Sun Storage 3310 Array
 

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

• got external SCSI bus reset
• Target x reducing sync. transfer rate
• Target x reverting to async. mode

These messages could further be followed by read/write errors such as:

Jan 16 04:04:21 larempdb scsi: [ID 107833 kern.warning] WARNING: /pci@1f,4000/scsi@4/sd@0,5 (sd76):
Jan 16 04:04:21 larempdb 	Error for Command: write                   Error Level: Retryable
Jan 16 04:04:21 larempdb scsi: [ID 107833 kern.notice] 	Requested Block: 153264                    Error Block: 153264

Jan 16 04:04:21 larempdb scsi: [ID 107833 kern.notice] 	Vendor: SUN                               Serial Number: 1ACACDB1-05
Jan 16 04:04:21 larempdb scsi: [ID 107833 kern.notice] 	Sense Key: Unit Attention
Jan 16 04:04:21 larempdb scsi: [ID 107833 kern.notice] 	ASC: 0x29 (power on, reset, or bus reset occurred), ASCQ: 0x0, FRU: 0x0

The above messages could be a result of:

Speed negotiation between the Ultra320 SCSI HBA and the Sun StorEdge 3310 SCSI array.

You can check the speed to each scsi target with:

 # 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
| NAME=mpt1 |
| NAME=glm0 |
:target0-sync-speed	 160000
:target1-sync-speed	 160000
| NAME=glm1 |
:target0-sync-speed	 160000
:target8-sync-speed	 160000
:target9-sync-speed	 160000
:targeta-sync-speed	 160000
:targetb-sync-speed	 160000

In this example the onboard controller of a Sun Fire[TM] V240 Server (glm1) is accessing the disks in one Sun StorEdge 3310 JBOD at 160MB/s, while the Ultra320 HBA (mpt0) is using a speed of 320MB/s to the other Sun StorEdge 3310 JBOD.

In such cases,the system may recover on its own, however performance will be degraded until then. It can also happen that one or more disks fails, becomes unavailable in format and causes messages to be logged, such as:

 disk not responding to selection

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

The following course of action is recommended.
This problem can be fixed by throttling the speed of the Ultra320 HBA in the mpt driver configuration file to the maximum speed supported by the Sun StorEdge 3310. It is also strongly recommended to upgrade the Sun StorEdge 3310 controller and/or SAF-TE firmware to the latest version.

To limit the HBA speed to 160MB/s, create the file /kernel/drv/mpt.conf with the following contents, then reboot the system.

• For an SE3310 RAID array: (there must be exactly 5 spaces between SUN and StorEdge)

      device-type-scsi-options-list =
          "SUN     StorEdge 3310", "SE3310-scsi-options";
      SE3310-scsi-options = 0x41ff8;

• For an SE3310 JBOD: (this example is for a dual-ported HBA with device paths /pci@1e,600000/scsi@3 and /pci@1e,600000/scsi@3,1)

      name="mpt" parent="/pci@1e,600000"
               unit-address="3"
               scsi-options=0x1ff8;
      name="mpt" parent="/pci@1e,600000"
               unit-address="3,1"
               scsi-options=0x1ff8; 

If the above does not help and Fujitsu (Allegro) drives are used, please refer to <Document: 1006888.1> - Fujitsu ( Allegro) drives in Sun StorEdge[TM] 3310 SCSI array connected with Ultra320 SCSI HBA can cause scsi bus resets and transport errors.

Notes:

- Minimum requirement is Solaris[TM] 8 7/03, Solaris 9 12/03 or Solaris 9 8/03 x86 Operating System
- Install Solaris 8 mpt driver patch 115275   and raidctl utility (Sun Fire[TM] V440 only) patch 115274
- Install Solaris 9 mpt driver patch 115665   and raidctl utility (Sun Fire V440 only) patch 115667
- Upgrade array controller firmware (RAID only) and SAF-TE firmware (RAID and JBOD): patch 113722-11   or later
- The mpt driver is included in Solaris 8 2/04 and Solaris 9 4/04.

This solution has no attachment