Introduction to SCSI over FCP for Linux on System z

Christof Schmitt <christof.schmitt@de.ibm.com>
IBM Deutschland Research & Development GmbH, Mainz
2010-10-20

Introduction to SCSI over FCP
for Linux on System z

© 2009 IBM Corporation

Trademarks
The following are trademarks of the International Business Machines Corporation in the United States, other countries, or both.
* All other products may be trademarks or registered trademarks of their respective companies.

Not all common law marks used by IBM are listed on this page. Failure of a mark to appear does not mean that IBM does not use the mark nor does it mean that the product is not
actively marketed or is not significant within its relevant market.
Those trademarks followed by ® are registered trademarks of IBM in the United States; all others are trademarks or common law marks of IBM in the United States.

For a complete list of IBM Trademarks, see www.ibm.com/legal/copytrade.shtml:

*, AS/400®, e business(logo)®, DBE, ESCO, eServer, FICON, IBM®, IBM (logo)®, iSeries®, MVS, OS/390®, pSeries®, RS/6000®, S/30, VM/ESA®, VSE/ESA, WebSphere®,
xSeries®, z/OS®, zSeries®, z/VM®, System i, System i5, System p, System p5, System x, System z, System z9®, BladeCenter®

The following are trademarks or registered trademarks of other companies.

Adobe, the Adobe logo, PostScript, and the PostScript logo are either registered trademarks or trademarks of Adobe Systems Incorporated in the United States, and/or other countries.
Cell Broadband Engine is a trademark of Sony Computer Entertainment, Inc. in the United States, other countries, or both and is used under license therefrom.
Java and all Java-based trademarks are trademarks of Sun Microsystems, Inc. in the United States, other countries, or both.
Microsoft, Windows, Windows NT, and the Windows logo are trademarks of Microsoft Corporation in the United States, other countries, or both.
Intel, Intel logo, Intel Inside, Intel Inside logo, Intel Centrino, Intel Centrino logo, Celeron, Intel Xeon, Intel SpeedStep, Itanium, and Pentium are trademarks or registered trademarks of
Intel Corporation or its subsidiaries in the United States and other countries.
UNIX is a registered trademark of The Open Group in the United States and other countries.
Linux is a registered trademark of Linus Torvalds in the United States, other countries, or both.
ITIL is a registered trademark, and a registered community trademark of the Office of Government Commerce, and is registered in the U.S. Patent and Trademark Office.
IT Infrastructure Library is a registered trademark of the Central Computer and Telecommunications Agency, which is now part of the Office of Government Commerce.

Notes:
Performance is in Internal Throughput Rate (ITR) ratio based on measurements and projections using standard IBM benchmarks in a controlled environment. The actual throughput that any user will experience will
vary depending upon considerations such as the amount of multiprogramming in the user's job stream, the I/O configuration, the storage configuration, and the workload processed. Therefore, no assurance can be
given that an individual user will achieve throughput improvements equivalent to the performance ratios stated here.
IBM hardware products are manufactured from new parts, or new and serviceable used parts. Regardless, our warranty terms apply.
All customer examples cited or described in this presentation are presented as illustrations of the manner in which some customers have used IBM products and the results they may have achieved. Actual
environmental costs and performance characteristics will vary depending on individual customer configurations and conditions.
This publication was produced in the United States. IBM may not offer the products, services or features discussed in this document in other countries, and the information may be subject to change without notice.
Consult your local IBM business contact for information on the product or services available in your area.
All statements regarding IBM's future direction and intent are subject to change or withdrawal without notice, and represent goals and objectives only.
Information about non-IBM products is obtained from the manufacturers of those products or their published announcements. IBM has not tested those products and cannot confirm the performance, compatibility, or
any other claims related to non-IBM products. Questions on the capabilities of non-IBM products should be addressed to the suppliers of those products.
Prices subject to change without notice. Contact your IBM representative or Business Partner for the most current pricing in your geography.

2 © 2010 IBM Corporation

Abstract

SCSI over FCP is an open, standard-based alternative and
supplement to existing ESCON or FICON connections. This
session will provide an introduction to the storage attachment
via the SCSI over FCP protocol. It includes setup
considerations, related features in Linux on System z and
troubleshooting basics.


IBM Presentation Template Full Version

Agenda

■ Introduction to SCSI/FCP storage attachment
– FCP channel
– Ports, LUNs
■ Setup considerations
– SAN, FCP devices, NPIV, Storage configuration
■ Related features in Linux on System z
– s390-tools
– Multipathing
– SCSI IPL
■ Troubleshooting basics
– Default kernel messages
– scsi_logging_level
– zfcp_show, zfcp_ping

Source If Applicable


Fibre Channel Storage Area Network (SAN)

■ System z connected
through FCP channel ports
FCP channel to SAN switch
ports
■ Storage connected through
ports
■ Each port has a unique 64
Bit “World Wide Port Name”
WWPN, e.g.
0x500507630318852a
■ Each disk (or tape drive, …)
Storage is represented as a 64 Bit
“remote” ports Logical Unit Number (LUN)

Disk volumes
(“LUNs”)

SAN topologies and System z
direct attached
point-to-point switched fabric
arbitrated loop

not
supported

Disk Disk Tape Disk Tape

■ A loop with two nodes looks like a point-to-point connection, but uses the unsupported loop
protocol. Check device specifications for details.

FCP Channel, FCP subchannel (“adapter”)

FCP subchannel,
“Virtual Adapter”
with devno, e.g.
0.0.181d

Physical FCP
channel ports with
Physical Channel
Id (PCHID)


IODF for FCP devices
• C HPID PATH=(C S S (0,1,2,3),51),S HAR E D, *
• NOTPAR T=((C S S (1),(TR X1),(=)),(C S S (3),(TR X2,T29C FA),(=)))*
• ,PC HID=1C 3,TYPE =FC P
• C NTLUNIT C UNUMB R =3D00, *
• PATH=((C S S (0),51),(C S S (1),51),(C S S (2),51),(C S S (3),51)), *
• UNIT=FC P
• IODE VIC E ADDR E S S =(3D00,001),C UNUM B R =(3D00),UNIT=FC P
• IODE VIC E ADDR E S S =(3D01,007),C UNUM B R =(3D00), *
• PAR TITION=((C S S (0),T29LP11,T29LP12,T29LP13,T29LP14,T29LP*
• 15),(C S S (1),T29LP26,T29LP27,T29LP29,T29LP30),(C S S (2),T29*
• LP41,T29LP42,T29LP43,T29LP44,T29LP45),(C S S (3),T29LP56,T2*
• 9LP57,T29LP58,T29LP59,T29LP60)),UNIT=FC P
• IODE VIC E ADDR E S S =(3D08,056),C UNUM B R =(3D00), *
• PAR TITION=((C S S (0),T29LP15),(C S S (1),T29LP30),(C S S (2),T29*
• LP45),(C S S (3),T29LP60)),UNIT=FC P

■ Defines only FCP adapter
■ WWPN and LUN configuration inside operating systems (e.g. Linux on System z)


N-Port Id Virtualization (NPIV)
NPIV subchannels ■ NPIV: Each virtual adapter has
have a unique its own WWPN in the SAN.
WWPN for the This is the foundation for
FCP device restricting access to storage.
■ SAN zoning: Only WWPNs in
same zone can communicate
Non-NPIV ■ “LUN masking” or “host
subchannels connections”: Each disk
share one volume is only available for a
WWPN specific list of WWPNs
■ With NPIV: Restricted
configuration possible where
SAN Zoning: disk volumes are defined for
Only WWPNs in each operating system
the same zone
can communicate ■ Without NPIV: First operating
system to access a disk
volume can use it.
“LUN masking”
or “host
connections”

NPIV

# lszfcp -a | grep port_name
permanent_port_name = "0xc05076ffe5005611"
port_name = "0xc05076ffe5005350"

■ “port_name” is the WWPN used by the FCP subchannel
■ “permanent_port_name” is the WWPN assigned to the FCP channel
■ Compare to find out if NPIV is in use:
– If both are the same, the FCP subchannel does NOT use NPIV
– If they differ, the FCP subchannel uses NPIV
■ Worldwide portname prediction tool
– http://publib.boulder.ibm.com/infocenter/zvm/v6r1/index.jsp?
topic=/com.ibm.zvm.v610.hcpa0/hcsf8c0014.htm


“Host connections” / “LUN masking” with NPIV

No
WWPN “A” WWPN “B” WWPN “C” storage
available
for
WWPN
“C”

Only defined WWPNs Host Host
can access storage: connections: connections:
Allow WWPN Allow WWPN
“A” “B”


NPIV requirements

+ +
• NPIV is available on System z9, System z10 and z/Enterprise
• z/VM
– z/VM 5.2 or newer
– z/VM 5.1 with the PTF for APAR VM63744
• Linux Distribution (LPAR mode or z/VM)
– SLES9 SP3, SP4, SLES10, SLES11, RHEL5, ...
• NPIV-Capable Switch
– only required for switch adjacent to System z
– Mostly firmware upgrades possible (e.g. McData, Brocade)


zfcp, differences to FICON and other Linux platforms

■ FICON: Channel subsystem device represents disk volume
■ FCP: Channel subsystem device represents the virtual adapter to the Fibre Channel SAN
■ Linux common code:
– Query available LUNs from storage server (“REPORT LUNS”)
– Attach all LUNs that are reported by the storage server
■ zfcp differences to other Linux platforms
– Only attach LUNs configured in Linux
– Required for FCP channel sharing in non-NPIV mode
(e.g. LUN 1 used by one system, LUN 2 by another)
■ Future (planned for Linux kernel 2.6.37): For FCP subchannels in NPIV mode, allow
automatic attachment through common code
– In NPIV mode, same behaviour as other Linux platforms


Manual LUN configuration with s390-tools
Show available FCP devices:

# lscss -t 1732/03,1732/04
Device Subchan. DevType CU Type Use PIM PAM POM CHPIDs
----------------------------------------------------------------------
0.0.181d 0.0.0010 1732/03 1731/03 80 80 ff 3b000000 00000000
0.0.191d 0.0.0011 1732/03 1731/03 80 80 ff 46000000 00000000

Enable FCP devices:

# chccwdev -e 0.0.181d,0.0.191d
Setting device 0.0.181d online
Done
Setting device 0.0.191d online
Done

Show online FCP devices and corresponding SCSI hosts:

# lszfcp
0.0.181d host0
0.0.191d host1


Manual zfcp LUN configuration
Attach ports, only required for older distributions (SLES10, RHEL5, ...):

# echo 0x500507630313c562 > /sys/bus/ccw/drivers/zfcp/0.0.181d/port_add
# echo 0x500507630300c562 > /sys/bus/ccw/drivers/zfcp/0.0.191d/port_add

Show available storage ports:

# lszfcp -P
0.0.181d/0x500507630313c562 rport-0:0-0
0.0.191d/0x500507630300c562 rport-1:0-0

Attach LUNs:

# echo 0x401040c300000000 >
/sys/bus/ccw/drivers/zfcp/0.0.181d/0x500507630313c562/unit_add
# echo 0x401040c300000000 >
/sys/bus/ccw/drivers/zfcp/0.0.191d/0x500507630300c562/unit_add

List available LUNs:

# lszfcp -D
0.0.181d/0x500507630313c562/0x401040c300000000 0:0:0:1086537744
0.0.191d/0x500507630300c562/0x401040c300000000 1:0:0:1086537744


Linux Tools: lsluns

lsscsi: Show SCSI devices and block device nodes:

# lsscsi
[0:0:0:1086537744]disk IBM 2107900 .280 /dev/sda
[1:0:0:1086537744]disk IBM 2107900 .280 /dev/sdb

lsluns: Which LUNs are available on the storage system for use?

# lsluns -c 0.0.181d -p 0x50050763031b0104
Scanning for LUNs on adapter 0.0.181d
at port 0x50050763031b0104:
0x4011400000000000
0x4011400100000000
0x4011400200000000
0x4011400300000000
0x4011400400000000
0x4011400500000000
0x4011400600000000
0x4011400700000000
0x4011400800000000
0x4011400900000000


zfcp configuration files

■ Configuring LUNs manually through sysfs is not persistent, will be lost after IPL
■ Use configuration mechanism provided by distribution for persistent LUN configuration
■ See documentation from distributions for details
■ zfcp configuration file in RHEL
– /etc/zfcp.conf
■ zfcp configuration files in SLES 10
– /etc/sysconfig/hardware/hwcfg-zfcp-bus-ccw-*
■ zfcp configuration files in SLES 11
– /etc/udev/rules.d/51-zfcp*


zfcp LUN configuration in RHEL 5.5

■ Dialog to add zfcp LUN configuration during installation (e.g. root filesystem disks)
■ Alternatively: Add zfcp LUN configuration to /etc/zfcp.conf config file (e.g. non-root filesystem
disks:
# cat /etc/zfcp.conf
0.0.3c00 0x500507630313c562 0x401040c300000000
0.0.3d00 0x500507630300c562 0x401040C300000000


YaST zfcp LUN configuration in SLES11 SP1


Multipathing

■ Use multiple paths from operating system to storage
■ Why multipathing?
– Avoid single points of failure
– Performance: I/O requests can be spread across multiple paths,
– Usually a requirement for SAN connected storage servers
• e.g. when one storage controller is in maintenance mode I/O continues to run
through second controller
■ Multipathing in Linux
– Implemented in Linux in multipath-tools package, together with the device-mapper in the
Linux kernel
– The default configuration is already included, configuration changes only for special
requirements
– SCSI device (“LUN”) in Linux represents one path to the disk volume on the storage
server
■ Multipath devices are block devices in Linux, can be used e.g.
– Directly for a filesystem
– Logical Volume Manager (LVM)
– Partitions


Multipathing overview
■ multiple paths to storage volumes
■ maintain access to storage volume
– during storage system update
FCP channels
– SAN fabric maintenance
– FCP channel maintenance
■ for FCP/SCSI managed by the
operating system (Linux)
■ WWID for storage volume, e.g.
36005076303ffc56200000000000010cc

1 2

storage
controllers © 2010 IBM Corporation
21

Linux on System z: SCSI stack and multipathing

Filesystem
Multipathing
Block Devices / Device Mapper / LVM in device mapper
Linux SCSI layer

zfcp

Linux qdio module inside Linux system

z/VM (optional)

FCP channel System z hardware

SAN


Multipathing setup

■ Config file /etc/multipath.conf
■ Ensure devices are not blacklisted and multipathd is running
■ Multipath tools include defaults for standard storage systems
■ Multipath devices are created automatically when SCSI LUNs are attached

# multipath -ll
36005076303ffc56200000000000010cc dm-0 IBM,2107900
size=5.0G features='1 queue_if_no_path' hwhandler='0' wp=rw
`-+- policy='round-robin 0' prio=1 status=active
|- 0:0:26:1087127568 sda 8:0 active ready running
`- 1:0:11:1087127568 sdb 8:16 active ready running
Resulting
Each, sda and sda multipath block
represent one path device
# mkfs.ext3 /dev/mapper/36005076303ffc56200000000000010cc
…

# mount /dev/mapper/36005076303ffc56200000000000010cc /mnt/


SLES11 multipathing setup through YaST


Multipath configuration
# cat /etc/multipath.conf
multipaths {
multipath {
wwid "36005076303ffc56200000000000010cc"
rr_min_io "10"
path_selector "service-time 0"
}
}

# multipath -l
36005076303ffc56200000000000010cc dm-0 IBM,2107900
size=5.0G features='1 queue_if_no_path' hwhandler='0' wp=rw
`-+- policy='service-time 0' prio=-1 status=active
|- 0:0:26:1087127568 sda 8:0 active undef running
`- 1:0:11:1087127568 sdb 8:16 active undef running

■ Default settings are usually good
■ Can be overwritten in /etc/multipath.conf
■ Example: Change path selector policy from default (“round robin”) to “service time”
■ See multipath.conf manpage and documentation from distributions for details


Root filesystem on SCSI multipath volume

■ Possibility to put root filesystem on SCSI multipath volume
■ Older zipl versions do not support writing on multipath device
■ Workaround: Use additional single path volume for /boot, use this volume for IPL
■ Newer Linux distributions (e.g. SLES 11 SP1) support root filesystem on multipath device
without workaround with s390-tools 1.8.3 or newer
■ http://www.ibm.com/developerworks/linux/linux390/s390-tools-1.8.3.html
–zipl: Add support for device mapper devices.
–zipl now allows installation of and booting from a boot record on logical
devices, i.e. devices managed by device mapper (or similar
packages), e.g. multipath devices.


SCSI IPL with z/VM

• setup load device
• WWPN
• LUN

• IPL from FCP device
• When using dedicated
/boot, this LUN
becomes IPL device


SCSI IPL with z/VM
00:
00: CP SET LOADDEV PORTNAME 50050763 0300C562 LUN 401040CD 00000000
00:
00: CP Q LOADDEV
PORTNAME 50050763 0300C562 LUN 401040CD 00000000 BOOTPROG 0
BR_LBA 00000000 00000000
00:
00: CP IPL 1701
00: HCPLDI2816I Acquiring the machine loader from the processor controller.
00: HCPLDI2817I Load completed from the processor controller.
00: HCPLDI2817I Now starting the machine loader.
01: HCPGSP2630I The virtual machine is placed in CP mode due to a SIGP stop
and
store status from CPU 00.
00: MLOEVL012I: Machine loader up and running (version v2.3).
00: MLOPDM003I: Machine loader finished, moving data to final storage
location.
Initializing cgroup subsys cpuset
Initializing cgroup subsys cpu
Linux version 2.6.27.42-0.1-default (geeko@buildhost) (gcc version 4.3.2 [gcc-
4_
3-branch revision 141291] (SUSE Linux) ) #1 SMP 2010-01-06 16:07:25 +0100
setup.1a06a7: Linux is running as a z/VM guest operating system in 64-bit mode
Zone PFN ranges:
DMA 0x00000000 -> 0x00080000
28 Normal 0x00080000 -> 0x00080000 © 2010 IBM Corporation

SCSI IPL LPAR from HMC


Troubleshooting

■ Check kernel messages that are possibly related to SCSI on Linux on System z:
– scsi (common SCSI code)
– sd (SCSI disk)
– rport (common SCSI code FC remote port messages)
– qdio (communication between Linux and FCP Channel)
– zfcp driver kernel messages
• See “Kernel Messages” book on
https://www.ibm.com/developerworks/linux/linux390/
– “device-mapper: multipath”
■ Other syslog messages
– Multipathd
■ zfcp driver traces available in /sys/kernel/debug/s390dbf/
■ Collect data with dbginfo.sh when reporting a problem to capture configuration, messages
and traces


Troubleshooting: scsi_logging_level
# scsi_logging_level -g • More SCSI output in kernel
Current scsi logging level:
dev.scsi.logging_level = 0 messages
• Higher levels can create lots of
SCSI_LOG_ERROR=0
SCSI_LOG_TIMEOUT=0 messages and slow down system
SCSI_LOG_SCAN=0
SCSI_LOG_MLQUEUE=0
SCSI_LOG_MLCOMPLETE=0
SCSI_LOG_LLQUEUE=0 # scsi_logging_level -s -a 1
SCSI_LOG_LLCOMPLETE=0 New scsi logging level:
SCSI_LOG_HLQUEUE=0 dev.scsi.logging_level =
SCSI_LOG_HLCOMPLETE=0 153391689
SCSI_LOG_IOCTL=0 SCSI_LOG_ERROR=1
SCSI_LOG_TIMEOUT=1
SCSI_LOG_SCAN=1
SCSI_LOG_MLQUEUE=1
SCSI_LOG_MLCOMPLETE=1
SCSI_LOG_LLQUEUE=1
SCSI_LOG_LLCOMPLETE=1
SCSI_LOG_HLQUEUE=1
SCSI_LOG_HLCOMPLETE=1
SCSI_LOG_IOCTL=1


zfcp_show / zfcp_ping
Query Fibre Channel nameserver about ports available for my system:

# zfcp_show -n
Local Port List:
0x500507630313c562 / 0x656000 [N_Port] proto = SCSI-FCP FICON
0x50050764012241e4 / 0x656100 [N_Port] proto = SCSI-FCP
0x5005076401221b97 / 0x656400 [N_Port] proto = SCSI-FCP

Query SAN topology, requires FC management server acccess:

# zfcp_show

Interconnect Element Name 0x100000051e4f7c00
Interconnect Element Domain ID 005
Interconnect Element Type Switch
Interconnect Element Ports 256
ICE Port 000 Online
Attached Port [WWPN/ID] 0x50050763030b0562 / 0x650000 [N_Port]
ICE Port 001 Online
Attached Port [WWPN/ID] 0x50050764012241e5 / 0x650100 [N_Port]
ICE Port 002 Online
Attached Port [WWPN/ID] 0x5005076303008562 / 0x650200 [N_Port]
ICE Port 003 Offline
...


zfcp_ping
Check if remote port responds (requires FC management service access):

# zfcp_ping 0x5005076303104562
Sending PNG from BUS_ID=0.0.3c00 speed=8 GBit/s
echo received from WWPN (0x5005076303104562) tok=0 time=1.905 ms

---------- ping statistics -----------
min/avg/max = 1.905/2.249/2.447 ms
--------------------------------------

zfcp_show and zfcp_ping are part of the zfcp-hbaapi 2.1 package:
http://www.ibm.com/developerworks/linux/linux390/zfcp-hbaapi-2.1.html


Summary

■ Use standard FCP/SCSI storage with Linux on System z
■ FCP subchannels defined in System z IODF; ports and LUNs managed in Linux
■ Use NPIV to enable storage management and access control
■ Use multipathing to avoid single points of failure (check storage requirements)
■ Pure SCSI setup possible with root on multipath SCSI device and SCSI IPL
■ Troubleshooting through error messages and utilities in s390-tools


Resources
■ Device Drivers, Features, and Commands
– Chapter 5. SCSI-over-Fibre Channel device driver
– http://www.ibm.com/developerworks/linux/linux390/documentation_dev.html
■ How to use FC-attached SCSI devices with Linux on System z
– http://www.ibm.com/developerworks/linux/linux390/documentation_dev.html
■ lsscsi utility for linux
– http://sg.danny.cz/scsi/lsscsi.html
■ SLES 11 SP1: Storage Administration Guide
– http://www.novell.com/documentation/sles11/stor_admin/?
page=/documentation/sles11/stor_admin/data/bookinfo.html
■ Red Hat Enterprise Linux 6: Storage Administration Guide
– http://docs.redhat.com/docs/en-
US/Red_Hat_Enterprise_Linux/6/html/Storage_Administration_Guide/index.html
■ System Storage Interoperation Center (SSIC)
– http://www.ibm.com/systems/support/storage/config/ssic/
■ System z FCP channels
– http://www.ibm.com/systems/z/hardware/connectivity/fcp.html


Thank You!

Questions?

christof.schmitt@de.ibm.com


Introduction to SCSI over FCP for Linux on System z

More Related Content

What's hot (15)

Similar to Introduction to SCSI over FCP for Linux on System z (20)

More from IBM India Smarter Computing (20)

Introduction to SCSI over FCP for Linux on System z