Hi,

 

Is unraid impacted by the 'skip_copy' problem described below?

 

https://forum.synology.com/enu/viewtopic.php?t=135059

http://www.sbsfaq.com/?p=4277&doing_wp_cron=1505991093.5444660186767578125000

http://www.sbsfaq.com/?p=4321

 

Regards,

Edited September 21, 20178 yr by themaxxz

Extremely unlikely. unRaid uses parity and not RAID5 logic.

27 minutes ago, bjp999 said:

Extremely unlikely. unRaid uses parity and not RAID5 logic.

 

You are probably right, but it would be nice to have confirmation from 'Lime Technology' directly.

E.g. synology mentions that RAID 5, RAID 6, RAID F1, or SHR with more than 3 disks are impacted.

UnRaid is different than RAID. UnRaid is not involved in the writing directly to the individual disks (although it must pass writes through from the md# device to the sdX device). It monitors those writes, computes parity, and writes parity.  No data is striped across drives. It is a very difference paradigm from Raid. If parity were being corrupted, our parity checks would be showing that corruption. That's the only part that unRaid manages.

 

So unless the issue is related to a problem with the handing off of writes, I doubt very much this affects unRaid. @limetech might be interested to know about this, and may be able to comment further..

First off, this does not affect unRAID becase the md/unraid driver diverged quite some time ago from 'stock' linux md driver.  Interestingly what 'skip_copy' optimization does in stock md is something I've wanted to put into md/unraid for quite a while - I guess I need to re-think that idea

 

'skip_copy' is an optimization in stock linux md driver.  I believe it was introduced here:

https://www.spinics.net/lists/raid/msg46364.html

 

The specific issue seen by QNAP and Synology (and all linux md based systems I suppose) is talked about here:

https://www.spinics.net/lists/raid/msg59084.html

 

This looks like a hairy bug to fix properly, though workaround is to disable skip_copy feature.  (Again, not talking about unRAID here.)

 

Edit: to be clear: I don't mean to imply anything about linux md integrity issue.  The 'bug' was not introduced merely by adding the 'skip_copy' feature, I think it probably got introduced sometime after by other changes in the block layer in later kernels.

Thanks for confirming there is no issue with unRAID.