LinuxQuestions.org - [SOLVED] Drive partition labels: why/how do they change, and why are there extra in /dev?

- Linux - Newbie (https://www.linuxquestions.org/questions/linux-newbie-8/)

- - Drive partition labels: why/how do they change, and why are there extra in /dev? (https://www.linuxquestions.org/questions/linux-newbie-8/drive-partition-labels-why-how-do-they-change-and-why-are-there-extra-in-dev-4175433930/)

Drive partition labels: why/how do they change, and why are there extra in /dev?

I'm ashamed to even be asking this, but I have to no matter how goofy it is.

I have a server with 8 SDDs in it. I installed CentOS 6.2 using Kickstart. The idea was to only format the first 2 disks and leave the rest as jbod (software requirement). Initially it failed with a "KeyError: /dev/sda" message during drive partitioning, and a colleague of mine suggested I change the bootloader line from "--driveorder=sda,sdb,sdc,etc." to start at sdc. I did, and the installation succeeded.

Now that I'm logged in when I do fdisk -l I see that sda and sdb were actually created and partitioned according to my scheme, so in effect I have sda1 and 2, and sdb1 and 2. The rest of the drives are untouched as I'd wanted. Why did I have to specify a starting point of sdc to successfully install? Why doesn't the partition table start at sdc then; why did it revert to sda?

Also, when I do fdisk -l I notice that the partitions run like this: sda, sdb, sdc, sdd, sde, sdf, sdi, sdj. If I look in /proc/partitions this matches as well. /dev directory and /sys/block have sdg and sdh as well though. Why did the installation skip those two and go to "i"? Are they actually being used for something else? How can I tell?

Thanks.

To add, I went ahead and ran two scripts that I would be using for deploying these servers. The first checks for the presence of an OS on the drives to make sure I didn't inadvertently screw with the wrong ones. It returned:

The OS disks:
/dev/sda1 /dev/sdb1 /dev/sda2 /dev/sdb2
-----
The non OS disks:
/dev/sda /dev/sdb /dev/sdc /dev/sdd /dev/sde /dev/sdf /dev/sdg /dev/sdh /dev/sdi /dev/sdj

The second script runs various hdparm commands on the result of ls on /dev. For example:

Quote:

/dev/sdf:
Issuing SECURITY_SET_PASS command, password="pasta", user=user, mode=high
security_password="pasta"

/dev/sdf:
Issuing SECURITY_ERASE command, password="pasta", user=user
security_password="pasta"

/dev/sdg:
Issuing SECURITY_SET_PASS command, password="pasta", user=user, mode=high
SECURITY_SET_PASS: Invalid exchange
security_password="pasta"

and

Quote:

/dev/sdf:
setting max visible sectors to 462997170 (permanent)
max sectors = 462997170/15647024(586072368?), HPA setting seems invalid (buggy kernel device driver?)

/dev/sdg:
setting max visible sectors to 462997170 (permanent)
HDIO_DRIVE_CMD(identify) failed: Invalid exchange
HDIO_DRIVE_CMD(identify) failed: Invalid exchange

This makes me think there must be something actually there, but I don't know what and I'm concerned about continuing this with the remaining servers since I'm not sure what I'm affecting.

And to further complicate things...upon reboot /dev/md0, which was using /dev/sda and /dev/sdb, is now using /dev/sda and /dev/sde. Fdisk -l shows sda and sde as the two formatted and partitioned drives as well. Should I now be moving this from the newb forum?

I found (I think) that the order in which Linux detects drives during boot determines what labels it uses for each drive, which is why the drive I installed the OS on may be /dev/sda during one session and /dev/sde on another. I'm opening a separate thread to discuss the hdparm output.

Have a read of this for info on your disk naming issues - and how to avoid it.
Can't help with your other thread.