Sun Storage 3310 Array
 
Sun Storage 3120 SCSI Array
 
Sun Storage 3320 SCSI Array
 

GCS>Sun Microsystems>Storage - Disk>Modular Disk - 3xxx Arrays
 

Feb 8 07:53:57 myhost unix: WARNING: ID[SUNWpd.glm.cmd_timeout.6017]
Feb 8 07:53:57 myhost unix: WARNING: /pci@1f,4000/scsi@3 (glm0):
Feb 8 07:53:57 myhost unix: Target 15 reducing sync. transfer rate

The reduced transfer rate affects only one target, target 15 in the message above. As it takes more time to transfer data to and from the target with the reduced speed, the target with the reduced speed ties up the bus and affects the performance of the entire bus. The affected target will stay at the reduced speed until the system is rebooted. There is no way to reset the speed of the affected target. To check the current bus speeds, you can use the prtconf -v command:

#prtconf -v
...
esp, unit #0
Driver software properties:
name <target1-sync-speed> length <4>
value <0x00002710>.
...

The "value" string contains the kilobytes per second transfer rate of the initiator/target pair, notated in hexadecimal. In this example, the value 0x00002710 is 10000 kb/second in decimal for the initiator/target pair esp, unit#0/target1. The SCSI sync speed is negotiated between initiators and targets. Each initiator/target pair can have its own sync speed.

#prtconf -v
scsi, instance #0
Driver properties:
name='target2-sync-speed' type=int items=1 dev=none
value=00013880
name='target1-sync-speed' type=int items=1 dev=none
value=00013880
name='target0-sync-speed' type=int items=1 dev=none
value=000001f4

SCSI sync speed for initiator/target pairs from above example

(scsi, instance #0/target0-sync-speed):0x1f4 = 500 Kb = 0.5 Mb
(scsi, instance #0/target1-sync-speed):0x13880 = 80000 Kb = 80 Mb
(scsi, instance #0/target0-sync-speed):0x13880 = 80000 Kb = 80 Mb

Some common root causes for a reduction in bus speed are bad, loose, or incorrect termination, bad or loose cables, or total cable length on the bus that exceeds the supported length. A firmware bug in some device on the bus is also a possible cause. Those are good places to start troubleshooting. Occasionally, a temporary one-time disruption in communications will reduce transfer rates, but will clear up permanently after a reboot.

Other known issues are incorrect settings of scsi_options and driver bugs. For JBODs its also possible that the HBA and the disks are supporting higher sync speeds than the JBOD chassis itself. In this cases the sync speed has to limit the maximum speed supported by the JBOD chassis using scsi_options. More information on those topics is available in the documents listed below.

Ultra[TM] 160/Ultra[TM] 320:Disk performance degradation
Technical Instruction <Document: 1012106.1> Ultra[TM] 160/Ultra[TM] 320:Disk performance degradation.

scsi bus sync rate set incorrectly
Technical Instruction < Solution: 210664 > scsi bus sync rate set incorrectly.

What do those scsi_options in /etc/system mean?
Techncical Instruction <Document: 1007761.1> What do those scsi_options in /etc/system mean?

There is another way available to check the sync speeds using prtpicl:
prtpicl -v | egrep "NAME=|sync-speed" | grep -v spindle
| NAME=ide-controller |
| NAME=mpt0 |
:target0-sync-speed 320000
:target8-sync-speed 320000
:target9-sync-speed 320000
:targeta-sync-speed 320000
:targetb-sync-speed 320000
:targetf-sync-speed 5000
| NAME=mpt1 |
| NAME=glm0 |
:target0-sync-speed 160000
:target1-sync-speed 160000

When a Sun StorEdge 3310 SCSI JBOD array equipped with Ultra320-capable disk drives is connected to the on-board SCSI port of a Sun Fire V440 system, the disks will attempt to negotiate at Ultra320. However, because of design and other limitations, this is not supported. The on-board SCSI port of the Sun Fire V440 system must be limited to Ultra160. This is done by creating a /kernel/drv/mpt.conf file with the following lines and then rebooting the system.

*See 3310 SCSI Array Release Notes:

http://www.sun.com/products-n-solutions/hardware/docs/html/816-7292-28/relnotes.html#pgfId-1260247

Sun StorEdge 3000 Family Installation, Operation, and Service Manual

Part No. 816-7290-18 May 2006, Revision A page D7

does not provide the hex value required for the SE3310 JBOD. To add some confusion the value is called "hexadecimal address".

Sun StorEdge" 3310 SCSI Array Release Notes

Part No. 816-7292-28 May 2006, Revision A

provides the proper value for scsi_options on page 17.

SE3310 JBOD : scsi_options=0x1ff8*

SE3310 RAID controller : no sync speed settings required, HBA and RAID controller will auto-negotiate the speed.

For details on setting the scsi_options for SE3310 JBOD have a look at:

SCSI bus reset & transport errors on Sun StorEdge[TM] 3310 SCSI array with Ultra320 SCSI HBA

Document ID: 80089 Jun 28, 2007 Symptoms and Resolutions

scsi_options defined in Solaris 10 1/06 s10s_u1wos_19a SPARC

#define SCSI_OPTIONS_LINK 0x10 /* Global linked commands */

#define SCSI_OPTIONS_TAG 0x80 /* Global tagged command support */

#define SCSI_OPTIONS_DR 0x8 /* Global disconnect/reconnect */

#define SCSI_OPTIONS_SYNC 0x20 /* Global synchronous xfer capability */

#define SCSI_OPTIONS_PARITY 0x40 /* Global parity support */

#define SCSI_OPTIONS_FAST 0x100 /* Global FAST scsi support */

#define SCSI_OPTIONS_WIDE 0x200 /* Global WIDE scsi support */

#define SCSI_OPTIONS_FAST20 0x400 /* Global FAST20 scsi support */

#define SCSI_OPTIONS_FAST40 0x800 /* Global FAST40 scsi support */

#define SCSI_OPTIONS_FAST80 0x1000 /* Global FAST80 scsi support */

#define SCSI_OPTIONS_FAST160 0x2000 /* Global FAST160 scsi support */

#define SCSI_OPTIONS_FAST320 0x4000 /* Global FAST320 scsi support */

#define SCSI_OPTIONS_NLUNS_MASK (0x70000)

#define SCSI_OPTIONS_NLUNS_DEFAULT 0x00000

#define SCSI_OPTIONS_NLUNS_1 0x10000

#define SCSI_OPTIONS_NLUNS_8 0x20000

#define SCSI_OPTIONS_NLUNS_16 0x30000

#define SCSI_OPTIONS_NLUNS_32 0x40000

Related Documents:

Problem Resolution <Document: 1012309.1> SCSI bus reset & transport errors on Sun StorEdge[TM] 3310 SCSI array with Ultra320 SCSI HBA

This solution has no attachment