Summary of issues with MHVTL

The objective is to understand what are all the issues we are facing with MHVLT and start addressing them:

1. problem to spawn multiple drive pods each with their own MHVTL
2. LTO9 support (RAO) not complete yet
3. MHVTL is a kernel-module, which means that it cannot be containerized. This complicates the setup as it is not possible to use any kubernetes cluster; the machines specifically need to have MHVTL installed.
4. MHVTL is not meant to be run across multiple machines. Even if we were to install mhvtl on each machine in a cluster, the tapes, drives and libraries are not shared between machines.
5. MHVTL is not maintained by us. In the past there have been problems where we had to package our own patched version as waiting for a release took too long.
6. We have had some performance issues with MHVTL in the past (see #884). I am not sure to what degree these have been addressed already
7. Releases not made frequently enough - issues when trying to compile latest mhVTL tag against Alma 9 - Mark fixed that and cut a new release, but releases are not made regularly. @poliverc added mhVTL compilation as part of CI pipeline in https://gitlab.cern.ch/cta/cta-dependencies in order to improve on this matter. Perhaps we could offer some CI to Mark Harvey or release management since we have more resources.

• ... ?