Last month's parity sync completed with 0 errors.  Since then, I've installed a new disk and SATA controller.  The disk is an unassigned device and is the only disk connected to the controller.  I cancelled this month's parity sync 3 minutes in because I was planning to zero and remove some smaller disks.  I zeroed two disks (4 & 5) using the dd method.  Each was zeroed one at a time.  Once completed, I stopped the array, performed a new config, preserved all assignments, set disk 4 and 5 unassigned, clicked parity is valid and started the array.  I started a manual parity sync.  When it completed, it reported 3 errors.  I'll re-run another parity sync, but any thoughts on possible cause?  This is first parity error on this server.

 

dd bs=1M if=/dev/zero of=/dev/md4p1 status=progress
dd bs=1M if=/dev/zero of=/dev/md5p1 status=progress

 

 

Was the array running in maintenance mode when you zeroed the disks?

14 minutes ago, JorgeB said:

Was the array running in maintenance mode when you zeroed the disks?

 

It wasn't in maintenance mode, but docker and VMs were stopped, and nothing was being written to the array.  Maintenance mode is a suggestion, correct?

Maintenance mode should always be used, unless you manually unmount those disks before doing it, if not a few sync errors are expected.

2 hours ago, JorgeB said:

Maintenance mode should always be used, unless you manually unmount those disks before doing it, if not a few sync errors are expected.

 

Thanks.  The docs procedure implies using maintenance mode gives better performance but, isn't necessarily crucial for maintaining parity during a disk clear.  

Quote

For best performance, make sure there are no reads/writes happening to the array. The easiest way to do this is to bring the array up in maintenance mode.