On 23/03/2020 18:13, Roger Heflin wrote:

Those default timeouts are usually at least 30 seconds, but in the
past the scsi subsystem did some retrying internally.  The timeout
needs to be higher than the length of time the disk could take.
Non-enterprise, non-raid disks generally have this timeout set 60-120
seconds hence MD waiting to see if the failure is a sector read
failure (will be a no-response until the disk timeout) or a complete
disk failure (no response ever).

https://raid.wiki.kernel.org/index.php/Timeout_Mismatch

The whole website is reasonably up-to-date, so it's worth a read.

Cheers,
Wol