How does UnRaid update parity with multiple disks? I gather it reads the parity drive and the target drive, calculates the new parity value based on the change in the target drive and then writes the target and parity drives? I assume this, as a write to disk 1 does not cause disk 2 to spin up. So it's not recalculating full parity for every write. I see this being good for two reasons, you're not spinning up disks you don't need and the parity calculation would be a dog as n got really big.

 

I guess the down side might be not knowing you had a disk going bad. I plan to implement one of the UnNotify scripts once I figure out which one is the latest  , but I gather this is only using SMART to monitor the drives. Does anyone schedule a parity check say once a month and email themselves the results?

 

 

How does UnRaid update parity with multiple disks? I gather it reads the parity drive and the target drive, calculates the new parity value based on the change in the target drive and then writes the target and parity drives? I assume this, as a write to disk 1 does not cause disk 2 to spin up. So it's not recalculating full parity for every write. I see this being good for two reasons, you're not spinning up disks you don't need and the parity calculation would be a dog as n got really big.

 

I guess the down side might be not knowing you had a disk going bad. I plan to implement one of the UnNotify scripts once I figure out which one is the latest  , but I gather this is only using SMART to monitor the drives. Does anyone schedule a parity check say once a month and email themselves the results?

 

 

 

for every block written, unRAID uses an XOR operation on data which already exists on the parity drive.

This is why when the array is first created, parity is built by reading every sector from every drive to build the parity sector for the matching drives.

Therefore if a single drive gets written to, only the parity for that drive is updated .

 

I believe the operation is something like this.

 

1. read of old data sector on drive.

2. read of old partiy sector.

3. xor out the old data from the parity sector.

4. write the new sector on the data drive.

5. xor in the new data to the partiy sector.

6, write the partiy sector to the drive.

 

 

Does anyone schedule a parity check say once a month and email themselves the results?

[/qiuote]

 

Many of us have monthly parity check jobs. THis is highly recommended. Even on a raid 5 array this is recommended. I believe they sometimes reference this as a scrub operation. There are threads with the cron job shell on the board.