TrueNAS Issue with ML
❓Question🗄️Server☑️Ready
Good morning,
I'm seeing no faces when I run face detection. The ML log is rather empty outside of some startup routine info, but in the server log I'm seeing:
Your docker-compose.yml and .env files.
I'm running TrueNAS and don't have access to these files. But I'm attaching config.txt which I hope will suffice.
Logs from all the containers and their status (see above).
All attached, including the docker ps -a command.
All the troubleshooting steps you've tried so far.
I tried using both the Default and the ROCm machine learning images, tried expanding CPU thread count and memory limit (current 16 threads and 24576 MBs or 24GB), and tried enabling and disabling the checkbox for GPU Configuration. All this is in the TrueNAS app config editor.
On the immich side, I tried swapping out the "local" IP with the actual IP of the server but that yielded no results. I also am noticing that based on the "docker ps -a" command as well as running "netstat -an | grep LISTEN" I'm noticing the port 32002 which is configured for the ML image is not shown anywhere. I feel like it probably should...
Any recent changes you've made to Immich or your system.
This is a new install. I've reverted back to ROCm ML image and the local "127" IP after troubleshooting.
Details about your system (both software/OS and hardware).
TrueNAS SCALE 25.10.1
Minisforum N5 PRO AI
64GB RAM
AMD Ryzen AI 9 HX PRO 370 w/Radeon 890M
Details about your storage (filesystems, type of disks, output of commands like fdisk -l and df -h).
4 x 16TB HDDs
Configured in RAID Z 1, with usable capacity of ~46 TB
fdisk and df are attached.
The version of the Immich server, mobile app, and other relevant pieces.
immich 2.5.2 (latest in TrueNAS)
Lastly, thank you! This project is awesome.
I'm seeing no faces when I run face detection. The ML log is rather empty outside of some startup routine info, but in the server log I'm seeing:
[32m[Nest] 33 - [39m02/04/2026, 11:30:42 AM [32m LOG[39m [33m[Api:MachineLearningRepository][39m [32mMachine learning server became unhealthy (http://127.0.0.1:32002).[39m
Your docker-compose.yml and .env files.
I'm running TrueNAS and don't have access to these files. But I'm attaching config.txt which I hope will suffice.
Logs from all the containers and their status (see above).
All attached, including the docker ps -a command.
All the troubleshooting steps you've tried so far.
I tried using both the Default and the ROCm machine learning images, tried expanding CPU thread count and memory limit (current 16 threads and 24576 MBs or 24GB), and tried enabling and disabling the checkbox for GPU Configuration. All this is in the TrueNAS app config editor.
On the immich side, I tried swapping out the "local" IP with the actual IP of the server but that yielded no results. I also am noticing that based on the "docker ps -a" command as well as running "netstat -an | grep LISTEN" I'm noticing the port 32002 which is configured for the ML image is not shown anywhere. I feel like it probably should...
Any recent changes you've made to Immich or your system.
This is a new install. I've reverted back to ROCm ML image and the local "127" IP after troubleshooting.
Details about your system (both software/OS and hardware).
TrueNAS SCALE 25.10.1
Minisforum N5 PRO AI
64GB RAM
AMD Ryzen AI 9 HX PRO 370 w/Radeon 890M
Details about your storage (filesystems, type of disks, output of commands like fdisk -l and df -h).
4 x 16TB HDDs
Configured in RAID Z 1, with usable capacity of ~46 TB
fdisk and df are attached.
The version of the Immich server, mobile app, and other relevant pieces.
immich 2.5.2 (latest in TrueNAS)
Lastly, thank you! This project is awesome.
df.txt6.47KB
fdisk.txt3.41KB
immich-permissions.log1.36KB
immich-redis.log2.86KB
immich-server.log51.37KB
docker_ps_-a.txt2.12KB
config.txt1.46KB
