I've been having the "the channel is scrambled or not available" issue that a number of people have reported as disk issues; however I've noticed through sysmon that there is always a CPU spike when this happens. I'm not a linux person, but I see that there's a package called procps. Is it possible to have top or ps logging to a file so I can get some more detail around what's burning the CPU at these times? Can someone tell me exactly how I would set up such monitoring?
Yes, I've seen those thanks; but those links simply mention the availability of the commands. As I mentioned, I'm not a linux person and was hoping that someone could help further with a bit of an idiots description of how to get these utils (or others) running and outputting data to files for diagnosis.
O.K. procps is not a runnable utility it is a collection of Unix / Linux programs that are added to Humax's operating system, some of them may already have been present in the busybox package, but this will add newer versions, the programs it adds are these :- Code: -r-xr-xr-x 1 root root 9800 Feb 28 2012 free -r-xr-xr-x 1 root root 13208 Feb 28 2012 kill -r-xr-xr-x 2 root root 13352 Feb 28 2012 pgrep -r-xr-xr-x 2 root root 13352 Feb 28 2012 pkill -r-xr-xr-x 1 root root 9896 Feb 28 2012 pmap -r-xr-xr-x 1 root root 80012 Feb 28 2012 ps -r-xr-xr-x 1 root root 5256 Feb 28 2012 pwdx -r-xr-xr-x 2 root root 13208 Feb 28 2012 skill -r-xr-xr-x 1 root root 10908 Feb 28 2012 slabtop -r-xr-xr-x 2 root root 13208 Feb 28 2012 snice -r-xr-xr-x 1 root root 6712 Feb 28 2012 tload -r-xr-xr-x 1 root root 57164 Feb 28 2012 top -r-xr-xr-x 1 root root 4296 Feb 28 2012 uptime -r-xr-xr-x 1 root root 20368 Feb 28 2012 vmstat -r-xr-xr-x 1 root root 13544 Feb 28 2012 w -r-xr-xr-x 1 root root 9644 Feb 28 2012 watch -r-xr-xr-x 1 root root 10600 Feb 28 2012 sysctl The one that is used in sysmon >> CPU Utilisation is probably 'top' , which, like all the other commands must be run from a Telnet window, e.g. :- Code: humax# top So you would have to run 'top' through a a filter like sed and then store the result in a file, which is basically what the sysmon program is already doing Take a look at Web-If >> Diagnostics >> Database Browser >> montor.db >> vmstat that is the database symon uses here is a bit of vmstat, it shows a spike, (of 31% SCPU useage) at epoch 1335139800 = Mon Apr 23 01:10:00 2012
It occurs during playback of either a Humax recording or of some other media file (can be anything). If I skip back, then the same portion plays just fine, so not actually anything to do with signal or the quality of the media file being played. I also see that if something was recording at that time (and no, there isn't always something recording when this happens), then that recording will have a gap, which is really annoying my better half when it's her soap. If I think to make a note of the time and check sysmon, then the tell-tale CPU spike will be there every time.
The latest procps package (3.2.8-3) has had a bug fix which enables stats for both cpu cores to be shown by 'top'. Press 1 in 'top' to enable this and 'W' if you wish to save this setting as the default.
The Broadcom BCM7405 / BMIPS4380 does claim to be dual core so I guess you're question is, which core is the current 'top' telling us about and what is the other core doing?, good point
Pester power! I know it desn't help, and we techies regard it as a matter of pride to track down the problem, but my stock answer is: watch it live if you prefer.
You're right and she does now usually watch when broadcast because of this, rather than when I would rather she watch it. i.e when I'm not there.
