Hey guys, Micha from DietPi here.
Some background about the setup of Emby on DietPi:
- It is disabled or ARMv6 RPi models (RPi1 and Zero), so the armhf packages are correct, pulling required dependencies from official Raspbian repo in case.
- The latest pre-compiled package from GitHub source is pulled, hence 18.104.22.168 at time of writing: https://api.github.c...releases/latest
- The emby user is added to shared dietpi group to have access to default media directory, else no special setup done, so basically a default setup like one would do on default Raspbian.
Since you observe the crash when having high RAM usage, and I also see in your video that used RAM is frequently jumping quite close to the total limit, I guess it is simply the OOM killer.
Please paste the output of the following:
journalctl -u emby-server
And also check journalctl + demsg for other system and kernel messages that match the time the service crashes.
Generally I would not run the system that way that regularly 80% and more of physical RAM are used. If there is some heavier cron job or systemd timer running concurrently, it will quickly run OOM. Even if it works, there is nearly nothing left for file system caching and similar, reducing performance in general. However to prevent OOM kills and generally verify that indeed memory limit is the reason, you can create a swap file:
This will create a 1 GiB swapfile on your system. "1" means it creates a swapfile with a size so that 2 GiB total memory are available, else you could use e.g. "512" for a 512 MiB swapfile or "0" to disable it again, or in case use dietpi-drive_manager to create it on a different drive, as this should no be on the SDcard.