As explained on this page, write vulnerabilities are Refers to the inconsistency between the array disks when the power is interrupted during the writing process. This page also explains that it affects both RAID-5/6 (if the power is interrupted after data is written but before parity is calculated) and RAID-1 (Data is written to one disk but not to other disks) and this is a hidden problem that can only be detected during resynchronization/scrubbing, or during the reconstruction of one of the disks (catastrophically)…but, most of the Other sources talk about it because it only affects the level of parity based on RAID.
According to my understanding, I think this may also be a problem with RAID-1, because from the disk containing this hole Reading will return garbage, so…is there a problem for every RAID level? Is it implementation dependent? Does it only affect the software RAID or the hardware controller? (Extra: how does mdadm perform in this regard?)
In the process of understanding the advantages of RAIDZ, I encountered the concept of writing loopholes.
As stated on this page As explained, write vulnerability refers to the inconsistency between array disks when the power is interrupted during the write process. This page also shows that it affects RAID-5/6 at the same time (if after writing data but calculating parity Previously powered off) and RAID-1 (data written to one disk but not written to other disks) and this is a secret problem that can only be detected during resynchronization/scrubbing, or during the reconstruction of one of the disks (disastrous Ground)…but, most of the other sources talk about it, because it only affects the level of parity based on RAID.
From my understanding, I think this may also be a problem with RAID-1 , Because reading from the disk containing this hole will return garbage, so…are there any problems for each RAID level? Is it implementation dependent? Does it only affect the software RAID or the hardware controller? (Extra: How does mdadm perform in this respect?)
This is the reason why raid requires cache batteries or some other cache consistency verification methods. All raid cards should have battery backup cache, and all storage controllers should have mirrored cache. For software raid, I don’t think there is a good answer. I think even attacking Z may fail due to power loss.