[3607] in linux-scsi channel archive

home help back first fref pref prev next nref lref last post

SCSI addressing - flexible?

daemon@ATHENA.MIT.EDU (=?iso-8859-1?Q?Mads_Christian_Rehh)
Mon Mar 30 08:40:29 1998

From: =?iso-8859-1?Q?Mads_Christian_Rehhoff_N=F8r?= <MCN@fleggaard.dk>
To: "'linux-scsi@vger.rutgers.edu'" <linux-scsi@vger.rutgers.edu>
Date: 	Mon, 30 Mar 1998 15:36:25 +0200

This message is in MIME format. Since your mail reader does not understand
this format, some or all of this message may not be legible.

------ =_NextPart_000_01BD5BF1.9AB23C60
Content-Type: text/plain;
	charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable

Hello - I have a problem with Linux system on a HP Netserver after
adding an external disk cabinet.
Please read the attached document, and reply if you have some idears...

ANY suggestions will be greatly appreciated, thank you.


Regards=20

Mads Chr. R. N=F8r
Fleggaard DanData
E-mail: mcn@fleggaard.dk




=20

------ =_NextPart_000_01BD5BF1.9AB23C60
Content-Type: text/plain;
	name="SCSI addressing.txt"
Content-Transfer-Encoding: quoted-printable
Content-Disposition: attachment;
	filename="SCSI addressing.txt"

How to address a specific SCSI device...

I would like to request possible solutions for the following =
situations.

The original setup is as follows:
I'm using a HP NetServer running Linux. The built-in drive bays =
contains 6 SCSI disks. There are two internal SCSI controllers (2 =
pieces Adaptec-78xx), whereof controller A is used solely for a cdrom =
drive and a DAT drive, and the second controller is not used.

Furthermore there is a DPT Raid IV controller, which controls the 6 =
disks in a RAID-1 configuration:
/dev/sda	Disk1 & Disk2 Mirrored (root-disk) (9 GB Total)
/dev/sdb	Disk3 & Disk4 Mirrored             (9 GB Total)
/dev/sdc	Disk5 & Disk6 Mirrored             (9 GB Total)


The order of initialization is at startup (bios control):
Controller A, controller B and finally the DPT. Since none of the =
built-in controllers have any disks on their bus, the disk with the =
lowest SCSI id on the DPT become the boot device.

New setup:
Now I want to extend the system with further disks, and have mounted an =
external cabinet: HP Netserver Rackstorage, capable of 8 hotplug SCSI =
disks.=20
To achieve greater bandwidth, I connect the external cabinet the =
internal SCSI controller B.

At boot time, as mentioned, the internal controller is first scanned, =
and one of the disks on bus B will become the boot device, but worse =
the previous "first device" or "sda" of the system moved to become =
"sdc" so now there is no reference to the Linux root partition.

The problem with the boot device can be worked around by disabling the =
SCSI bios of the built-in controllers, but when the Linux boots, it =
does it own scanning for SCSI devices and enumerates the disks =
according the bus initialisation order: bus A, bus B and the DPT bus.


Every time there is a change in the number of disks in the external =
cabinet, e.g. by adding or removing a disk, or a disk may become =
faulty; the enumeration of the disks on the DPT bus changes, which =
again changes logical order, hence the root drive of my Linux =
installation changes, which leaves me with a un-bootable system and bad =
references to the data partitions.


As far as I can see it would be of great advantage if it would be =
possible to specify the SCSI bus and then the logical devices, e.g. =
instead of sda, sdb etc. it would be preferable with s0da, s0db, or =
s1da, s1db etc.,e.g. where s0da is SCSI bus 0 disk 1 (lowest SCSI-id), =
and s1db is SCSI bus 1 disk 2 (second-lowest SCSI-id).

I believe this could lead to a more flexible and foolproof system, less =
vulnerable to hardware changes. How else can you handle a scaleable =
system expected to grow and grow with approximately 2 x 9 GB disks =
every 3 mounth. Just think of the risk to forget to redirect a device =
number in /etc/fstab etc. This could prove the system less scaleable =
than it had to. Here it becomes clear that Linux has it's roots within =
users having only one or very few disks on their system. In fact it is =
the way Hewlett Packard does it on their HP 9000 Serie 800 Machines =
running HP-UX. It has been designed to handle lots of disk capacity. On =
HP-UX there also exists a command "ioscan" that shows existing and new =
hardware connected to the mashine even on the fly. No special knowledge =
is needed, just write "ioscan -f -n" and you will get all you need. =
That would be a nice command to have on linux as well. I know that we =
can use the /proc and that there is a way to cat informations about =
add/removing devices/disks to the system, but I think the usage can be =
made a lot easier. At our place we prefere the KISS method (Keep It =
Simple, Stupid).


Below I have included a printout of the detected SCSI devices, showing =
the assigned logic device numbers, before and after adding devices =
harddisks to the system. The second printout is modified to show what =
would happen if the mentioned changes to the addressing method where =
implemented.



Kind regards

Mads Chr. R. Nor
Fleggaard DanData
Denmark
e-mail: mcn@fleggaard.dk


(The lines showing the detected device are marked).



Today:

Before EXPANSION:

Registered HBAs:                                                        =
       =20
HBA no. Boardtype    Revis  EATA Bus  BaseIO IRQ DMA Ch ID Pr QS  S/G =
IS       =20
scsi2 : PM3334UW     v07L.0 2.0c PCI  0xecf0   9 BMST 2  7  N  64 252 Y =
       =20
scsi0 : Adaptec AHA274x/284x/294x (EISA/VLB/PCI-Fast SCSI) 4.1.1/3.2.1  =
       =20
scsi1 : Adaptec AHA274x/284x/294x (EISA/VLB/PCI-Fast SCSI) 4.1.1/3.2.1  =
       =20
scsi2 : EATA (Extended Attachment) HBA driver                           =
       =20
scsi : 3 hosts.                                                         =
       =20
scsi0: Scanning channel A for devices.                                  =
       =20
  Vendor: TOSHIBA   Model: CD-ROM XM-5701TA  Rev: 0317                  =
       =20
  Type:   CD-ROM                             ANSI SCSI revision: 02     =
       =20
Detected scsi CD-ROM sr0 at scsi0, channel 0, id 5, lun 0               =
        -------
scsi1: Scanning channel A for devices.                                  =
       =20
  Vendor: DPT       Model: RAID-5            Rev: 07L0                  =
       =20
  Type:   Direct-Access                      ANSI SCSI revision: 02     =
       =20
Detected scsi disk sda at scsi2, channel 0, id 1, lun 0                 =
        <<<<<<<
scsi2: queue depth for target 1 on channel 0 set to 64                  =
       =20
scsi : detected 1 SCSI cdrom 1 SCSI disk total.                         =
       =20
SCSI device sda: hdwr sector=3D 512 bytes. Sectors=3D 88864000 [43390 =
MB] [43.4 GB]=20

After EXPANSION:

Registered HBAs:                                                        =
       =20
HBA no. Boardtype    Revis  EATA Bus  BaseIO IRQ DMA Ch ID Pr QS  S/G =
IS       =20
scsi2 : PM3334UW     v07L.0 2.0c PCI  0xecf0   9 BMST 2  7  N  64 252 Y =
       =20
scsi0 : Adaptec AHA274x/284x/294x (EISA/VLB/PCI-Fast SCSI) 4.1.1/3.2.1  =
       =20
scsi1 : Adaptec AHA274x/284x/294x (EISA/VLB/PCI-Fast SCSI) 4.1.1/3.2.1  =
       =20
scsi2 : EATA (Extended Attachment) HBA driver                           =
       =20
scsi : 3 hosts.                                                         =
       =20
scsi0: Scanning channel A for devices.                                  =
       =20
  Vendor: QUANTUM   Model: XP34550J          Rev: LXY4                  =
       =20
  Type:   Direct-Access                      ANSI SCSI revision: 02     =
       =20
Detected scsi disk sda at scsi0, channel 0, id 1, lun 0                 =
        >>>>>>> (new disk)
  Vendor: TOSHIBA   Model: CD-ROM XM-5701TA  Rev: 0317                  =
       =20
  Type:   CD-ROM                             ANSI SCSI revision: 02     =
       =20
Detected scsi CD-ROM sr0 at scsi0, channel 0, id 5, lun 0               =
        -------
scsi1: Scanning channel A for devices.                                  =
       =20
  Vendor: DPT       Model: RAID-5            Rev: 07L0                  =
       =20
  Type:   Direct-Access                      ANSI SCSI revision: 02     =
       =20
Detected scsi disk sdb at scsi2, channel 0, id 1, lun 0                 =
        <<<<<<<
scsi2: queue depth for target 1 on channel 0 set to 64                  =
       =20
scsi : detected 1 SCSI cdrom 1 SCSI disk total.                         =
       =20
SCSI device sda: hdwr sector=3D 512 bytes. Sectors=3D 88864000 [43390 =
MB] [43.4 GB]=20

Future:

Before EXPANSION:

Registered HBAs:                                                        =
       =20
HBA no. Boardtype    Revis  EATA Bus  BaseIO IRQ DMA Ch ID Pr QS  S/G =
IS       =20
scsi2 : PM3334UW     v07L.0 2.0c PCI  0xecf0   9 BMST 2  7  N  64 252 Y =
       =20
scsi0 : Adaptec AHA274x/284x/294x (EISA/VLB/PCI-Fast SCSI) 4.1.1/3.2.1  =
       =20
scsi1 : Adaptec AHA274x/284x/294x (EISA/VLB/PCI-Fast SCSI) 4.1.1/3.2.1  =
       =20
scsi2 : EATA (Extended Attachment) HBA driver                           =
       =20
scsi : 3 hosts.                                                         =
       =20
scsi0: Scanning channel A for devices.                                  =
       =20
  Vendor: TOSHIBA   Model: CD-ROM XM-5701TA  Rev: 0317                  =
       =20
  Type:   CD-ROM                             ANSI SCSI revision: 02     =
       =20
Detected scsi CD-ROM s0r0 at scsi0, channel 0, id 5, lun 0              =
        -------
scsi1: Scanning channel A for devices.                                  =
       =20
  Vendor: DPT       Model: RAID-5            Rev: 07L0                  =
       =20
  Type:   Direct-Access                      ANSI SCSI revision: 02     =
       =20
Detected scsi disk s2da at scsi2, channel 0, id 1, lun 0                =
        <<<<<<<=20
scsi2: queue depth for target 1 on channel 0 set to 64                  =
       =20
scsi : detected 1 SCSI cdrom 1 SCSI disk total.                         =
       =20
SCSI device sda: hdwr sector=3D 512 bytes. Sectors=3D 88864000 [43390 =
MB] [43.4 GB]=20

After EXPANSION:

Registered HBAs:                                                        =
       =20
HBA no. Boardtype    Revis  EATA Bus  BaseIO IRQ DMA Ch ID Pr QS  S/G =
IS       =20
scsi2 : PM3334UW     v07L.0 2.0c PCI  0xecf0   9 BMST 2  7  N  64 252 Y =
       =20
scsi0 : Adaptec AHA274x/284x/294x (EISA/VLB/PCI-Fast SCSI) 4.1.1/3.2.1  =
       =20
scsi1 : Adaptec AHA274x/284x/294x (EISA/VLB/PCI-Fast SCSI) 4.1.1/3.2.1  =
       =20
scsi2 : EATA (Extended Attachment) HBA driver                           =
       =20
scsi : 3 hosts.                                                         =
       =20
scsi0: Scanning channel A for devices.                                  =
       =20
  Vendor: QUANTUM   Model: XP34550J          Rev: LXY4                  =
       =20
  Type:   Direct-Access                      ANSI SCSI revision: 02     =
       =20
Detected scsi disk s0da at scsi0, channel 0, id 1, lun 0                =
        >>>>>>> (new disk)
  Vendor: TOSHIBA   Model: CD-ROM XM-5701TA  Rev: 0317                  =
       =20
  Type:   CD-ROM                             ANSI SCSI revision: 02     =
       =20
Detected scsi CD-ROM s0r0 at scsi0, channel 0, id 5, lun 0              =
        -------=20
scsi1: Scanning channel A for devices.                                  =
       =20
  Vendor: DPT       Model: RAID-5            Rev: 07L0                  =
       =20
  Type:   Direct-Access                      ANSI SCSI revision: 02     =
       =20
Detected scsi disk s2da at scsi2, channel 0, id 1, lun 0                =
        <<<<<<<=20
scsi2: queue depth for target 1 on channel 0 set to 64                  =
       =20
scsi : detected 1 SCSI cdrom 1 SCSI disk total.                         =
       =20
SCSI device sda: hdwr sector=3D 512 bytes. Sectors=3D 88864000 [43390 =
MB] [43.4 GB]=20


------ =_NextPart_000_01BD5BF1.9AB23C60--

-
To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
the body of a message to majordomo@vger.rutgers.edu

home help back first fref pref prev next nref lref last post