Mailinglist Archive: opensuse (929 mails)

< Previous Next >
Re: [opensuse] SCSI Problem
  • From: Meike Stone <meike.stone@xxxxxxxxxxxxxx>
  • Date: Fri, 14 Mar 2014 12:32:20 +0100
  • Message-id: <CAFNHiA_PcRMac36buDnAD1GbjeUXZu4m=rQ5s7LQs46UGyzvtg@mail.gmail.com>
2014-03-14 11:12 GMT+01:00 Carl Hartung <opensuse@xxxxxxxxxxxxx>:
On Fri, 14 Mar 2014 10:28:07 +0100
Meike Stone wrote:

Hello,

I've a problem with the scsi subsystem
We have tape library connected over SAN with ten tape devices.
The devices are seen with lssci and the mappings between the generic
scsi devices and the scsi addresses (H:B:T:L) with sg_map -x.
This looks like:
..
/dev/sg235 7 0 0 0 12
/dev/sg236 7 0 0 1 8
/dev/sg237 7 0 0 2 8
/dev/sg238 7 0 1 0 1 /dev/nst0
/dev/sg239 7 0 2 0 1 /dev/nst1
/dev/sg240 7 0 2 1 8
/dev/sg241 7 0 3 0 1 /dev/nst2
/dev/sg242 7 0 4 0 1 /dev/nst3
...

But sometimes I get the messages from sg_map:
"Strange, could not find device /dev/nstxx mapped to sg device??"
on top of the output and at the end the strange "scsi address" (see
sg254) /dev/sg252 9 0 1 0 1
/dev/sg253 10 0 0 0 1 /dev/nst11
/dev/sg254 -2 -2 -2 -2 -2
/dev/sg255 10 0 2 0 1

The device file /dev/sg254 exist in the /dev directory.
If I disconnect the device, the device file will be removed and if I
reconnect again, the device file is create automatically.

Does anyone has a clue, why this strange behavior occurs?

thanks Meike

Hello Meike,

Do you know the manufacturer, model & serial numbers for the subsystem?
For the tape drives? These devices usually have firmware, too, so you
want those version numbers as well. Is the subsystem single-ended SCSI
or is it differentially driven? And which SCSI flavor? (1, 2, fast,
fast-wide, and so on...?) The same information is helpful to have for
the host controller, as well, including firmware version. If it turns
out that all the hardware, including controllers, cables, connectors and
terminators, are in good repair you'll need this information to start
debugging at the driver / firmware / software level.

Hello,
there are 10 tape devices from HP (Ultrium 4-SCSI und 6-SCSI)
the are connected via FC, so no electrical wiring and terminating necessary.

The FC HBAs are:
Fibre Channel: QLogic Corp. ISP2532-based 8Gb Fibre Channel to PCI
Express HBA (rev 02)

With udevad I get following Informations for a failing device:

================================================================
~#udevadm info --attribute-walk --name /dev/sg254

Udevadm info starts with the device specified by the devpath and then
walks up the chain of parent devices. It prints for every device
found, all possible attributes in the udev rules key format.
A rule to match, can be composed by the attributes of the device
and the attributes from one single parent device.

looking at device
'/devices/pci0000:20/0000:20:03.0/0000:21:00.1/host10/rport-10:0-1/target10:0:1/10:0:1:0/scsi_generic/sg254':
KERNEL=="sg254"
SUBSYSTEM=="scsi_generic"
DRIVER==""

looking at parent device
'/devices/pci0000:20/0000:20:03.0/0000:21:00.1/host10/rport-10:0-1/target10:0:1/10:0:1:0':
KERNELS=="10:0:1:0"
SUBSYSTEMS=="scsi"
DRIVERS=="st"
ATTRS{device_blocked}=="0"
ATTRS{type}=="1"
ATTRS{scsi_level}=="7"
ATTRS{vendor}=="HP "
ATTRS{model}=="Ultrium 6-SCSI "
ATTRS{rev}=="J3KZ"
ATTRS{state}=="running"
ATTRS{timeout}=="900"
ATTRS{iocounterbits}=="32"
ATTRS{iorequest_cnt}=="0x714a5e"
ATTRS{iodone_cnt}=="0x714a5e"
ATTRS{ioerr_cnt}=="0x2c"
ATTRS{modalias}=="scsi:t-0x01"
ATTRS{evt_media_change}=="0"
ATTRS{dh_state}=="alua"
ATTRS{queue_depth}=="16"
ATTRS{queue_ramp_up_period}=="120000"
ATTRS{queue_type}=="none"

looking at parent device
'/devices/pci0000:20/0000:20:03.0/0000:21:00.1/host10/rport-10:0-1/target10:0:1':
KERNELS=="target10:0:1"
SUBSYSTEMS=="scsi"
DRIVERS==""

looking at parent device
'/devices/pci0000:20/0000:20:03.0/0000:21:00.1/host10/rport-10:0-1':
KERNELS=="rport-10:0-1"
SUBSYSTEMS==""
DRIVERS==""

looking at parent device
'/devices/pci0000:20/0000:20:03.0/0000:21:00.1/host10':
KERNELS=="host10"
SUBSYSTEMS=="scsi"
DRIVERS==""
ATTRS{fw_dump}==""
ATTRS{optrom}==""

looking at parent device '/devices/pci0000:20/0000:20:03.0/0000:21:00.1':
KERNELS=="0000:21:00.1"
SUBSYSTEMS=="pci"
DRIVERS=="qla2xxx"
ATTRS{vendor}=="0x1077"
ATTRS{device}=="0x2532"
ATTRS{subsystem_vendor}=="0x103c"
ATTRS{subsystem_device}=="0x3263"
ATTRS{class}=="0x0c0400"
ATTRS{irq}=="68"
ATTRS{local_cpus}=="00000000,000000f0"
ATTRS{local_cpulist}=="4-7"
ATTRS{modalias}=="pci:v00001077d00002532sv0000103Csd00003263bc0Csc04i00"
ATTRS{numa_node}=="1"
ATTRS{dma_mask_bits}=="64"
ATTRS{consistent_dma_mask_bits}=="64"
ATTRS{enable}=="1"
ATTRS{broken_parity_status}=="0"
ATTRS{msi_bus}==""

looking at parent device '/devices/pci0000:20/0000:20:03.0':
KERNELS=="0000:20:03.0"
SUBSYSTEMS=="pci"
DRIVERS=="pcieport"
ATTRS{vendor}=="0x8086"
ATTRS{device}=="0x3c08"
ATTRS{subsystem_vendor}=="0x103c"
ATTRS{subsystem_device}=="0x18a8"
ATTRS{class}=="0x060400"
ATTRS{irq}=="69"
ATTRS{local_cpus}=="00000000,000000f0"
ATTRS{local_cpulist}=="4-7"
ATTRS{modalias}=="pci:v00008086d00003C08sv0000103Csd000018A8bc06sc04i00"
ATTRS{numa_node}=="1"
ATTRS{dma_mask_bits}=="32"
ATTRS{consistent_dma_mask_bits}=="32"
ATTRS{enable}=="2"
ATTRS{broken_parity_status}=="0"
ATTRS{msi_bus}=="1"

looking at parent device '/devices/pci0000:20':
KERNELS=="pci0000:20"
SUBSYSTEMS==""
DRIVERS==""
================================================================

Looks like that all is ok...

The sg_map command tries to gather informations from the device via
systemcall ioctl request code SG_GET_SCSI_ID after opening the device
with open.

But the open fails with -1 :-O ...

Thanks Meike
--
To unsubscribe, e-mail: opensuse+unsubscribe@xxxxxxxxxxxx
To contact the owner, e-mail: opensuse+owner@xxxxxxxxxxxx

< Previous Next >
Follow Ups