Amanda-Users

Re: AMANDA reports no slots, but there ARE slots!

2007-03-08 00:36:22
Subject: Re: AMANDA reports no slots, but there ARE slots!
From: FL <lengyel AT gmail DOT com>
To: sgw AT amanda DOT org
Date: Thu, 8 Mar 2007 00:00:11 -0500


On 3/7/07, Stefan G. Weichinger <sgw AT amanda DOT org> wrote:
FL schrieb:
>
>
> On 3/6/07, *Stefan G. Weichinger* < sgw AT amanda DOT org
> <mailto:sgw AT amanda DOT org>> wrote:
>
Glad that thinks worked out. Might be a nice thing to write as small
udev-related subsection for the HOWTOs ...

Greets, Stefan.

Maybe I can add to it -- when I have a break.  I've just
gone through the same exercise on a CentOS 4 system (I intend to upgrade it to RHEL 4); on that system it was necessary to run modprobe sg to see the devices, and the udev syntax (which I believe is the same as Fedora Core). Perhaps the need to run modprobe is a bug. 
 
After running
 
udevinfo -a -p /sys/class/scsi_generic/sg2
 
 

looking at class device '/sys/class/scsi_generic/sg2':
SYSFS{dev}="21:2"

follow the class device's "device"
looking at the device chain at '/sys/devices/pci0000:00/0000:00: 0e.0/host0/target0:0:2/0:0:2:0':
BUS="scsi"
ID="0:0:2:0"
SYSFS{detach_state}="0"
SYSFS{device_blocked}="0"
SYSFS{model}="215 "
SYSFS{queue_depth}="2"
SYSFS{rev}="2201"
SYSFS{scsi_level}="3"
SYSFS{state}="running"
SYSFS{timeout}="0"
SYSFS{type}="8"
SYSFS{vendor}="SPECTRA "

looking at the device chain at '/sys/devices/pci0000:00/0000:00: 0e.0/host0/target0:0:2':
BUS=""
ID="target0:0:2"
SYSFS{detach_state}="0"

looking at the device chain at '/sys/devices/pci0000:00/0000:00:0e.0/host0':
BUS=""
ID="host0"
SYSFS{detach_state}="0"

looking at the device chain at '/sys/devices/pci0000:00/0000:00:0e.0':
BUS="pci"
ID="0000:00:0e.0"
SYSFS{class}="0x010000"
SYSFS{detach_state}="0"
SYSFS{device}="0x8478"
SYSFS{irq}="169"
SYSFS{subsystem_device}="0x7884"
SYSFS{subsystem_vendor}="0x9004"
SYSFS{vendor}="0x9004"

looking at the device chain at '/sys/devices/pci0000:00':
BUS=""
ID="pci0000:00"
SYSFS{detach_state}="0"

[root@amanda rules.d]#

This suggests that the new rule for matching the correct scsi_generic (sg) device and creating the correct node in /dev
(togther with a symbolic link, with the right ownership and permissions should be based on the "device" entry above. Here is the rule:

KERNEL="sg2", BUS="scsi", ID="0:0:2:0", SYSFS{vendor}="SPECTRA ", OWNER="root", GROUP="disk", MODE="0660",
SYMLINK="changer"

The program udevtest shows the result of adding this rule:

[root@amanda rules.d]# udevtest /sys/class/scsi_generic/sg2
version 039
looking at '/class/scsi_generic/sg2'
configured rule in '/etc/udev/rules.d/50-udev.rules' at line 83 applied, added symlink 'changer'
creating device node '/dev/sg2', major = '21', minor = '2', mode = '020660', uid = '0', gid = '6'
[root@amanda rules.d]#


Though it is still necessary to run modprobe sg