MD raid scrubbing and NVMe

[Hannes Reinecke]

Hi all,

now there's a question from one of our partners, who was complaining
that running 'mdcheck' on an NVMe RAID array was particularly slow.
While this could be resolved with fine-tuning MD parameters, it kept
me wondering:
How useful is RAID scrubbing on NVMe?
 From my perspective scrubbing is useful if one wants to avoid excessive
recovery latency on bad sectors (eg TLER).
But the FTL on NVMe already does a quite heavy remapping / rewriting /
garbage collecting for every write, and so it should be detecting
(and correcting) faulty sectors far earlier.
So how effective is RAID scrubbing for NVMe? Can't we delegate the
task of detecting faulty sectors to the FTL and _not_ scrub at all?

(Especially as modern file systems like btrfs do an internal scrubbing
_anyway_...)

Cheers,

Hannes
-- 
Dr. Hannes Reinecke                  Kernel Storage Architect
hare at suse.de                                +49 911 74053 688
SUSE Software Solutions GmbH, Frankenstr. 146, 90461 Nürnberg
HRB 36809 (AG Nürnberg), GF: I. Totev, A. McDonald, W. Knoblich