POGO error after migrating to 8.3 from 8.1

[UseCog]

Hi,

Currently we are doing UAT on Cognos 8.3 migration. The TEST server is having Cognos 8.3 installed. The environment was perfect for test runs.

During UAT, our users got report POGO error. When we restart the service, we got following error
7. 17:49:57, 'com.cognos.pogo.monitoring.jmx.PogoMBeanServer', 'pogo', 'Failure'.
DPR-DPR-1035 Dispatcher detected an error.. But other services started successfully.

Some reports are giving follolwing error and some reports are running.

DPR-ERR-2077

      The request failed because the associated report service process is unavailable.

when i restarted the congos services, i got an error

[Launching a JVM using 'Maximum memory in MB']
1. Successfully launched a test JVM with the memory setting of '1152'. Note that this does not guarantee that Cognos 8 will start and run successfully.

To see which JVM options are based on this setting, view c8_location/bin/bootstrap_<OS>.xml and see your JVM documentation for an explanation of those options.

[Archive Local File System Root]
1. Since the value is empty, the feature is disabled. There is nothing to test.

[Start Service]
1. 12:15:14, 'LogService', 'StartService', 'SUCCESS'.
2. 12:15:14, 'LogService', 'StartService', 'Success'.
3. 12:15:22, 'AuditDatabaseConnection', 'Execute', 'Success'.
4. 12:15:21, CM-SYS-5159 Content Manager is running in active mode.
5. 12:15:21, 'ContentManagerService', 'StartService', 'Success'.
6. 12:15:22, DPR-DPR-1002 Successfully registered the dispatcher http://hostname:9300/p2pd in Content Manager.
7. 12:15:23, 'com.cognos.pogo.monitoring.jmx.PogoMBeanServer', 'pogo', 'Failure'.
DPR-DPR-1035 Dispatcher detected an error.
com.cognos:type=queue,service=dataMovementService
8. 12:15:27, 'ReportDataService', 'StartService', 'Success'.
9. 12:15:27, 'PresentationService', 'StartService', 'Success'.
10. 12:15:27, 'DataMovementService', 'StartService', 'Success'.
11. 12:15:27, 'ReportService', 'StartService', 'Success'.
12. 12:15:27, 'BatchReportService', 'StartService', 'Success'.
13. 12:15:27, 'AgentService', 'StartService', 'Success'.
14. 12:15:27, 'SystemService', 'StartService', 'Success'.
15. 12:15:28, 'MonitorService', 'StartService', 'Success'.
16. 12:15:28, 'DeliveryService', 'StartService', 'Success'.
17. 12:15:27, 'EventService', 'StartService', 'Success'.
18. 12:15:27, 'JobService', 'StartService', 'Success'.
19. 12:15:28, 'CPS Producer Registration Service', 'StartService', 'Success'.

please offer your thoughts on this to come out of this issue.

Regards
UG

[UseCog]

Hi All,

unfortunately Cognos left this issue completely empty-hand, saying that this is a producat issue(no fix for C83) and adviced to go for 8.4 instead of 8.3. If anybody interested for more details please visit Problem: 550361, Title: Monitoring error message when starting DMS from Cognos support.

Thanks for your thoughts,
UC

[jue]

We had a similar problem when we migrated to 8.3 which resulted in serious instability which was connected to how 8.3 utilises JMX for its new system metrics. Your problem sounds similar....

Our testing showed that the use of the new system metrics in a large environment :
a) increases the number of background WebSphere threads by over 100%.
b) causes components to fail to talk to each other reliably.
c) prevents services starting up.
d) causes the Cognos Administration section to load and refresh very slowly indeed.

The JMX driven system metrics can be disabled in 8.3 by :
1) Locate and open the p2pd_deploy_defaults.properties file.
   > cd /usr/cognos/JVM-Name/webapps/p2pd/web-inf/
   > vi p2pd_deploy_defaults.properties
2) Disable jmx by uncommenting the #noJMX=true line in the file located in each component.
3) Restart the Cognos component.

Making these changes returns the System view in the Cognos Adminstration section to the 8.1 view. This loads and refreshes very quickly after implementing the change.

This was done on WebSphere 5.1. JRE 1.4.2 SR9. AIX 5.3 ML07. It may only be an issue with WebSphere 5.1 but it is an option worth trying in WebSphere 6.x if WebSphere threads are running high.

Cognos also cnfirmed it is a bug and it is resovled in 8.4 by giving you a toggle to turn off detailed system metrics without having to edit xml files. I've not seen how its been implemented in 8.4 yet though.