Sun Storage 6130 Array
 

GCS>Sun Microsystems>Sun Alert>Criteria Category>Availability
 
GCS>Sun Microsystems>Sun Alert>Release Phase>Resolved
 

Vdisk failure on a Sun StorEdge 6130 array may cause volumes on healthy vdisks located on the same controller to become inaccessible.

This issue can occur on the following system:

• Sun StorEdge 6130 Array

on the following platforms:

SPARC Platform

• solaris 8 without patch 108974-48
• Solaris 9 without patch 113277-41
• Solaris 10 without patch 119374-08

x86 Platform

• Solaris 10 without patch 119375-08

Note: Fiber Channel configuration is not supported on Solaris 8 or Solaris 9 on the x86 platform.

If the described issue occurs, messages similar to the following will be seen upon vdisk failure:

    Jun 23 10:01:39 xxx scsi: [ID 107833 kern.warning] WARNING:
    /scsi_vhci/ssd@g600a0b800017b043000016b142691bc1 (ssd17):
    Jun 23 10:01:39 xxx      Error for Command: write(10)   Error Level:
    Retryable
    Jun 23 10:01:39 xxx scsi: [ID 107833 kern.notice]   Requested Block:
    123739552                 Error Block: 123739552
    Jun 23 10:01:39 xxx scsi: [ID 107833 kern.notice]   Vendor: SUN
    Serial  Number:
    Jun 23 10:01:39 xxx scsi: [ID 107833 kern.notice]   Sense Key:
    Hardware  Error
    Jun 23 10:01:39 xxx scsi: [ID 107833 kern.notice]   ASC: 0x84
    (<vendor>  unique code 0x84  ), ASCQ: 0x0, FRU: 0x0

In the above example, "ssd17" is the RAID0 volume which is unavailable upon a disk failure. "0x84/0x0" indicates the requested command or "Mode Select" operation is not allowed with the logical unit in this state.

When this occurs, the current behavior of the target driver is that it keeps resetting the target which is on the controller.  As a result, the volumes hosted on the controller from all the other vdisks get reset.

The above messages are followed by messages similar to the following:

    Jun 23 10:01:39 xxx scsi: [ID 243001 kern.warning] WARNING:
    /ssm@0,0/pci@18,600000/SUNW,qlc@1/fp@0,0 (fcp0):
    Jun 23 10:01:39 xxx     FCP: WWN 0x200500a0b817aff0   reset
    successfully

There is no workaround. Please see the "Resolution" section below.

This issue is addressed in the following release:

SPARC Platform:

• solaris 8 with patch 108974-48 or later
• Solaris 9 with patch 113277-41 or later
• Solaris 10 with patch 119374-08 or later

x86 Platform

• Solaris 10 with patch 119375-08 or later

• State: Resolved
• Updated Contributing Factors, Relief/Workaround, and Resolution sections

• Changed State from "Resolved" to "Workaround"
• Updated Contributing Factors, Relief/Workaround, and Resolution sections

• State: Resolved
• Updated Contributing Factors, Relief/Workaround, and Resolution sections

References

This issue is similar to bugIDs 4511840 and 4842793 which were fixed for the Sun StorEdge 3500. It wasn't able to handle a condition returned by the array and the fix was to implement a condition which checks this using the unique vendor id for the SE3500. The same fix is needed for the Sun StorEdge 6130.

On the SE6130 the vendor id has changed so the code path taken by the driver to implement the fix on the SE3500 is not taken.  A new bugID was opened with the fix being to make the target driver implement the fix for both the SE3500 and the SE6130.

A new modified ssd driver is available as a temporary fix which resolves this issue. Please contact [email protected] or [email protected] for the binary fix (IDR).

References

This solution has no attachment