HP ProLiant - GNU/Linux - Can't find RAID drives

Question

I have a ProLiant DL360 G5. I would like to install Arch Linux onto this machine, but I'm experiencing the following difficulties:

I have booted Arch Linux installation CD, but I can't seem to access the hard drives (visibly, there are 6). There is a /dev/cciss/c0d0, but running fdisk c0d0; yields "fdisk: cannot open c0d0: No such device or address". There is /dev/sda{,1,2}, but they correspond to the installation media. There are no /dev/hd*. dmesg | grep -i cis yields "HP CISS Driver (v 3.6.26)" and "cciss 0000:06:00.0: cciss0: <0x3230> at PCI 0000:06:00.0 IRQ 24 using DAC". lspci yields "RAID bus controller: Hewlett-Packard Company Smart Array Controller (rev 04)".

I have no experience with ProLiant (or even other "server room"-type machines). Also, this is a used machine, and I don't have access to the original owners/configurators. One of the hard drives has an orange light showing on it in the front...should I take it out?

Based on my reading, the smart array drivers have been included in the linux kernel for a while now, but perhaps I need some drivers anyway?

Let me know if I need to provide any more information, and if you have any tips on getting archlinux installed onto this machine.

Solution

I probably wouldn't be using ArchLinux, since as a distribution, it's not really in the realm of support of any of the hardware vendors I'm aware of. There are reasons Red Hat/CentOS, Debian and SuSE are vendor compatibility targets.

But the real issue here is the age of the server, coupled with the (presumed) newness of your ArchLinux distribution and kernel.

HP Smart Array RAID controllers used the Linux cciss kernel driver for many years (~2001-2010). The corresponding block device was /dev/cciss/cXdY, where X is the controller, enumerated from 0 on, and Y the logical drive number.

This scheme changed with the introduction of the hpsa driver, which moved back to standard Linux SCSI /dev/sdX naming.

The newest Linux kernels have limited support for the older generation of HP controllers. See the following question:

CentOS 7 x64 and HP Proliant DL360 G5 SCSI Controller compatibility

---

In this situation, we need to know the health of your array. An orange or amber light indicates a failed disk. You can configure the RAID controller and logical drives at a basic level using the HP Option ROM Configuration for Arrays (ORCA). Press F8 when prompted during the server's POST process. Here's the related documentation.

In addition, you can use the offline HP Smart Storage Administrator or the HP Service Pack for ProLiant to run more initial configuration steps for the system and array. You'll also be able to update the system's firmware and a few other critical items if you use the latter.

Either way, I'd remove the bad disk or work with the people who sold you the system to arrange for replacement. If the drive is failed, it won't be usable in the creation of a logical drive.

Comments

[lmat - Reinstate Monica]

Exactly. I got a different keyboard so that I could 'F8' at the smart array POST time (I guess), and built a new raid. I replaced the drive and now all 6 drives are good to go. I created two logical volumes, etc.... I'll create a new server fault question as this question has been decidedly answered!

[lmat - Reinstate Monica]

The followup question is here: serverfault.com/questions/684042