As far as I know, there is no circumstance in which you would not want to run a parity check immediately after a parity sync completes.  The sync writes all the parity data, the check makes sure it can be read back again.  In the sense of being sure that you can recover from a single drive failure, your array is not protected until both of these steps are complete (because unreadable parity data is useless, and you don't know if it can be read until the parity check completes without errors).

 

So why not combine them into one operation?  A parity sync should be immediately followed by a parity check, without any user input needed.

 

Any reason not to do this?  Sure this one operation will take about twice as long as either half, but it is completely necessary, so why put it off?

your array is not protected until both of these steps are complete

Actually, it is not confirmed to be protected, but baring bad sectors on the parity disk, it is. 

 

The problem is, we have absolutely no way to know of those bad sectors until you try to read them back.  you cannot even assume you have no un-readable sectors because you pre-cleared the parity disk, since the un-readable sector might be perfectly fine, but the "write" to it occurred when the disk head was too high (high-fly-write) and therefor, it is marginal when  we try to read it back to use/verify parity.

So why not combine them into one operation?  A parity sync should be immediately followed by a parity check, without any user input needed.

 

Any reason not to do this?  Sure this one operation will take about twice as long as either half, but it is completely necessary, so why put it off?

I think you have a good idea.  Perhaps a checkbox to "Verify parity after initial parity calculation" might work, with the default to be checked.

 

Joe L.

I think you have a good idea.  Perhaps a checkbox to "Verify parity after initial parity calculation" might work, with the default to be checked.

 

Sure, more options are always better, but I just can't think of a reason to not want to do this besides maybe just saving time (and putting your data at risk as a result).