tony013

No, that is no acceptable solution. apart from that, it's about the first impression during installation. I have two NAS (rockpi4) both running armbian. The installation of the first one was a few years ago. I invested a lot of time to be able to control the fan of the tower. When I installed the second NAS, I found that the newer armbian prohibits PWM fan control. I then decided to use an external fan controller and stayed with armbian. Recently I had to install a raspberry4 (for the company). Armbian was of course my first choice. But the new installation routine frustrated me so much that I threw armbian in the garbage can and switched to raspi os. Raspi OS allows one to switch the keyboard to language in use before having to enter a password. With the password, everything is accepted. You just get a hint that the password may be too insecure to operate on the internet. With Armbian, the first thing you have to do is enter a strong password. With a keyboard setting that you don't even know! And even after the installation it is still very very difficult to switch the keyboard. At least for someone who doesn't know Linux that well. If you compare the impression the first installation leaves on the user, a raspi os says: Welcome. Thank you for using raspi os. The message of armbian on the other hand is: fuck off! Armbian is not for you!

Hm, the default power supply of the sata tower package has 60W and the old PC power supply has 500W According to rockpi-wiki a ssd needs about 5W - so about 25W should already be enuf ... I suspect that the problem is more likely to be voltage spikes that could be transmitted from one part to the other via the power supply line. That's why I would have thought of diodes rather than capacitors. What do you think?

Hi, I built a second nas from rockpi4 and as I used the 12V plug to power the tower, I thought, I could use an old pc power supply to drive both. But when I start both nas from the same power supply, both will crash. Any idea, how I could avoid that crash and stil use only one power supply? Maybe add some electronic components?

I used to have a real linux-server with raid-systems and there I had regular failures of this kind. The problem is that the hard disks that are sold as single disks are getting worse and worse. I had a few times even the case that a newly purchased disk was already defective. Well, I could exchange it via RMA, but that costs time and is annoying. A while ago I discarded the server and built a NAS based on rockpi4 with sata-extension. 4x 2.5" SSD - and since then there is peace on the disk front. Everything works as desired.

Hi, I installed a fresh Armbian_22.05.4_Rockpi-4a_sid_edge_5.18.0 first on SD-card, then I used nand-sata-install to copy system to emmc. After that I rebooted and set up the nas (rockpi-4a with 4 ssd-drives). I already used ssh remote shell to perform apt dist-upgrade and copying of about 6TB ... When all setup was finished, I shutdown the system, closed the tower and restarted again. On that restart, the network came up without inet-address. ifconfig showed eth0 and l0, but eth0 had a mac-address (not the one, I configured in armbianEnv.txt) but no ip-address. Here are some lines from syslog: Aug 28 15:57:01 mynas NetworkManager[1659]: <info> [1661695021.5656] device (eth0): state change: disconnected -> unavailable (reason 'carrier-changed', sys-iface-state: 'managed') Aug 28 15:57:02 mynas NetworkManager[1659]: <info> [1661695022.7361] device (eth0): carrier: link connected Aug 28 15:57:02 mynas NetworkManager[1659]: <info> [1661695022.7370] device (eth0): state change: unavailable -> disconnected (reason 'carrier-changed', sys-iface-state: 'managed') Aug 28 15:57:02 mynas kernel: [ 327.420987] rk_gmac-dwmac fe300000.ethernet eth0: Link is Up - 100Mbps/Full - flow control rx/tx Aug 28 15:57:02 mynas kernel: [ 327.421037] IPv6: ADDRCONF(NETDEV_CHANGE): eth0: link becomes ready Aug 28 15:57:02 mynas NetworkManager[1659]: <info> [1661695022.7393] policy: auto-activating connection 'Armbian ethernet' (9362e2eb-3533-439f-acd4-3804f33c01a5) Aug 28 15:57:02 mynas NetworkManager[1659]: <info> [1661695022.7409] device (eth0): Activation: starting connection 'Armbian ethernet' (9362e2eb-3533-439f-acd4-3804f33c01a5) Aug 28 15:57:02 mynas NetworkManager[1659]: <info> [1661695022.7413] device (eth0): state change: disconnected -> prepare (reason 'none', sys-iface-state: 'managed') Aug 28 15:57:02 mynas NetworkManager[1659]: <info> [1661695022.7425] manager: NetworkManager state is now CONNECTING Aug 28 15:57:02 mynas NetworkManager[1659]: <warn> [1661695022.7434] platform-linux: do-change-link[2]: failure changing link: failure 16 (Device or resource busy) Aug 28 15:57:02 mynas kernel: [ 327.429651] rk_gmac-dwmac fe300000.ethernet eth0: Link is Down Aug 28 15:57:02 mynas NetworkManager[1659]: <warn> [1661695022.7460] platform-linux: do-change-link[2]: failure changing link: failure 99 (Cannot assign requested address) Aug 28 15:57:02 mynas NetworkManager[1659]: <warn> [1661695022.7461] device (eth0): set-hw-addr: failed to set-cloned MAC address to AB:CD:47:11:FE:D0 (AB:CD:47:11:FE:D0) (NME_UNSPEC) Aug 28 15:57:02 mynas kernel: [ 327.433545] rk_gmac-dwmac fe300000.ethernet eth0: PHY [stmmac-0:00] driver [RTL8211E Gigabit Ethernet] (irq=POLL) Aug 28 15:57:02 mynas kernel: [ 327.484253] rk_gmac-dwmac fe300000.ethernet eth0: Register MEM_TYPE_PAGE_POOL RxQ-0 Aug 28 15:57:02 mynas kernel: [ 327.495328] rk_gmac-dwmac fe300000.ethernet eth0: No Safety Features support found Aug 28 15:57:02 mynas kernel: [ 327.495354] rk_gmac-dwmac fe300000.ethernet eth0: PTP not supported by HW Aug 28 15:57:02 mynas kernel: [ 327.495370] rk_gmac-dwmac fe300000.ethernet eth0: configuring for phy/rgmii link mode Aug 28 15:57:02 mynas NetworkManager[1659]: <info> [1661695022.8109] device (eth0): state change: prepare -> failed (reason 'config-failed', sys-iface-state: 'managed') Aug 28 15:57:02 mynas NetworkManager[1659]: <info> [1661695022.8121] manager: NetworkManager state is now DISCONNECTED Aug 28 15:57:02 mynas NetworkManager[1659]: <warn> [1661695022.8128] device (eth0): Activation: failed for connection 'Armbian ethernet' Aug 28 15:57:02 mynas NetworkManager[1659]: <info> [1661695022.8139] device (eth0): state change: failed -> disconnected (reason 'none', sys-iface-state: 'managed') Aug 28 15:57:08 mynas NetworkManager[1659]: <info> [1661695028.8165] device (eth0): state change: disconnected -> unavailable (reason 'carrier-changed', sys-iface-state: 'managed') I see the error, but I have no idea, what to do.

No worries. I am already an adult. To be more precise, I'm already over 60 years old. To use your words, I was scared by the tone and expectations of you developers. It's the first forum I've experienced that was designed for developers and not for users (3 of 18 sections are for users). A beginner has to come to terms with that first. You constantly write about how much effort the development is and that you want support - but when one is ready to cooperate, he will be left out in the cold. It is not yet 3 months since I bought my first rockpi4. Before that I had no contact at all with such small parts- and for so, I don't know anything about its requirements or handling. From there - my question was meant seriously. But I also do not have to impose myself. I have almost reached my goals, so I am quite satisfied.

some more feedback about stumbling around NAS and mediaplayer ... I was running rockpi4 and penta-sata-tower with armbian current, whereas the mediaplayer is a rockpi4 with armbian legacy multimedia. As I already wrote, NAS got frozen sometimes watching movies When it happened 3 out of 5 times, I kicked armbian from the NAS and startet testing official images from radxa. I started with ubuntu focal server. Streaming works pretty stable - no freezes any more. But - ubuntu wasn't able to shutdown the NAS. Very anoying, as I had to unplug power stage to turn of the NAS. I did some more tests and found out, that shutdown on ubuntu works having the hat connected but no drives attached. As soon as I attach the drives, no more shutdown. So I checked official debian - but that behaves the same. Hm, on armbian shutdown worked without issue. So I started to build a server image of armbian. Long time tests have to show, whether I could get a system that does not freeze ... On the mediaplayer front I did lot of testings too. Lately I thought, that LE has the best player support, but that's not true. I have few movies with VC-1 codec and armbian legacy multimedia is the only one, that plays the movies some kind of acceptable (stil some freezes, but overall experience is more than movie than diashow). But change between kodi and desktop is unusable for me (my beamer screen misses about 10% on top and bottom, so no menubar ...). That was the trigger to start searching for alternatives. Sadly I had to realize, that there was none - so I had to develop one on my own. With all that testing, I killed two sdcards and an emmc ... but I almoust met my expectations.

The greeter can be downloaded from github.

Hi, I did some more tests and it turned out, that there is little difference between LE and armbian legacy for x264 or avc movies - but I found no movieplayer, that worked half as good as armbian legacy with vc-1 movies. Don't have that many movies with vc-1 but some much-loved like inception ... I already wrote, that I use rockpi as movieplayer together with a beamer, so session selection of armbian legacy is suboptimal for me, as about 10-15% on top and bottom of the screen are invisible (outside of screen). Lightdm/xfce is so dumb, that icons in topbar will not be scaled to bar height, which works fine for second bar ... Sadly I didn't find any other greeter, that runs on armbian and is more attractive like lightdm-gtk-greeter. So I started to build my own, which took lot more of time than expected ... any way - I finally have a working sample: is uses gtk too, so it works with armbian. I check the available sessions and if there's no session definition for kody-gbm, the kodi-button will be disabled (as in the screenshot taken from my desktop computer). Font- and button-size have been increased for a screen distance of 3m approximately. Last login will be cached, so user will be prefilled and hitting ENTER in the password field will login to the last session. Left most button starts kody and right most button the desktop.

I finally was able to kill armbian-spammer, so don't need to search for other os any more :) No, I'm not interested in 4k. Don't have equipment for 4k. I'm happy with 2k and the equipment that I own. But I do have movies, that none of rockpi images is able to play (like inception). Some movies have wrong colors on playback or you see jpg-artifacts or the like ... But I couldn't detect any difference between armbian legacy and twisteros. Anyway - as I can't use twisteros any more, armbian legacy is the only desktop os with acceptable movie playback. I found the settings for login top bar and changed fontsize to 100 or so, now I can use the menu. Having that bar moveable like it is possible for the second bar would be wonderful. I'll keep searching.

Thank you for that pointer! That forum looks like a lovely place for users. Looks like you did a very good job. Very nice!

I'm not convinced, that it is true. TwisterOS has same kodi-version like armbian legacy and movie playback gives the same impression like armbian legacy - smooth video and fresh sound. Firefox-implementation of armbian pushed me towards twisteros. ... and I'm already completely angry about firefox experience on armbian. Really, that sort of spam turns me into a spree killer :( Well, I tried it. Very disappointing experience. Kodi on armbian current is 3 generations back and has no hardware support for video and audio is the same very poor. I guess, you should give twisteros a try. They use current kernel (same version like armbian current and LE) and have hardware acceleration for video and desktop - and best of all, switch between desktop and kodi in a wink. Don't know, if that propaganda is true, but from twisterOS you can read: Latest PanfrostGPU drivers with desktop hardware accelerationKodi and MPV media players with built-in hardware acceleration support

Yes, but I found a way around that. I use the second bar with iconsize of 100 and changed conky to have all informations, I otherwise want in top bar ... With twister I could start cody from desktop without changing lightdm session. That was beautiful :) But as I can't use twister any longer, I have to live with your solution Sorry, I was unprecise. Of cause I tried the hints from #1 to disable lightdm, but I don't know, how to recover desktop. So I had to reinstall armbian from sdcard. Ok, I'll try that. Thank you

@JMCC could you please explain me, how cody is startet, or what needs to be done to start kodi? After changing rockpi model, twisterOS does not boot anymore. Not even from a freshly prepared sdcard. So I had to step back to armbian legacy. Things work, but as I already wrote, top menu is outside of my screen. Therefore I would like to select between desktop and kodi by using different users or something like that. Could you please shine me a light for that way? Or would it be possible to start kody right from desktop, like twisterOS is doing it?

Don't know - could it be possible, that some "misunderstandings" about edid could lead to disabled gpu?

Hi, I startet with conky-grapes from TwisterOS, which I wanted to tweak to show cpu frequency for every core. But that showed, that conky is not able to do consistent text formatting, so I startet to write my own theme. As I like the armbian title font, I asked for the font, but it turned out, that it did not look as good as expected. So I discarded that variant and stayed with the previous selected fonts. Main purpose is to use that monitor in home cinema, therefore I used bigger fonts, which are readable from distance. Feedback is welcome.

good news :) how can I help testing?

Thank you for showing your script. Beautiful work! I love verbose coding :) Thanks for that command! Haven't read about fdtoverlay before. That's a very nice way (with little additional scripting) to enable overlays for cases where dynamically overloading is not supported. Good job! This is by far the first reason for using "new" gpiod layer. Your key sample didn't convince me. I prefer "my" way of button handling. But finding pins by pin-name is sexy :) Do you mind to publish the source of your overlay? I asked dtc for the source, and it looks terrible. Can't believe, that you had to write such ugly stuff. Do you know about a way to find conflicts between muxed modules, which share the same pin? A developer told me, that pwm-pin for the fan control of penta-sata-hat conflicts with hdmi sound (I2S ?). I'd like to verify that. But I can't I use both - the fancontrol on the same rockpi4 which generates the sound for my AVR (over HDMI) and I can't confirm any conflicts as I don't hear any problems.

Hi, I have a really strange situation, which I don't understand and for so - I need help to solve it. I had 2 rockpi, one with sata hat which plays the nas role and one without sata drives, which is my frontend for home cinema (connected to AVR and beamer). Last few days I had the effect that the frontend froze and after close examination it turned out that all memory was used up. Fortunately, I already had a new rockpi with 1,4Gb memory. I thought I'd just change the eMMC and go on as before. But that was probably a big mistake. The new one did not start at all with the eMMC from the 1.2Gb rockpi. That's why I thought I had to completely rebuild the bigger one. So I inserted the backup on sd card - and nothing did. I tried little by little all test sd cards, but with none started the rockpi. At least that's what I thought. When I opened the network overview, the rockpi was reported as active and lo and behold, I could log in via ssh. On the screen still dead pants. Then I tried the computer in the home theater and there I had a picture. So connected to a computer monitor - no picture, connected to a beamer - picture is ok. Thought, well, maybe the monitor is over. So I connected the monitor to the NAS. Clean picture including startup procedure - everything as usual. Thought maybe I just imagined it, so ran all the tests again - same picture. Just no picture! Stupidly, the previous frontend also no longer has a picture on the computer monitor. Only on the beamer. I am glad about every tip!

Sorry, but I don't understand your reply. You write that the pin should not be available for GPIO. I completely agree with you. But not with the output of gpioinfo! - gpioinfo states the pin as unused (and for so as available) Of cause - the pin is muxed - according radxa docs, the fan pwm line is the clock signal of a spi-channel Hm, that looks of cause very interesting. Is that an overlay or what is it? Who processes that file? I would not have that key labeled "Power-key". ... anyway: does that "wakeup-source" mean, that the rockpi4 is able to wakeup on keypress after issuing a "poweroff"? and last not least - how can I act on key-events in userspace? Actually I recognize 3 different button states in my driver: single key press, double press and long hold press. Together with the config file, it is possible to call a python function, or a system process So I can execute 3 different things with a single button. Is that possible with gpio-keys too?

Interestingly the pwm, which I use for fan control is not recognized by gpioinfo. Fan is attached to Pin 33, which is GPIO2_B4 or pin-num 76... So I looked at line 12 from chip2, but that is noted as unused ... Of cause. I did it - and the result is unbelievable ;) command: gpiomon gpiochip4 18 I counted the lines: 308 events for ONE single button press. Well the button is not of good quality and some kind of flaky, but that many events? Puh Now I understand, why I had so many problems in detecting the right button states ... :D

That may of cause be true. ... but I can't follow your calculation. radxa writes: In my case GPIO4_C2 is 146 (you can find that number on the wiki-page from above) ... so I "only" need to take out the chip-calculation, which is 4*32 = 128 So 146 - 128 = 18 ... and forward calculation leads to: C (2) * 8 + Pin 2 => 16 + 2 = 18

Some notes about why gpiod may not be the "right" tool for research and small user ideas ... I was curious about gpiod, installed it on my armbian rockpi4 and did the usual tests. gpioinfo shows all gpio lines and which of them are used ... Now I wanted to see, what gpio says about my penta-sata top-panel button. According to docs from radxa the button is connected to GPIO4_C2 - and following the pin numbering scheme from radxa, it has number 146 libgpiod breaks the numbering scheme and uses line-numbering for each chip. So I have to select chip4 - which reduces the pin-number from above by 4*32 and leads to 18 Looking at output from gpioinfo, line 18 is used by "sysfs" and gpiomon fails with gpiomon: error waiting for events: Device or resource busy My button class from last post works without issues. If I want to achieve the same with gpiod, I probably have to modify kernel or at least some kernel modules - which is nothing I want to play with :(

Thank you for insisting! I didn't paid attention to this on your first posting :( u r right - but if the pessimist would have been right, I could not have succeeded in my research. As I wanted to have driver support for penta-sata on LibreElec, I had to do things, I never wanted to ... LE does not have gpiod support. So I'm happy that sysfs stil works :) No, I wrote that shellscript just for you as a sample, how to test gpio pins without any use of foreign library ... If you're interested, my button-class to handle gpio buttons

You won't need external tools. You can do all you need with tools provided by any system. to enable a gpio pin with number 13: echo 13 > /sys/class/gpio/export to verify the pin direction: cat /sys/class/gpio/gpio13/direction Result should be "in" (the default) to check the gpio pin state (value): cat /sys/class/gpio/gpio13/value From what I read so far, 3V should be a reasonable voltage for digital pins (1.8V for analog pins). The button from tower works with a pullup resistor, so default value of unpressed button is "1" Pressed button results in "0" When your test pin has a default value of "0", check whether active_low is enabled cat /sys/class/gpio/gpio13/active_low active_low should have a value of "0" A 10k resistor between gpio-pin and 3V will activate the pin and the switch should connect the pin with GND. You could solder the resistor and the switch with a cable from connection between switch and resistor, so you have 3 cables. Connect resistor to 3V and switch to GND, then the intermediate cable is your pin-checker. Last not least a little helper script, so you don't have to repeat the cat command: #!/bin/bash while grep -q 1 /sys/class/gpio/gpio13/value do echo -n . sleep 1 done echo 'Ok, got it'