Marvell PCIe SATA Controller

[kennelm]

I need your advice on debugging a problem with my server.  I have a Marvell SATA controller to expand the server from 6 to 8 hard drives (I now know that I shouldn't be using such a cheap controller, so I plan to look into that).  

 

Here is the hardware report from lspci:

 

# lspci | grep SATA
00:17.0 SATA controller: Intel Corporation 200 Series PCH SATA controller [AHCI mode]
02:00.0 SATA controller: Marvell Technology Group Ltd. 88SE9215 PCIe 2.0 x1 4-port SATA 6 Gb/s Controller (rev 11)

 

I noticed that in the BIOS, the 6 drives connected directly to the motherboard are reported, but the 2 drives connected to the Marvell controller are not reported. Yet, upon boot, fdisk -l reports all 8 drives (plus usb flash and m.2 cache).  This seems odd, no?  Shouldn't the BIOS see all eight drives?

 

I'm pretty sure that the Marvell controller was working with 4TB drives, but recently, when I attempted to upgrade to an 8TB drive, the drive keeps going offline with a red X. 

 

Do we know if this Marvell 9215 SATA controller has trouble with 8TB drives?  If this is likely the issue I am having, then I'll be moving quickly to locate and install something better.   Is there a SATA controller that is recommended?  Or, should I be looking at a SAS HBA like the LSI 9211-8i??

 

Thanks,

Larry

Edited September 29, 2021 by kennelm
typo

[ChatNoir]

6 hours ago, kennelm said:

Shouldn't the BIOS see all eight drives?

Perhaps the extension card only works with a driver, thus the drives are not visible at BIOS.

 

6 hours ago, kennelm said:

Is there a SATA controller that is recommended?  Or, should I be looking at a SAS HBA like the LSI 9211-8i??

There are both SATA controllers and HBAs recommendations here :

 

 

[kennelm]

Thanks for the list.  I'll be ordering a new card from it.

 

Can installation of a driver at the O/S level affect the BIOS?  Seems like any awareness of hardware at the BIOS boot screen would require flashing the BIOS itself?   

 

Just for my sanity, is it possible for the Marvell 9215 to cause the flaky behavior I described?  I preclear the drive, add it to the array, parity rebuilds it, and then it pretty much immediately flips to a red X.  This seems to be the pattern now with drives larger than 4TB.