unr41dus3r

Would it be possible to move the license from an single USB to the Internal Boot Devices (which are 2 USB for me) or one of the USB devices in the new USB Pool?

One thing i want to mention and i also could not find this information in the announcement? The internal boot Pool will count in to the License of maximal Array devices. I only was informed after moving to dedicated internal boot pool, that i can't start my array anymore as i do not have the correct license... I would understand if i use a NON dedicated boot pool, but i moved from 1 USB device to the 2 USB device pool. @JorgeB maybe you could implement an check that checks the current array count with the planned boot pool count and verifys the current license if its enough afterwards. Went now from Pro to Plus (or the other way arround) as i really needed the array but yeah, sucks a little bit (again, probably i overead it anywhere) Or overthink the decision that an dedicated boot pool will be added to the license count. Edit: Found the mentioning in this thread from JorgeB about the license count, but nothing official?

Hey, i am in the same boat ;) I installed thors Kernel and the is "better" recognized, but is nearly working the same as before. SR-IOV is not working. I thought the kernel is missing some features, for my understand what would be necessary for the unraid team to update to get SR-IOV working?

About your experience with the FAN, did you try to plug something into the Display Port? I plugged in an Monitor first (now an DP Dummy) and the FAN is completely normal and i can't hear it in daily tasks. As you mentioned, it is really buggy at the moment, still i got it running in Jellyfin and also in the Ollama Arc Version, which was one of my main reasons to get a B50. At the moment i don't use ollama, because of the crashes if the GPU is mounted in different containers.

An Jellyfin integration is not planned as i guess? It would be really aweseome, if this would be possible

Thanks for your response, but this is exactly what i did. You can reproduce it yourself if you want, create an new container (does not matter which one or from app store) Set the container to autostart (check mark in the Docker GUI) and remove the container with the webgui. If you check now the "/var/lib/docker/unraid-autostart" file, the autostart config for this container name is still there. Now, you add the same container in your Compose File and if you use the same container name, this setting is still persistent. So after an reboot (if the container exist while array starting) the container will still be started, as the container was set to autostart in the autostart file. The problem, Unraid is not updating this file if you remove an container from the web gui. I disabled the autostart for all containers in the GUI before removing, but for some it did not stick and was not edited into the file, but this is a different problem with unraid. As i remember i tried this already and i can see in the syslog it writes in the Log "docker compose starting stack 12345" (not exactly the log) but the containers does not start and i see not more output. I will try it again, and report back if i see more. Edit: this is my log with debug enabled, i only obscured the names of compose names. All stacks are only have compose files and only one uses and ENV. The /config/ folder is on my cache drive If i use a userscript to start the containers while array start it works without any problem. ... Sep 25 09:16:48 tower emhttpd: nothing to sync Sep 25 09:16:48 tower root: Starting compose stack: stack0 Sep 25 09:16:48 tower root: docker compose -f '/mnt/user/config/docker/stack0/docker-compose.yml' -f '/mnt/user/config/docker/stack0/docker-compose.override.yml' -p stack0 up -d Sep 25 09:16:48 tower root: Starting compose stack: stack1 Sep 25 09:16:48 tower root: docker compose -f '/mnt/user/config/docker/stack1/docker-compose.yml' -f '/mnt/user/config/docker/stack1/docker-compose.override.yml' -p stack1 up -d Sep 25 09:16:48 tower root: Starting compose stack: stack12 Sep 25 09:16:48 tower root: docker compose -f '/mnt/user/config/docker/stack12/docker-compose.yml' -f '/mnt/user/config/docker/stack12/docker-compose.override.yml' -p stack12 up -d Sep 25 09:16:48 tower root: Starting compose stack: stack123 Sep 25 09:16:48 tower root: docker compose -f '/mnt/user/config/docker/stack123/docker-compose.yml' -f '/mnt/user/config/docker/stack123/docker-compose.override.yml' -p stack123 up -d Sep 25 09:16:48 tower root: Starting compose stack: stack1234 Sep 25 09:16:48 tower root: docker compose -f '/mnt/user/config/docker/stack1234/docker-compose.yml' -f '/mnt/user/config/docker/stack1234/docker-compose.override.yml' -p stack1234 up -d Sep 25 09:16:48 tower root: Starting compose stack: stack12345 Sep 25 09:16:48 tower root: docker compose -f '/mnt/user/config/docker/stack12345/docker-compose.yml' -f '/mnt/user/config/docker/stack12345/docker-compose.override.yml' -p stack12345 up -d Sep 25 09:16:48 tower root: Starting compose stack: stack123456 Sep 25 09:16:48 tower root: docker compose -f '/mnt/user/config/docker/stack123456/docker-compose.yml' -f '/mnt/user/config/docker/stack123456/docker-compose.override.yml' -p stack123456 up -d Sep 25 09:16:48 tower usb_manager: Info: virsh called VM12345 005 002 Device attached successfully ... My Userscript i set at "Array Start" which works without any problem echo "START CONTAINERS START" sleep 15 #stack0 cd /mnt/user/config/docker/stack0 docker compose start

I had some more problems with the migration to docker-compose on unraid. I will leave some more information here, because this problem cost me hours of debug and searching for the solution in the wrong place. After moving all my containers to compose, i still had some containers who did an autostart while everything was disabled. I disabled before moving to compose all autostart flags in the webgui. After moving to compose, i moved all templates-user files (unraid docker xml files) to a new place, to be sure that they are not the problem. Then i searched hours, in the compose as i thought i removed all old settings. I was wrong, in unraid the container autostart are saved in a different file, after removing them from this place, the autostart stopped. /var/lib/docker/unraid-autostart Another problem, some containers never stopped automatically while disabling the docker service or array stop. I could find the following line in the logs: Sep 25 08:43:17 server root: Stopping compose stack: I have much more compose files and normally it looks like this: Sep 25 08:43:17 server root: Stopping compose stack: compose1234 So something was not correct running with the compose stack and the plugin. My simple solution at the end was, backup the compose, delete the compose stack in gui and reconfigure it. My last and still existing problem Not a single of my Docker-Compose Stacks is auto starting, so this is not related to one container or one stack, same for all. After reboot all stopped. The only idea i have is that my config files are not in the default folder, but who knows. I will probably fix this with an userscript and have one benefit i can also control the order when a compose start.

I am so blind, yes you are right, thanks!

Little feedback from my side. Jellyfin works fine. Tested ollama specially for arc https://github.com/eleiton/ollama-intel-arc Which also works fine, BUT i experienced now already 2 times my unraid server completly crashed in the moment i shared the GPU with different containers. When i used jellyfin OR ollama, everything fine, if both uses the GPU, it is a matter of time till the server was stuck and had to do a hard reboot. As mentioned before, when i add the GPU to my Virtual Maschine, the GPU also will be removed from /dev/dri what could cause problems. So for me it is OK, but if you want use different services with the GPU at the same time, there could be problems. About Plex, i can't add anything as i don't use it.

Another finding, a maybe a little bug fix worth. If Docker is disabled and the array starts, as i can see in the logs docker compose still trys to start the compose stacks, but docker host is not running. Could be checked before. Still have the problem that my compose files are not starting all containers and still searching what is causing this. Is it a problem if the compose folder is on the array instead of the USB drive? thanks for the work and the great plugin!

As it looks like, jellyfin is working fine, tried now also immich and nextcloud/go-vod There i get no streams, will see whats going on, but already great use jellyfin is running

Your command shows a stream and i think this is the same as the gpu_stat is doing. In GPU stat i have not much data but i can see if a app uses the card

I make a new post because i can report back, for me transcoding looks like it is working fine with Jellyfin ! I am still on 7.1.4 and installed the intel_gpu_top plugin, no other special changes. Mapped the card to jellyfin and as i can interpret the logs it is transcoding fine. As intel_gpu_top is not working i am not 100% sure how to verify. As my testing skills are limited, to verify the Intel GPU is doing something, i tried HDR tonemapping and it also works. If i would point the QSV device mapping to my IGPU it fails correctly (to verify it is really the doing of the intel card) Next step will be to try a different container, to check QSV is working (on my AMD CPU/APU system)

How is the fan on your side? For the moment it would fine for me if i could use one container with the card, but at the moment i need to connect it to the VM and can't use it anywhere else. Sadly i will send the card back if the fan can't be fixed, because i have no idea how long i would have to wait. Edit: Found a fix for my problem, i had no display plugged in into the Intel Arc Pro and set my onboard GPU to default. I still have my onboard GPU set to default but plugged in an dongle into the Intel Arc Card and now it stays silent. Which looks for me like an bug, as long the card has no output, the fan spins 100% I really have to say, i am really "shocked" that such basic problems are still here. Will test now transcoding and report back if anything works