|
Re: A non-amanda issue: the abysmal misery of LUNs
2007-03-26 20:19:42
On 3/26/07, Gene Heskett <gene.heskett AT verizon DOT net> wrote:
On Monday 26 March 2007, FL wrote:
>Executive summary:
>mt status works if no tape is loaded!
> ...
>
>Then I created a new initial ram disk with
>
>
>mkinitrd /boot/initrd-2.6.9-42.0.10.ELsmpLUN.img 2.6.9-42.0.10.ELsmp
>
>and modified /boot/grub/menu.lst accordingly.
>
>But nothing sort of worked until I located the SCSI terminator.
This is very important. Missing or improper terms has caused the wasted
sacrifice of lots of virgins over the last 30 or so years we've had a
scsi spec. Few people understand that this is in fact a transmission
line, and as such it _must_ be terminated at both _ends_ of the physical
cable.
Yes, I was actually well aware of this (it was kind of a joke -- I even have a ham license).
It is crucially important to terminate the SCSI bus, but sometimes a person's SCSI
terminator falls off, and one proceeds anyway. I probably should have omitted
these remarks, but now that I've broadcast it to the entire world, it will be held against me.
Using the next connector on the cable, and leaving another foot
or so curled up and unused at the end will cause data trashing echo's and
cost you any religion you may have thought you had.
In that case, I owe the world a religion.
I'm wondering if RHEL has continued the practice of only scanning the scsi
bus for LUN=0, in which case many libraries and changers will be missed.
This requires a rebuild of the kernel, with the 'scan all luns' set to 'y'
in the scsi menu.
AHA (behold the last line):
[root@amanda 2.6.9-42.0.10.EL-smp-i686]# grep SCSI .config
CONFIG_CISS_SCSI_TAPE=y
CONFIG_BLK_DEV_IDESCSI=m
# SCSI device support
CONFIG_SCSI=m
CONFIG_SCSI_PROC_FS=y
# SCSI support type (disk, tape, CD-ROM)
CONFIG_SCSI_DUMP=m
# Some SCSI devices (e.g. CD jukebox) support multiple LUNs
# CONFIG_SCSI_MULTI_LUN is not set
Oh well. I figured it would come to this.
I would however, scan the /var/log/dmesg file to see if /dev/nst2 is the
correct address for the tape drive, which may not be the same as the
changer robot you are running with the mtx command. You're use
of /dev/sg4 to address the robot, but /dev/nst2 for the drive looks like
something I'd want to check.
Here is an additional piece of information (what I call a factlet: a crucial fact that
you need to know that no one will tell you unless you ask):
I have another SCSI card to which my spectra logic 2K is attached.
Drives 0 and 1 of the 2K are nst0 and nst1, respectively. The Exabyte drive
is nst2.
[root@amanda 2.6.9-42.0.10.EL-smp-i686]# cat /proc/scsi/scsi
Attached devices:
Host: scsi0 Channel: 00 Id: 00 Lun: 00
Vendor: SONY Model: SDX-300C Rev: 04c7
Type: Sequential-Access ANSI SCSI revision: 02
Host: scsi0 Channel: 00 Id: 01 Lun: 00
Vendor: SONY Model: SDX-300C Rev: 04c7
Type: Sequential-Access ANSI SCSI revision: 02
Host: scsi0 Channel: 00 Id: 02 Lun: 00
Vendor: SPECTRA Model: 215 Rev: 2201
Type: Medium Changer ANSI SCSI revision: 02
Host: scsi1 Channel: 00 Id: 06 Lun: 00
Vendor: HP Model: Ultrium 2-SCSI Rev: S33U
Type: Sequential-Access ANSI SCSI revision: 03
Host: scsi1 Channel: 00 Id: 06 Lun: 01
Vendor: EXABYTE Model: MAGNUM 224 Rev: C118
Type: Medium Changer ANSI SCSI revision: 04
[root@amanda 2.6.9-42.0.10.EL-smp-i686]#
And,
[root@amanda 2.6.9-42.0.10.EL-smp-i686]# dmesg | grep -A 1 -B 4 EXABYTE
(scsi1:A:6):
160.000MB/s transfers (80.000MHz DT, 16bit)
Vendor: HP Model: Ultrium 2-SCSI Rev: S33U
Type: Sequential-Access ANSI SCSI revision: 03
Vendor: EXABYTE Model: MAGNUM 224 Rev: C118
Type: Medium Changer ANSI SCSI revision: 04
[root@amanda 2.6.9-42.0.10.EL-smp-i686]#
--
Cheers, Gene
"There are four boxes to be used in defense of liberty:
soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
Everything that you know is wrong, but you can be straightened out.
Many thanks!
|
|
|
