0

I have SAS drives in RAID1. The RAID1 is done by a PERC H200 card on a Dell PowerEdge T310. This card has four sockets: 0,1,2,3. The RAID1 uses slots 0 and 1.

When inserting third disk in slot 2, the system goes on working, but the disks in slots 0 and 1 become very busy (sizzling). The disks in slot 0 and 1 are 4TB spinning at 7200 rpm, whilst the disk in slot 2 is a 600GB one at 15000 rpm.

I want to check what is happening, and make sure that RAID1 is not extended to the third disk. I also want to check that the RAID1 is correctly working for disks 0 and 1. For these reasons, I would like to dump the first track of each disk behind the RAID1 card.

The virtual disk in RAID 1 is dev/sda. The physical disks are dev/sg1, dev/sg2.

Smartmontools succeeds passing through the RAID (PERC H200) card to see the disks individually: sudo smartctl -a /dev/sg1 (respectively sg2, sg3)

But I could not dump the first sectors of each disk:

sudo dd if=/dev/sg1 of='dump_of_1st_disk.bin' bs=512 count=63

or

sudo dd if=/dev/sg1 bs=512 count=63 ¦ hexdump -C

Is the SAS controller preventing that or is something wrong with the syntax of the commands?

Is there any way to access the hexadecimal content of the drives behind the card?

OuzoPower
  • 347
  • 1
  • 4
  • 11
  • You should install the approriate monitoring/management tool for that RAID card and use that to check what is going on. Alternatively you can connect the disk to a SAS/SATA only port and make your dump then. But I doubt that you can derive any usable information from that. – Thomas Mar 25 '18 at 09:53

1 Answers1

0

You can not directly access a disk forming an hardware RAID array. The smartctl syntax does not grant access to the raw user data; rather, it only ask the controller for the disk's SMART value.

Anyway, no RAID1 can be automatically extended on a smaller disk; it simply does not work in this manner. Moreover, if I remember correctly, LSI-based card support 2-way RAID1 only (with n-way RAID1 out of question).

shodanshok
  • 47,711
  • 7
  • 111
  • 180
  • I was aware that smartctl is to ask for the disk SMART values and perform tests. I mentioned it only to demonstrate that such tool is able to access each physical disk behind the RAID controller. I'm also aware that RAID1 should not extend to a smaller disk (or such would be only a partial mirror), The two disks in RAID1 got quiter after about 30 seconds. No idea why they are so busy when I plug one more drive. It appears that the additional HDD is simply recognized as standard drive. That said, the ability to bypass the RAID card is still of interest, for erasing disk, accessing data, a.s.o. – OuzoPower Mar 24 '18 at 20:24