Jens Bauer

I've now gotten myself a new network card in my Mac, so now I've finally got network access again (the two built-in ports were fried). I have two Espressobin boards; one (remote) is currently running Xenial, the other one (local) is running Bionic (18.4). There seem to be a difference in how network needs to be set up on each of those. Xenial allows me to configure wan alone, but if I try the same setup on bionic, I never get an IP-address on the WAN interface. But using lampra's settings works; this sets the IP-address of the bridge to my desired IP-address. That will require me to only connect one cable and use one IP-address, but I can live with that. Update: Unfortunately, after applying lampra's settings, DNS-lookup does not seem to work (eg. curl http://someserver.com/ does not succeed in looking up the host).

Yesterday, I recorded the commands and the outputs ... -All I can see about systemd is 'processing triggers for systemd', so I'm not sure systemd is at fault directly. I can see a lot of library updates, so I would expect that this could be caused by an updated library that systemd is using. -only a simple guess, though.

I haven't tried debian - and I do not have much time on my hand, because my server needs to be set up completely within a few days, as I'm going back home. Sadly, it seems the Ubuntu I'm running does not allow me to "reboot" without hanging, so I can't maintain much via SSH from home (because it'll cost me 6 hours to travel each way plus around $200, which I currently cannot afford). -I just thought I'd like to mention this, so that others will have fewer problems and the maintainers may know about it.

Using Ubuntu, I just found out during the last three days that if I do a fresh install of 5.50 on my EspressoBIN and as the first thing issue apt-get update && apt-get upgrade, followed by a reboot, then the network will never come up and I'll have to log in using a serial console.

After struggling with this for 3 days (this was absolutely no fun at all), I now finally came to a conclusion: apt-get update && apt-get upgrade -causes the network connection to be lost - right after a fresh installation, so my problems were completely unrelated to the modification of the interfaces.

I know the MacchiatoBIN does consume more power, but it's also a way faster creature, so I accept that. -It's not easy to find anything that gets near the MacchiatoBIN (especially due to its two 10Gbit connections). As for energy consumption, the MacchiatoBIN can easlily run on solar panels with a battery bank and a backup from mains power; that's what I'm aiming for. The EspressoBIN (when the board is perfect) is good - CPU performance might not be in the top, but it is better than a fair amount of SBCs. I could wish for a better design - I especially hate that the 2.5Gbit connection to the Topaz switch was sacrificed in order to get a USB3 port (I'm way against anything USB), but also the problem with unstable boards is a concern. -AND that GlobalScale could make themselves find some cheap USB cable manufacturer that is just as unreliable as their boards does not add anything positive to the experience. I've also seen that Turris now use GlobalScale for their new board - and that scares me away from purchasing one. I'm a bit dissapointed with SBCs; it all started out good in the early Raspi days, but the imagination of the board vendors are just limited to what someone else made. I very much expected to see that native SATA would become standard on the boards (not SATA via USB) - and I also expected multiple Gbit Ethernet ports; I also expected performance to grow and the amount of RAM to grow. -But the vendors keep making boards with similar specs to previous boards - especially TV-box vendors. If I were to design a board, I'd pick one hgih-performance CPU (like the 7040 or 8040) and add a whole bunch of cheap 64-bit China chips as slaves, then inter-connect them via an on-board network switch/PHY; this would make a great compute-cluster (and build-farm). Each slave should have at least 1GB RAM (perferrably 2), one SPI-flash and one SATA connector per slave as well. Yes, it'd be fairly expensive if the board had 16 or more slaves, but I think it'd be worth it. I very much look forward to see Cortex-A76 (and even Cortex-A75) boards. I hope that the chip manufactures will add built-in support for both at least multiple 6G SATA and multiple PCIe 3.0 lanes, so we can have some "WOW"-board - yes, I'm dreaming quite a bit..

Hopefully they don't provide cheap china-cables (like the broken cable shipped with my bad espessobin). How I wish EspressoBIN was made by Solid-Run instead of GlobalScale - I'm pretty sure they would have done a proper job. I'll be going for the MacchiatoBIN soon; I don't know if I'll succeed in setting up the Espresso, but at best it'll get a job as NAS server, but I don't trust it enough to make it a mail or web server.

I know 'vi' will cause me additional frustrations, so I'm trying to avoid using it. -But I just re-installed ubuntu 4.17.3, then did the following as the very first thing: sudo add-apt-repository ppa:chris-lea/redis-server sudo add-apt-repository ppa:nginx/stable sudo apt-get update sudo apt-get install redis nginx sudo apt-get upgrade ... then I applied your modifications to 10-br0.network saved and rebooted. As soon as the board is in uboot, both LEDs are on (which is what they should be), but later during boot, the right-hand LED is turned off (which indicates that things won't work). I do not get any IP-address. If I log in via the terminal (now I got some portable PC with Windows and 'RealTerm', since I'm worn out moving my Mac back and forth more than 30 times), then ifconfig -a | grep addr won't show me any IPv4 address; not the static one I wanted and neither one assigned by DHCP. The router does not indicate a dynamically assigned IP address either. If I copy the original files back and reboot, the board stays off the network; it seems I can't get it online by restoring settings. -Do I need to "kick" something in anyway; eg. restart systemd, tell it to reload or flush or something like that ?

That sounds very likely. -And likely that's why they're now shipping boards with a single 2GB chip instead of two 1GB chips.

I just realized that my bogus .txt-files were causing the above, so I deleted them (even before I reached that you mentioned I should delete them). -I used head and tail to construct the configuration files, since I can't use nano in my terminal. I tried this, and I'm pretty sure that the files in systemd/network match the ones you gave me. (I checked using hexdump -C), and ensured that line-endings were 0x0a only. As far as I understand, I should not delete 10-eth0.network or 10-br0.net*, correct ? I'm willing to try a re-install, since I haven't done much customization yet - but that would unfortunately not help others getting in the same situation...

I just recalled that before the network stopped working, I did the following: add-apt-repository for redis-server add-apt-repository for nginx apt-get update apt-get upgrade modify nginx configuration (I didn't think about mentioning it ealier, because the network was of course working until I made a reboot) -I think that the 'apt-get upgrade' might have something to do with my network no longer being brought up.

I entered a few of the commands you listed and got the following output: (sorry, some of the lines seem to be truncated; especially parts with important information): -so zram loads (I don't think this is fatal), but it seems that there's some trouble loading modules. I noticed that resolv.conf has been changed at some point (it suddenly contained google's DNS instead of OpenDNS, which I or the router had provided earlier via DHCP). 'could not open moddep file' looks like a clue.

Thank you so much again for your help. I've tried your setup and made a #reboot and also a hard reset for rebooting, but still no luck. Unfortunately, it seems something else must have gone wrong, because I noticed that when things worked, both the left and right LEDs on the WAN connector were lit, but currently only the left one is lit (seen from the side where you plug in the wire). Sadly I'm color-blind, so Ican be specific about the color. What should I look for in the terminal when it boots ?

... And suddenly it seems it doesn't work anymore (no matter how many times I reboot).

That is a perfectly valid solution and I understand completely. -Fortunately it seems I have one good board, but I also have one which is flakey. Yeah, it seems some of the boards suffer from production issues; my guess would be that they're probably not soldered perfectly. (eg. I've heard some people having problems running at higher frequencies, while others have no problems). -My good board is stable, while my bad board crashes at random. GlobalScale does not reply to my emails after I told them that I had a bad board (I've always kept a nice tone). I know there's nothing wrong with the Armada; it's a high quality CPU. The RAM is likely also of good quality, but the placement of the RAM is, if you ask me, not a good choice. -I'd say that some of the PCB-design is not done very well; ESPECIALLY that idiotic Molex connector. You never supply power on a male connector (basic electric/electronic knowledge) - because they easily short circuit. You don't see the pins come out of wall-sockets either - for a good reason. -And now Odroid seem to want to copy this stupid ritual, which means I don't need to think about purchasing anything Odroid. ... By the way, I got the EspressoBIN to boot directly from the low-cost JMB321 port multiplier; the boot device just need to be connected to interface 0 (which is located in the center on most low-cost Addonics/Syba/StarTech/Deltaco/DatOptic/iocrest/whatever port multipliers). -I plan on changing to a JMB575, though - just wanted to make sure that I could use a PM before purchasing the JMB575.

Thank you for the quick reply. Your answer did give me ideas on what to search for and I ended up with the following:, which seems to work for me - just in case others need a starting-point: $ cat 10-wan.network # see: # <https://www.freedesktop.org/software/systemd/man/systemd.network.html> [Match] Name=wan [Network] Bridge=br0 DHCP=no Address=10.0.1.2/8 Gateway=10.0.0.1 I've kept my DNS in resolv.conf (since systemd falls back to that if no DNS is specified).

Just in case someone else experience this problem, this might be a solution. In order to try and get my EspressoBIN to work with a port multiplier, I figured that I'd boot via a microSD card, which then does the real boot from SATA. But I had plenty of problems writing to my microSD card, which worked fine for the CubieBoard2 a few minutes ago. The thing I did wrong was to write (armbian) to the /dev/mmcblk1 using the DD command. eg. dd if=/some/path/Armbian_5.50_Espressobin_Ubuntu_xenial_next_4.17.3.img of=/dev/mmcblk1 bs=1048576 ... I also tried cp /some/path/Armbian_5.50_Espressobin_Ubuntu_xenial_next_4.17.3.img /dev/mmcblk1 -but that failed as well. So I decided to use a $1-china-crap USB-cardreader and insert in the USB2.0 port. I then used cp /some/path/Armbian_5.50_Espressobin_Ubuntu_xenial_next_4.17.3.img /dev/sdb ... and to my big surprise, this works! I do not know if there's something wrong with the microSD card slot on the board (it reads fine from it), or if it's something else. The only goal of this post is to tell others who experience problems with "wrong fs type" during mount attempts, even though lsblk shows it's "ext4", that they can try writing using a USB-card reader instead. - and just use 'cp' instead lf 'dd'; it does the job just as well (still you have to be careful that you're writing to the correct device, always think twice when you're writing to anything in /dev).

This happened for me too when running GlobalScale's ubuntu image. I have not yet tested if 5.50 does the same thing, but next time I'm using a console, I'll try and see if I can remember testing for it.

I've been running the Ubuntu from GlobalScale until now. -I finally had a chance for switching to Armbian, so I did; now I'm running 5.50: Ubuntu Xenial (4.17.3). So far, I've installed it onto my SATA harddisk, it boots, network is available via DHCP, web-server is up, SSHD is up, I can update via apt-get, etc... But there's still one important piece missing; I need to make the IP-address static (which was one of the main reasons I switched to Armbian). On GlobalScale's forum, I get the impression that Armbian is much easier to configure static IP address for, but I have the impression that I should not touch anything /etc/network/interfaces*. Instead it seems I'm supposed to write something in /etc/systemd/network/ - I just have no clue what to write where. I'm assuming I'll be using the "WAN" interface, so my guess is I should change /etc/systemd/network/10-wan.network. -If my netmask is 255.0.0.0 and the static ip address is 10.0.1.2 and my gateway is 10.0.0.1, what's supposed to go where ? Currently, my 10-wan.network is unchanged and looks like this: [Match] Name=wan [Network] Bridge=br0 I've tried searching for Network manager and how to configure using it, but I just ended up more confused. :/ (I also tried selecting 'static ip' in Armbian-config, but it does not seem to do anything).

If anyone is interested in lowering the heat from the board, I've used a 5.9V/3.8A switchmode power supply with great success. A 5.2V/2A PSU should be enough, though - my PSU just happened to be 3.8A. The Voltage range for the PSU should be above 5.2V and below 12.5V. There is an on-board 13V zener, this is in order to try and protect any attached harddisk from a spike. As for current, anything above 2A should be fine. If you're planning on powering a 3.5 inch harddisk from the board, stick with a 12.2V PSU. -If you're planning on using a 2.5 inch harddisk, you can run on low input voltage between 5.2V and 12.5V. Personally, I'll be using 2.5 inch WD Red drives because they consume very little power. I might also add a liquid cooler (cheap aluminum blocks from eBay and a silent <5dB, <6W, 0 .. 8L/min pump) - I hate noise. Note: I have not had any crashes or other issues with my 2GB boards - not even while running the boards on a 12.25V PSU with a 3.5 inch harddisk; the cooling is likely not even necessary, as the boards will be installed in an always-cold room. If, however, I add the cooling, I'll run the pump slowly, so it'll likely use less than 1W anyway.

Just tried a 'strings flash-image-2g-1000_800_boot_sd_and_usb.bin' and I could not find 'load_script' anywhere. So I searched for image_name and it says: -Which indicates that this image is not what's on github. If I search for get_images, I find this: ... Does this mean that the file I downloaded is an older file, which is actually not relevant ?

Before I flashed the image onto the SPI-flash, this was my U-boot version: After re-flashing, this is what I get: -So it's definitely changed.

Before doing anything else, on my Mac, I made a backup-copy of the image that I flashed. Then I downloaded the image again, and did a 'diff oldimg newimg' and diff didn't say that the images were different, so the images match. I removed the old image from the USB-stick and copied the new image there anyway. I tried re-flashing the bootloader; CRC check pass and it even said that it didn't write anything this time. When I boot, I no longer get the CRC-warning I got earlier (?!??), but otherwise things stay the same. I also tried checking if the load_script environment variable was set...

It never said CRC-error before, so something has changed. Perhaps this is a fail-safe / backup uboot ? Alright, I'll try re-downloading and re-flashing. Is there any way I can verify that the .bin file is not corrupt ? (I'd prefer downloaded a tar.bz2 file, so the archiver checks the archive integrity for me).

I've attempted to install Ubuntu_xenial_default_4.4.112. In order to get this to work, I tried to follow the instructions on the download page as close as I could... Downloaded flash-image-2g-1000_800_boot_sd_and_usb.bin (because I have a 2GB model, which reports CPU-speed 1000 MHz and DDR 800 MHz. Formatted an USB-stick as FAT, copied flash-image-2g-1000_800_boot_sd_and_usb.bin to that stick, unmounted it. Inserted the USB-stick in the EspressoBIN, rebooted the EspressoBIN. Via the terminal I typed ... When it was done updating and returned to the "Marvell>>" prompt, I pressed the RESET button on the board. The board attempts to reboot, I get the board info, and right after the SF line, I get a warning and later a few errors... CRC-error - does that mean that the boot image was not written properly to the SPI flash ? Note: The bootcmd *has* changed; now it reads: ... As I understand it, it seems it's attempting to netboot via tftp (which is why I looked at the IP adresses). Note: I can still do a ... ... which displays the directory on /dev/sda1 (boot partition) I assume that the boot image should be "Image", but I have no clue which of the 28 .dtb files I should use. (and whether to load them from /dtb/marvell or /dtb-4.4.112-mvebu64/marvell.