New Ambient Lighting system - no knowledge (2 Viewers)

[The_Stig]

Could be network related, do you have the rPI wired to a gigabit network?

HTPC and Pi wired gigabit.

best to set the logging to debug in the Mediaportal configuration

Will do!

and maybe rPI just in case

How do I check this?

CPU load on the Windows machine

Windows machine is bored...its an i5 with more than enough ooomph...

 

[Rick164]

How do I check this?

In Raspian by default it has a program called "top" which is very basic so would install a better one like this:

- Open SSH session to rPi
- Execute command: "sudo apt-get install htop"
- Once installed execute command: "htop"
- Press Q to exit it

Should get a nice overview of every process and at the top a load average (current / history of machine load) and current cpu / mem usage, normally the load average should be below 2.0 with the rPi2.

 

[The_Stig]

Just tested from away: starting TV leads to around 20% CPU-Load for the Pi on one core...so no problem at all. Will check if the lagging is there again...thanks @Rick164 (what would I do without you)

 

[The_Stig]

Had the problem yesterday again but unfortunately no chance to check the CPU load. It was after the ambilight had been running for aprox 1 hour. Logs attached, it happened at 20.01.2016 at around 21.40. I see many PS-messages in the log, but nothing more. @Rick164, do you see more?

What happened was after a while: LEDs getting out of sync and somewhen they didn't react anymore. Stopping the video didn't let the LEDs go off.

Could it be an overheating issue of the Pi? Its directly attached between wall and TV. Has it got a temperature sensor?

 

[The_Stig]

Ok, now I know why:

Hyperion is maxing out one CPU-Core after a while.

But: what to do now?

 

[edsche]

Ok, now I know why:
View attachment 175110
Hyperion is maxing out one CPU-Core after a while.

But: what to do now?

You can trace a process. Maybe there you can see what's going on.

Try: strace - p pid

Replace pid with the process Id.

 

[The_Stig]

Hrm, what would be typically to see in such a trace? I mean, its clear that its hyperion which is causing it. In other words: what additional info are you expecting there?

 

[edsche]

Database selects or file/device locks. You can see there, for example, which file/device/ causing the 100% load.

 

[The_Stig]

Tested strace now (as its not yet maxing out the CPU). I get wildly running messages which do not stop. Is this supposed to be? When I cancel it, it seems that repeating always something like

ioctl(9, FIONREAD, [12317])             = 0
read(9, "\0\0000\31\10\2Z\224`\0102\20@\30@\"\200`\247\244\245\250\245\246\250\2                     43\244\246\243\244\250\243"..., 12317) = 12317
gettimeofday({1453450833, 623517}, NULL) = 0
gettimeofday({1453450833, 624041}, NULL) = 0
write(3, "\1\0\0\0\0\0\0\0", 8)         = 8
write(9, "\0\0\0\2\10\1", 6)            = 6
write(3, "\1\0\0\0\0\0\0\0", 8)         = 8
clock_gettime(CLOCK_MONOTONIC, {47988, 692066031}) = 0
poll([{fd=3, events=POLLIN}, {fd=11, events=POLLIN}, {fd=12, events=POLLIN}, {fd                     =9, events=POLLIN}], 4, 8) = 2 ([{fd=3, revents=POLLIN}, {fd=9, revents=POLLIN}]                     )

 

[edsche]

Tested strace now (as its not yet maxing out the CPU). I get wildly running messages which do not stop. Is this supposed to be? When I cancel it, it seems that repeating always something like

ioctl(9, FIONREAD, [12317])             = 0
read(9, "\0\0000\31\10\2Z\224`\0102\20@\30@\"\200`\247\244\245\250\245\246\250\2                     43\244\246\243\244\250\243"..., 12317) = 12317
gettimeofday({1453450833, 623517}, NULL) = 0
gettimeofday({1453450833, 624041}, NULL) = 0
write(3, "\1\0\0\0\0\0\0\0", 8)         = 8
write(9, "\0\0\0\2\10\1", 6)            = 6
write(3, "\1\0\0\0\0\0\0\0", 8)         = 8
clock_gettime(CLOCK_MONOTONIC, {47988, 692066031}) = 0
poll([{fd=3, events=POLLIN}, {fd=11, events=POLLIN}, {fd=12, events=POLLIN}, {fd                     =9, events=POLLIN}], 4, 8) = 2 ([{fd=3, revents=POLLIN}, {fd=9, revents=POLLIN}]                     )

Yes that's expected, but you need to run it when hyperion is constantly at 100% CPU load. You can exit with strg + c.

The widely running messages is a good sign. It means the process is running. Else the process would hang.

Maybe we can find then a hint what's going on when the process is at 100% CPU load.