Sun Cluster 3.0 U1 Release Notes

Known Problems

The following known problems affect the operation of the Sun Cluster 3.0 U1 release. For the most current information, see the online Sun Cluster 3.0 U1 Release Notes Supplement at http://docs.sun.com.

Bug ID 4314698

Problem Summary: After installing Solstice Disksuite software, the scgdevs(1M) command must be run for the Solstice Disksuite device links to appear in the global namespace.

Workaround: Run the scgdevs command manually to make sure that Solstice Disksuite device nodes are created.

BugId 4358349

Problem Summary: Do not create Sun Cluster HA for NFS resources in a resource group that contains a SharedAddress resource. Sun Cluster software does not support the use of SharedAddress resources with that data service.

Workaround: Add the desired logical hostname resources into the failover resource group.

You must set up a LogicalHostname resource with this step. The hostname used with Sun Cluster HA for NFS cannot be a SharedAddress resource.

# scrgadm -a -L -g resource-group-name -l hostname,...

-a -L -g resource-group-name: Specifies the failover resource group into which to place the logical hostname resources.
-l hostname,...: Specifies the network resources (logical hostnames) to be added.

Bug ID 4359321

Problem Summary: The scinstall utility enables you to specify /global as the directory name for the global devices file system. However, because the mount point for the global devices file system is /global/.devices/node@nodeid, this specification should not be enabled.

Workaround: Re-install the node using /globaldevices as the directory name for the global devices file system.

Although not preferred, fixing the entries in the /etc/vfstab files, rebooting the cluster, and then running the scgdevs command is a possible workaround. Check that each /global/.devices/node@nodeid entry in each/etc/vfstab file has the global mount option set.

Bug ID 4362925

Problem Summary:

nodeA# scshutdown -g0 -y
scshutdown: Unmount of /dev/md/sc/dsk/d30 failed: Device busy.
scshutdown: Could not unmount all PxFS filesystems.

The Networker packages were bundled and installed during the Oracle installation. Therefore, the nsrmmd daemon is running and mounting the /global/oracle directory, which prevents the unmount of all cluster file systems.

nodeA# umount /global/oracle
umount: global/oracle busy
nodeA# fuser -c /global/oracle
/global/oracle: 335co 317co 302co 273co 272co
nodeA# ps -ef|grep 335
 root 335 273 0 17:17:41 ?       0:00 /usr/sbin/nsrmmd -n 1
 root 448 397 0 17:19:37 console 0:00 grep 335

This problem occurs during Sun Cluster shutdown when the shutdown tries to unmount a cluster file system that the process nsrmmd is still referencing.

Workaround: Run the fuser(1M) command on each node to establish a list of all processes still using the cluster file systems that cannot be unmounted. Check that no Resource Group Manager resources have been restarted since the failed scshutdown(1M) command was first run. Kill all these processes with the kill -9 command. This kill list should not include any processes under the control of the Resource Group Manager. After all such processes have terminated, rerun the scshutdown command, and the shutdown should run to successful completion.

Bug ID 4369668

Problem Summary: When the system administrator edits the Nodelist property of a managed resource group, the Resource Group Manager should run the INIT method on all resources in the resource group that have the property Init_nodes=RG_PRIMARIES, on all nodes that have been added to the node list. The Resource Group Manager should run the FINI method on such resources, on nodes that were deleted from the node list.Similarly, if the Installed_nodes property of a resource type is edited, the Resource Group Manager should run the INIT or FINI method on all resources of that type that reside in managed resource groups and have the property Init_nodes=RT_installed_nodes.

Currently, the Resource Group Manager does not run INIT or FINI methods when these updates are performed. As a result, the resources might not be properly initialized or cleaned up on these nodes.

Workaround: Using the scswitch command, unmanage and then re-manage the affected resource groups. Unfortunately, this process requires that the administrator take the resource group offline. Alternatively, the administrator can run the equivalent INIT or FINI actions manually (without unmanaging the resource group), if such procedures are documented for the resource types that occur within the group.

This workaround is unnecessary if none of the resources in the group have INIT or FINI methods. The only Sun-supplied resource types that use INIT and FINI methods are:

SUNW.SharedAddress (shared address resource type)
SUNW.apache (Apache Web Server) configured in scalable mode
SUNW.iws (iPlanet/Netscape web server) configured in scalable mode

Resource types that customers or third parties implement might also use INIT or FINI methods. If so, this workaround is necessary for resource groups that contain such resource types.

Note -

All scalable services implicitly use INIT and FINI methods, even if such methods are not explicitly declared for the resource type.

Bug ID 4370760

Problem Summary: You cannot remove the last node from a Solstice DiskSuite diskset unless you first take the device group offline.

Workaround: To remove the last node from a diskset, take the device group offline first. To remove the last node, run the following two commands as superuser from the nodeto be removed.

# /usr/cluster/bin/scswitch -m -D disksetname
# metaset -s disksetname -d -h nodename

Bug ID 4373911

Problem Summary: If you do the following:

Configure multiple Sun Cluster HA for NFS resources on the cluster.
Switch an HA-NFS resource on or off and leave another online on the same node.

The HA-NFS fault monitor might display the following warning message.

clnt_tp_create_timed of program statd failed:RPC:Program not registered

Workaround: No workaround is necessary. The warning message can safely be ignored.

BugId 4376171

Problem Summary: Placing a FC-AL SBus Card (FC100/S) and a Sun Quad FastEthernet^TM 2.0 (SQFE/S) on the same SBus might cause unexpected resets on the QFE card.

Workaround: Avoid configuring cluster nodes with a FC-AL SBus Card (FC100/S) and a Sun Quad FastEthernet 2.0 (SQFE/S) on the same SBus.

Bug ID 4378553

Problem Summary: A resource group's Nodelist property is an ordered list of nodes that can master the resource group, with the most-preferred node listed first. The Resource Group Manager should always host a resource group on the most-preferred node that is available. However, when an administrator reboots the cluster (when all nodes are rebooting at once), managed resource groups might end up being mastered on nodes other than the most-preferred node. This problem occurs only upon reboot of the entire cluster.

Workaround: After rebooting the cluster, use the scswitch command to switch resource groups onto the desired nodes. The Nodelist preference order will be enforced automatically from that point onward, as long as the cluster remains up.

BugId 4396185

Problem Summary: The usage message for pnmstat returns unsupported options.

Workaround: Use the pnmstat man page to determine supported options.

BugId 4449906

Problem Summary: Sun Cluster HA for Sybase ASE encounters several problems, including core dump, when the Wait_for_online extension property is set to TRUE.

Workaround: Always set the Wait_for_online extension property to its default value, FALSE.

BugId 4459021

Problem Summary: Oracle installation requires the following Solaris packages: SUNWarc, SUNWbtool, SUNWhea, SUNWlibm, SUNWlibms, SUNWsprot, and SUNWtoo. These packages must be present before running the SunPlex Manager Oracle data service installation.

Workaround: Install these packages manually, as described in the Oracle documentation.

BugId 4460312

Problem Summary: The SunPlex Manager installation of Oracle does not configure tnsnames.ora on the cluster nodes. abase.

Workaround: If the cluster nodes are to be used as clients of the Oracle database, the appropriate tnsnames.ora files must be manually updated.

BugId 4461863

Problem Summary: The SunPlex Manager will not function properly if the hostname of one or more cluster nodes is multi-homed, has more than one IP address assigned to it. However, multi-homed logical addresses may be safely used for data services.

Workaround: Comment out the BindAddress line in /opt/SUNWscvw/conf/httpd.conf, and then start the server with /etc/init.d/initspm start.

BugId 4462957

Problem Summary: The following message may be encountered during initial boot of a cluster node.

 
cl_runtime: WARNING: Major number for driver (did) does
not match the one on other nodes.  Confirm that the /etc/name_to_major
files are in sync on all cluster nodes

Workaround: Contact your authorized Sun service provider to determine whether a workaround or patch is available.

BugId 4464476

Problem Summary: A 2.2 to 3.0 Sun Cluster upgrade for Netra machines requires the use of the -O option of scinstall (1M) during the initial phase of the upgrade. Otherwise, the Netra hardware platforms might not be recognized causing the upgrade to fail.

Workaround: Use the -O option of scinstall (1M) during the initial phase of the upgrade. This option overrides hardware checking.

BugId 4464870

Problem Summary: Calling scds_free_netaddr_list when there are more than two ports specified in the portlist causes the program to dump core.

Workaround: If you specified more than two ports in the portlist, do not call scds_free_netaddr_list to free memory allocated by scds_get_netaddr_list. This memory will be automatically reallocated when the process terminates.