Wheezy FCP problem

[PROT Pierre-François]

hello,

I'm having trouble running Wheezy in z/VM 6.2. I try to access scsi lun with fcp driver but this doesn't work.
Wheezy is installed running on eckd dasd, but now I'd like to test FCP lun.
I have s390-tools version 1.16.0-2 installed. Kernel 3.2.0-4-s390x
We use NPIV. Storage and SAN seems wel l configured.

On z/VM side in the USER DIRECT C, I have
dedicate 3400 3400
dedicate 4400 4400

On debian side, when I made a "vmcp q fcp all", I see my 2 devices :
FCP 3400 ATTACHED TO DEBIAN3 3400 CHPID 30
WWPN C05076E7E3800000
FCP 4400 ATTACHED TO DEBIAN3 4400 CHPID 40
WWPN C05076E7E3800680

But a chccwdev -e 3400 or chccwdev -e 4400 does nothing.
I have the /sys/bus/ccw/drivers/zfcp/0.0.3400 directory, but not the <wwpn> subdirectory.

I referred to the IBM's doc "Linux on System z Device Drivers, Features, and Commands Development stream (Kernel 3.2)" SC33-8411-14.

Also made a "lszfcp -P -H -D -V" and the result is the following.
/sys/devices/css0/0.0.0002/0.0.3400
/sys/class/fc_host/host2
/sys/class/scsi_host/host2
Error: No fcp devices found.

I also see in the doc a lsluns command, but a't find it on my system ?
Do you have any idea on how to solve this problem ? Thank you for your help.

Best Regads
Pierre-François Prot

[Philipp Kern]

Hi,

Post by PROT Pierre-François
But a chccwdev -e 3400 or chccwdev -e 4400 does nothing.
I have the /sys/bus/ccw/drivers/zfcp/0.0.3400 directory, but not the <wwpn> subdirectory.

please try to write 1 into /sys/bus/ccw/drivers/zfcp/0.0.3400/port_rescan.
Apart from that the output of dmesg would be helpful to see why no port
has been detected.

Kind regards
Philipp Kern

[PROT Pierre-François]

Hello,

I have already made a port rescan, but the result is the same (no more <wwpn> subdirectory and nothing in the log).

Here are some results of few command :

lsscsi -H -v
[0] zfcp
dir: /sys/class/scsi_host//host0
device dir: /sys/devices/css0/0.0.0003/0.0.4400/host0
[1] zfcp
dir: /sys/class/scsi_host//host1
device dir: /sys/devices/css0/0.0.0002/0.0.3400/host1

The result in dmesg when I chccwdev -e 3400
[ 315.807811] qdio: 0.0.3400 ZFCP on SC 2 using AI:1 QEBSM:1 PCI:1 TDD:1 SIGA: W A

lszfcp -b 0.0.3400 -a
0.0.3400 host1
Bus = "ccw"
availability = "good"
card_version = "0x0006"
cmb_enable = "0"
cutype = "1731/03"
devtype = "1732/03"
failed = "0"
hardware_version = "0x00000000"
in_recovery = "0"
lic_version = "0x0000040f"
modalias = "ccw:t1731m03dt1732dm03"
online = "1"
peer_d_id = "0x000000"
peer_wwnn = "0x0000000000000000"
peer_wwpn = "0x0000000000000000"
status = "0x5400040a"
uevent = "DRIVER=zfcp"
Class = "fc_host"
active_fc4s = "0x00 0x00 0x01 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00"
dev_loss_tmo = "60"
maxframe_size = "2112 bytes"
node_name = "0x5005076400cdc8a6"
permanent_port_name = "0xc05076e7e3801101"
port_id = "0x20ba41"
port_name = "0xc05076e7e3800000"
port_state = "Online"
port_type = "NPIV VPORT"
serial_number = "IBM840000000DC8A6"
speed = "8 Gbit"
supported_classes = "Class 2, Class 3"
supported_fc4s = "0x00 0x00 0x01 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00"
supported_speeds = "2 Gbit, 4 Gbit"
symbolic_name = "IBM 2818 840000000DC8A6 PCHID: 0110 NPIV UlpId: 00300D00 DEVNO: 0.0.3400 NAME: debian3"
tgtid_bind_type = "wwpn (World Wide Port Name)"
Class = "scsi_host"
active_mode = "Initiator"
can_queue = "4096"
cmd_per_lun = "1"
host_busy = "0"
megabytes = "0 0"
proc_name = "zfcp"
prot_capabilities = "0"
prot_guard_type = "0"
queue_full = "0 156597"
requests = "0 0 0"
seconds_active = "439"
sg_prot_tablesize = "0"
sg_tablesize = "538"
state = "running"
supported_mode = "Initiator"
unchecked_isa_dma = "0"
unique_id = "13312"
utilization = "0 0 0"


The strange thing is that on the Storage, my host is well seen (symbolic name for example), but I my host the device is not seen.

Should I create an udev rule or something like that ? Did I missed some parameters when loading the zfcp driver ?


-----Message d'origine-----
De : Philipp Kern [mailto:***@debian.org]
Envoyé : lundi 18 mars 2013 21:44
À : PROT Pierre-François
Cc : debian-***@lists.debian.org
Objet : Re: Wheezy FCP problem

Hi,

Post by PROT Pierre-François
But a chccwdev -e 3400 or chccwdev -e 4400 does nothing.
I have the /sys/bus/ccw/drivers/zfcp/0.0.3400 directory, but not the <wwpn> subdirectory.

please try to write 1 into /sys/bus/ccw/drivers/zfcp/0.0.3400/port_rescan.
Apart from that the output of dmesg would be helpful to see why no port has been detected.

Kind regards
Philip

[Philipp Kern]

Hi,

Post by PROT Pierre-François
I have already made a port rescan, but the result is the same (no more <wwpn>
subdirectory and nothing in the log).
[ 315.807811] qdio: 0.0.3400 ZFCP on SC 2 using AI:1 QEBSM:1 PCI:1 TDD:1 SIGA: W A

that looks correct. In the hope that there weren't more clarifying messages in
dmesg, I'd suggest that you send your question to LINUX-***@VM.MARIST.EDU. It
might be Debian-specific in the sense of the kernel version being special,
but it might be a general problem, too.

Sadly I currently lack time and a wheezy test environment, but that's what
I get on squeeze with a working zfcp:

[ 3.106777] qdio: 0.0.2000 ZFCP on SC 0 using AI:1 QEBSM:1 PCI:1 TDD:1 SIGA: W AO
[ 3.177469] scsi 0:0:9:1: Direct-Access IBM 2145 0000 PQ: 0 ANSI: 6

I.e. the next message is already the result of the probe. To automate the
startup one writes the LUN into a special file, there should be no need to
create manual udev rules:

/etc/sysconfig/hardware # cat config-ccw-0.0.2000
ZFCP_DEVICES=(0xaaaaaaaaaaaaaaaa:0x0001000000000000)

(WWPN masked.)

This setup does not use NPIV, so I'm a bit at a loss. The WWPNs of the peers
should turn up in /sys/bus/ccw/drivers/zfcp/0.0.3400 in any case, if it's
properly configured. Maybe somebody on LINUX-390 knows more about it.

Kind regards
Philipp Kern