Subject:
You can create a level of redundancy on a zpool by using a mirrored set of VDEVs.
If you have a pool that is using mirrored devices then it can use the "bit-rot" repair feature and recover from most data errors.
This allows ZFS to try to repair damage to data when it finds a checksum that doesn't match the data it is reading.
If each of the disks shown in 'zpool status' output is a single physical disk and not a RAID 1 array, then losing one disk or having corrupted writes will mean all data is lost on the pool. The only recovery method will be to destroy the pool and recover from a backup of the data.
Adding SSD cache for write operation will not prevent data errors or help with redundancy.
Detail:
If you have a ZFS RAIDZ or RAID10 pool then the filesystem will do automatic repairs during read/write operations so you generally don't need to do anything to repair bit-rot, it's automatic.
However, if you have a pool with no ZFS RAID VDEV's that provide Data Redundancy (common when layering over hardware RAID), then the Storage Pool will be unable to automatically make repairs when the filesystem detects a block with a bad checksum.
Additional detail can be found here:
QuantaStor_Troubleshooting_Guide#Fixing_ZFS_bad_sectors_.2F_ZVOL_checksum_repair_process
Solution:
To avoid the risk of data corruption, using disks as pass-through (RAID0) presented to QuantaStor is recommended. A storage pool can then be created on top of the disks.
Comments