Open sharpeye opened 2 months ago
if the comparition fails, perform a resync.
this part will make the code more complex I decided to simply return E_REJECTED if the data is the same in 2 out of 3 replicas, one of the next retry attempts will hit those 2 replicas and succeed there's no point optimizing this scenario since it should never happen (or almost never) the discrepancy will be eventually fixed by the scrubbing process
In addition to the regular reading operation, mirror-3 disk could read a checksum from another replica, compare cehcksums and, if the comparition fails, perform a resync.