NVMe I/O errors on Proxmox

I migrated my system from an (unstable) NUC to a home-build AMD-based tower system with a little bit more space for cooling. After a couple of days the system crashed reporting I/O errors on the secondary NVMe.

After some investigation this seems to be related to a power saving mechanism that puts the NVMe (partially) to sleep. To disable this power saving mechanism I had to change this file /etc/default/grub:

The problem was of a SSD features, the Autonomous Power State Transitions(APST) was causing the freezes.


GRUB_TIMEOUT=10

GRUB_DISTRIBUTOR=`lsb_release -i -s 2> /dev/null || echo Debian`

GRUB_CMDLINE_LINUX_DEFAULT="quiet splash nvme_core.default_ps_max_latency_us=0"

GRUB_CMDLINE_LINUX=""