Sun Storage 3510 FC Array
 
Sun Storage 3310 Array
 
Sun Storage 3511 SATA Array
 

GCS>Sun Microsystems>Sun FAB>Standard>Controlled Proactive
 

Sun StorEdge 3310/3510/3511 SCSI Timeout Values may be set too low resulting in intermittent timeouts due to arbitration, retries, or bad block reallocation. In an I/O intensive application environment, intermittent drive failures may also occur due to lack of SCSI arbitration fairness if the timeout value is set too low.

For all SE3000 Arrays:

The default setting for "SCSI I/O Timeout is 30 seconds for Controller Firmware revisions 4.11I and later. Earlier firmware versions have this value set according to array type. The recommendation is to verify that this value is set to 30 seconds for all SE3000 Arrays.

For SE3310 Arrays:

The default setting was 15 seconds for revisions 3.25S and later. Earlier firmware versions have this value set to 7 seconds. The recommendation is to verify that this value is set to 30 seconds for all 3310 arrays regardless of the firmware revision installed.

For 3510 Arrays:

The default setting for "SCSI I/O Timeout" is 30 seconds for Controller Firmware revisions 3.27P and later. Earlier firmware versions had this value set to 15 seconds. The recommendation is to verify that this value is set to 30 seconds for all 3510 arrays regardless of the firmware revision installed.

For 3511 Arrays:

The default setting for "SCSI I/O Timeout" is 30 seconds for Controller Firmware revisions 3.27R and later. The recommendation is to verify that this value is set to 30 seconds for all 3511 arrays regardless of the firmware revision installed.

To determine the SCSI I/O Timeout value for an array:

In a telnet session:

select "View and edit Configuration parameters" in the main menu

select: "Drive-side SCSI parameters", and then enter return

select: Select "SCSI I/O Timeout"

and the current value will get displayed. Alternately, with the sccli (comand line interface), you may run the "show config" command, and the SCSI I/O Timeout value will be displayed.

The "SCSI I/O Timeout" is the time interval for the controller to wait for a drive to respond. If the controller attempts to read data from or write data to a drive but the drive does not respond within the SCSI I/O timeout value, the drive will be considered a failed drive.

The specified timeout values are not long enough in some cases for a response on the SCSI bus. The controller makes 3 attempts to select the drive on the SCSI bus, and will timeout the command if a response is not returned.

Setting the timeout to a lower value will cause the controller to judge a drive as failed while a drive is still retrying or while a drive is unable to arbitrate the SCSI bus.

This can also be a issue with bad block management due to the amount of time the controller requires to replace a bad block. The controller issues a number of commands to complete the replacement of a bad block. The SCSI command timeout value must be large enough to allow all of the commands to be completed along with the initial I/O command received by the controller.

Here is a sample of error messages which may be seen if this issue is encountered:

   [1117] LG:0 CHL:2 ID:0 Block Reassignment Succeeded
Thu Dec 4 09:20:38 2003
[1112] LG:0 CHL:2 ID:0 Drive HW Error
Sun Dec 21 11:09:27 2003
[1105] LG:0 CHL:2 ID:0 Time Out Waiting for I/O to Complete
Sun Dec 21 11:09:34 2003
[110F] LG:0 CHL:2 ID:0 SCSI Bus Reset Issued
Sun Dec 21 11:09:34 2003
[1117] LG:0 CHL:2 ID:0 Block Reassignment Succeeded
Sun Dec 21 11:09:36 2003
[1112] LG:0 CHL:2 ID:0 Drive HW Error
Sun Jan 4 03:55:26 2004

Follow the procedure to verify/modify the SCSI Timeout Value. Changing this value should be done during a scheduled maintenance window, as a reset is required.

In the telnet session:

select: "View and edit Configuration parameters" in the main menu

select: "Drive-side SCSI parameters", and then enter return

select: Select "SCSI I/O Timeout"

For SE3310 with 4.13B or later firmware:

select "Default (30 seconds), and then enter return

For earlier firmware versions select 30 seconds, and then enter return.

Please install patchId 113722-12 (or later) after performing above procedure.

For 3510 with 3.27P or later firmware:

select "Default (30 seconds), and then enter return

OR

For 3510 with firmware prior to 3.27P:

select 30 seconds, and then enter return.

For this value to take affect, this will require a reset to the array.

Please install patchId 113723-15 (or later), after performing the above procedure.

For 3511 with 3.27R or later firmware:

Please install patchId 113724-08 (or later), after performing the above procedure.

Select YES, when asked the question: "Do you want to reset the controller now?"

OR

Explicitly reset the controller for this setting to take effect.

• Updated FIN to add 3511 product and appropriate patches.

None.

• Manual: 817-3711-xx Sun StorEdge 3000 Family RAID Firmware 4.1x User's Guide

This solution has no attachment