Packaging broken

[aleksandaratanasov]

Running Ubuntu 22.04 with Nvidia proprietary driver 515. It appears that the APT packages break between 510 and 515 regarding NSCQ. While I did not install the package myself (probably came along with CUDA, cuDNN or some other ML/GPU computing related library/toolkit), my system is currently broken due to libnvidia-nscq-xxx. I am unable to update, upgrade, fix or install anything due to this package.

Attempting to install the package results in the following error:

(Reading database ... 228886 files and directories currently installed.)
Preparing to unpack .../libnvidia-nscq-515_515.86.01-0ubuntu0.22.04.1_amd64.deb 
...
Unpacking libnvidia-nscq-515 (515.86.01-0ubuntu0.22.04.1) ...
dpkg: error processing archive /var/cache/apt/archives/libnvidia-nscq-515_515.86
.01-0ubuntu0.22.04.1_amd64.deb (--unpack):
 trying to overwrite '/usr/lib/x86_64-linux-gnu/libnvidia-nscq.so.515.86.01', wh
ich is also in package libnvidia-nscq-510 515.86.01-0ubuntu0.22.04.1
dpkg-deb: error: paste subprocess was killed by signal (Broken pipe)
Errors were encountered while processing:
 /var/cache/apt/archives/libnvidia-nscq-515_515.86.01-0ubuntu0.22.04.1_amd64.deb
E: Sub-process /usr/bin/dpkg returned an error code (1)

I was told that the problem lies in moving from multiverse to restricted. I have restricted enabled yet it does not solve the issue.

[kmittman]

Based on the apt policy output on stackexchange, this .deb package appears to be coming from Canonical's repository rather than NVIDIA's CUDA repository. I will reach out to the Ubuntu maintainer to see if this is expected.

Out of curiosity, can you please elaborate on your hardware setup? NSCQ is an optional dependency for DCGM and intended for DGX (and similar) multi-GPU machines, see: https://docs.nvidia.com/datacenter/tesla/hgx-software-guide/#nscq