I have 1 HP DL360 machine.

- It has 2 disks (both green LED indication)and has RHEL 4 installed. Machine is runnig fine.

If I do : df -k, I get this :

Filesystem 1K-blocks Used Available Use% Mounted on
/dev/cciss/c0d0p2 16126420 2279576 13027532 15% /
/dev/cciss/c0d0p1 101086 11358 84509 12% /boot
none 1037384 0 1037384 0% /dev/shm
/dev/cciss/c0d0p3 8151384 1019412 6717900 14% /u01
/dev/cciss/c0d0p5 3929312 40608 3689100 2% /u02
/dev/cciss/c0d0p7 4640964 2749940 1655272 63% /var
[root@twodisk]#


For me this means that RHEL uses both disks as one disk only (c0d0) and installed the partitions on it.
But is there somwehere a command that tells that there are actually 2 disks in the system via CLI ?

cat /proc/mdstat gives me following output.


[root@twodisk]# cat /proc/mdstat
Personalities :
unused devices: <none>

I know that HP has an inbuild HW RAID controller, that has to be configured in the BIOS.

How can I see that HW RAID is enabled from CLI without rebooting (and watch in the BIOS) the box ?
if cat /proc/mdstat gives no output, is that sufficient to conclude that HW RAID is not active or is this command only used in Software RAID ?

Thanks in advance,

Mark.

Try:

fdisk -l

That won't do it. He is asking how many "real" disks are making up the "logical" disks (LUNs) that fdisk would show as a single disk.

That is to say if you have a Hardware RAID controller that makes 4 "real" disks into a single RAID 5 "logical" disk (LUN) then in Linux fdisk would see that LUN as a single disk. It wouldn't know the underlying disk information.

It would be more complicated if you had the RAID controller doing say 1 RAID 1 (mirror) of 2 disks and another RAID 5 of 4 disks (we have this on some of our Dells with PERC controllers). That would be 6 real disks but only 2 logical disks - fdisk -l would show only 2 disks.

Off the top of my head I don't know any utility that would do this if it weren't written for the specific hardware RAID. I'm not familiar with the cciss stuff but it may be that relates to a specific driver and if so there may be utilities associated with that driver that would directly query the RAID controller to give you the information you need.

You might do it by inference:
1) Run "fdisk -l" to see how many disks Linux thinks it has and the "size" of those disks.
2) Run lspci to see what controllers/drivers are in the system - if one is for a RAID controller it might help you.
3) Guestimate RAID and number of "real disks" by knowing the size of the "real disks". e.g. if you knew you'd only installed 73 GB "real disks" and you saw a "logical" disk from fdisk -l of approximately 219 GB you could assume that you had 4 real 73 GB disks in a RAID 5 config (RAID 5 = n + 1 so your RAID 5 is "4", 4 -1 = n = 3, 3 * 73 = 219 GB)
Not an exact science to be sure.

Its even worse if its not just a controller in the host but rather an external RAID array attached via controllers. (e.g. Clariion or Symmetrix arrays attached via SAN to Qlogic fibre adapters - in that case the adapter is just passing "logical" information from the array rather than creating the array.

Most newer dl360s (g3 and above for sure) have iLO (integrated lights out monitoring) and support for the hp insight agent stuff. The insight agent will give you web access to the entire configuration of the box (memory, HDs, bios status, etc.)
Check out HP's PSP for your linux build. They typically have the insight stuff in it.

Thanks for you reactions

Mark.