So, as I'm sitting here and doing my research, I'm wondering a how unRAID handles a major concern of mine: silent data corruption

 

Since unRAID can be most comparable to a RAID 4, how is it able to handle data corruption, namely silent data corruption comparability to ZFS?

 

This has been my major concern, among others (massively degraded performance), before building out my NAS.

 

Any insight would be helpful!

 

 

So, as I'm sitting here and doing my research, I'm wondering a how unRAID handles a major concern of mine: silent data corruption

It doesn't. So any check and correction strategy will need to be implemented separately. Generating par volumes for each folder is one way to do it for static content, I can't think of a good way off the top of my head to handle it for dynamic content. Your question is rather open ended though, because "silent data corruption" has not been well defined here. I can think of a couple ways for content to get mangled before it is even seen by the NAS, so the NAS would have no way of knowing or correcting anything anyway. The chances of data that has been correctly written to unraid changing without any errors showing up is extremely small.

unRAID does not directly address data rot. This concern is mitigated with periodic parity scans (I use the unMenu monthly option). Potentially, a future change from RieserFS to BTRFS will address your concern.

 

I have read threads here on others using probably better data scrubs, but unMenu meets the bar for me.

UnRAID doesn't have an integral mechanism to address this; but as noted, it's VERY unlikely that a parity check wouldn't catch any deteriorated data.

 

Whenever a parity check results in a non-zero "corrected" count, I run a full compare between my UnRAID array and my backups ... this takes several days -- but only a couple minutes of "my time" to start the comparison.    I've NEVER found any data errors when I've done this ... but it's nice to be certain 

 

I've NEVER found any data errors when I've done this ...

 

Good to know, indeed - it supports Tom's assertion that it is (almost?) always the parity drive which gets out of sync.

I've NEVER found any data errors when I've done this ...

 

Good to know, indeed - it supports Tom's assertion that it is (almost?) always the parity drive which gets out of sync.

 

If there were actual data errors on the data drives, I'd expect errors in the disk "errors" column ... which (knock on wood) I've never had except when a cable needed re-seating.  But I have had a couple of parity checks over the years that had non-zero sync counts.

 

So, as I'm sitting here and doing my research, I'm wondering a how unRAID handles a major concern of mine: silent data corruption

It doesn't. So any check and correction strategy will need to be implemented separately. Generating par volumes for each folder is one way to do it for static content, I can't think of a good way off the top of my head to handle it for dynamic content. Your question is rather open ended though, because "silent data corruption" has not been well defined here. I can think of a couple ways for content to get mangled before it is even seen by the NAS, so the NAS would have no way of knowing or correcting anything anyway. The chances of data that has been correctly written to unraid changing without any errors showing up is extremely small.

 

Sorry, it was rather open-ended. AFAIK, and correct me if I'm wrong, in essence unRAID is a software RAID with single parity check (RAID 4). With one parity, only one detected error can be corrected by RAID, which is usually sufficient. If an error is undetected, the single parity situation would not be able to determine which disk is at fault so it just corrects the parity and checksum information. In a two parity RAID, the undetected error can be corrected as opposed to just a single parity disk.

 

ZFS requires that each data block be verified against an independent checksum, after it has arrived in the host's memory. This end-to-end checksum principle allows for silent data corruption to be corrected and fixed.

 

I've just had issues in the past where data was corrupted and never detected, and as time passed, the files eventually became unusable.

 

I just wonder if there is something else you can do with unRAID, such as additional parity disks or plugins that do some sort of end-to-end checking such as ZFS.

I run a full compare between my UnRAID array and my backups...

 

What do you mean "full compare"? What, exactly, do you do?

 

Sent from my DROID RAZR using Tapatalk 2

 

 

What do you mean "full compare"? What, exactly, do you do?

 

I'd think it's fairly self-explanatory ... I run a comparison program (FolderMatch) and compare all of the files on the UnRAID server with my backups.

 

I just wonder if there is something else you can do with unRAID, such as additional parity disks or plugins that do some sort of end-to-end checking such as ZFS.

 

unMenu has the monthly_parity_check, which is a bit like the ZFS scrubbing.