SE3510 FC Array LUNs in Specific Configurations May Fail to Work as Quorum Devices

---

Impact

When using specific LUN/Logical Volume/Controller combinations as "quorum" devices, Sun Cluster nodes may panic which could result in the entire cluster being unavailable.

Contributing Factors

This issue can occur in the following releases:

SPARC Platform

• Sun StorEdge 3510 FC (Dual Controller) Array

This issue can occur when all of the following LUN configurations are present:

• the LUN is used with Sun Cluster 3.0 or 3.1, and
• the LUN is used as a "quorum" device, and
• the LUN is comprised of a Logical Volume, rather than a Logical Drive, and
• the LUN is presented to the host via the array's secondary controller

Symptoms

When a cluster node is shut down, the "quorum" device fails to respond properly and the secondary node will panic with the following output to the console:

    cl_runtime: WARNING: CMM: Issuing a SCSI2 Tkown ioctl failed on quorum device
    with error 5
    cl_runtime: WARNING: CMM: Preempting node 2 from quorum device /dev/did/rdsk/d4s2 failed
    with error 5

    panic[cpu0]/thread=30001f3a800: CMM: Cluster lost operational quorum; aborting

Note: In a cluster configuration, if a cluster node is shut down, the secondary node must have the "quorum" device to continue operation.

Workaround

From the SE3510 Array, ensure all LUNs are created directly from the Logical Drives rather than using Logical Volumes.

Please see the "StorEdge 3000 Family RAID Firmware User's Guide" available from http://docs.sun.com, for complete details on the above procedure.

Resolution

The above "Relief/Workaround" is the final resolution for this issue.

Modification History




Attachments

This solution has no attachment