SLES11 / udev rules / tape media changer

[xunil321]

Dear all,
at our SLES11 server there is an iscsi tape roboter from OVERLAND attached.
From this great forum we learnt to use udev rules in order to fix the device name
for the media changer /dev/sgN where N is altering after each reboot.
Here is the udevinfo output of our tape changer:

#udevinfo -a -p $(udevinfo -q path -n /dev/sg5)

looking at device
'/devices/platform/host4/session1/target4:0:0/4:0:0:2/scsi_generic/sg5':
KERNEL=="sg5"
SUBSYSTEM=="scsi_generic"
DRIVER==""

looking at parent device
'/devices/platform/host4/session1/target4:0:0/4:0:0:2':
KERNELS=="4:0:0:2"
SUBSYSTEMS=="scsi"
DRIVERS==""
ATTRS{device_blocked}=="0"
ATTRS{type}=="8"
ATTRS{scsi_level}=="6"
ATTRS{vendor}=="OVERLAND"
ATTRS{model}=="NEO Series "
ATTRS{rev}=="0510"
ATTRS{tgps}=="1"
ATTRS{state}=="running"
ATTRS{timeout}=="0"
ATTRS{iocounterbits}=="32"
ATTRS{iorequest_cnt}=="0xe"
ATTRS{iodone_cnt}=="0xe"
ATTRS{ioerr_cnt}=="0x3"
ATTRS{modalias}=="scsi:t-0x08"
ATTRS{evt_media_change}=="0"
ATTRS{queue_depth}=="32"
ATTRS{queue_type}=="none"

Therefore we built a new rule called 100-local.rules in /etc/udev/rules.d with
this line:

KERNEL=="sg*", SUBSYSTEMS=="scsi", ATTRS{rev}=="0510", SYMLINK+="loader0"

After running /etc/init.d/boot.udev restart we would expect to see
/dev/loader0. But it does'nt exist, why?

THX for any hint advice,
Rainer

[argon99]

Dear all,
at our SLES11 server there is an iscsi tape roboter from OVERLAND attached.
From this great forum we learnt to use udev rules in order to fix the device name
for the media changer /dev/sgN where N is altering after each reboot.
Here is the udevinfo output of our tape changer:

#udevinfo -a -p $(udevinfo -q path -n /dev/sg5)

looking at device
'/devices/platform/host4/session1/target4:0:0/4:0:0:2/scsi_generic/sg5':
KERNEL=="sg5"
SUBSYSTEM=="scsi_generic"
DRIVER==""

looking at parent device
'/devices/platform/host4/session1/target4:0:0/4:0:0:2':
KERNELS=="4:0:0:2"
SUBSYSTEMS=="scsi"
DRIVERS==""
ATTRS{device_blocked}=="0"
ATTRS{type}=="8"
ATTRS{scsi_level}=="6"
ATTRS{vendor}=="OVERLAND"
ATTRS{model}=="NEO Series "
ATTRS{rev}=="0510"
ATTRS{tgps}=="1"
ATTRS{state}=="running"
ATTRS{timeout}=="0"
ATTRS{iocounterbits}=="32"
ATTRS{iorequest_cnt}=="0xe"
ATTRS{iodone_cnt}=="0xe"
ATTRS{ioerr_cnt}=="0x3"
ATTRS{modalias}=="scsi:t-0x08"
ATTRS{evt_media_change}=="0"
ATTRS{queue_depth}=="32"
ATTRS{queue_type}=="none"

Therefore we built a new rule called 100-local.rules in /etc/udev/rules.d with
this line:

KERNEL=="sg*", SUBSYSTEMS=="scsi", ATTRS{rev}=="0510", SYMLINK+="loader0"

After running /etc/init.d/boot.udev restart we would expect to see
/dev/loader0. But it does'nt exist, why?

THX for any hint advice,
Rainer

The reason is udev absolutely sucks!!! I am here looking for the same problem but with a nic card. I have three interfaces in my box, eth0 eth1 and eth2. All the rules in udev/rules.d are correct. But udev just decided to change the name of the interfaces. This totally screws up my bridging script. The rules are set by mac address that doesn't change, I checked, and should be just fine. BUT NO udev is crap. Gives admins all sorts of head aches. I think udev wasn't tested properly before it was included in the kernel. Obviously.