average should jump up to 15? Even if no processes are > waiting after the resume, the one-minute average won't go back down for > ... a minute. > > Have you actually tried a benchmark, instead of looking at the system > load? Something like > > $ for i in $(seq 10) ; do date ; time head -c 1000000 /dev/urandom | > md5sum ; done 2>&1 | grep ^real > $ sudo suspend ; for i in $(seq 10) ; do date ; time head -c 1000000 / > dev/urandom | md5sum ; done 2>&1 | grep ^real > > This would give you an idea of how long soon after the suspend your > machine gets back to its normal speed. The "load average" may not mean > what it usually does here. > > I haven't followed the thread closely, apologies if this is duplicate > information. very nice deduction. I'll run that next time I suspend... thanks A
Attachment:
signature.asc
Description: Digital signature