Hello, first time posting here, as my experience with OMV is quite recent but has been mostly good so far. Here is what happened:
1. I have an OMV 5.6.10-1 box set up with an EXT4 RAID0 with docker handling plex, radarr, sonarr, etc., which have had their issues but I managed to sorta fix that
2. I also have a quadro m2000 in there that I want to use for plex transcoding. I initially followed this thread: How to setup Nvidia in Plex docker for hardware transcoding? and it seemed to be working but I was getting very bad performance - only around 15% GPU utilisation when transcoding 4k HEVC to 1080p. (opposed to 60-80% when I was testing the card under windows with the same file)
3. I thought it was something with the drivers and the cuda/nvenc versions so I dug in to find alternative methods, and came across this: [HowTo] Nvidia hardware transcoding on OMV 5 in a Plex docker container - I can see the person that created this guide is marked as "Beginner" but the guide is very well laid out and also corresponds to other forums where people are saying the correct way to do it is through Debian Sid and backport drivers, not through the official nvidia drivers as instructed in the first thread.
4. So I added Sid to sources and followed the steps, making sure to clean the previous installation first (apt-get purge *nvidia*)
5. The first try I got the driver installing and was getting the prompts as expected. Before the driver started installing I got a prompt saying there are modules that are currently active and need to be suspended in order to get the installation going, so I selected yes. I think that was my first mistake as during the driver install I lost SSH access and couldn't get it to work again until I went in the OMV GUI and disabled/re-enabled SSH.
6. After the initial fail, I did a purge again and repeated the steps, but this time there were no prompts and no messages - it seemed the driver installed without an issue, but when I tried nvidia-smi I got "NVIDIA-SMI has failed because it couldn't communicate with the NVIDIA driver. Make sure that the latest NVIDIA driver is installed and running." and no matter what I tried, it was just not picking it up. I also noticed the second time the driver version was older than the first, not sure exactly but was something like latest 460.84 the first time vs 430.xx on the consecutive tries.
7. I repeated 6) couple more times with the same result and was almost hopeless I'll get it to work until I found on multiple threads an apt upgrade might help so did that. There were quite a few updates a lot of which installed with no issue until it reached the libpam-modules. It failed and suggested apt --fix-broken install which resulted in a full-screen prompt saying pam_tally and pam_tally2 modules are deprecated in pam 1.4.0-7 and all their instances in the configs have to be replaced with pam_faillock before the upgrade can continue.
8. I dug in to understand what it will take to do that and if it will break everything, and found this: https://bugs.debian.org/cgi-bi…port.cgi?att=1;bug=982530 - it seemed pretty serious and potentially installation-breaking, so I did not touch any of the files in /etc/pam.d/ (grep-ing showed 4 configs containing the modules)
9. Instead, I tried to revert what I did - removed Sid from sources and did omv-aptclean as suggested elsewhere. This didn't seem to do much and after restart I'm now only getting the below error - no prompts or messages, just that.
10. I still have full access to the box and everything seems to be still working (GUI, SSH, containers, etc.), but I cannot install/remove any modules and all apt commands result in the same error below:
root@HTPC:~# apt --fix-broken install
Reading package lists... Done
Building dependency tree... Done
Reading state information... Done
Correcting dependencies... failed.
The following packages have unmet dependencies:
libpam-modules : PreDepends: libpam-modules-bin (= 1.3.1-5) but 1.4.0-7 is installed
E: Error, pkgProblemResolver::Resolve generated breaks, this may be caused by held packages.
E: Unable to correct dependencies
Any suggestions that would not result in a reinstallation of the whole box are greatly appreciated. Thanks!