Hi everyone,

I have recently managed to move to unraid 5.0 rc10. Unfortunately, some time later, i got a red balled data drive which i know is due to a cable problem. I understand that the trust my array procedure does not work anymore.

1. Somewhere in the forum i show that I could use utils->new configuration and then there will be a choice to trust the parity. does that work?

2. If, instead, I want to rebuild my data drive, using the trusted parity, how would I do that in unraid 5?

Start the array with the red drive unassigned. Then start the array with the drive assigned.

Start the array with the red drive unassigned. Then start the array with the drive assigned.

 

Thanks dgaschk, I already did that and, indeed, it worked fine.

 

But the other question is still valid, how would i go about implementing a 'trust my array' procedure?

So, there is no way to trust my array in unraid 5; Or is it not safe yet?

If your drive red-balled because it could not be written to and then you used the trust my array procedure you'd end up losing the data written between the time it red-balled and the time you quit writing data. So, you'd end up with missing or corrupted data.

 

However, the data which didn't get written to the red-balled data drive is still stored via the parity drive so rebuilding the red-balled data drive recovers without any data loss.

 

Which method do you want to use to fix the red-balled drive, the one that can lead to data corruption or the one that will recover all the data?

 

FYI, the trust my parity procedure does still work.

If your drive red-balled because it could not be written to and then you used the trust my array procedure you'd end up losing the data written between the time it red-balled and the time you quit writing data. So, you'd end up with missing or corrupted data.

 

However, the data which didn't get written to the red-balled data drive is still stored via the parity drive so rebuilding the red-balled data drive recovers without any data loss.

 

Which method do you want to use to fix the red-balled drive, the one that can lead to data corruption or the one that will recover all the data?

 

You are obviously generally right. But there might be few chances that a drive got red balled while reading from it (that is what happened in my case, the parity drive was spun down and I knew it was a matter of a cable and not a problematic disk, although it was the first time i saw a red ball while reading).

 

Additionally, I can imagine a situation where you feel that a second disk is in danger and you are in hurry to restore the red balled one lest you lose the other (even if that costs you a few missing bytes).

 

IfFYI, the trust my parity procedure does still work.

 

That is good. I was just wondering because I saw some posts that said the 'trust my array' procedure does not work.

E.g.: http://lime-technology.com/forum/index.php?topic=14830.msg139730#msg139730

 

UnRAID v5.0-rc10 includes a fix for the "mdcmd set invalidslot" command, which had not worked for most of the v5.0 beta series.  The v5.0-rc10 change log includes the following:

 

emhttp: fix issue where user-entered "mdcmd set invalidslot" gets ignored upon array start

UnRAID v5.0-rc10 includes a fix for the "mdcmd set invalidslot" command, which had not worked for most of the v5.0 beta series.  The v5.0-rc10 change log includes the following:

 

emhttp: fix issue where user-entered "mdcmd set invalidslot" gets ignored upon array start

 

Thanks, that is good to know in case an unexpected situation arises.

 

I am currently rebuilding parity using a new 3TB HD and I though of another instance that could be useful: If during a parity rebuild, a data disk became redballed (not because it were faulty but because a cable moved), I would have to trust my array and rebuild parity.