Sun Storage 6920 System
 

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

When both nodes of a two-node cluster are directly connected to a Sun StorEdge (SSE) 6920, the SSE 6920 may not handle the release of SCSI-2 (FC-AL) disk reservations properly. The SSE 6920 may become inaccessible, and one of the two cluster nodes may return a "SCSI reservation panic" on bootup.

This issue can occur in the following release:

• Sun StorEdge 6920 with System Processor (SP) version 2.0.5 and earlier (with Sun Cluster 3.x for Solaris 8 and 9)

Note: This issue can only occur when both nodes of a two-node cluster are directly connected to the SSE 6920.

To verify the Storage Service Processor Image revision:

1. Log in into the Storage Automated Diagnostic Environment web console. (This can be done with the guest login/password).
2. Click on the Storage Automated Diagnostic Environment link that is under the "Storage" category.
3. Click on the "Monitor" tab, then click on the "Device" tab (under the Monitor tab).
4. In the "Monitored Devices" list that comes up, find in the "Type" category an entry titled "Summary" that has for its Device entry 6920-spx (where x equals the sp number for this site; typically 0).
5. Click on that 6920-spx link.
6. In the window that comes up, insure that the "Page:" window has "Summary Report" selected, or if not, select "Summary Report" and click the "Go" button next to it.
7. In the upper left hand window of the window that comes up there will be a "Name:" field with the 6920 spx information in it. Immediately below that is a "Revision:" field that will contain the current System Service Processor version.

The system processor version can be determined from "/etc/release" with the following command (run on the SP):

    [sp0]# cat /etc/release
    Solaris 9 s9_58shwpl3 SPARC
    ...
    Sun StorEdge(tm) 6920
    Version: 2.0.2

Cluster node host will experience a SCSI reservation panic on bootup.

To work around this issue, use fibre channel switches between the cluster nodes and the SE6920 and have all the SAN connections operate in fabric mode (which can be confirmed with a simple status check at the fibre channel switches). To perform this workaround, do the following:

1. Make sure all SCSI reservations on the storage have been removed.
2. Shut down the cluster using "scshutdown -g0 -y" [WARNING: Avoid shutting down the cluster one node at a time. This will cause SCSI reservations to be placed on the disks again]
3. Boot both cluster nodes in non-cluster mode using "boot -x".
4. Run "format" on both nodes to confirm that no SCSI reservations are on the LUNs.
5. Obtain at least 2 fibre channel switches for the connections from the cluster nodes to the SE6920.
6. Establish the new connections between the cluster nodes and the SE6920 such that all connections use fabric mode, and not arbitrated loop mode, and so that there are no direct connections from the cluster nodes to the SE6920.
7. Re-map the LUNs in the SE6920 and the fibre channel switches, as necessary, to this new hardware configuration.
8. Use cfgadm and devfsadm, as necessary, and confirm the storage is accessible from both cluster nodes. [Note: It should be possible to make these changes without altering the paths to the storage. This should also leave all cluster global device entries unchanged as well.]
9. IMPORTANT... Reboot the DSP in the SE6920.
10. Reboot both cluster nodes into cluster mode and test that each has access to the storage.
11. Test that the workaround has been completed properly by shutting down the cluster one node at a time, until the cluster is completely down, and then rebooting both cluster nodes into cluster mode again [Note: The SCSI reservations should be released and the nodes should be able to form a cluster again at the end of this step.]

Complete details of this process can be found in "Sun Cluster 3.x With Sun StorEdge 6920 System Manual for Solaris OS" at http://docs.sun.com/app/docs?q=817-6747-11.

This issue is addressed in the following release:

• Sun StorEdge 6920 with System Processor (SP) version 3.0.0.25 and later

Information for upgrade procedures for the 6920 can be found in the following document: http://www.sun.com/products-n-solutions/hardware/docs/pdf/817-5229-14.pdf

29-Nov-2005:

• Updated Contributing Factors and Resolution sections, re-release as Resolved

The version 2.0.6 was dropped as there was a lack of customer's awaiting this release. Therefore bug 2123693, will not be fixed, since there are no more 2.x releases on the books. The fix was implemented in the 3.0.0.16 build, and is in either of the current releases of the 6920(3.0.0.25 or 3.0.0.30).

The bug for this fix is actually 6230109.

This solution has no attachment