SOLVED
AF-Series Arrays have Spare chunks on every SSD. If we have 24 SSDs in the Array,
1st RAID set will be (20+3) meaning 1st chunk on 24th SSD is not taken as it is marked as Spare chunk.
2nd RAID set will be (20+3) meaning 2nd chunk on 23rd SSD is not taken as it is marked as Spare chunk.
So it goes in this fashion. When a drive fails, spare chunk works much the same way as a spare drive does.
The data and parity within failed drive are recalculated using remaining data and parity blocks and then stored in the spare chunk. This is performed by working through every stripe with valid data, rebuild missing data and write the data to the appropriate spare chunk.
Same process will be applied even if there are 3 simulateneous drive failures. If 4th drive also fails before the rebuilding completes, Intra spare (which exists inside every chunk) will be used to reconstruct that data.
In HF-Series, there is NO Spare drive or Spare chunk. So if a drive fails and once the drive is replaced, data is reconstructed and written to the replaced drive. If 3 drives in a HF-series fail before any of the failed drive is replaced, Array will automatically shutdown to prevent failure of any more drives which is a configurable option.
Rebuild to Replacement (Physical) Drives in AF-Series
Walk through each stripe with valid data
Copy data/parity from spare chunks back to replacement drive if available, OR
Rebuild data/parity from the rest of the drives and write to replacement drive
After rebuild process is complete, every spare chunk is erased and returned to the drive as “overprovision” capacity
I work for HPE
