I've got two servers, and 8-disk and a 17-disk.  I understand that the array only functions as fast as the slowest drive.  Over the last few months I've upgraded disks across both servers, and now am running 5900+ rpm NAS disks and 7200 rpm NAS and standard disks.

 

Both servers contain an LSI 92xx series controller -- an 8-port and a 16-port.  Both LSIs are in x8 slots on the motherboards.

 

The 16-drive system has two parity disks on motherboard SATA3 ports, member disks on the LSI.

 

The 8-drive system has one parity disk and all members on the LSI.

 

For parity syncs and rebuilds, the 8-disk system averages 150MB/s.  As I write this, its rebuilding the array using a new (to it) 8TB Seagate Archive, which is arguably to worst disk in the history of disks, and its currently doing 178MB/s.

 

Conversely, the 17-disk system is currently rebuilding a new HGST 12TB 7200rpm NAS disk, and its doing 49MB/s.  The last time a parity sync was done (finished overnight) it reported an average of 109MB/s.

 

Wondering what is causing the reduced throughput on the larger array.  I'm slowly doing new-configs on the larger system to remove the last remaining 8TB archives in favor of 12TB HGSTs, and will eventually have ever disk in the system on the LSI, with only a couple of NVMe disks for cache and dockers. 

 

Could the MB SATA3 ports be the bottleneck?

 

Thanks.

22 hours ago, tucansam said:

Could the MB SATA3 ports be the bottleneck?

Very unlikely, unless they are sharing the DMI bandwidth with other things, start by posting the diags during a rebuild/check.