Linux kernel doesn't care about your disk health
a day ago
- #linux-monitoring
- #disk-health
- #storage-protocols
- Adding disk health monitoring to the simob agent revealed complexities due to legacy storage protocols and kernel permissions.
- SMART data is messy with vendor-specific attributes, making interpretation difficult without tools like smartmontools.
- Accessing SMART data requires root privileges because it involves direct hardware communication via ioctls, unlike other kernel-mediated stats.
- Udisks2 can act as a middleman for SMART data but offers limited granularity and often fails with NVMe drives on older versions.
- NVMe drives use a standardized health log instead of SMART, but udisks2 may not support it without recent updates, leaving a monitoring gap.
- The Linux kernel lacks built-in disk health monitoring, requiring external tools and highlighting a blind spot in system observability.