DaveKimble

My house is 350 metres from the meter at the front gate, with the cable in a 600 mm trench, so 240 VAC becomes 220 V all the time, 208 V with the kettle on, and even less with the toaster on too. The area frequently has thunderstorms so surge protection is essential. 5V powerpacks usually have the Amps rating on them, but without some voltage data from the manufacturer and direct feedback from the board, the whole thing is pointless. Apart from learning that "it doesn't have to be exactly 5.000 V", still no one has answered the question "What is the acceptable range?". And if the board doesn't tell you continuously what it has actually got, ANYTHING can attributed to a "power failure".

I appreciate why you need a good power supply and SD card, but if the mains is browning out there is no way that a transformer will give exactly 5.000 V, so is there a file in filestore, maybe in /sys/class/power/ , that says what voltage it actually has got? What are the acceptable limits (while under load)?

While the inclusion in armbian-config of install to SATA is welcome, the calculation of the size of rootfs is wrong in some cases - it presumably (?) adds up the sizes of /bin, /dev, /etc, /home ... without taking account of volumes mounted on /home/<user>/, (my encrypted files) but the following rsync tries to recursively copy them to / and ends up blowing the size of /'s partition. Simplest solution would seem to be another warning page that says to remove all removable devices and mounted partitions, clear caches, etc.

11:57:30: 1056MHz 1.14 20% 1% 16% 0% 2% 0% 69.8°C 2/8 Time CPU load %cpu %sys %usr %nice %io %irq CPU C.St. 11:57:35: 1008MHz 1.21 27% 1% 26% 0% 0% 0% 69.2°C 3/8 11:57:40: 1056MHz 1.11 33% 1% 29% 0% 1% 0% 69.6°C 2/8 11:57:45: 1056MHz 1.10 20% 1% 17% 0% 1% 0% 69.5°C 2/8 11:57:50: 1008MHz 1.09 26% 1% 23% 0% 1% 0% 70.9°C 3/8 11:57:55: 1008MHz 1.08 27% 1% 26% 0% 0% 0% 70.8°C 3/8 11:58:01: 1008MHz 1.00 27% 1% 24% 0% 1% 0% 70.2°C 3/8 11:58:06: 1008MHz 1.00 29% 2% 27% 0% 0% 0% 70.3°C 3/8 11:58:11: 1056MHz 0.92 24% 2% 21% 0% 1% 0% 69.8°C 2/8 11:58:16: 1056MHz 0.84 18% 1% 17% 0% 0% 0% 69.6°C 2/8 11:58:21: 1008MHz 1.02 20% 1% 17% 0% 2% 0% 70.5°C 3/8

When I started with Opi PC over a year ago, I naturally started with the Sunxi links to an OS, and ran into overheating problems. Searches eventually led to Armbian where the people seemed to know what they were talking about, so I went there for my Opi Zero and now Opi Prime OS. My problem is that the wiki doesn't give an overview of the problem(s), but is at a "user-reference level" for people who already know what they are talking about. So I don't know what I'm doing, and asking simple questions like "At what temperature is the board considered to be overheating?  And when it suddenly reboots, how can I tell if it was overheating that caused it?" and get no answers. I would suggest that you all spend some time bringing the wiki up to date (for beginners) and maybe that will set more up people for doing testing. Sunxi ought to get their board naming up to date by going to "chip.RAM.ethernet" (e.g. H5.2GB.1000), and putting big heat sinks on everything. /rant While typing this I had 2 crashes: Time CPU load %cpu %sys %usr %nice %io %irq CPU C.St. 08:24:18: 1056MHz 1.40 32% 2% 28% 0% 0% 0% 67.6°C 2/8 08:24:23: 1056MHz 1.29 33% 2% 29% 0% 0% 0% 67.6°C 2/8 08:24:28: 1104MHz 1.27 31% 2% 17% 0% 11% 0% 63.9°C 1/8 08:24:34: 1056MHz 1.24 32% 2% 27% 0% 2% 0% 67.6°C 2/8 08:24:39: 1056MHz 1.22 31% 2% 27% 0% 1% 0% 68.0°C 2/8 08:24:44: 1056MHz 1.13 28% 3% 24% 0% 1% 0% 68.0°C 2/8 08:24:49: 1056MHz 1.04 28% 2% 25% 0% 0% 0% 66.2°C 2/8 08:24:54: 1056MHz 0.95 28% 3% 25% 0% 0% 0% 65.9°C 2/8 08:24:59: 1056MHz 0.96 26% 2% 23% 0% 0% 0% 66.5°C 2/8 08:25:04: 1056MHz 0.96 28% 2% 22% 0% 3% 0% 65.7°C 2/8 08:25:09: 1056MHz 1.04 29% 2% 24% 0% 1% 0% 66.1°C 2/8 08:25:14: 1056MHz 1.12 29% 2% 26% 0% 0% 0% 66.4°C 2/8 08:25:19: 1056MHz 1.19 26% 2% 20% 0% 2% 0% 65.1°C 2/8 08:25:24: 1104MHz 1.18 25% 2% 23% 0% 0% 0% 65.0°C 1/8 08:25:29: 1056MHz 1.08 24% 2% 21% 0% 0% 0% 65.9°C 2/8 Time CPU load %cpu %sys %usr %nice %io %irq CPU C.St. 08:25:35: 1056MHz 1.07 30% 2% 24% 0% 2% 0% 66.4°C 2/8 08:25:40: 1056MHz 1.15 31% 2% 26% 0% 1% 0% 66.8°C 2/8 08:25:45: 1056MHz 1.14 32% 2% 27% 0% 2% 0% 65.7°C 2/8 08:25:50: 1056MHz 1.13 31% 3% 28% 0% 0% 0% 67.0°C 2/8 Time CPU load %cpu %sys %usr %nice %io %irq CPU C.St. 10:00:38: 1056MHz 1.62 49% 4% 44% 0% 0% 0% 65.7°C 2/8 10:00:43: 1056MHz 1.89 70% 5% 63% 0% 2% 0% 68.6°C 2/8 10:00:48: 1008MHz 1.98 73% 6% 66% 0% 0% 0% 70.3°C 3/8 10:00:53: 1056MHz 2.14 84% 7% 74% 0% 1% 0% 69.6°C 2/8 10:00:58: 1056MHz 2.45 58% 4% 52% 0% 1% 0% 69.2°C 2/8 10:01:03: 1056MHz 2.34 41% 4% 37% 0% 0% 0% 66.3°C 2/8 10:01:08: 1056MHz 2.55 43% 4% 38% 0% 0% 0% 68.5°C 2/8 10:01:13: 1056MHz 2.50 30% 2% 25% 0% 1% 0% 65.5°C 2/8 10:01:18: 1104MHz 2.38 28% 1% 23% 0% 2% 0% 64.6°C 0/8 10:01:24: 1104MHz 2.19 27% 1% 25% 0% 0% 0% 65.8°C 1/8 10:01:29: 1104MHz 2.18 29% 2% 14% 0% 11% 0% 65.8°C 1/8 10:01:34: 1104MHz 2.16 26% 1% 24% 0% 0% 0% 65.9°C 1/8 10:01:39: 1152MHz 2.07 21% 1% 17% 0% 2% 0% 63.0°C 0/8 10:01:44: 1104MHz 2.06 26% 2% 24% 0% 0% 0% 65.7°C 1/8 Interpretations please.

Ah, that makes sense, thanks. At what temperature is the board considered to be overheating? And when it suddenly reboots, how can I tell if it was overheating that caused it? "armbianmonitor -m 1": Time CPU load %cpu %sys %usr %nice %io %irq CPU C.St. 17:19:09: 1152MHz 0.25 7% 2% 4% 0% 0% 0% 62.7°C 0/8 17:19:10: 408MHz 0.25 6% 1% 4% 0% 0% 0% 63.2°C 0/8 17:19:11: 1152MHz 0.25 8% 2% 4% 1% 0% 0% 63.2°C 0/8 17:19:12: 408MHz 0.23 6% 1% 4% 0% 0% 0% 63.3°C 0/8 17:19:13: 1152MHz 0.23 7% 2% 3% 0% 0% 0% 63.3°C 0/8 17:19:14: 408MHz 0.23 7% 2% 4% 0% 0% 0% 62.9°C 0/8 17:19:15: 1152MHz 0.23 8% 3% 3% 0% 0% 0% 62.9°C 0/8 17:19:16: 408MHz 0.21 6% 1% 4% 0% 0% 0% 62.9°C 0/8 17:19:17: 1152MHz 0.21 7% 2% 4% 0% 0% 0% 62.9°C 0/8 Seems very hot for doing not very much. Currently no heatsink, lying on the desk. "sysbench --test=cpu --num-threads=4 --cpu-max-prime=100000 run" drives temperature up to ~77°C but doesn't seem to break it, yet other things do.

If anyone wants to check this out, here is my PHP script: <?php // print temperature and cpu freqencies every second $log = fopen("/home/dk/temperature.log", "w"); loop: $temp = file_get_contents("/sys/class/thermal/thermal_zone0/temp")/1000; $freq0 = file_get_contents("/sys/devices/system/cpu/cpu0/cpufreq/cpuinfo_cur_freq")/1000; $freq1 = file_get_contents("/sys/devices/system/cpu/cpu1/cpufreq/cpuinfo_cur_freq")/1000; $freq2 = file_get_contents("/sys/devices/system/cpu/cpu2/cpufreq/cpuinfo_cur_freq")/1000; $freq3 = file_get_contents("/sys/devices/system/cpu/cpu3/cpufreq/cpuinfo_cur_freq")/1000; $line = date("H:i:s ").$temp."°C $freq0 $freq1 $freq2 $freq3\n";; print $line; fwrite($log, $line); sleep(1); goto loop; ?> At idle, room temperature 23°C : 12:23:17 59.532°C 1152 1152 1152 1152 12:23:18 64.493°C 1152 1152 1152 1152 12:23:19 64.493°C 1152 1152 1152 1152 12:23:20 64.493°C 1152 1152 1152 1152 12:23:21 61.589°C 408 408 408 408 12:23:22 61.589°C 1152 1152 1152 1152 12:23:23 59.169°C 1152 1152 1152 1152 12:23:24 59.169°C 1152 1152 1152 1152 12:23:25 59.169°C 1152 1152 1152 1152 12:23:26 59.169°C 1152 1152 1152 1152 12:23:27 59.169°C 1152 1152 1152 1152 12:23:28 59.169°C 408 408 408 408 12:23:29 59.169°C 1152 1152 1152 1152 12:23:30 59.169°C 1152 1152 1152 1152 12:23:31 59.169°C 1152 1152 1152 1152 12:23:32 59.29°C 1152 1152 1152 1152 12:23:33 59.29°C 648 648 648 648 12:23:34 59.411°C 1152 1152 1152 1152 12:23:35 59.411°C 1152 1152 1152 1152 12:23:36 59.895°C 1152 1152 1152 1152 12:23:37 59.895°C 1152 1152 1152 1152 12:23:38 59.895°C 1152 1152 1152 1152 12:23:39 59.895°C 1152 1152 1152 1152 12:23:40 59.532°C 408 408 408 408 12:23:41 59.532°C 1152 1152 1152 1152 12:23:42 59.532°C 1152 1152 1152 1152 12:23:43 57.233°C 408 408 408 408 12:23:44 57.233°C 1152 1152 1152 1152 12:23:45 57.959°C 1152 1152 1152 1152 I can't see any logic that would produce this output. Is the temperature control working?

> Where is cpufreq? - it seems to be at /sys/devices/system/cpu/cpu*/cpufreq/cpuinfo_cur_freq (* = 0 - 3) So I've written a PHP script to print out the temperature and cpufreq* every second. While at idle: 15:42:22 47.553°C 408 408 408 408 15:42:23 47.553°C 1152 1152 1152 1152 15:42:24 51.062°C 1152 1152 1152 1152 15:42:25 51.062°C 1152 1152 1152 1152 15:42:26 51.788°C 1152 1152 1152 1152 15:42:27 51.788°C 408 408 408 408 15:42:28 51.788°C 408 408 408 408 15:42:29 51.062°C 1152 1152 1152 1152 15:42:30 51.062°C 1152 1152 1152 1152 15:42:31 52.998°C 408 408 408 408 15:42:32 52.998°C 408 408 408 408 15:42:33 49.489°C 1152 1152 1152 1152 15:42:34 49.489°C 1152 1152 1152 1152 15:42:35 52.877°C 408 408 408 408 15:42:36 52.877°C 408 408 408 408 15:42:37 49.973°C 1152 1152 1152 1152 15:42:38 49.973°C 1152 1152 1152 1152 15:42:39 49.973°C 408 408 408 408 15:42:40 52.756°C 408 408 408 408 15:42:41 52.756°C 408 408 408 408 This doesn't look like what I was expecting. There would be some lack of synchronisation between ticks on the system adjustment and ticks in my script, but it is spending half the time fully throttled, and missing out on all the intermediate levels. A view (since last reboot ?) shows it does spend some time at other levels, but mostly at 408. /sys/devices/system/cpu/cpu*/cpufreq/time-in-state: 408000 169548 648000 8380 816000 68 912000 26 960000 14 1008000 408 1056000 1307 1104000 830 1152000 42870

OK, nice clean reinstall. Didn't remove XFCE this time. LXDE installed OK. lxpanel configured with multiple launcher buttons: Couldn't get armbian-config to do the rootfs to USB-SSD it reboots at 0 - 3% of transfer, I presume that's because of heat. "sysbench test=cpu" runs OK for 4 minutes at 100% - 71C. Numerous reboots, some at 64C. The temperature is at /sys/class/thermal/thermal_zone0/temp and I think armbian adjusts cpu freqency to control temperature, but where is cpufreq ? What are the critical limits? /var/log/syslog isn't any help because it seems to be from some other session. pinta (runs under mono) doesn't load: dk@orangepiprime:~$ pinta Can't find custom attr constructor image: /usr/lib/pinta/Pinta.Core.dll mtoken: 0x0a0003d8 due to Could not load file or assembly or one of its dependencies. assembly:Mono.Addins, Version=1.0.0.0, Culture=neutral, PublicKeyToken=0738eb9f132ed756 type:<unknown type> member:<none> Unhandled Exception: System.IO.FileNotFoundException: Could not load file or assembly or one of its dependencies. File name: 'glib-sharp, Version=2.12.0.0, Culture=neutral, PublicKeyToken=35e10195dab3c99f' dk@orangepiprime:~$ glib-sharp: E: Unable to locate package glib-sharp

Solution: ifconfig eth0 up ifconfig // now shows inet 192.168.0.4 apt install xinit startx - starts GUI without my lxpanel's configuration, and everything else's app configuration, and user=root. Will probably start again from scratch.

The local power grid workers decided to replace the fuses on something today, so the power went off and on several times in quick succession. When the Prime came back up it was in "CLI maintenance mode" and Control-Alt-F7 didn't do anything. I tried several ways of getting the GUI up and did some CLI investigations, which showed it hadn't connected with my router, or got an IP address from the router, or successfully resolved any domain names. Is there a simple way to get to my LXDE desktop? I have tried : startx (not installed) "w": USER TTY FROM LOGIN@ IDLE JCPU PCPU WHAT /dev/tty7 : permission denied chvt 7 : goes to start of CLI maintenance cat /var/log/syslog | grep NetworkManager : seems to show it getting the correct IP from router, but router disagrees and resolving domains doesn't work.

That works OK. Thank you.

Does that amount to deleting lines 75 to 81? ( I'm assuming "-a" is "AND" ).

Thanks - that didn't change anything, h3disp still complains. Fiddling with NoMachine settings allows some scaling of the window, which gives the impression of a bigger window at least.