What Is Monitored
How It Works
Deploy
KPIs
Overview
GermainUX reads the logs of your Oracle BI Publisher (BIP) servers and checks that they’re running. You see every error and warning the BI Publisher application and its WebLogic servers write, from failed reports and delivery errors to stuck threads, and how many WebLogic servers and Node Managers run on each host. An SLA tells you about each new error, so a broken data source or a bursting job that can’t send email is found before the report recipients notice. Your team stops searchingbipublisher.log and the WebLogic logs host by host.
Report and Delivery Errors
WebLogic Server Events
bi_server1 logs, with their WebLogic message ID, such as BEA-000337 for stuck threads.Servers Running
Every Host in One Place
Questions It Answers
- Which BI Publisher reports failed today, and with what error?
- Is
bi_server1running on every host right now? - Did the Node Manager start after the host was patched?
- Why did
bi_server1restart? - Are threads getting stuck on a BI Publisher server?
- Did a bursting job fail to deliver its reports by email?
- Which users ran into report errors this week?
What Is Monitored
Analytics, Alerts and Automation
AI Analysis
Ask AI Explore questions such as “Which BI Publisher errors happened most this week?” and get the answer without building a report. Error categorization groups similar messages, so a new error stands out from the known ones.Dashboards and Reports
Build dashboards and reports on the BI Publisher KPIs: events per host and module, the WebLogic message IDs that occur most, and the server counts over time. Click a value to open it in the Analysis screen. In the example below, threads got stuck onbi_server1, and the WebLogic server events spiked with BEA-000337 messages.

Analysis of BIP Server Events During Stuck Threads - GermainUX
Analysis
Open a BI Publisher KPI in the Analysis screen to see what changed in the selected period. The top row compares the period with the same hours on the previous days, and shows the hourly trend. The relationships chart shows the KPI over a longer time range, with the selected period highlighted. The segmentation breakdown lists the hosts, users and message types that changed most, and labels each one, for example as a spike.Investigate a Node Manager That Didn’t Start
After a host is patched, the WebLogic servers can come back while the Node Manager doesn’t. Open the WebLogic Node Manager Count KPI in the Analysis screen. Compare the average count with the previous days, and use the hourly trend to see when it fell to0.
In the example below, bip-01 was patched on 9 November, and its Node Manager wasn’t started again. From 09:00 the count is 0. Over the day shown, the average count is 0.264, down 73.6% on the 1.000 of the previous days, and the breakdown marks bip-01 as a drop. The relationships chart shows the count at 0 until the Node Manager was started on the afternoon of 10 November. The SLA on the KPI flags the host as soon as no Node Manager runs, so an alert can tell you on the morning of the patch.

Aggregate Analysis of WebLogic Node Manager Count After Host Patching - GermainUX
Investigate a Server Restart
Whenbi_server1 crashes and the Node Manager restarts it, the WebLogic Server Count of the host drops for a few minutes. Open the KPI in the Analysis screen to see when it dropped and for how long.
In the example below, bi_server1 on bip-01 ran out of heap at 15:20. Before it stopped, WebLogic logged low-memory warnings (BEA-310003) in the BIP Server Event KPI, and BI Publisher logged java.lang.OutOfMemoryError: Java heap space in BIP Diagnostic Event. The server count fell from 2 to 1 until the Node Manager restarted the server 25 minutes later. Over the five hours shown, the average count is 1.917, down 4.2% on the 2.000 of the previous days, and the hourly trend dips in the afternoon. The count never reaches 0, so the SLA on the KPI isn’t breached. Open the event KPIs for the same period to see why the server stopped.

Aggregate Analysis of WebLogic Server Count During a Restart - GermainUX
Investigate Reporting Database Errors
When the reporting database can’t be reached, every report that reads it fails. Open the BIP Diagnostic Event KPI in the Analysis screen. The top row shows how far the errors rose above the previous days. The segmentation breakdown shows whether they come from one user or from all of them. In the example below, the reporting databaserpt-db-01 was unreachable for 45 minutes. The data engine logged a connection error for data source RPT_DB for every report that ran, and WebLogic logged BEA-001129 for the RPT_DB connection pool in the BIP Server Event KPI. The analysis shows 146 diagnostic events against 13 at the same hours on the previous days. The breakdown marks the host bip-01 and the ERROR message type as a spike, and spreads the errors over the users ops_lead01, sched_batch, fin_analyst01 and fin_analyst02. Errors from every user point to a shared cause, such as the database, rather than one report. The errors stop when the database is back.

Aggregate Analysis of BIP Diagnostic Events During a Database Outage - GermainUX
Investigate Failed Report Delivery
A bursting job that can’t send its email logs a delivery error for each report it tries to send. Open the BIP Diagnostic Event KPI in the Analysis screen for the time the job ran. The segmentation breakdown shows the user the job runs as. In the example below, the nightly bursting job couldn’t reach the SMTP serversmtp.example.com for 40 minutes after 02:00. Each delivery failed with Could not connect to SMTP host: smtp.example.com, port: 25. The analysis shows 133 events against 3 at the same hours on the previous days, and the breakdown marks the host bip-01 and the user sched_batch as a spike. The SLA on the KPI flags each delivery error, so an alert can tell you before the report recipients notice that nothing arrived.

Aggregate Analysis of Failed Report Delivery - GermainUX
SLAs and Alerts
The event KPIs ship with an SLA that flags every new event. The server-count KPIs ship with an SLA that flags a host where no WebLogic server or no Node Manager runs. Get an alert by email, SMS, Slack or Teams when one is breached, and adjust the SLAs to the events you care about.Correlation with Other Technologies
Correlate BI Publisher events with the Oracle WebLogic servers it runs on, the Oracle Database behind its reports, and OBIEE when BI Publisher runs as part of it.How It Works
Before You Start
- A GermainUX Engine runs on the BI Publisher host.
- The Engine’s account can read the AdminServer log directory and the
bi_server1log directory, for example<DOMAIN_HOME>/servers/AdminServer/logsand<DOMAIN_HOME>/servers/bi_server1/logs. - The BI Publisher managed server is named
bi_server1, as in a default installation. The log monitor picks up its log files by that name. - The host is a host data source in GermainUX.
Deploy
Open the BIP App Server wizard
Select the BI Publisher host
Enter the log directories

BIP App Server Profile Fields - GermainUX
Validate
Check the server counts
2 for the AdminServer and bi_server1.Check the events
Troubleshooting
No BI Publisher events appear
No BI Publisher events appear
bi_server1.log and bi_server1-diagnostic.log. Log files of a managed server with another name aren’t picked up.WebLogic Server Count shows 0
WebLogic Server Count shows 0
Only some log messages appear
Only some log messages appear
Security and Privacy
Oracle BI Publisher monitoring data includes personal data:- Identity data: the BI Publisher user ID of the request that logged the event
- Free text: the log message in each event, which can contain report names, parameters, email addresses and host names. Treat it as possibly containing personal or customer data.
KPIs
GenericEvent, and the count KPIs GenericMetric. The count KPIs are shared with OBIEE. See KPIs and SLAs.
Related Documentation
Oracle WebLogic
File and Log Monitoring
Automation
Get Help
The GermainUX team can help you monitor BI Publisher installations with other managed server names or log locations.Contact GermainUX Support
Can't Find What You Need?
The content of this document is proprietary to GERMAIN SOFTWARE.

