Yeah, I wasn't asking for diagnostics, I realise that this issue could be for a myriad of reasons, just I was sure (before) that it was hardware related, now I'm not so sure. FWIW, it has just happened again less than an hour after boot, so I'm going to pull the card and chroot in to analyse. (I'm a software developer, so am well aware of the 'it don't go, what's wrong?' approach to tech support 😉) Interestingly, I was still logged in on an ssh shell, and I realised there was an issue when it was telling me that `cat` was an unknown command (clearly not!). I logged out, and now got the dreaded "Connection reset by 192.168.1.44 port 22". I also ran a full badblocks pass (4 real passes) on the SD card prior to imaging just to 100% make sure the card is fine. My main concern is that it will not have flushed logs to disk after the issue, so I suspect I will find nothing. ____ Edit: It looks like Armbian has rsync on a schedule to hold logs in RAM and perhaps periodically copy back to HDD. so unfortunately I have no useful info here. Going to look into the Armbian docs and see if I can get it to log directly to disk whilst I test. ______________ Another edit: I managed to disable ramlog, but unfortunately nothing of note has been written to the log files. I have tried connecting a monitor via HDMI, but am getting no output. I am sure this is fixable, but not sure how much I'm prepared to fiddle around with a box that just wants to lock up after some undetermined amount of time. I'm going to install the official Ubuntu 18.04 image (urrrgh) and leave it running for a long while. If that displays the same symptoms, then I can safely say this board is e-waste. If it does not, then I will come back to it and try to get myself an output.

@torz77 Without logs and diagnostic information it's hard to tell what is going on because what you're saying is the equivalent of "it doesn't work" Nonetheless I had the same symptoms when my SDcard was dying (bad sectors here and there). The kernel and network layer was loaded in to memory OK, so pings would work, but ssh calls upon extra binaries and files from the SD card and I was getting I/O errors, hence would result in SSH dying which is the connection refused the client sees. I had to connect mine to a display for the duration of a diagnostic session but thankfully it was easily reproducible so the error glared out. A replacement SD card got me there.

I wonder if that is related to the issue I've been having. Not long after I posted my last message, I hit an issue where the Odroid would stop allowing me to SSH in (connection refused) but would still respond to pings. I tried with many different SD Cards (and USB Hard disks) and the issue would not go away. Could happen after a few hours or a week or so. In the end I just took the device out of service, as I assumed that it was just dying. My Odroid is headless, so can't connect to console to see any potential error messages. I can't remember how far I dug into logs, but I am sure I must have. I have now re-imaged it again, with latest Armbian Stable on Trixie, and will monitor it for further issues. I still suspect the device is dying, but your message has given me hope that it is not. I would be interested to know if anyone else is experiencing similar symptoms. v25.11 rolling for Odroid C1 running Armbian Linux 6.12.47-current-meson Packages: Debian stable (trixie) Support: DIY (custom image)

@Rodrigo Campos @kazekami_him I think @nitrolack and @afiftyp had the same Transpeed M98-8k board. It booted fine with my transpeed image. (other than wifi) They didn't have a board that had secure boot. Check if your board is the same. Maybe you could try my older images. It could be the newer u-boot and patches broke the ram detection? Maybe your board is using a different ram?

Hi and Thank @jock That's what I thought. I used ledconf-3 because that's the configuration you made specifically for my MX10 box. I ran a memtest, which went smoothly. The system is working, and I was able to compile rockchip mpp without any problems. The system is running Bookworm. Would it be better to update to Trixie or revert to a previous version? What do you think? I think the problem comes either from rk3318-box.dtb or from the overlays. Thanks again. 🤜🤛