IV36415: NAVIGATOR STATUS INDICATOR NOT MATCH LOWER LEVELS
Fixes are available
Closed as program error.
Over time, status indicators on higher level navigator nodes become inconsistent with lower level node indicators. A ClassCast exception appears in the ras1 log some time before the inconsistency is observed. RECREATE INSTRUCTIONS: The exact cause of this problem has still not been identified. Before resolution can be completed, an additional diagnostic build must be run in the customer environment to gather more details on the the exception believed to underly the reported problem
None identified to date
Over time, status indicators on higher level navigator nodes become inconsistent with lower level node indicators. A classCast exception may appear in the RAS1 log some time before the inconsistency is observed. Navigator node state augmentation anomalies may appear when navigator nodes are affected by high severity acknowledged events combined with lower severity events. The problems appear when event state changes occur after the portal client is started but before affected nodes are expanded. A classCast exception may appear in the client RAS1 log some time before the inconsistency is observed although this may not be directly related to the problem.
lthough the external symptoms are the same, internally there are two different problems. In the first case, the lower severity event becomes true, while in the second case it becomes false (is removed), after the portal client is started. For the first case, the logic was changed to ensure accumulated events and acknowledgements are always processed in chronological order. The classCast exception was eliminated by changing the event handling logic to ensure all events are processed appropriately. For the second case, the logic was changed to ensure state removal events are saved for propagation to lower level nodes during node expansion. However, because of performance implications, this second part of the fix must be explicitly enabled via the "tep.save.removed.events" run-time parameter. This parameter controls how removed events are saved for propagation to expanding nodes. It may be set to one of the following case-insensitive values (if not specified, it is assumed to be NONE): NONE - disables the option. Removal events are not saved and the second part of the problem is not corrected. This setting has no adverse performance impact. ALL - causes all removal events to be saved for the duration of the portal client execution. Each portal client instance maintains its own record of removal events. This option should prevent all state anomalies but, in an active environment, these events may eventually consume an excessive amount of memory. Any resulting performance problems must be corrected by restarting the affected portal client instance. NEEDED - saves removal events until all impacted nodes have been fully expanded. This option should prevent all state anomalies and will release memory as nodes are expanded so there should be a lower chance of performance problems. However, if navigator nodes are never fully expanded, saved removal events will accumulate in the same way as for the ALL option and could consume excessive memory. Any resulting performance problems must be corrected by restarting the affected portal client instance. The NONE option is recommended for customers who have not encountered navigator state inconsistencies. The NEEDED option is recommended for customers who still see anomalies when using the NONE option. The ALL option is a last resort choice and is not recommended unless anomalies are seen with the NEEDED option and the customer is able to tolerate any memory consumption issues that may arise. The Manage Tivoli Monitoring Services application can be used to configure the setting. According to your environment, set the desired value as follows: For desktop clients: Select Advanced | Edit Variables... from the context menu, double-click the "tep.save.removed.events" symbol, enter "NEEDED" in the input field and select the "In use" checkbox. For browser clients: Perform the same steps as for the desktop clients or select Advanced | Edit ENV File... and insert the following line: document.writeln( '<PARAM NAME="tep.save.removed.events" VALUE="NEEDED">' ); before the following line: <!--END OF PARAMS--> For Java Web Start clients: On the portal server, edit the $CANDLEHOME\Config\tep.jnlpt file and insert the following line: <property name="tep.save.removed.events" value="NEEDED"/> before the following line: <!--$COOKIES$--> and reconfigure the portal server. To avoid the reconfigure step, the same change can be applied to both the $CANDLEHOME\CNB\tep.jnlp and $CANDLEHOME\Config\tep.jnlpt files. Adjust the file locations as needed for Linux or UNIX portal servers. In either case, use the Internet Explorer Internet Options function to delete the browser temporary files to ensure the modified tep.jnlp file is reloaded. The fix for this APAR is contained in the following maintenance packages: | fix pack | 6.2.3-TIV-ITM-FP0004 | fix pack | 6.3.0-TIV-ITM-FP0002
Reported component name
Reported component ID
Last modified date
APAR is sysrouted FROM one or more of the following:
APAR is sysrouted TO one or more of the following:
Fixed component name
Fixed component ID
Applicable component levels