Sun StorEdge 3510 FC Array in Loop Mode May Cause Attached Hosts to Experience Excessive SCSI Timeouts Upon a Reboot

---

Impact

A specific loop configuration of a Sun StorEdge 3510 FC Array may cause attached hosts to experience excessive SCSI timeouts when a single host is rebooted. As a result, the hosts may experience performance degradation.

Contributing Factors

This issue can occur in the following release:

SPARC Platform

• Sun StorEdge 3510 FC Array without patch 113723-10

Note: This issue only occurs when the Sun StorEdge 3510 FC Array is configured in a loop mode in a SAN configuration.

Symptoms

Should the described issue occur, messages similar to the following are logged to the "/var/adm/messages" file of the other hosts when one of the hosts is rebooted:

    Apr 27 nsvr-88  /scsi_vhci/ssd@g600c0ff00000000000272a68ad13490a 
    (ssd75): Command Timeout on path /pci@1e,600000/SUNW,qlc@4,1/fp@0,0(fp3)

Workaround

To recover from the described issue, power cycle the array or fail, then unfail the controller on the array.

Note: "fail" and "unfail" are sccli(1M) commands which "disable" and then "enable" the controller.

To workaround the described issue, change from "loop mode" to "fabric point-to-point mode" on the Sun StorEdge 3510 FC Array by doing the following:

1. Telnet to the array

2. Select the following from the menu:

View and Edit Configuration Parameters ----> Host-side SCSI Parameters ----> Fiber Connection Option - Loop only

3. Change Fiber Connection Option - "Loop only" to "Point to Point only"

4. Reset and re-map the channels and luns on the array

Resolution

This issue is addressed in the following release:

SPARC Platform

• Sun StorEdge 3510 FC Array with patch 113723-10 or later

Modification History


Date: 25-MAY-2005

Change History

25-May-2005

• State: Resolved
• Updated Contributing Factors and Resolution sections

Attachments

This solution has no attachment