In those rare conditions, when one of my unRAID server crashes due to problems with an adapter card or cabeling, unRAID starts with a parity check that does corrections to the parity disk.

 

In 100% of these conditions this trashed the content of the parity drive.

 

I would like to see an option like this: Automatic parity check corrective [yes|no] (default yes).

 

I saw a different topic regarding Automatic parity checks but this describes a specific and different topic.

 

Thanks for listening.

 

Having fallen foul of this myself on more than one occasion, I agree. The assumption is always that if there is an inconsistency between the parity drive and the rest of the array then the error must be in the parity, but that simply isn't always the case. What we also really need is the option of having multiple parity drives, in order to pinpoint the actual problem. But that's another topic 

What we also really need is the option of having multiple parity drives, in order to pinpoint the actual problem. But that's another topic 

We know that dual parity is a feature of the 6.2 release but I have not seen any feedback as to whether this allows the problem disk to be identified in such cases.

7 years ago, I had a loosened adapter in one unRAID box. Several thousand sync errors were the result of an automatic parity check after the box came up again. All sync errors were applied to the parity disk. I didn't give much attention to that in 2008, the implications were not clear to me. During all those years I did swap many drives. All these drives received these errors from the parity disk. It took me several years to find out why I occasionally find corrupt files. Last year I replaced all 2008 files from some backup DVDs I found in the garage.

 

Since that, whenever I restart an unRAID box, I immediately look if an automatic parity check has been started. I always cancel them and restart with an unchecked "Write corrections to parity disk". If there's a single error, I will restart with an checked "Write corrections to parity disk". If there are lots of errors, I do check the box first.

 

So my wish is a global setting like that found in the topic.

 

Regards.

 

Doesn't this already exist in the scheduled parity check option?

Doesn't this already exist in the scheduled parity check option?

Only applies to the scheduled parity check. The request has to do with the automatic parity check which follows an unclean shutdown.

...During all those years I did swap many drives. All these drives received these errors from the parity disk. It took me several years to find out why I occasionally find corrupt files.

Which is why things like MD5 checksum's exist.  After a rebuild, you can go through and compare the checksums of the files to see if anything is corrupted. 

The request below is a more complete analysis of the overall "automatically correcting parity check" issue and it also gives those of us who have 95 degree daytimes and 50 degree nighttimes to schedule our parity checks in a way that brings longevity to the drives.

 

http://lime-technology.com/forum/index.php?topic=44012.0