GitLab

After attending a Frontdesk meeting it was made very clear to me that #17834 (closed) was a major pain for users. The Frontdesk team is worried that we are moving too fast with kernel updates.

I think it might be worth revisiting our guidelines for kernel upgrades, and perhaps do a blameless post-mortem of #17786 (closed) where we proceeded with the kernel upgrade to 5.7.0-1 despite knowing about this regression. Perhaps we want to switch to another model, for instance, to skip the x.y.0-1 kernels and wait for the x.y.0-2 or higher when the some of the more painful regressions are fixed, although that would makes us lag on security fixes since we use Sid kernels => there are no backported security fixes.