Discussion:
[smartmontools-support] smartd performance at scale
Exec Unerd
2016-11-28 19:23:21 UTC
Permalink
I'm looking for best practices for optimizing smartmontools performance on
a large system with >1000 drives. I’d hate for this simple widget to swamp
the CPUs or memory, or slow disk IO especially under regular operational
loads, but I don’t have a lab system big enough to test it. Are there any
metrics that would allow me to extrapolate system load effects with
smartmontools at this level?

My goal is to just monitor disk health basics in the background, which
would let me flag a pass/fail event and then let me check logs or run more
extensive tests manually when there’s a problem.

smartd config (logging plus weekly short tests)
/dev/sda -H -o on -S on -l selftest -l error -f -s (S/../.././02)
Loading...