PowerHA fix information

Contents of this Readme

• Service Pack 1 important content
• APAR IV56377: allow syslog to manage cluster.log instead of clcycle
• Critical Volume Group handling
• Disabling automatic restart of ERS
• Other related APARs
• Fix information

This service pack includes important updates for the following features:

SAP Smart Assist
Hyperswap Support

You can view and install individual fixes ased on your specific configuration.

The PowerHA utility clcycle manages most log files generated by the product. clcycle will automatically manage cluster log files such that individual logs are limited to a maximum size and are removed after they reach a certain age or are overwritten by newer version. In general, the cluster logs are written to directly by the cluster components, and the maintenance performed y clcycle is coordinated with those components. One exception is the cluster.log file which is written to using the AIX syslog facility.

In order to manage the cluster.log, clcycle needs to refresh the syslogd subsystem, which can cause syslogd to lose track of other refresh functions that were scheduled by other system components.

With APAR IV56377, clcycle will no longer manage cluster.log by default, and the syslog facility will e configured to manage cluster.log instead. Installing this update will automatically update the /etc/syslog.conf configuration file to configure syslog management of cluster.log. If you want clcycle to continue to manage cluster.log, you can override this behavior as described in the following section.

With this update, the default behavior of clcycle can also be customized as follows:

In general, PowerHA will default to these rules for all log files:

1. maximum size - log files that are over 1MB will be cycled
2. maximum number of outdated logs - no more than 7 prior versions of the file will be preserved
3. maximum age - log files older than one day will be cycled

If you wish to customize these values, you can override them by specifying different values in the /etc/environment file on each cluster node. Add the following entries to override the default(s):

CLCYCLE_MAX_SIZE = <size in bytes>
add this entry to limit the maximum size of any saved log file
CLCYCLE_MAX_LOGS = <number of old files to save>
add this entry to change the number of old log files preserved by clycle
CLCYCLE_MAX_DAYS = <cycle log files older than this number of days>
add this entry to change the age at which log files will be cycled
CLCYCLE_CLUSTER_LOG = <FALSE|TRUE>
if you want clcycle to manage the cluster.log file instead of syslog, you can set this value to TRUE

PowerHA SystemMirror includes special volume group monitoring which can be customized with a specifice response when the group fails. You can now define a volume group as a "critical volume group" and select a response when a failure occurs:

• Notify only.
• Halt the node.
• Fence the node (node remains up) but cannot access the group.
• Shut down cluster services and bring all resource groups offline.

This facility is useful for monitoring and protecting Oracle RAC voting disks.

1. smitty cl_vg
2. Select "Create a Volume Group"
3. From the pick list, select all nodes in the cluster
4. From the pick list of candidate disks, select the set of disks for the volume group, for example, the Oracle RAC voting disks
5. From the pick list of volume group types, select "Scalable"
6. Enter a Resource Group Name
7. Set "Enable Fast Disk Takeover or Concurrent Access" to "Concurrent Access"
8. Set "CRITICAL volume group?" to "yes"
9. Make any other changes necessary, then create the volume group
10. Synchronize the PowerHA SystemMirror cluster

In order for the SAP smart assist to correctly manage failovers, you must disable automatic restart of ERS.

To disable automatic restart of ERS, you will need to edit the profile for the ERS instance located in /usr/sap/<SID>/ERS<nn>/profile/

1. comment out or remove the line containing "Autostart = 1"
2. modify the line containing:
   Restart_Program_00 = local $(_ER) pf=$(_PFL) NR=$(SCSID)
   to
   start_Program_00 = local $(_ER) pf=$(_PFL) NR=$(SCSID)

It is important to keep your entire installation up to date with the most recent and important fixes for SysteMirror, AIX, CAA, RSCT, and any other components on which you rely. The following list is not all inclusive and may not be applicable to your installation, but you should consider applying these fixes as needed.