This section contains the following topics: ApplicationStorage I/O graph does not filter according to Application and Tier Storage I/O graph does not filter according to Application and Tier, resulting in differences in I/O values between the Storage Units table and the graph. Solution: None |
Precise FrameworkPrecise FocalPointOutOfMemory exceptions appear in the logs/infra.i3fp.out file In a large environment the following error may appear in the logs/infra.i3fp.out file: java.lang.OutOfMemoryError Also the Precise FocalPoint may crash. This is caused due to the default memory limits of the Precise FocalPoint process. Solution: Workaround To increase the Precise FocalPoint memory limit: - Open the products/i3fp/bin/psin_i3fp_init.xml file.
- Change the line:
<option>-Xmx128m</option> to: <option>-Xmx256m</option> or even for very large environments: <option>-Xmx512m</option>
The Tomcat GUI Web Server service fails to start Sometimes, if you do not have sufficient free disk space on the disk and you start the Tomcat service, Tomcat corrupts its configuration files, and as a result, the Tomcat service fails to start again. (PR1092773) Solution: Workaround To verify this is the problem: - Check if the server.xml, tomct-users.xml, and web.xml files in the <i3_root>\products\gui\website\conf folder are valid XML files (this can be done by trying to open them in Internet Explorer and checking if an error message is displayed by Internet Explorer).
- If a file is corrupt, extract the package named psui_tomcat_deploy_ALL.zip to the <i3_root> folder again.
- Restart the Tomcat service.
CommunicationCommunication may get stuck on AIX machine due to IPV6 Host Address lookup Communication may get stuck on AIX machine due to IPV6 Host Address lookup. This may cause many basic operations to stop functioning. When running a thread dump, the report should point out that the JVM is waiting for the following API: at java/net/Inet6AddressImpl.lookupAllHostAddr(Native Method) at java/net/InetAddress$2.lookupAllHostAddr(InetAddress.java:927) at java/net/InetAddress.getAddressFromNameService(InetAddress.java:1295) at java/net/InetAddress.getLocalHost(InetAddress.java:1473) (QCID 6941) Solution: Workaround Do the following: - Edit the file: products/i3fp/bin/psin_i3fp_init.xml
- Add the following option: <option>-Djava.net.preferIPv4Stack=true</option> to the: <jvm-options> section.
- Restart the Precise FocalPoint.
x00142 error or many processes hang Many Precise processes issue a Windows x00142 error and/or get permanently stuck. This may be caused by one or both of the following: - A bug in the JRE for Windows causes processes that are started from java processes to have their STDIN/STDOUT handles stuck.
- The processes exceed the Windows predefined registry setting for the amount of memory allocated to each console application started by a service.
Solution: Workaround The following solution addresses the two possible causes. It should only be performed with help from Precise Technical Support. |
- Replace the JRE with the latest version from www.java.com.
Increase the threshold for the console applications in the Windows registry: Be careful when you change these registry settings. Make sure only to change the mentioned values according to instructions. |
If you use the Registry Editor incorrectly, you may cause serious problems that may require you to reinstall your operating system. Microsoft cannot guarantee that you can solve problems that result from using the Registry Editor incorrectly. Use Registry Editor at your own risk. |
- At a command prompt, type REGEDT32.EXE to start Registry Editor.
- In Registry Editor, locate the following registry key:
HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\Session Manager\SubSystems - In the right pane of Registry Editor, click Windows.
- On the Edit menu, click Modify.
In the Edit String dialog box, locate the SharedSection parameter string in the Value data box, and then specify a larger value for the SharedSection parameter. The SharedSection parameter specifies the system and desktop by using the following format: SharedSection=<xxxx>,<yyyy>,<zzzz> where <xxxx> defines the maximum size of the system-wide heap (in kilobytes), <yyyy> defines the size of the per desktop heap, and <zzzz> is the size of the desktop heap for each desktop that is associated with a non-interactive Windows station.
|
- Click OK.
- Restart the server.
PatchesAdminPointComputer processes get stuck on HP 64-bit JRE computers On HP 64-bit computers, processes get stuck. To enable HP computer support of HP 64-bit JREs, you can apply a patch. To identify the problem, perform the following steps. - From the i3_root directory, run the following:
./java/JRE/bin/java - If the following error message appears:
aCC runtime: dlmodinfo: Invalid descriptor version 58 then this problem exists.
Solution: Workaround To resolve the problem, you can apply one of the following patches: - for HP 11.0 - PHSS_28869
- for HP 11.11 - PHSS_28871
A reference for a similar problem can be found in: http://forums1.itrc.hp.com/service/forums/questionanswer.do?threadId=680743 You can also temporarily bypass the problem by editing the following scripts: - infra/bin/bin/psin_jav.sh
- infra/bin/psin_java_by_xml.sh
Then add a comment to the following line: JVM64FLAG=-d64 Time difference from GMT is reported incorrectly in logs Precise log entries may show a time difference of “GMT+0X” (X representing any number) instead of “GMT+01” during the daylight savings end date and until it restarts. NOTE The time written is correct, but the GMT related string is wrong. Let us assume that daylight savings time ends on October 1 at 01:00. For this example, the clock must be moved back one hour. The correct time written in the log one minute before DST ends is: Time: 2007-04-01 01:59:04.044 GMT+02 GMT+02 is correct. The time written in the log after DST ends is: Time: 2007-04-01 01:00:00.052 GMT+02 GMT+02 is incorrect and should be GMT+01. Let us assume that daylight savings time starts on April 1 at 02:00. For this example, the clock must be moved forward one hour. The correct time written in the log one minute before DST starts is: Time: 2007-10-01 00:59:05.044 GMT+02 GMT+02 is incorrect and should be GMT+01. The time written in the log after DST starts is: Time: 2007-10-01 02:00:00.052 GMT+02 GMT+02 is correct. As you can see in both examples the time was written correctly in the log, but the GMT related string is wrong. This error will continue throughout the entire non-daylight-savings period. Solution: None The machine's time zone is not displayed correctly in some of the Precise agents The machine's timezone is not displayed correctly is some of the Precise agents (mainly FocalPoints and Alerts Point). This may be due to incomplete timezone settings or a change on the machine. The agent may show wrong time zone, even though displaying the GMT time and local time using other tools (for example, on UNIX: `date` and `date -u`) does show the correct time. (ID 1200869) Solution: Workaround To overcome this problem, set the TZ environment variable to the required zone. For example on UNIX: export TZ=GMT Precise LoginError message displayed because default language is not defined The “An HTTP object failed in the send method” error message appears in the browser because no default language has been defined. Solution: Workaround From Internet Explorer, select Tools > Internet Options, and click the Languages button. If the list is empty, add a language. Error message Object does not support this property or method is displayed The “Object does not support this property or method” error message appears in the browser, because Precise products cannot work on an Internet Explorer version earlier than version 6. Solution: Workaround Upgrade Internet Explorer. Apache Tomcat fails to start In a multiple installation, where more than one Precise FocalPoint is installed on the same server, Apache Tomcat fails to start. Solution: Workaround Change the internal Apache Tomcat shutdown port, as follows: Go to the <i3_root>/infra directory. Open the setup.xml file and locate the line with the tomcat shutdown port node, which should be similar to the following line: <tomcat-shutdown-port>20768</tomcat-shutdown-port> Change the node value to the number of a port not used by any other process on the server. This is especially important on UNIX and Linux servers with many installations of Precise version 7.x. Restart Precise FocalPoint. NOTE This is an internal port used by Apache Tomcat, not the port used to access the product’s user interface (both can be set using Precise Installer). GeneralError in agents' verification for errors found in the out file The Precise product uses the SQL Server JDBC driver. In some cases, this driver prints errors to the agent's out file. This error is reported in the verification process of the agent in the following manner: errors found in log file XXXX in the last 24 hours. Check the log file for more details The error in the out file has the following format: com.jnetdirect.jsql.x: DBComms.transmit exception:[java.net.SocketException: Connection reset by peer: socket write error] context:[(1) [Thread[i3comm(pw-data-retrieve),5,main], IO:22671, Dbc:1192]]] at com.jnetdirect.jsql.x.a(Unknown Source) at com.jnetdirect.jsql.h.a(Unknown Source) at com.jnetdirect.jsql.ax.a(Unknown Source) at com.jnetdirect.jsql.ai.new(Unknown Source) at com.jnetdirect.jsql.ai.for(Unknown Source) at com.jnetdirect.jsql.ai.execute(Unknown Source) (QCID 6721) Solution: None Ignore these errors. All downtime settings are updated one hour forward or backward Downtime settings are affected by daylight saving time. In case the FocalPoint server machine time moves one hour forward or backward, all downtime settings need to be manually fixed to match the previous settings. Solution: Workaround To update all downtime settings one hour backwards, run the following CLI on the main Precise FocalPoint server: infra\bin\psin_cli.bat -i3-user admin -i3-clear-password admin -action downtime-update -offset -1 To update all downtime settings one hour forward, run the following CLI on the main Precise FocalPoint server: infra\bin\psin_cli.bat -i3-user admin -i3-clear-password admin -action downtime-update -offset 1 Save option in print preview Sending an htm file that was saved from the print option is possible only within the same network of the installation, otherwise it may include broken image links. (IN 33583) Solution: None PMDBOracleCore dump on HP-UX 64-bit servers On HP-UX 64-bit servers, the Oracle JDBC driver may cause a core dump in the PMDB FocalPoint. Solution: Workaround Copy the .hotspot_compiler file from the <i3_root>/products/dbms/files/etc directory to the <i3_root> directory on every Precise server. Restart all Precise agents on all Precise servers. SQL Server instance cannot be connected When the PMDB FocalPoint cannot connect to an SQL Server instance (usually when the instance is down), the 3rd party JDBC driver may write an error to the pw.fp.out log file. Solution: None Problem when the default database is dropped When you drop the default database of the PMDB user, the PMDB stops working and is not able to connect to the database. (PR1009283) Solution: None Warehouse ProcessesPMDB Load process fails with database deadlock error message The load process fails with the following error message: Transaction (Process ID x) was deadlocked on lock resources with another process and has been chosen as the deadlock victim. Rerun the transaction. (PR895778, PR982755, PR1074722) Solution: None The next run will work fine. Baseline process performance reduced On very large environments, the Baseline process may run for a very long time and during that time, the load processes performance is dramatically reduced. (IN 35421) Solution: None Oracle data load process fails with ORA-01461 The Load process sometimes fails when the PMDB runs the Load Performance Data process of Precise for Oracle. The following Oracle error is displayed in the pw.fp.trc log file: ORA-01461: can bind a LONG value only for insert it into a LONG column The problem only happens in a PMDB with the “AL32UTF8” character set. (PR 811481) Solution: Workaround If you encounter this problem, contact Precise Technical Support. Load process fails with ORA-00054 The Load process sometimes fails when the PMDB is running the Summary or Maintain process at the same time. The following Oracle error is displayed: ORA-00054: resource busy and acquire with NOWAIT specified Solution: None The next Load will work fine. AlertsFocalPointHigh CPU usage on SNMP Server Alerts uses a lot of CPU on an SNMP server. This may happen because the messages sent are too long for the SNMP browser to handle, thus locking the Alerts FocalPoint process and consuming extensive CPU. Solution: Workaround Not all SNMP browsers can handle the same message length. In general, before applying SNMP traps, configure the traps according to your SNMP browser and test them. To test the message size, put a defined length string in the <test-message> section in the Alerts FocalPoint registry file. Then perform a test for the SNMP trap, using the Test button on the SNMP tab of the Metric Properties dialog box. If the sent messages are too long and the Alerts FocalPoint process is locked, do the following: Stop Alerts FocalPoint. Open the Alerts FocalPoint registry file. Reduce the size of the message sent to your SNMP browser by adjusting the <max-message-size> value. The maximum default size is 900 characters, which is the maximum size supported by Alerts. Deadlock may appear A deadlock may appear after you disconnect an instance from Alerts. If this happens, the sync operation inside Alerts FocalPoint does not end and the instance will remain in the Alerts database. Solution: Workaround Click the Instances tab in the Settings dialog box. Select the instance and click Edit. The Instance Settings dialog box opens. Clear the Do not report alerts for this instance check box and click OK. Open the Instance Settings dialog box again for the same instance and select the Do not report alerts for this instance check box. To verify that the instance has been disconnected, execute the following SQL query: select instance_id, instance_name, server_name from pulse_nodes Under instance_name, the result of this query should not return the removed instance. If it does, execute the following SQL delete statements, where instance_id should be the number retrieved from the previous select query: DELETE FROM PULSE_ALERT_MATRIX WHERE INSTANCE_ID = <instance_id> DELETE FROM ALERTS_SAMPLES WHERE INSTANCE_ID = <instance_id> DELETE FROM PULSE_ACTION_LOG WHERE INSTANCE_ID = <instance_id> DELETE FROM PULSE_ACTION_SETTINGS WHERE INSTANCE_ID = <instance_id> DELETE FROM PULSE_INDICATOR_SETTINGS WHERE NODE_ID IN (SELECT NODE_ID FROM PULSE_NODES WHERE INSTANCE_ID = <instance_id>) DELETE FROM PULSE_NODES WHERE INSTANCE_ID = <instance_id> LaunchingLaunch does not work correctly from the Events and History tabs Launch from Alerts metric view into Precise/Insight/Report Manager products may not work correctly from the "Events" tab and the "History" tab.(PR989899) Solution: Workaround Launch should be performed only from the "Current" tab. Launch does not work correctly after sorting the items The launch from the Alerts metric view to Precise, Insight, or Report Manager products may not work correctly if items in the current view are sorted differently from the original sorting. (PR811075) Solution: Workaround Do not sort items in the current tab of the metric view before launching. Launch from Process Availability metrics to Insight is incorrect Launch from Process Availability metric should go to the Insight Savvy for OS. Instead, it goes to the AppTier of the metric. An empty screen is displayed. (PR895936) Solution: None MetricsAfter adding customized metric only able to save for all instances When adding a new metric, saving the settings, and switching to another tab and then to the Customized tab, you are only able to save for all instances. (QCID 5531) Solution: Workaround Follow these steps: Make sure the Alerts Agent is installed and started in a required server where the instance should be sampled. Reopen the screen. Instance does not sample all its metrics An instance is not sampling all its metrics. Solution: Workaround If the InformPoint has not been started yet or has not been started up completely, allow additional time. For the first time, most metrics are only sampled after a while, according to their sampling parameters (sampling rate and sampling base). The InformPoint samples all instances whose FocalPoints are installed on the same server as the InformPoint. For example: If the InformPoint is installed on server A, and both Precise for SQL Server FocalPoint and Precise for Web FocalPoint are installed on this server, the InformPoint samples all SQL Server instances and all Web instances, no matter on which server they are installed (but provided they are connected to Alerts). Metric value reaches over 100% The values of metrics in the Performance set and the Load set may reach over 100% because each item is rounded to an integer value for the following AppTiers: Tuxedo, Operating System, WebSphere MQ, Oracle Applications, and Other. The problem occurs in Insight metrics because Insight rounds the data it sends to Alerts. For example, if User1 was running for 4.5 seconds and User2 for 0.5 seconds, Alerts will calculate a total time of 5 seconds (4.5+0.5). Insight, however, rounds up the time to 5 seconds for User1 and 1 second for User2. Therefore, when you calculate the percentage, Alerts will reach the following results: User1—100% (User1 time/total time = 5/5) User2—20% (User2 time/total time = 1/5) Together, this amounts to 120%. Solution: None Creating the 500th metric fails In a single Precise installation, the number of metrics that can ever be defined for the Customized set is 499. Trying to create the 500th metric leads to an Oracle error (Primary Key constraint). Solution: None Consider only and Ignore the following items functionalities not supported in the SQL Server AppTier The SQL Server AppTier does not support the Consider only the following items and Ignore the following items functionalities on the Thresholds tab in the Metric Properties dialog box for the following metrics: Top Activities, child metrics “Top Statements” and “Top Ad hoc Batches” (Performance set) Frequent Activities, child metrics “Frequently Used Statements” and “Frequently Used Ad hoc Batches” (Load set) Solution: None Customized metric hangs If a Customized metric hangs, the InformPoint cannot terminate it. Solution: None Oracle UDF - procedure does not run Execution of a User Defined Metric with an Oracle procedure, on a server with Interpoint installed but with no Precise for Oracle FP server, fails. (ID 17082) Solution: Workaround From the Precise for Oracle FP server, copy the following file: <i3_root>\products\dbms\files\predefined_queries\or_pulse.xml to the Interpoint server, to path: <i3_root>\products\dbms\files\predefined_queries\. Restart the Alerts InformPoint. Empty values for dynamic parameters Dynamic parameters for Customized metrics may return empty values, so that they are passed incorrectly to the sampling script. For example: When the Consider only the following items dialog box on the Thresholds tab in the Metric Properties dialog box is empty, the @INCLUDE_LIST parameter returns an empty string. Solution: None Line breaks do not appear in free text Line breaks may not appear in free text in an E-mail or Message action defined in the Metric Properties dialog box. Solution: Workaround Add more lines. Then click the Test button to view the result. Message actions are delayed Message actions are displayed with a delay of up to 15 minutes after an alert was triggered. Solution: None Results when you add a new customized metric are unexpected When a new customized metric is added, the following issues may occur: The $instance_name value of this metric can be incorrect. When the Apply All button in the Metric Properties dialog box is clicked while the InformPoint sampling the Customized metric is down, the error message that appears can show an incorrect instance name. When creating the metric, warning 3570217 may appear in the log file (no effect on the outcome). When deleting the metric, warning 3570085 may appear in the log file (no effect on the outcome). When testing an action that is not an SNMP action, warning 3570084 may appear in the log file, and the action will not be triggered. Solution: Workaround After creating a new Customized metric, restart the Alerts FocalPoint. You can also resample the new Customized metrics without restarting the Alerts FocalPoint and this will solve the issue with the warnings. Max, Average, and Min metric values are rounded When Alerts is installed with an SQL Server database, Max, Average, and Min metric values are rounded to integer values. For example, a customized metric with a constant value of 1.57 is rounded to a constant Max, Average, and Min value of 2.00. Solution: None Customized metrics are shown as disabled After you add a new instance or environment or reconnect an instance to be monitored by Alerts, some of the customized metrics of that instance are shown as disabled in the left pane of the Metrics tab, while the metrics are actually sampling. Opening the Metric Properties shows that the metric is enabled. This inconsistency between the customized metric's enabled state and the metric’s actual sampling state can also be detected by refreshing the Alerts Metrics tab. The sampling “Timestamp” of the disabled metric is updated although the metric is disabled. The problem occurs when you set customized metrics in the Metrics Properties dialog box to be sampled and use Apply all to update the setting on all instances. After you add new instances, the customized metrics will be added automatically to the new instances. Those metrics will be sampled, but shown as disabled in the left pane of the Metrics tab. Solution: Workaround Change the metric state to disabled in the Metric Properties dialog box. This will cause the metric to stop sampling. The metric can be enabled again when required. () Not Sampled status for Performance and Load metrics that sample AppTiers The Performance and Load metrics that sample AppTiers monitored by Insight (Oracle Applications, Tuxedo, Websphere MQ, or Other) report the “() Not Sampled” and running the Verification script for Alerts FocalPoint in AdminPoint may result in warnings about those metrics. The Event viewer and logs include the following error message: No data has been found for the period sampled. If an activity has occurred in the monitored instance during the period sampled, start AdminPoint and use the Verify option to check the status of Insight FocalPoint and the relevant agent. If no activity has occurred in the monitored instance, ignore this message. Solution: Workaround If you use Oracle Applications, make sure that in the Metric Sampling Properties dialog box the metric is sampled over the last 30 minutes (insert 30 in the m field in the Analyze metric over the last parameter). Check that data of the relevant instance is available in Insight. If so, make sure that the Load process does not linger more than 5 minutes after the end of the time slice. Actions’ text is not translated to Japanese The content of the actions is not translated to Japanese.(PR1006773) Solution: None Editing the disabled Process Availability metric causes errors in the alerts log file Process Availability metrics are available for editing from the Settings>General Settings>Metric dialog box, even if they are not enabled and configured. While editing them, you may see errors in the Alerts FocalPoint log. These errors can be ignored. (PR966397) NOTE By editing a disabled "Process Availability" metric in the properties screen of Alerts, you do not make it available. To configure this metric, you should define its configuration using the Settings>Process Availability dialog box. Solution: By design Actions defined for all instances are not duplicated for new added instances Actions defined on the metric for all instances are not duplicated for new added instances. If you define an action for a metric and save it for all instances, the action will be defined for all available instances at that moment. Instances added after the action definition will not have the action for the metric. (PR894976) Solution: Workaround To define the action for new added instances, edit the action details and save it again for all instances. Agents Metrics not updated If using alerts metrics to sample Precise agent status, the agent status is updated only once every 15 minutes. If you perform an agent operation using the AdminPoint (start/stop/refresh), the operation is not reflected in the alerts Precise agent status until an update is made. (IN 35124) Solution: By design MOMMOM password cannot be updated If Alerts is integrated with MOM server, and the MOM user's password changes, it cannot be updated in Alerts MOM settings. MOM actions will fail. Solution: Workaround There are two ways to bypass this issue: Change the user name each time a password change is needed. Enter an invalid server name to register to it (causing unregistering from the old server), and then reregister to the real server with the new password. Report ManagerFocalPointJVM out of memory error while running reports A JVM out of memory error may occur while running reports with a large amount of data. (PR1071865) Solution: Workaround Open the <i3_root>\products\foresight\bin\psfs_focal_init.xml file for editing. Look for the <jvm-options> tag. The default size of maximum memory Report Manager FocalPoint JVM can consume is now 320m. <option>-Xmx320m</option> Enlarge this size. Problem downloading files when using SSL Configuration When the Apache Tomcat is configured with the secured SSL protocol, there is a problem downloading and uploading files between the FocalPoints machines and the clients. When the user tries to create a new User Defined Report (UDR), he needs to move the defined entities from the tree (left bar) to the right tables. This means the selected entities are now included in his Excel report. After defining all the report columns and fields, the user clicks on the Execute button. When executing the report, the client tries to download a UDR template file which contains a signed macro, and the process fails. In a normal situation, Report Manager modifies this signed template with all its data. Error message received: An error occurred. Extended Error info: Error #1004 - 'C:\Documents and Settings\<username>\Local Settings\Temp\xlUDR.xls' could not be found. Check the spelling of the file name, and verify that the file location is correct. If you are trying to open the file from your list of most recently used files on the File menu, make sure that the file has not been renamed, moved, or deleted. Solution: Workaround Before clicking the Execute button, the user needs to copy the UDR.xls file from <i3_root>\products\gui\website\webapps\i3\Foresight\udr to his temp directory (%temp%). LaunchingMismatch of values between Precise products and Report Manager All Report Manager reports are based on daily tables and display data for each day in Overtime graphs. The summary level of Insight and the different Precise products depends on the settings that you defined. Only the default settings match the time frame of Report Manager default report. If you customize the time frame of a report or in Insight or a Precise product, this can cause a mismatch. Solution: None Launch not working outside the Precise environment When Report Manager reports are viewed offline, not through Report Manager, the launch options to Insight and Precise products do not work. Solution: None ReportsAvailability reports show a mismatch of unavailability times Availability reports sometimes show a mismatch between unavailability times in Report Manager and Precise products. This happens due to a lack of data in the availability tables. (PR810228) Solution: None Summary data mismatch Small differences between the summary data in the reports and the summary data in the products may occur because the PMDB saves values without fractions. Solution: None Duplicate time records in graphs When the Y-axis represents time, the reports display this as “hh:mm:ss.” Because the data also contains milliseconds, non-linear gaps will occur between the points on the axis. Solution: None Launching error with Acrobat Reader version 7.0.5 or higher but lower than version 8 Launching Precise/Insight/Inform products from an Report Manager report causes an error on a server with Acrobat Reader version 7.0.5 or higher but lower than version 8. (PR828075) Solution: Workaround Uninstall Acrobat Reader and install version 8. Y-axis labels are overriding each other and look like black boxes When the amount of data is higher than Peta (such as petabytes and petaseconds), the y-axis labels override each other and look like black boxes. (PR996640) Solution: None Report Manager customized report (UDR) with Excel 2007 opens in a new Excel window When viewing a customized report with Excel 2007, the report is opened in a new Excel window instead of in the Precise browser. When rerunning the report, the Precise browser is stuck. Working with the customized report in Edit mode (using the Edit button) is fine. (IN 35559) Solution: Workaround Open Windows explorer Open Folder Options in the Tools menu Go to the File Types tab Find .XLS type and click on the 'Advanced' button Mark the Browse in the same window check box Click OK twice and try to view the Excel report again Formula values are not updated when running the report based on the scheduler When creating formulas in the "report-design" sheet of a customized report's Excel file, formula values are not updated when running the report based on the time set in the scheduler. The data in the "FS-*" query sheets is updated correctly when running the report. Running the report using the Edit button and then the Execute button is working fine. (IN 30831) Solution: None UIError during access to Report Manager when the Report Manager FocalPoint is down When entering the Report Manager user interface while Report Manager FocalPoint is down, the following message is displayed: Error returned from Report Manager FocalPoint. Upon clicking OK, the screen is left blank. (IN 35703) Solution: Workaround Close the Explorer, reopen it to go to the AdminPoint and start the Report Manager FocalPoint. Data of non-Latin language may be displayed as ???? Data of non-Latin language may be displayed as ????. (PR1004841) Solution: Workaround To change the regional settings of the browser's machine: Open the settings dialog: Start>Settings>Control Panel>Regional and Language Options. Click the Advance tab and select the correct language (Hebrew/Turkish/Japanese) in the drop down menu to be the language of non-Unicode programs. Restart the machine. InsightFocalPointChanges not reflected in Insight License updates or installation changes (via Installer), such as installing a new AppTier, are not reflected in Insight unless it is opened in a new browser window. Solution: None Problems with text in graphs The graphs in Insight do not properly display texts when the language used in the graphs does not match the default character set on the client server. Example: On a server with a default English character set, Insight replaces Japanese characters with a question mark. Solution: Workaround Change the Display toolbar in the AppTier area from All to a specific entity, such as URLs. Data of non-Latin language may be displayed as ???? Data of non-Latin language may be displayed as ????. (PR1004841) Solution: Workaround To change the regional settings of the browser's machine: Open the settings dialog: Start>Settings>Control Panel>Regional and Language Options. Click the Advance tab and select the correct language (Hebrew/Turkish/Japanese) in the drop down menu to be the language of non-Unicode programs. Restart the machine. GeneralCorrupted or empty insight.cfg file An Insight Savvy fails to start due to a corrupted or empty insight.cfg file, and the following error is written to the insight.agents.server file. Key pcs_dir is missing from section [general] in configuration file insight.cfg Solution: Workaround Make sure all Insight Savvies on the server are down. Delete the current insight.cfg file. Restart the Insight Savvies. Failure to read insight.cfg file during startup of Insight Savvies When you try to start Insight Savvies, the following may appear as a result of irrelevant or corrupted files in the Precise registry: error message 1714—“Failed during operation: read insight.cfg” Solution: Workaround Insight reads information from the following directories (including their subdirectories): <i3_root>/products/i3fp/registry/products/Insight <i3_root>/products/i3fp/registry/instances Check if a backup directory or files have been added to the listed directories. If so, move them to another location. Check if the files contained in the listed directories have an invalid XML format. If so, correct the XML code to validate the files. Data counters loaded more than once Data counters monitored by Insight Collectors (OS, Websphere MQ, Tuxedo, Oracle Applications, Network and J2EE Insight info) are loaded twice. If, in a Precise installation, there is a server whose name would become a substring of the name of another server existing in the same installation (as a result of a concatenation with an underscore), the collected data of one of the servers will be loaded twice. For example, “server1” would be OK because "server_" (“server” becomes “server_” after concatenation) is not a substring of server1. This is the case with the name “server_1” because "server_" is a substring of server_1 and the monitored data on server_1 would be loaded twice. (PR 2382) Solution: Workaround Rename one of the servers to a name which solves the described problem (for example using the IP address) according to the Precise server rename procedure or contact Precise Technical Support. Network“The kernel queue is full” error The error message “The kernel queue is full. Loosing packets …” appears due to the following reasons: If it is displayed immediately after startup, psi_poll is waiting for the psi_ccs initialization to complete. To verify that this is not a temporary problem, send the signal USR1 to psi_poll to check if the message is rewritten to the log file. This message can also appear a few times after startup. The reason for this is that the Insight network kernel module increases the kernel queue size step-by-step, till it reaches the maximum size, as is configured by the kernel_queue_size parameter. Every time the queue is increased, this message will appear. The Network queue size is not big enough for the site’s heavy network traffic. To verify that this is not a temporary problem, send the signal USR1 to psi_poll to check if the message is rewritten to the log file. Solution bullet 1: Workaround Ignore the message. Solution bullet 2: Workaround Gradually increase the kernel_queue_size parameter in the Precise FocalPoint registry file. Stop all Insight Savvies on the server. On UNIX servers: Start Insight Savvy for Network; then stop it. Restart the kernel network module as follows: Go to the products/Insight/<hostname>/kernel directory and run the pr_config.sh –u command as root user. Restart all Insight Savvies on the server. (The insight.cfg file is rebuilt.) Insight Savvy fails to start If you kill psi_poll using the task manager or similar (on a Windows server), Insight Savvy may fail to start. Solution: Workaround Reboot the Windows server. No network data is collected for servers with dynamic IP addresses Insight Savvy for Network cannot monitor the network activity of servers that changed their IP address after the installation of Insight Savvy for Network (for example a server which is assigned a different IP address at boot time using a DHCP server). The list of monitored IPs is determined by Insight Savvy for Network at installation time. If the IP of the server is later changed, Insight Savvy will not monitor TCP/IP activity on the new address. If the server changes its IP address frequently, Insight Savvy for Network cannot monitor the server. Solution: Workaround If the IP address change is permanent, perform the following steps to monitor the Insight Savvy for Network configuration: Log on to Precise FocalPoint server. Edit the <i3_root>/products/i3fp/registry/products/insight/hosts/<HOSTNAME>.xml file (replace <HOSTNAME> with the name of the monitored server). The IP addresses are listed under the <ips> section; you can modify this section to reflect the changed or new IP addresses which are in use by the monitored server. NOTE The recommended way to do this is to modify an existing IP definition (to reflect IP change) or to Copy-Paste an existing IP definition and modify it (to reflect an addition of an IP address). NOTE Each IP address is represented as an XML element, the digits of an IP address must be preceded with two underscore characters, make sure to keep this format. Format for each IP address definition: For UNIX: < XXX.XXX.XXX.XXX>YES</ XXX.XXX.XXX.XXX> Replace XXX.XXX.XXX.XXX with the IP address. The content of the XML element is the constant string "YES" (without the quotes) For example: <__9.148.241.116>YES</ 9.148.241.116> For Windows: < XXX.XXX.XXX.XXX > <product_name>PRODUCT_NAME_STRING</product_name> <service_name>NETWORK_CARD_ID</service_name> <description>NETWORK_CARD_DESCRIPTION</description> </__XXX.XXX.XXX.XXX > Replace XXX.XXX.XXX.XXX with the IP address. The content of the XML element is three sub-elements: product_name, service_name and description, their values are irrelevant and do not affect the operation of Insight Savvy for Network. NOTE You can copy-paste these elements with their values from another card definition on the server. For example: < 166.98.16.45> <product_name>%% NO_VALUE %%</product_name> <service_name>{4E7FBFB9-3F4C-41E6-9901-7EB01ECCA5BF}</service_name> <description>Broadcom NetXtreme Gigabit Ethernet</description> </__166.98.16.45> NOTE Always make sure when you edit the <ips> section to change both the IP address at the opening and closing tags of the XML element and to precede the IP address with two underscore characters. If possible, open the XML file with Internet Explorer to verify that the XML file is valid. To apply the configuration changes, stop all Insight Savvies running on the monitored server, and restart them. High CPU consumption on AIX 5 systems On AIX 5 systems, Insight Savvy for Network can cause high CPU consumption and possibly significant performance degradation of the server, when it monitors very active TCP ports. This problem has occurred when you monitor very active Oracle database ports. Solution: Workaround Stop Insight Savvy for Network and uninstall it from the monitored server. SEVERE-3000335 If service time is prolonged, Insight Savvy for Network will write the following message to the insight.agents.server.log file: ID : SEVERE-3000335 Msg :Agent received invalid record. Invalid field name: max average time . (PR 1222) Solution: Workaround Ignore this message Insight Savvy for Network cannot be started Insight cannot monitor the same port number for two Precise instances on the same server. When you configure two instances in the Installer with Insight Savvy for Network selected and at least one identical port number appears in the configuration of both instances, one of the following messages is displayed in AdminPoint while trying to start Insight Savvy for Network: Cannot start Insight agent to monitor <Technology Name> [Instance name]: Port number <Port Number> because current port is already defined for <Technology Name> [Instance name]. Cannot start Insight Network agent because port <Port Number> is defined both for <Technology Name> [Instance name] and <Technology Name> [Instance name]. (PR 1980) NOTE The [Instance name] is optional (it will not appear when Insight Savvy monitors multiple instances). Solution: Workaround If the two instances are not really sharing the same port number and the duplicate is a configuration mistake, correct it and make sure that port numbers are unique per instance on the same server. If the instances are using the same port number for separate IPs, refer to Tech Note 275773 (http://support.veritas.com/docs/275773). If the instances actually share ports (for example, two Oracle instances that share one Oracle Listener agent with a shared port for all instances): Configure the port number for just one of the instances (preferably the Front-End instance. For example, if you have Web and J2EE instances choose the Web instance). The network will be shared between them when viewed through the Insight UI. Do not integrate the instances with Insight Savvy for Network, and define an “Other” instance that monitors the shared ports. lsof utility may consume too many resources The lsof utility which is activated by Insight Savvy for Oracle Applications and Insight Savvy for Network may consume many resources on a loaded server. lsof is executed every three to five minutes (configurable), so a runtime of one minute on a busy server is not acceptable. This problem is likely to occur on HP-UX servers. (PR 1081) Solution: Workaround To reduce lsof resource consumption, the following parameter needs be added to the <i3>/products/i3fp/registry/products/insight/<host-name>.xml file: Add the following parameter to the <oa> section: On HPUX, Solaris and Aix: <oa_lsof_command>./bin/lsof -a -c [process-name] -l -n -P | grep ESTABLISH | awk '{print $2,$9}' | uniq</oa_lsof_command> On Linux: <oa_lsof_command>./bin/lsof -a -c <process-name> -l -n -P | grep ESTABLISH | awk '{print $2,$8}' | uniq</oa_lsof_command> Replace [process-name] with the prefix of the program-name. Use “f45” for Oracle Applications 11.0 and “f60” for Oracle Applications 11i. NOTE To apply the change, restart all Insight Savvies on the relevant server. Error when you start Insight Savvy for Network on Windows After you install or upgrade Insight Savvy for Network on a Windows environment and starting it, one of the following errors is written to the insight.agents.server.trc file: ID : SEVERE-3000261 Process : psi_poll.exe (ID:3004) Msg : Error in service handling while starting service Psi_drv_v71. SysErr : 2:The system cannot find the file specified. Or: ID : SEVERE-3000259 Process : psi_poll.exe (ID:5856) Msg : Error in service handling while opening service Psi_drv_v71. SysErr : 1060:The specified service does not exist as an installed service. The message appears in Windows 2000 environments and later when a previous installation of Insight Savvy for Network existed on the server (regardless of whether or not it was uninstalled cleanly). This occurs because Insight Savvy for Network uses a device driver (psi_drv) and the operating system does not allow access to the newly installed device driver without a reboot. (PR 3345) Solution: Workaround Reboot the server with the Windows environment. Network data is not shown IIS provides you with a way to define two different Web sites on the same server with the same port (using Host-Headers). Precise for Web can be installed for each of these Web sites separately, but the Network agent cannot collect network data for each of these Web sites separately. Therefore- when adding Precise for Web instances in this configuration, the option "Associate this information with information about network performance" should not be selected. (PR999518) Solution: None Network data files are not harvested Network data files are not harvested from the host (global zone) when no instances are defined on the host (global zone) that is integrated with the Insight Savvy for Network. (Incident 1168576) Solution: Workaround Define a dummy instance on the host (global zone). (For example: Another instance that monitors a dummy port which is not in use). Hosted (local zone) Savvy unable to start Hosted (local zone) Savvy unable to start. (Incident 1166364) The following error messages appear in the agents log and the verification of the hosted (local zone) Savvy: Key host_server_name is missing from section [ccs] in configuration file insight.cfg. Function failed in duty: kmMainInit. Starting process. Insight Network agent is returning a fault status. Check the agent’s log for details. In distributed mode, the host server name is not configured Make sure the Insight Network Savvy is installed on the host server, that it is configured correctly and that it is running. Solution: Workaround Restart the Insight Savvy for Network on the host (global zone). Network settings are not refreshed in distributed zone environment Network settings are not refreshed in distributed zone environment. Changes on the Local zones are not getting updated fully on the Global zone. (IN 35750) Solution: Workaround Restart all Insight Agents on the Global zone after making any changes on the local zones regarding OA or Network settings TuxedoNo information is displayed in Insight’s Tuxedo tab when the “Smartlink is On” button is active When either a Tuxedo instance, or an Oracle instance which serves a Tuxedo instance, are part of two or more Precise environments, opening Insight’s Tuxedo tab and turning ‘Smartlink is On’ results in a screen showing no data. (QCID 8902) NOTE This known problem applies if you have bundle patch BU_IN_870_100 installed. Solution: Workaround Remove the instance from all the Precise environments except one. Insight Savvy for Tuxedo does not sample data Insight Savvy for Tuxedo does not sample any data because the physical host name differs from the Tuxedo logical machine ID (LMID). This problem may occur if Insight Savvy for Tuxedo is installed on a shared disk of a cluster and the Tuxedo LMID is the logical host name of the cluster. Solution: Workaround Disable Insight’s autodetect functionality by directly specifying the shared memory key if Insight Savvy for Tuxedo in the Precise registry. Availability of the Tuxedo domain shows a status of down in Alerts, even when it is up The Availability of the Tuxedo domain shows a status of down in Alerts, even when it is up. This occurs when the Tuxedo domain is configured in the Installer with an incorrect path for the Tuxedo configuration file, for example using a soft-link for the file while the domain is configured to use the real path. The following error message may appear when verifying the Insight Savvy for Tuxedo for the domain in AdminPoint: TMADMIN-CAT:1607 Unable to initilize public key sub system. (PR 2055) Solution: Workaround Edit the Tuxedo domain properties and change the Tuxedo configuration filename to be exactly what is used by Tuxedo with the Installer. Use the exact value of the TUXCONFIG environment variable of the Tuxedo domain user. Insight Savvy for Tuxedo may not sample data Insight Savvy for Tuxedo may not sample any data because the physical host name differs from the PMID defined in the UBBCONFIG file. Solution: Workaround For each Tuxedo instance: Open the <i3_root>/products/i3fp/registry/instances/tuxedo/<server name>/<instance name>.xml file. Add the following entry to the XML file: <tux_physical_name>[PMID]</tux_physical_name> Where [PMID] is the physical host name as configured in the Tuxedo configuration file for the specific instance (in the MACHINES section). Start the Insight Savvy for Tuxedo on the server. Insight Tuxedo server processes Queue Time (Avg) value is truncated Server Processes Queue Time (Avg) is rounded (truncated) though it should display 2 digits after the decimal point. (IN 34657) Solution: None SmartLinkSSA files arrive too late to the Insight FocalPoint machine Precise is installed on VM and the clock time changes dramatically every hour (back and forth). This affects Precise common time and agents’ files are stamped too late. Files harvesting is performed after the timeslice was handled by the Insight FocalPoint. When the files arrive to the Insight FocalPoint they are deleted. (IN N/A) Solution: Workaround The following procedure must be approved and performed by the IT people: VM configuration: Check the ‘Time synchronization between the virtual machine and the console operating system’ check box to make the VM responsible for time synchronization Windows Services – Stop and disable the ‘Windows Time’ service. This action must be performed by IT (because if the VM is not synchronized with the domain controller, login to the machine will fail). J2EEMonitoring all loaded methods does not work The monitored application does not work when the following Custom.xml is used to monitor all loaded methods: <instrumenter-config> <custom-config> <java-classes> <java-class> <invocation-type>Test_1</invocation-type> <class-name>*</class-name> <methods> <method> <name>*</name> </method> </methods> </java-class> </java-classes> </custom-config> </instrumenter-config> Solution: Workaround Use the bootstrap classpath instead of the Java agent. Cannot monitor Felix OSGi When installing Precise for J2EE with the application server using the Felix OSGi, the implementation stops the application from starting. Solution: Workaround Add com.precise, com.precise.* to the org.osgi.framework.bootdelegation list inside the following files: <GLASSFISH_INSTALLATION_DIR>\osgi\equinox\configuration\config.ini <GLASSFISH_INSTALLATION_DIR>\osgi\felix\conf\config.properties Missing transactions in TPM In certain cases, methods with custom instrumentation become a root nodes, causing the entry points of the transaction (like HTTP) to be a second method and hence not to be reported to the Smartlink engine. Solution: Workaround Turn off custom instrumentation first and verify that the transaction is correctly reported. After that, identify the method that breaks the correlation and add it into the ignore list. Increasing Appache CXF Web services causes Tomcat to stop After installing Precise for J2EE 9.5 or upgrading from a previous version, Apache CXF web services client calls may stop working, with the following error message: javax.xml.ws.soap.SOAPFaultException:No binding operation info while invoking unknown method with params unknown. This problem is an Apache CXF bug, that can be reproduced even on an unmonitored server by adding an empty LogicalHandler to the Binding's execution chain using any of the methods described in the JAX-WS specification. Solution: Workaround If you want to apply this workaround to all the instances on a certain machine, perform the next steps on the file: <I3_ROOT>\ products\j2ee\config\instrumenter\WebService_JAXWS_AOP.xml on this machine. If you want to apply it only to one instance, separate its Jax-WS instrumentation configuration, and perform the next step on the separate file: Copy the file <I3_ROOT>\ products\j2ee\config\instrumenter\WebService_JAXWS_AOP.xml from the monitored JVM's machine into <I3_ROOT>\products\i3fp\registry\clusters\j2ee\<CLUSTER_NAME>\ on the FP machine. Edit the WebService_JAXWS_AOP.xml file on the FP machine. Add an enclosing < WebService_JAXWS_AOP > element to the entire file contents. In the file <I3_ROOT>\products\i3fp\registry\clusters\j2ee\<CLUSTER_NAME>\InstrumenterConfigList.xml, change <config-file>WebService_JAXWS_AOP.xml</config-file> to <config-file>${indepth.j2ee.home}/config/${indepth.j2ee.server_id}/WebSer vice_JAXWS_AOP.xml</config-file>. In the file WebService_JAXWS_AOP.xml, locate the lines: com.precise.javaperf.aop.jaxws.HTTPHandlerChainAspect and com.precise.javaperf.aop.jaxws.SOAPHandlerChainAspect and change them to: com.precise.javaperf.aop.jaxws.SOAPHandlerChainAspect. NOTE There are several such lines. Restart the application server. Some Precise for J2EE nodes in Smartlink flow display as an Additional Invocation instead of an Activity name In the SmartLink flow workspace area, all J2EE nodes in the tree are displayed as Additional Invocation instead of as the Activity name. The following exception message in the j2ee.fp.trc log file indicates a correlation problem between Smartlink and Precise for J2EE: Caught Exception in getLastSmartlinkTransferTimestamp() (QCID 7519) Solution: Workaround Restart the Precise for J2EE FocalPoint. Websphere MQErrors in Websphere MQ log after installing Insight Savvy for WebSphere MQ API Exit If a Websphere MQ application fails to connect to a Queue Manager (for example, the Queue Manager is down or shutting down or the communication channel is down), then Websphere MQ wrongly reports that Insight Savvy for WebSphere MQ API Exit is the root cause of these errors. The same connection attempt should fail without Insight Savvy for WebSphere MQ Exits installed, but in this case it will not be reported to the logs. The following error or similar Websphere MQ connection errors, with other reason codes such as 2063 and 2035, may occur in the Websphere MQ error log: ----- amqzuax0.c : 6207 --------------------------------- 10/14/2005 11:01:00 AMQ7216: An API Exit initialization function returned an error. EXPLANATION: The API Exit 'VeritasI3Agent' function 'PSI_EntryPoint' in the module 'C:\Program Files\VERITAS Software\products\insight\bin\psi_mq_exits.dll' ('/var/mqm/exits/psi_mq_exits' on UNIX) returned CompCode 2 and ReasonCode 2059. ACTION: Correct the problem with the API Exit 'VeritasI3Agent' ----- amqzuax0.c : 6207 ---------------------------------- Solution: Workaround Usually these errors can be ignored and should stop appearing once the root cause of the problem is solved. To validate this, verify that the root cause of the problem is solved (for example start the Queue Manager or the communication channel), and make sure no additional errors are written to the error log. If the problem persists, remove Insight Savvy for WebSphere MQ API Exit, and check if the applications can connect to Websphere MQ successfully with no errors reported in the Websphere MQ log. Field values trimmed or replaced by dots Display of non-English or binary characters in the 'MQ Reader application', 'MQ Writer application', and 'Message Body' fields is not supported by Insight Savvy for WebSphere MQ. As a result, in some cases the field value will be trimmed in the first non-English or binary character, while in others the non-English characters will be replaced by dots. (PR 906) Solution: None Setting the extended collection mode does not work correctly When setting the extended collection mode in the PMDB parameters, the agent starts to collect the message body texts. Though during the summary it does not add the message body texts of other messages and as a result only the message body text of the first message is shown. Solution: Workaround In the registry of the WebSphere MQ instance set the km_summarize to 0 so that no summary is performed. The registry file is located on the Precise FocalPoint server in the products\i3fp\registry\instances\MQ folder. Oracle Applicationslsof utility may consume too many resources The lsof utility which is activated by Insight Savvy for Oracle Applications and Insight Savvy for Network may consume many resources on a loaded server. lsof is executed every three to five minutes (configurable), so a runtime of one minute on a busy server is not acceptable. This problem is likely to occur on HP-UX servers. (PR 1081) Solution: Workaround To reduce the lsof resource consumption, add the following parameter to the <i3>/products/i3fp/registry/products/insight/<host-name>.xml file: Add the following parameter to the <oa> section: On HPUX, Solaris and Aix: <oa_lsof_command>./bin/lsof -a -c [process-name] -l -n -P | grep ESTABLISH | awk '{print $2,$9}' | uniq</oa_lsof_command> On Linux: <oa_lsof_command>./bin/lsof -a -c <process-name> -l -n -P | grep ESTABLISH | awk '{print $2,$8}' | uniq</oa_lsof_command> Replace [process-name] with the prefix of the program-name. Use “f45” for Oracle Applications 11.0 and “f60” for Oracle Applications 11i. NOTE To apply the change, all Insight Savvies should be restarted on the relevant server. OSI/O Bytes counter displays a value of 0 On Windows NT 4 and Linux, the I/O Bytes counter is not available. A value of “0” is displayed instead. Solution: None Load to the PMDB fails If Insight Savvy for OS is installed on a UNIX server and it monitors processes which include attributes with non-English characters, Insight Savvy for OS may send the process data to Insight FocalPoint without converting it to “UTF-8” encoding. Since Insight FocalPoint expects the OS data to be encoded in “UTF-8”, the data may fail to be loaded into the PMDB and the entire OS data file may be rejected and moved to the <i3_root>/products/insight/Focal Point/data_files/error directory. The following message will be written to the insight.focal.trc file: ID : SEVERE-3010229 Msg : An error occurred in the Insight FocalPoint Loader thread. root cause: sun.io.MalformedInputException at sun.io.ByteToCharUTF8.convert(ByteToCharUTF8.java(Compiled Code)) […stack trace continues] (PR 1887) Solution: Workaround In the <i3_root>/products/i3fp/registry/products/insight directory on your Precise FocalPoint server, edit the focalpoint.xml file and add the following line under the <input-encoding> section: <OS_[Server-Name]> [Encoding-Name]</ OS_[Server-Name]> Where: [Server Name] is the monitored server name as known by Precise. [Encoding -Name] is the name of the encoding used by Insight Savvy. For example: Add the following line to the focalpoint.xml file: <OS_mhstbs1>ISO-8859-1</OS_mhstbs1> Restart Insight FocalPoint. Insight Savvy for OS may fail to sample process information on Windows The Insight Savvy for OS may fail to sample process information on Windows and add the following message to the Savvy for OS log file (insight.agents.server.log): ID: WARNING-3000358 Msg: Unable to poll processes information for current sample. (PR 1876) Solution: None The Insight Savvy for OS fails on some HP-UX servers The Insight Savvy for OS may fail on certain HP-UX 11.11 or 11.23 servers with the following error in the log: OSLib: pstat_getprocessor failed In such case a kernel patch needs to be installed on the server. (PR1052029) Solution: Patch Use the relevant patch for the HP-UX kernel: HP-UX 11.11 - PHKL_24582 HP-UX 11.23 - PHKL_34912 Known Problems in the Insight Savvy for OS Savvy for OS over time graphs, compare by CPU %, summary level Monthly, with drill to program - produces error Savvy for OS over time graphs, compare by server load, with drill to server group - displays "N/A" values Savvy for OS over time graphs, compare by concurrent threads, with drill to server group - displays an error message (PR963305, PR963309, PR1065537) Solution: None Insight Savvy for OS shows incorrect Citrix Client Latency values Insight Savvy for OS shows bloated values of Client Latency for Citrix Applications, Users and Clients. (PR1059330) Solution: None Add-on AppTier (SDK)64-bit ARM library does not send data to AIX 5.3 servers The 64-bit ARM library fails to send correct data to the ARM Collector on AIX 5.3. This can be verified by running the `oslevel -g` command on the server. There is a problem with version 5.3.0.10 of bos.rte fileset. The required version is 5.3.0.30. Solution: Patch Upgrade the bos.rte fileset with a patch to version 5.2.0.30. The SDK AppTier help windows are empty When asking for help on an SDK AppTier in Insight, an empty help window is shown. (PR1059133) Solution: Workaround Open the Insight user guide and go to SDK section. Add-on AppTier name does not appear in graph titles in Cross-AppTiers station All graphs in the Cross-AppTiers station should be titled with the name of the AppTier followed by the name of the entity. Graphs of the Add-on AppTier are only titled with the name of the entity. (PR1058956) Solution: None Hidden errors in plug-in of an Add-on AppTier may cause Insight to stop loading files from all technologies It is possible that plug-in of an Add-on is not successful, yet no error message is be displayed to the user. In some occasions, these hidden errors can cause Insight to stop loading files from all technologies to the PMDB. (PR1059806) Solution: Workaround Go to the products/insight/FocalPoint/etc folder under the <i3_root> directory where the Insight FocalPoint is installed. Look for the loadFieldsInfo_XX.dat file where XX is the technology code of the Add-on Apptier Rename this file, changing its extension to something other than .dat. Restart the Insight FocalPoint. Send the corrupted loadFieldsInfo file to Precise Enterprise Support. vCenter Server agentvCenter Server agent does not show statistics after installation vCenter Server agent does not show statistics after installation. Agent runs with no verification errors. Solution: Workaround Stop all insight agents running on the same machine and start them again. The vCenter may need a second restart. If the machine is the insight FP server then insight FP should be restarted as well. Precise for WebMonitoring Siebel on IISTransaction names not collected Transaction names (Siebel views) are not collected in HTTP POST transactions. Defect 17106 Solution: None Siebel on IIS - Missing application user-name and business transaction name When monitoring Siebel on IIS with p4web, Application user is missing from the GUI.In addition, transactions do not have a clear business name, they just show up as URLs. This is because post parameters on the Siebel virtual directories are not monitored by p4web. Solution: Workaround To monitor (collect) post parameters on your Siebel applications do the following: Identify which “virtual directories” (application's prefix - /callcenter_enu), you want to monitor: Launch the IIS manager GUI. Expand the tree of the web site called Default web site to view all the “virtual directories”. For each "virtual directories" you want to monitor do the following: Open a command line window. cd to <precise-root> on the monitored IIS machine. Run the following: products\www\install\psww_insert_isapi_handlers.bat <precise_root> <instance_id> <site_name>/<virtual-directory-name> add-handler PreciseWebHandler <filter_path> where: <filter_path> is the path to the instance bin folder concatenated with "IISFilter", without the _32.dll or _64.dll extension. For example, c:\precise\products\www\instances\2643\bin\IISFilter. Example for running the command: products\www\install\psww_insert_isapi_handlers.bat E:\Nightly95 2807 "Default Web Site/callcenter_ENU" add-handler PreciseWebHandler E:\Nightly95\products\www\instances\2807\bin\IISFilter. FocalPointAvailability sampling fails If a Web server cannot be accessed using the server name of an instance, availability is not sampled. Solution: Workaround Edit the instance via the UI and change the IP in the Domain Information screen to be an IP/domain-name that is accessible from the Precise for Web FocalPoint machine. This can also be done by modifying the network element in the instance registry file: products\i3fp\registry\instances\www\<server-name>\<instance-name>.xml on the Precise FocalPoint machine. Open AdminPoint and restart the Precise for Web FocalPoint. No availability on Windows 2008 when firewall is on When the Windows 2008 firewall is running, there may not be availability data in the Precise for Web GUI. (IN 34708) Solution: Workaround Open the firewall for dataloader availability requests. UIData of non-Latin language may be displayed as ???? Data of non-Latin language may be displayed as ????. (PR1004841) Solution: Workaround To change the regional settings of the browser's machine: Open the settings dialog: Start>Settings>Control Panel>Regional and Language Options. Click the Advanced tab and select the correct language (Hebrew/Turkish/Japanese) in the drop-down menu to be the language of non-Unicode programs. Restart the machine. Web CollectorCollection of application user does not support special characters When a monitored application uses special characters for user login (e.g. '#', '$'), the collected application user appears distorted in the Precise for Web GUI. Precise for Web does however support an email address as the user login. (QCID 8135) Solution: None Web agents are not refreshed when changes in Locations are done while the PMDB is down. If you make changes to Locations definitions (add/delete/change) while the PMDB is down, the Precise for Web agents will not be aware of the changes. (QCID 6720) Solution: Workaround Restart the PMDB and then restart all Precise for Web agents on all servers. Error on page message on the browser window When the monitored Web application pages are composed of VBscripts only (no JavaScripts), a collision between the injected Precise JavaScript and the application VBScripts may occur. This is reflected by an "error on page" message on the browser window. The reason - when the JavaScript is executed in Internet Explorer before the VBScript, the VBScript is not executed properly. (PR1006333) Solution: Workaround To inject an empty VBScript (with the Dynamic Instrumentation Agent) before the Precise JavaScript: Edit the <i3_root>\products\www\instances\<instance-name>\etc\ClientS1_dynamic.tpl file and add the following text at its beginning: <script language="VBScript"></script> Stop the Dynamic Instrumentation Agent. Start The Dynamic Instrumentation Agent AIX OS: no server-side data in Activity workspace The following conditions must hold: Server-side agent installed on AIX OS. Port 20999 is occupied by a third party application. Server-side agent status is OK. Web Collector was restarted, either directly or as a result of an all Web agents restart. There is no server-side data in the Activity workspace. (IN 35367) Solution: Workaround The following workaround must be applied each time the Web Collector is restarted, either directly or as a result of an all Web agents restart: On the Precise FocalPoint machine, edit the instance registry file: products\i3fp\registry\instances\www\[hostname]\[instance-name].xml Search for the <report-thread-long-reconnect-timeout-ms> tag. If it exists, increase the value inside the tag by one. If it does not exist, add it under the <server-side> section, like this: <report-thread-long-reconnect-timeout-ms>15001</report-thread-long-reconnect-timeout-ms> Save the file. Restart the server-side agent from the AdminPoint. SmartLink for PeopleSoft is not working well under heavy load The Collector is closing and signing TAC files before the end of the time-slice and before the URL and Page data files are written and signed. As a result, the Dataloader fails to load the TAC data. (IN 35825) Solution: Workaround Add the following to the PeopleSoft server machine registry file found in the directory: products\i3fp\registry\products\indepth-web under the root element: <web-collector> <smartlink> <tac> <server-side> <max-recs-in-file>10000000</max-recs-in-file> </server-side> <client-side> <max-recs-in-file>10000000</max-recs-in-file> </client-side> </tac> </smartlink> </web-collector> Data CollectionCorrelation does not work for PeopleSoft In a PeopleSoft installation, client server correlation does not work for extensions that contain an “_”. Solution: Workaround Edit the <i3_root>\products\www\filter\etc\pssExtTable.bin file and add the extensions that are used in the monitored application as “page” extensions. If the extension includes an underscore, the extension must be wrapped in the @-character, for example, @abc_efg@. Network data is not shown IIS provides you with a way to define two different Web sites on the same server with the same port (using Host-Headers). Precise for Web can be installed for each of these Web sites separately, but the Network agent cannot collect network data for each of these Web sites separately. Therefore- when adding Precise for Web instances in this configuration, the option "Associate this information with information about network performance" should not be selected. (PR999518) Solution: None Access Count Number field is empty when monitoring WebSphere 6.1 This occurs because WebSphere statistics collection is disabled. To enable it, follow the instructions below. (IN 34522) Solution: Workaround To enable statistics collection on WebSphere 6.1 and higher: Open the URL: ibm/console/login.do Expand the task Monitoring and Tuning and click Performance Monitoring Infrastructure (PMI) On the right pane, click the server you want to enable statistics on (for example, server1) On the screen displayed, select the Runtime tab On the Runtime tab, select the Basic, Extended, All, or Custom statistic set NOTE If you choose the custom statistics set, make sure the CommittedCount counter under Transaction Manager is enabled. Click Apply and OK. PeopleSoft monitoringNo activity appears in IIS when Web collector is restarted manually When Precise for Web monitors a PeopleSoft instance, and post parameters are defined to be collected, PeopleSoft users fail to log in through the PeopleSoft login page, even when the username and password are correct. This occurs because PeopleSoft doesn’t take the post parameters in a standard J2EE way, so while Precise for Web can access these parameters, the PeopleSoft application cannot, resulting in unsuccessful login attempts. (QC ID 3609) Solution: Workaround Disable the post parameters collection for the PeopleSoft instance: Change the following registry key (both in main & proxy) to false: <collect-post-parameters>false</collect-post-parameters> The key is located at the following two locations: products\i3fp\main\registry_proxies\<any proxy>\registry\instances\www\<server>\<instance>\settings.xml and: products\i3fp\registry\instances\www\<server>\<instance>\settings.xml From AdminPoint, restart the Web server-side agent. ProcessesNo client-server correlation for several pages Several pages do not have client-server correlation in SmartLink and Web because the URL seen by the browser and by the Web server is different. (QC ID 5970) For example: Client side sees http://rnd-nightly-8:7565/bank/ Server side sees /bank/index.jsp Solution: Workaround The browser activity needs to be done with the exact URL (http://rnd-nightly-8:7565/bank/index.jsp for the example above). Page/URL size is always 0 in the Precise for Web Activity workspace of a monitored IIS 7 Precise for Web collects page size as part of its statistics on a monitored server. Only on IIS 7, this field is always collected as 0. (IN 35500) Solution: None Data not shown OS data for Linux is not displayed in the Statistics workspace. Solution: None NOTE OS statistics are not supported on Linux. Cache Hit Ratio is 0.00% on IIS 7 Cache Hit Ratio is 0.00% on IIS 7 (IN 35462) Solution: workaround Do the following: On the IIS machine, open the file products\www\instances\xxxx\etc\Counters.conf (while xxxx is the instance id) Change \Internet Information Services Global\URI Cache Hits %;AVG to \Web Service Cache\URI Cache Hits %;AVG Web Server AgentAbnormal behavior in Maximo application In Maximo© you can define a link which can run a program on the client computer. If Precise for Web is installed, you will have to click the link twice (instead of once) to activate the program. Solution: None Installation of two instances fails You cannot install two instances (different Precise environments) on the same Web site. Solution: None Abandoned pages may not correlate with Network activity Heavy URLs that are abandoned while loading may not correlate with their network activity. Solution: None Precise for Web connection attempts show as severe in the logs The connection attempts of the instance statistics process to the Web server are shown as severe in the events viewer. Solution: None Correlation does not work properly on cached URLs The 304 status (page not modified) on any Web server will cause an invalid correlation for this page. Solution: None Correlation on IIS does not work if logging is disabled The Web server agent does not do any monitoring if the Log Visits check box is not selected on the Home Directory tab of the Properties page in IIS. Solution: None An extra instance statistics JVM runs on UNIX If on UNIX you start an agent that is already started (from AdminPoint), another instance statistics Java process is created and the old one is not killed. Potentially, this may cause many redundant processes to be running for no reason. (PR1059846) Solution: Workaround Always stop an agent before starting it again. IIS 6 fails to load more than 50 ISAP filters When installing a large amount of Precise for Web instances on a single application pool (the exact number depends on the physical server configuration) on IIS 6, some of the Web sites may fail to load. (PR811480) Solution: Workaround Configure the ISS Web sites to work on different application pools (no more than 10 Web sites per application pool). No statistics on IBM HTTP Server 1.3 The IHS1.3.x (Apache 1.3.x) httpd.conf file does not include the following details for the statistics collection: No "ExtendedStatus On/Off" entry No mod_status.so file in the modules directory (PR1025929) Solution: Workaround Copy the mod_status.so file to the <IHS_Root>/modules directory. The mod_status.so file may be copied from a regular Apache 1.3 installation (<Apache13_Root>/modules) Add the following line to the httpd.conf file: LoadModule status_module modules/mod_status.so Add the following line to the httpd.conf file below the line LoadModule status_module: ExtendedStatus On Dynamic InstrumentationWebsphere Dynamic Instrumentation agent fails to start/stop In case action items for modifying the web.xml file are performed manually for Websphere’s Dynamic Instrumentation agent, it fails to start/stop. (IN 35470) Solution: Workaround Use psww_web_desc_modify utility to modify the web.xml file, as specified in action items. Dynamic Instrumentation breaks/changes Web page layout In certain cases, Dynamic Instrumentation breaks/changes Web page layout (or blank pages). Solution: Workaround Change the configuration as follows: Open the page source. If it has the tag <meta http-equiv="X-UA-Compatible" ... >, the solution is to change the script location inside the page: If every page contains the compatibility tag (Sharepoint or other sites based on template), change the web instance registry as follows: Restart the Web DI agent. Refresh the page and test again. If only a few pages contain the tag, change the web instance registry as follows: Use "script" + "before" to insert before first script. or Use "/head" + "before" to insert at the end of head section Restart the Web DI agent. Refresh the page and test again GeneralServer-network correlation is low in Precise for Web activity on SAP Netweaver Web servers When installing a Precise for Web instance on SAP J2EE, network data is available but the server-network correlation is low. (QCID 5579) Solution: None No Client-Server correlation There is no Client-Server correlation when installing an instance where the Web client collector, Web FocalPoint, and the instance are on the same machine. (QCID 6704) Solution: Workaround Follow these steps: Change the <aging-period-when-using-client-collector> in the server registry file to 90000. Restart the Web FocalPoint. No network data in Precise for Web when using "user-defined client-ip" Network data correlation is based on IP matching. Because the IP is changed in Web – to the custom collected one, the correlation breaks. This means the network time from the Load-balancer/Proxy to the Web server machine will not be seen in the Web user interface. (QCID 5473) Solution: None Partial statistics data in Precise for Web when IIS site name contains local (non-ascii) characters When the IIS site name (what we call real name) contains non-ascii characters - for example "Site Web per default"- IIS statistics works partially. The web.instancestatistics log for that instance shows this error: ID : SEVERE-6561422 Time : 2009-03-17 16:06:47.108 GMT+0200 i3 Time : 2009-03-17 14:06:47.108 GMT Process : psww_iis_instance_stat.exe 1085 Site-Web-default (ID:3676) Src : IISInstanceStat.c:1910 Logger : web.instancestatistics.qapifa16fr Msg : PDH_AddCounter failed for some counters (increase log level and restart agent to see details). When you increase the log level, you see the actual problem (attached): ID : FINE-6561414 Time : 2009-03-17 17:46:04.116 GMT+0200 i3 Time : 2009-03-17 15:46:04.116 GMT Process : psww_iis_instance_stat.exe 1085 Site-Web-default (ID:3236) Src : IISInstanceStat.c:1886 Logger : web.instancestatistics.qapifa16fr Msg : PDH_AddCounter failed for Counter '\\QAPIFA16FR\Web Service(Site Web par d?©faut)\Total Method Requests' Type 'SUM'. (QCID 3917) Solution: None Static instrumentation instances do not report client data after changing client collector port If you have a monitored instance with manual static instrumentation, it will not be updated with the new client collector configuration (new port number or host name). (IN 33022) Solution: Workaround If some of the instances are configured to use manual instrumentation, run the following script on the monitored machine, after changing the client collector configuration: Log on to each of those servers and perform the following step: Change directory to the <i3_root> on the monitored machine. Windows: Run the following script: products\www\scriptinstaller\standalone\bin\psww_instrumentation_update_<instance name>.bat UNIX: Set execute permission to: products/www/scriptinstaller/standalone/bin/psww_instrumentation_update_<ins tance name>.sh Run the following script: products/www/scriptinstaller/standalone/bin/psww_instrumentation_update_<ins tance name>.sh Precise for SAPUINo database server indication shown When the database server has more than one logical name, the table in the Association area of the Statistics workspace may not display a check mark to the left of the database server. Solution: Workaround Update the dbServer field in the pw_spid_ai_sids table with the logical name of the application server. SAP JCO RFC error in batch workspace The Precise for SAP Batch workspace will display an RFC error instead of current active batches when SAP JCO 3.0 or later is used. Solution: Workaround Use a SAP JCO version 2.1.3 - 2.1.8 (check the SAP Website). Zero transaction count in Activity workspace In the Activity workspace of Precise for SAP, the number of transactions column may contain a zero value for a transaction. This occurs when a single SAP transaction is executed and has not finished when Precise for SAP monitors it (the transaction is split between the times it was monitored). In such a case, the number of transactions should be considered as one (instead of zero). Solution: None Data of non-Latin language may be displayed as ???? Data of non-Latin language may be displayed as ????. (PR1004841) Solution: Workaround To change the regional settings of the browser's machine: Open the settings dialog: Start>Settings>Control Panel>Regional and Language Options. Click the Advance tab and select the correct language (Hebrew/Turkish/Japanese) in the drop down menu to be the language of non-Unicode programs. Restart the machine. SAP Agentdev_rfc.trc log file in the <i3_root> directory When the Precise for SAP agents are running, a log file with the name “dev_rfc.trc” is created in the <i3_root> directory. This is a log file of the SAP RFC (the interface used by the agents to connect to SAP). Solution: Workaround To place the file in a different directory, a system environment variable should be added to the windows system where the Precise for SAP agents are running. The name of the variable is RFC_TRACE_DIR and the value should be the directory in which this file should be located. After the variable is defined, restart the Precise for SAP agents. Workload CollectorServer statistics of standalone database not shown Precise for SAP does not display server statistics for a SAP standalone database server. Solution: None Availability CollectorOrganization availability data not shown After a new SAP instance has been installed, no availability data is collected for the Organization entity during the first day because the PMDB process “Organizational Mapping” automatically runs for the first time only at 11:00 PM. Solution: Workaround Make sure organizations and locales have been set up. Then use AdminPoint to run the Organizational Mapping process manually. Locale CollectorCannot see the target device On the Locale Availability tab in the Locale Settings - Edit dialog box, if the name of a host cannot be translated to an IP address on the server that the ping is made from, the target device is ignored. Solution: Workaround If available, enter the IP address instead of the name. GeneralIn the step list, the filter on the transaction name is not working and the filter on program and user is always in upper case. When viewing the step list in the Activity workspace, you cannot filter the step list by transaction name. Additionally, filtering on program and user is only possible for upper case values. (IN 35539) Solution: Workaround Optional solutions: Before going to the step list view, drill on the relevant transaction in the transaction list. It will give you the same step list result as filtering by transaction. Use the work type list box in the screen header to filter the step by the relevant work type. If you need to filter on a program or user that contains upper and lower case characters, use the like option with a value that contains the upper case or special characters. LaunchingLaunch from Precise for SAP running jobs to the Current workspace in Precise for Oracle fails when the db server name is different. Launch from Precise for SAP running jobs to the Current workspace in Precise for Oracle can fail when the SAP database server name in Precise for SAP is different than the database server name in Precise for Oracle. When you install SAP instance, Precise for SAP installer automatically detect the SAP database server and uses this name for launching to Precise for Oracle. If the Precise BA was installed on this server with a different name (different from the name Precise for SAP installer detects) the launch from Precise for SAP running jobs to the Current workspace in Precise for Oracle can fail and Precise for Oracle may show data of another instance. For example: DB server in SAP AppTier is "db-server-name" and Precise BA is installed on this server with another name like "db-server-name.precise.com" or using IP. (IN 35785) Solution: Workaround There are two optional workarounds: Reinstall the Precise BA and the Precise for Oracle collector with the db server name in Precise for SAP. Update filed SPID_SERVER in table PW_SPID_AI_SIDS for the relevant SAP instance. Set the value to the server name in Precise BA. Precise for BWFederationLaunch to framework node from mail does not work When launching from an email to the Precise for BW user interface, no information is shown. This happens only in framework node installations when launching to an instance in a proxy environment. (QCID 7965) Solution: None When launching to Precise for BW in a federated installation (Framework node), using the node's URL generates an error message displayed after the login page When launching to Precise for BW in a federated installation (Framework node), using the node's URL will generate the following error message displayed after the login page: There was an unhandled failure on the server. Com.precise.nuf.services.serviceInvockationException QCID 7371) Solution: Workaround Use the URL to the Main node to login to Precise for BW or use the link from the StartPoint. GeneralMTE names in CCMS Alerts screen do not match RZ20 transaction In the CCMS Alerts MTE tree, names are sometimes shown that do not match the names shown in the RZ20 transaction. This happens only when the RZ20 transaction displays the "Context" property of the MTE instead of the "Short Name" property. (QCID 7668) Solution: None Precise for OracleFocalPointError indicating the HOME Environment variable is not set If the environment variables for the user of the Precise Listener process does not contain HOME, the Precise for Oracle FocalPoint and the Precise for Oracle Collector will not start. The “The HOME environment variable is not set. Please set it to your home directory.” error is shown. (PR997690) Solution: Workaround Stop the Precise Listener process. Set the HOME variable to the login folder. Restart the Precise Listener process. Restart the Precise for Oracle FocalPoint and the Precise for Oracle Collector. Every 4 hours, the Collect Instance Statistics process times out after one minute The Collect Instance Statistics process also collects segment statistics from V$SEGSTAT. There is an open SR (6378216.99) in Oracle, which describes the case where simple queries on the above table perform full table scans on certain installations (until now it was reproduced only on Oracle 10.1.0.2). because this query runs by default every 4 hours, it may result in loading the customer's production environment. Therefore we added a timeout mechanism which stops the process if the query runs over 60 seconds. As a result, the Collect Instance Statistics process reports the following error every 4 hours: ERROR: v$segstat query caused timeout exception, consider disabling segment statistics collection and contact Indepth/Oracle support. (PR1081167) Solution: Workaround You may either disable segment statistics collection for that specific instance or upgrade the oracle instance. To disable segments collection perform the following steps: Go to AdminPoint. Click warehouse processes. Click the 'collect instance statistics' process. Click parameters. Click the instance in error. Click Edit. Change the value of 'Collect segment statistics every:' text box from 4 to 0. Statements do not appear after the time slice If the "Top-N Statements" default is changed and set to be high (around 2000) the statements are loaded and seen only after the next time slice. Note that the higher the number, the higher the risk of occurrence. (ID 31285) Solution: Workaround Follow these steps: In the AdminPoint click Warehouse>Processes Select the Oracle "Load Performance Data" process and press the Scheduling button. Select the relevant instance(s) and change the number to "4" before the text "minutes after the slice". Precise for Oracle FocalPoint After Roles/Permissions definition, it can take up to 10 minutes (in federated environments) for the Precise for Oracle FocalPoint to be refreshed with the new definitions. Meanwhile, the user may not see his changes in Precise for Oracle or may not see any instances in the Dashboard (only the "All" row is displayed). (IN 35170) Solution: By design Wait 10 minutes for the refresh process to complete. UIRun statement with captured bind set that was not fully captured is not available When a bind set was captured, but not all of the bind values were captured, Running the statement by choosing this bind set will fail. (QCID 4901) Solution: None Error message when you use the Statistics/Usage tab in the cluster key entity In the Objects workspace, drill down from Cluster to Cluster key, click the Statistics/Usage tab in the Association area and an error message is displayed. (PR 15) Solution: None SQL text longer than 64kb gives error message In the SQL workspace, trying to edit a statement in SQL that is longer than 64kb will result in an error message. (PR542) Solution: None Table name contains ~ In the What If workspace, click on add table and choose user. The table appears as RNF. (PR 731) Solution: None Owner appears twice When you associate to Triggers in the Objects workspace, the owner appears twice. (PR 1395) Solution: None SQL text contains Chinese characters The Parser does not support Chinese characters. This means that you can't add a statement in SQL and extended collection containing Chinese characters. (PR 1399 + 1401 + 1554) Solution: None Add dialog box in What-If only shows first 50 items When in the What If workspace you use the ADD action, the combo-boxes are limited to 50. (PR 798) Solution: None Statement that has been run returns no row When the Run Statements dialog box in the SQL workspace is used to run a statement that uses a blank as a bind variable, no rows are returned. (PR 5548) Solution: None Alias is not collected When the text of a statement that is imported from a file in the Activity workspace contains an alias to the object, the alias is not collected. (PR 7195) Solution: None Negative values in the Statistics workspace Statistics workspace: Negative values may be displayed for Oracle statistics that are not cumulative, such as the current login statistics. (PR 8953) Solution: None Statement text format may fail Loaded statements may not always be formatted (such as INSERT ALL statements). (PR 9433) Solution: None File size and Max Blocks are 0 The maximum file size and maximum number of blocks for Oracle datafiles is 0 when a file is not auto extensible. Solution: Workaround This is a known problem, but not a real “bug”. If a file is not auto extensible, there is no meaning for the values of maximum file size and maximum number of blocks. No data for LOBs when you drilldown from objects When you try to drilldown from objects and the object type is LOB, the result shows no data. Solution: None CPU utilization seems abnormal The statistics for the CPU utilization graph in the Statistics workspace > Operation System statistics are gathered from v$osstat. Due to a known Oracle bug (5104594) the view may contain invalid data. (PR 4439) Solution: None Wait events count and time grow inconsistently in the Statistics workspace In the Block Contention association view, there are situations in which the Waits column shows a number while the Time Waited column does not. This is an accepted behavior by Oracle because the two metrics are not calculated at the same time (Waits are counted at the beginning of a wait and Time Waited at the end) and Time Waited is summed to 1/100 of a second only. (PR 4215) Solution: None Wrong object name is displayed in Execution Plans/All Changes In the SQL workspace, when choosing the Executions Plans in the tree, you can see a list of schema changes in the details area. When checking Index changes, you will see that a wrong name is displayed. (PR 4683) Solution: None Data of non-Latin language may be displayed as ???? Data of non-Latin language may be displayed as ????. (PR1004841) Solution: Workaround To change the regional settings of the browser's server: Open the settings dialog: Start>Settings>Control Panel>Regional and Language Options. Click the Advance tab and select the correct language (Hebrew/Turkish/Japanese) in the drop down menu to be the language of non-Unicode programs. Restart the server. Script Error after sorting in the Activity workspace In the Tree View of the Activity workspace, trying to sort a table on the right side (by clicking its header) may cause a script error. (PR1084543) Solution: None The number of Statements in the Activity Workspace is less than expected The number of Statements is not displaying correctly in the GUI (The default number is 50 but 50 are not displayed). Additionally, changes in the number of Statements are not correctly displayed. (ID 1268328) Solution: Workaround This issue can occur due to re-installation of the collector when the user has not deleted the existing data collector folder. To solve the problem please follow these steps: Connect to the PMDB using SQL*Plus as the Precise schema owner and execute the following: Select 'update PW_orsn_statement_info t1 set orsn_instance_id=' || inst_new.ince_id || ' where orsn_instance_id=' || inst_del.ince_id || ' and orsn_shv not in (select orsn_shv from pw_orsn_statement_info where orsn_instance_id = ' || inst_new.ince_id || ');' from PS_INCE_INSTANCE inst_del, PS_INCE_INSTANCE inst_new where inst_del.ince_deleted = 'T' and inst_new.ince_deleted = 'F' and inst_del.ince_name = inst_new.ince_name and inst_del.INCE_INSM_ID = inst_new.INCE_INSM_ID If this query returns rows, each line represents an update statement that fixes a specific instance. Execute each row as an SQL statement to fix all instances. Don't forget to commit. To verify that the issue was fixed, navigate to the Activity workpsace and look for the missing statements. You should be able to see them now. Wrong error message when no permission to run recommend When the user presses the recommend button without having permissions to run Recommend/Explain, the wrong error message is displayed. ORA-01405 is displayed instead of a clear message stating the user does not have permissions to run recommend. (IN 33851) Solution: None TunerTable/index statistics dropped during recommend operation Existing statistics may be dropped when performing an "Index recommendation". Relevant cases are tables that are part of the evaluated statements when: The table has no statistics and indexes have. The table is a partitioned table and statistics were never gathered on the table level, but only at some of the partition levels. (PR1015663, PR1015626) Solution: Workaround Set two parameters which affect the way "Index recommendation" works: Set "gather-miss-stats" to N Set "del-gathered-stats" to N All parameters can be changed in the <i3fp directory on the Indepth for Oracle FocalPoint machine>\products\i3fp\registry\products\indepth-oracle\<HOST>\oracle\<SID>.xml file. After the change, the file's content should look like this: <oracle SID name goes here> <recommend> <gather-miss-stats>N</gather-miss-stats> <del-gathered-stats>N</del-gathered-stats> Restart the Precise for Oracle FocalPoint. CollectorAfter upgrade or after stopping the Collector, Interpoint for Oracle Applications does not start again After upgrade and sometimes after stopping the Collector, the Interpoint for Oracle Applications fails to start (the status in AdminPoint shows error on the Collector) and a message in the oracle.collector.trc log file states that the server is already up. This means there is another active Interpoint for Oracle Applications on the same instance. There is no possibility to activate more than one Interpoint for Oracle Applications at a time. If the Interpoint for Oracle Applications process has been terminated unexpectedly, its session is still active, and will not allow a newer Interpoint for Oracle Applications to start. (QCID 7374) Solution: Workaround If there is no other Interpoint for Oracle Applications running, you need to kill the Interpoint for Oracle Applications session manually using the alter system kill session command. To detect the session ID and serial number: Select sid, serial# from v$session where client_info like '%VERITAS%' To kill the session with the previous statement sid, serial#: Alter system kill session '<sid>, <serial#>' Host user name is truncated to 15 Bytes The Host user name is cut off after 15 characters in the Activity workspace (host users) and Current workspace (session overview). (PR 563) Solution: None SQL texts for statements activated by dbms_job.submit are not collected in Oracle 10g Due to an Oracle bug in Oracle 10g, the v$session table is not updated with the SQL text information when the statement is activated by the dbms_job.submit function. As a result, the text is not presented in Precise for Oracle because the Collector samples the text information according to the information that exists in the v$session table. (PR 4444) Solution: None SQL text is not formatted in the Current workspace In the Current workspace the SQL text is not formatted. Therefore the statements that contain carriage return characters in the original text will be displayed in the Association area as if they are cut after the end of the first line. (PR 4610) Solution: By design There are three ways to see the complete text: Roll the mouse over the text and see it in the ToolTip. Drill to the statement and view its text in the Main area. Wait until the statement is loaded into the PMDB and see it in the Activity workspace. Current workspace does not work after restarting Precise for Oracle or its Listener The Current workspace displays an error message after Precise for Oracle is restarted or after starting the Listener agent. This error is caused by a premature manual start of the Collector that causes an internal conflict between the manual start processes and the automatic restart process. When one of the processes updates the precise.init file, the other overwrites it with partial data. This file is essential for the operation of the Current workspace. (PR 5305) Solution: Workaround Stop the Collector. Make sure that there are no further restart processes (ps -ef | grep pss_restart.sh). If such processes exist, -kill them and restart the Collector. Extended collection crashes with a Windows alert The Extended collection in a Windows environment crashes and reports missing DLL files in TrcStart.exe. (PR5577) Solution: Workaround Add the <i3_root>/infra/bin directory to the PATH environment variable to the monitored host for the administrator user. Log on to the Windows server with the user with administrator rights. Open the Control Panel. Double click the "System" icon. Click the "Advanced" tab. Click "Environment Variables". Choose the "PATH" variable under the "System Variables" section. Click "Edit" of System Variables section. At the end of the current PATH value, add a semicolon and afterwards the <i3_root>/infra/bin folder. Cannot create Extended collection due to crontab problems When creating a new Extended collection you can get the following error message: Cannot schedule extended collection due to crontab problems. please check that Precise/i3 Unix user can create crontab This may even happen when you use a user that has the permissions for using crontab. This problem occurs if the Precise UNIX user doesn't have an existing crontab file (this means that the user has never activated a crontab operation). (PR5738) Solution: Workaround To create a crontab file for the Precise UNIX user: In a UNIX session of the Precise user, run the following commands: touch empty_crontab.txt crontab empty_crontab.txt rm empty_crontab.txt The Precise for Oracle Collector installed on a Windows platform does not start The Precise for Oracle Collector installed on a Windows platform fails to start because the PerfProc performance counters are not enabled. The following error message is written to the log file: Msg : Invalid performance data (object 232). (PR6057) Solution: By design NOTE You can enable and disable the PerfProc performance counters by using the Extensible Counter List utility. This utility is distributed through the Windows Resource Kit and is also available for download at http://www.microsoft.com/windows2000/techinfo/reskit/tools/existing/exctrlst-o.asp. You see more objects than expected when running a statement from dba_segments/extents If Oracle is configured with local management tablespaces and if you access the dba_segments/extents view, you will see more objects than expected. When we access the db_segments/extents, Oracle performs I/O operations for all tables located on the local management tablespaces. Oracle retrieves all extent’s table information from the header of the table. (PR913958) Solution: By design Error indicating the HOME Environment variable is not set If the environment variables for the user of the Precise Listener process does not contain HOME, the Precise for Oracle FocalPoint and the Precise for Oracle Collector will not start. The “The HOME environment variable is not set. Please set it to your home directory.” error is shown. (PR997690) Solution: Workaround Stop the Precise Listener process. Set the HOME variable to the login folder. Restart the Precise Listener process. Restart the Precise for Oracle FocalPoint and the Precise for Oracle Collector. Form names in Japanese are not shown Precise for Oracle Applications adds the ability to see which database activity relates to which ERP form. When the form name is in Japanese, Precise for Oracle Applications is not able to display the name. (PR847892) Solution: Workaround Stop the Precise for Oracle Collector and Oracle Applications agents. Connect to the Precise for Oracle schema (usually precise_i3_orcl) on the monitored server using SQL*Plus and run the following statement: create or replace view fnd_form as select form.application_id application_id, form.form_name form_name, form_tl.form_id form_id, 'US' language, form_tl.user_form_name user_form_name from fnd_form_syn form, fnd_form_tl_syn form_tl where form.form_id = form_tl.form_id and form.application_id = form_tl.application_id and form_tl.language = 'JA' union select form.application_id application_id, form.form_name form_name, form_tl.form_id form_id, 'US' language, form_tl.user_form_name user_form_name from fnd_form_syn form, fnd_form_tl_syn form_tl where form.form_id = form_tl.form_id and form.application_id = form_tl.application_id and form_tl.language = 'US' and not exists (select 1 from fnd_form_tl_syn form_tl_in where form_tl_in.application_id = form.application_id and form_tl_in.form_id = form.form_id and form_tl_in.language = 'JA') Restart the Precise for Oracle Collector and Oracle Applications agents. Object monitoring reports on partitions for Undo I/O and Temp I/O In the Activity workspace, associating with Objects may display Temporary I/O and Undo I/O with partitions. This has no affect on the functionality and the partitioning indication can be ignored. (PR1039817) Solution: None Object monitoring displays object ID details and not object names When new objects are caught by the Collector, they may appear in the Activity workspace in Objects with their object ID and not with their name and owner. This is due to the fact that the Object IDs file is not updated (used to be updated every 12 hours). Solution: Workaround Restart the Collector, this will update the Object IDs file. Oracle collector does not start on 64-bit Windows environment Oracle collector does not start on 64-bit Windows operating system (OS). Receive the following error in the oracle.collector.trc log file: ======================================== ID : SEVERE-5002250 Time : 2008-12-03 14:58:09.958 GMT+0200 i3 Time : 2008-12-03 12:57:39.573 GMT Process : pss_start.9i.exe ORA10204 (ID:4372) Src : sql.c:1324 Logger : com.precise.oracle.collector.log.chagall02.OR.ORA10204 Msg : SQL error occured, see the following messages: ORA-06413: Connection not open. ======================================== Occurs when the Precise Listener is installed in a directory that contain parenthesis. For example: "C:\Program Files (x86)\..." (IN 34467) Solution: Workaround Uninstall the Oracle collector and the Precise Listener, Relocate the Precise Listener to a directory without any parenthesis in the path. See also: http://forums.microsoft.com/MSDN/ShowPost.aspx?siteid=1&PostID=159581 Oracle collector does not start on Windows environment Oracle collector does not start on Windows operating system (OS). Receive the following error in the oracle.collector.trc log file: ======================================== ID : SEVERE-5002411 Time : 2008-11-03 19:29:48.078 GMT+0200 i3 Time : 2008-11-03 17:29:48.078 GMT Process : ORACLE.EXE ORCL (ID:392) Src : kernelNt.c:180 Logger : com.precise.oracle.collector.log.w-yochai-u.precise.com.OR.ORCL Msg : Invalid performance data (object 0). SysErr : 2:No such file or directory ======================================== (IN 33654) Solution: Workaround This is a Windows known issue of corrupted performance counter in the registry. The problem can be caused by strange numbers in the perfmon utility. To fix this issue do the following: Copy the c:\windows\system32\PerfStringBackup.INI file from a PC where perfmon works correctly. Change directory to c:\windows\system32 Performance Data LoadLoad process fails with ORA-01461: can bind a LONG value only for insert into a LONG column message on UTF-8 database Oracle Load process fails to upload statements data to the PMDB when the database uses multi-bytes character set. The process fails with "ORA-01461: can bind a LONG value only for insert into a LONG column" error message. Precise for Oracle The JDBC driver issues an ORA-01461 when a varchar bind variable value length may exceed the field length, even though the actual string length in bytes is less than the maximum. Solution: Workaround Backup the <i3_root>/products/oracle/bin/load_config.xml file on the Precise for Oracle FocalPoint server. Edit the load_config.xml file as follows: Locate: <TXV name="ORST_TEXT_VARCHAR" type="STRING" nullable="true" max-len="4000"/> Change this line to: <TXV name="ORST_TEXT_VARCHAR" type="STRING" nullable="true" max-len="1333"/> Save the file. Restart the Precise for Oracle FocalPoint. From time to time the error message Connection-pool attribute not received may appear in the Load Performance Data warehouse process From time to time the error message "Connection-pool attribute not received" may appear in the Load Performance Data warehouse process. This message is displayed when the load performance data process has not completed and a new one is executed. (IN 35221) Solution: Workaround Wait for the previous load to complete. ProcessesSGA Utilization metric appears all the time when using sga_target When using the sga_target in Oracle, the SGA utilization is always between 99% and 100%. Since it is not possible to define the threshold infractions, for example 99.5%, even if the near critical is 99% and the critical threshold is set to 100%, the SGA Utilization always exceeds the near critical thresholds. (QC ID 5857) Solution: None To avoid this metric, disable it when using sga_target on your Oracle instance. Expanded text is not available when using "Select * from View" When statement is "Select * from View", the expanded text with the view body is not available in the SQL workspace. (QCID 5007) Solution: None The Schema Changes process may fail with Russian character Set Sometimes the Schema Changes process may fail when using the Russian character set. This is due to different length of characters. QCID (4687) Solution: None Sending import request without setting the cabinet/folder It is possible to send import requests of statements without setting the cabinet/folder. This is achieved by clearing their value in the “Catalog” tab. (PR5412) Solution: Workaround Enter the required information prior to sending the request. The Object statistics changes process takes a very long time and fails The 'Object statistics changes' process queries the dba_free_space view for details of the tablespaces. The dba_free_space view queries the recycle bin (10g) which may take a long time if the recycle bin contains dropped objects. The bug only applies to Oracle versions 10.1.0.1 and 10.1.0.2. Oracle relevant notes/bugs in metalink: note - 271169.1 bug - 3492682.8 (PR915225) Solution: Workaround Clear the recycle bin with the PURGE command. The preferred method is to purge the entire recycle bin (dba_recyclebin). It is also possible to disable the recycle bin entirely using the ALTER SYSTEM command: ALTER SYSTEM SET NOTE Keep in mind that disabling the recycle bin does not clear it. recyclebin = OFF; The Collect Actual Execution Plans process fails with: ORA-600 internal error code, arguments: [504], [0xC00000006DAABA88], [32], [4], [row cache objects]… Queries against the V$SQL_PLAN may fail with the ORA-600 error due to a know Oracle bug. The 'Collect Actual Execution Plans' process queries that view to get the actual execution plan performed for the statements. Related MetaLink docs: 340090.1 28104.1 (PR1037909) Solution: Workaround For Oracle 10.1.x instances try the following: Set _cursor_plan_unparse_enabled=false: At session level: * alter session set "_cursor_plan_unparse_enabled"=false At database level: * In the init.ora or spfile.or a parameter file used to start up the database. * Bounce the database to have the new parameter taken into account. For Oracle 9i instances the problem cannot be solved. Precise for SQL ServerUIMismatch between information displayed on the blocking tree of the Locking view and the information shown in the Association area In the Current workspace, there can be a mismatch between the information displayed on the blocking tree of the Locking view and the information shown in the Association area. This happens because the information on the blocking tree is retrieved by a different request than the information for the rest of the page. Solution: None Mismatch in blocking information In an active environment where locks are released very fast, it can occur that a session was not locked when the Collector sampled the SQL Server instance, but at the time the blocking tree request was issued, the session was waiting for a resource. Solution: None Maintenance workspace does not work well for an SQL Server 2005 instance with a long alias name When you monitor an SQL Server 2005 instance with a long alias name, the Maintenance workspace does not work well. For example, no jobs are displayed in the workspace, even though there are jobs defined for the instance. Solution: None The error The database_id:32767 doesn't exist appears in the SQL workspace in SQL Server 2005 In the SQL workspace, the error “The database_id:32767 doesn't exist” occurs while trying to explain some statements that are run in SQL Server version 2005. This problem occurs when there is a reference to objects that are related to the new SQL Server 2005 internal database called “mssqlsystemresource”. The database is an internal database. When it tries to reference some statements during the explain, an error is returned. Solution: None Precise for SQL Server schema changes will only monitor one schedule if the schedule names are the same In SQL Sever 2005, there is an option to define more than one schedule for the same job. However, if the schedules are using the same name - the Precise for SQL Server Schema changes will monitor and report changes only for one of those schedules. Solution: None Column content is truncated There are cases in which the column content is truncated in the UI. For example, in the Statistics workspace, the content of the Description column in the Performance Counter entity is truncated. This problem occurs due to UI limitations. To see the full text use the column ToolTip. Solution: None Access plan changes are not displayed In the SQL workspace, if a statement is inserted through the Statement Workshop, the Access Plan changes will not appear under the History tab, Association area, All Changes and Access Plan Changes lists. The problem occurs due to a bug in the queries of the lists mentioned above. Solution: None The SQL Server error log view does not show all messages in the Objects workspace In the Objects workspace, the SQL Server Error Log view does not show all messages that are in the current SQL Server log, even though there are relevant message in the log for the requested timeframe. This problem occurs when the Precise for SQL Server user is using a different language than us_english. Solution: Workaround Change the default user language. Total number of objects mismatch In the SQL workspace, Operations tab under Plan view, the total number of objects in the “Major Operations” of the “Other” Major Operation type does not add up to the total number of objects displayed in the “Operations” grid of the Major Type “Other”. Solution: None Displaying graphs for symmetric device fails In the Objects workspace, displaying graphs for the Symmetric Device fails because of a wrong conversion. This problem occurs when the Precise for SQL Server user is using a different language than us_english. Solution: Workaround Change the default user language. Statement is not formatted and text is not highlighted In the SQL workspace statements are not formatted and the text is not highlighted when you examine it. Reason for it is that the parser does not support queries containing hints like the following: FROM my_table WITH (ROWLOCK INDEX(my_index)) nor batches containing a semicolon (;). Solution: None SQL Server error in Maintenance workspace when SQL Server 2005 agent is down In the Maintenance workspace, select an instance of SQL Server 2005. When the SQL Server agent is down the following error message appears: SQL error while Directly executing statement. [Microsoft][ODBC SQL Server Driver][SQL Server]SQL Server blocked access to procedure 'sys.xp_sqlagent_enum_jobs' of component 'Agent XPs' because this component is turned off as part of the security configuration for this server. A system administrator can enable the use of 'Agent XPs' by using sp_configure. For more information about enabling 'Agent XPs', see “Surface Area Configuration" in SQL Server Books Online.” (PR 1229) Solution: Workaround Run the following commands on the relevant instance (master): sp_configure 'show advanced options', 1; GO RECONFIGURE; GO sp_configure 'Agent XPs', 1; GO RECONFIGURE GO Save As shows incorrect SQL text In the session level, Save as in the Current workspace not always shows the correct SQL text. Drill down to the Session entity in the Current workspace, choose the text tab and select Save As. You may see a wrong SQL text in the dialog box. This occurred when you run a batch with more than one statement. The text that will be seen in the Save As dialog box is the SQL text that was running at the moment you drilled down to Session entity. (PR 490) Solution: None A batch is not highlighted in the compared access plan When you compare plans in the SQL workspace, they do not work on batches with more than one statement. (PR 965) Solution: None SQL Server Error Log metric fails The SQL Server Error Log metric does not always work. The metric will not work if one of the following occurs: The language is other than English, German, French, Chinese, or Japanese and the default login language is not English. One of the previous languages is used and the message does not contain one of the following keywords that indicate a problem: Error, Warning, fatal, raising, assertion:, assert , using 'dbghelp.dll' (PR2992) Solution: Workaround Translate the list of words that appears in the <i3_root>/products/SQL-Server/<instance name>/etc/ErrorKeywords.txt file. Add the translated list to the file. The translation should match what appears in the SQL Server error log. Make sure that the default login language is set to English. Missing information on Maintenance workspace for long named aliases in SQL Server 2005 The Maintenance workspace does not display entity details, such as operators, jobs and alerts for long named aliases in SQL Server 2005. (PR 1453) Solution: None Graphs display future data In some graphs, when you select a time range that is later than the current date, the last value is displayed until the end of the selected range. (PR 3204) Solution: None Mismatch in Availability data The report of the Availability in the Activity workspace is not always identical to the information in the SQL Server error log. (PR 1300) Solution: None Data of non-Latin language may be displayed as ???? Data of non-Latin language may be displayed as ????. (PR1004841) Solution: Workaround To change the regional settings of the browser's machine: Open the settings dialog: Start>Settings>Control Panel>Regional and Language Options. Click the Advance tab and select the correct language (Hebrew/Turkish/Japanese) in the drop down menu to be the language of non-Unicode programs. Restart the machine. The Statistics workspace shows illogical numbers in the counters The Precise pssq_perf process samples the performance counters of the OS and SQL Server and shows them in the Statistics workspace. Sometimes the sampled numbers are illogical (negative or very high). You can also see the incorrect counters in perfmon. (PR813586) Solution: None There is no data for a number of counters There is a problem of calculation of the counters values in the sys.dm_os_wait_stats dynamic viewing MS-SQL2005. Our process calculates the delta once every 15 minutes. The delta value stored in the PMDB. In some circumstances, the current sample of a counter may be less than the previous (-1). In this case a NULL value will be inserted into the PMDB. (PR968393) Solution: By Design When selecting MS-Plan in the SQL workspace an error message is displayed When selecting MS-Plan in the SQL workspace with instance SQL Server 2008 the following error message is displayed: Cannot connect to instance. The version of Microsoft Management Studio can only be used to connect to SQL Server 2000 and SQL Server 2005. (IN 35382) Solution: None Error displayed when adding a new statement to the SQL workspace and the Database or User contains special characters The following error is displayed in the SQL workspace: SQL received an error while %p1%: %p0%' or 'SQL received an error while executing the following error The problem occurs in a different language when adding a new statement to the SQL workspace and the Database or User contains special characters. (IN 35065) Solution: None The Recommend role does not go hand-in-hand with the What-If role In the Role Settings of the SQL-Server, when you check Tune.What-If, the Tune.Recommend is not checked although it should. Both roles of Tune.Recommend and Tune.What-If should have a dependency between them. (IN 29299) Solution: Workaround In the Role Settings of the SQL-Server, when you check Tune.What-If check also Tune.Recommend Error displayed when drilling to Physical Disks entity - Invalid character value The following error is displayed when drilling to the Physical Disks entity in the Objects workspace: SQL error while executing the prepared statement. Message text from [Microsoft][ODBC SQL Server Driver]Invalid character value for cast specification. (IN 35732) Solution: None FocalPointFailed Jobs metric fails when using include/exclude like %sxxxxx%. The Failed Jobs metric fails when using include/exclude like "%sxxxxx%". We get the following error: SQL error while Executing the prepared statement. [Microsoft][ODBC SQL Server Driver][SQL Server] Conversion failed when converting datetime from character string. SQLCODE=241. Erroneous fetched (Row/Column): (1,-1) (PR 1103467) Solution: Workaround Don't use include/exclude in a Failed Job metric with "%sxxxxx%". Use only "%xxxxxx%". LaunchingDatabase Files report cannot launch to Precise for SQL Server In the SQL Server AppTier in Report Manager, the Database Files report located in “Profile” does not enable the launch to Precise for SQL Server. Solution: None Mismatched data in the Details area of the Batch entity When you launch the “Ad-Hoc Batches” report from Report Manager to Precise for SQL Server, there are cases in which the information in the Details area of the Batch entity shows mismatched data. This problem occurs when the same batch runs in different databases. Therefore, the “Properties” section in the Details area only shows information of one of the two batches and the In MS SQL Breakdown graph shows summary information of the two batches. Solution: None Mismatch of information shown in the Availability metric In Alerts, the information shown for the Availability metric (Status set) on the Overview tab of the Main area is not refreshed in sync with the information in the Metric Value area. This occurs because the refresh requests are sent sequentially, not in parallel, when the Refresh button on the Precise Bar is clicked. Solution: None Runtime error when you launch from the Failed Jobs metric When you launch from the Failed Jobs metric in Alerts into Precise for SQL Server, a runtime error is displayed. (PR1925) Solution: None Runtime error when you launch from the Data Files without Enough Space to Grow metric When you launch from the Data files without enough space to grow metric in Alerts into Precise for SQL Server, a runtime error is displayed. (PR 1927) Solution: None ProcessesThe pssq_sfw process reports an "unsupported file system" to the pss_collector log file and exits If the Veritas Volume Manager Enterprise Edition is installed on a Windows server, the pssq_sfw process cannot bypass it to find the name the file system of the locally mounted drives, therefore it exits. (QCID 4119) Solution: None Mutex on the settings file fails to be created when the login user of the service is not the administrator When the login of the collector service is not the local system or administrator on the collector server, a Failed to create mutex ...... error appears. This occurs because the user that logs on does not have access to the mutex object created by the collector processes. (PR1053984) Solution: Workaround Use the same login for the FocalPoint service and Collector service. SQL Error Log display errors in future time In the Objects workspace, the SQL Error Log may show error logs in future time. (PR1020122) Solution: None Restart was detected message on dmv_waits_stats process On SQL Server 2005, a Restart was detected message will appear consistently in the dmv_waits_stats.trc file, due to unproportional growing of some latch counters on the DMV mechanism. In this case, statistics data are not collected for the Latch waits, displayed on the Activity workspace. (PR1043078) Solution: By Design SQL Server Error in the SQL workspace Objects tab You receive an SQL Server error in the SQL Workspace Objects tab when the cost is 0. This happens when you launch from a Batch entity with more than one statement in the Activity workspace to the SQL workspace. Select the Statements tab and select a statement to tune. If you now select the Objects tab, you may receive the following error message if the cost is 0: SQL Server received an error while Executing the prepared statement. (PR 1703) Solution:None Schema ChangesViolation of PRIMARY KEY error In SQL Server 2005 there is an option to define more than one jobs with the same name. This causes the Schema Change process to fail on the uniqueness of the Job name with the following error: Violation of PRIMARY KEY constraint 'IW_SQCJ_01'. Cannot insert duplicate key in object 'PW_SQCJ_MTNC_CHANGE_JOB'.. SQLCODE=2627. (PR 2225) Solution: Workaround Change the name of one of the jobs. Precise for SQL collector crashes on pssq_pcs.exe process exception The problem occurs when master DBs sysprocesses view has many rows with the same spid due to SQL Server parallelism. In the log file pss.trc/pss_collector.trc (version 8.0) the following line will be displayed: Changing pss_max_sessions from [X] to [Y]. Solution: Workaround The workaround is to reconfigure the Precise FocalPoint registry of the instance to be higher than the maximum session number with a safety margin (for example, Y+1000) <i3_root>\products\i3fp\registry\products\indepth-sql-server\ Go to <agent name>.xml Edit the "pss_max_sessions" value under the correct instance The Precise for SQL Server Collector cannot remotely sample the performance counters of a 32-bit SQL Server running on a computer with a 64-bit platform The collector process pssq_perf samples the server performance counters. A 32-bit SQL Server installed on a computer with a 64-bit platform will place the counters on the 32-bit registry path. The Microsoft PDH API used to sample the registry cannot remotely access the 32-bit registry on a computer installed with a 64-bit platform (by Microsoft design). (Incident 1236116) Solution: Workaround Install the Collector locally and use the 32-bit version pssq_perf process. pssq_proc process does not start for SQL Server 2005 that was upgraded from SQL Server 2000 pssq_proc process does not start for an SQL Server 2005 that was upgraded from SQL Server 2000. The following error message is displayed in the pss_collector.trc log file: ======================================== ID : INFO-4502025 Time : 2008-12-05 10:57:27.621 GMT-0600 i3 Time : 2008-12-05 16:57:27.621 GMT Process : pssq_proc.exe SDDBSERV14$SDDSQLP01 (ID:13256) Src : odbc_util.c:428 Logger : com.precise.sql-server.inspect.log Msg : ODBC message - SQLExecute(hstmt) Failed szSqlState = 42000 pfNativeError = 8180 szErrorMsg = [Microsoft][ODBC SQL Server Driver][SQL Server]Statement(s) could not be prepared. pcbErrorMsg = 82 ======================================== ID : INFO-4502025 Time : 2008-12-05 10:57:27.621 GMT-0600 i3 Time : 2008-12-05 16:57:27.621 GMT Process : pssq_proc.exe SDDBSERV14$SDDSQLP01 (ID:13256) Src : odbc_util.c:439 Logger : com.precise.sql-server.inspect.log Msg : ODBC message - declare @num int;select @num=count(*) from (select distinct sql_handle from master.sys.dm_exec_query_stats) a cross apply master.sys.dm_exec_sql_text(a.sql_handle) b select @num,sql_handle, b.encrypted, b.text from (select distinct sql_handle from master.sys.dm_exec_query_stats) a cross apply master.sys.dm_exec_sql_text(a.sql_handle) b ======================================== (IN 34772) Solution: Workaround If the SQL Server 2005 database was configured with SQL Server 2000 compatible mode, you should not start the proc process. The procedure cache views do not exist under a database that works with the SQL Server 2000 compatible mode, therefore the proc process should not be initialized. To disable the proc process, add the following entry to the settings.xml file of the instance: <pss_exec_proc>NO</pss_exec_proc> NOTE The following command returns the database compatibility mode sp_dbcmptlevel <database_name>: Database compatibility modes: 60 = SQL Server 6.0 65 = SQL Server 6.5 70 = SQL Server 7.0 80 = SQL Server 2000 90 = SQL Server 2005 Verify (IVP)Agent is in Fault status when it is down When the pssq_service.exe process of the Collector terminates unexpectedly, the status of the agent is the “fault” state. However, verification of the agent returns that it is down. (PR 2385) Solution: Workaround Stop and restart the Collector. CollectorWhen installing SQL Server 2008 with SP1 and sampling it locally with the 2008 client, the Collectors fail with an ODBC timeout error or severe ODBC error. SQL Server 2008 SP1 presents an update in the SQL Server ODBC driver (the one the collector uses) which causes the collector sampling processes to not work correctly with the ODBC functionality. (QC ID 5931) Solution: Workaround Perform remote monitoring with a previous version of SQL Server client. Update the machine ODBC client to a previous MDAC driver. Database or Precise for Storage collector failed to start on Windows environment The collector fails to start as a result of low system resources. The problem can be determined by the following example message within the trace file: ======================================== ID : SEVERE-5002233 Time : 2009-01-22 15:34:45.528 GMT+0200 i3 Time : 2009-01-22 13:34:45.927 GMT Process : pss_symmetrix.63.exe symmetrx -START (ID:5020) Src : iSocketNt.c:206 Logger : com.precise.oracle.collector.log.storage-srv-1.OR.symmetrx Msg : Operation create on a Socket(unix)/Pipe(win) failed SysErr : 231:All pipe instances are busy. ======================================== (IN 35582) Solution: Workaround Change the system and desktop heaps based on this document: http://support.microsoft.com/kb/126962 Process pssq_perf crash due to corrupted perfmon counters Collector process pssq_perf crashes at collector startup due to corrupted perfmon counters. When using perfmon all counters appear as numbers instead of counter name. For example refer to: http://exchangepedia.com/blog/2008/06/perfmon-counters-show-up-as-numbers.html (HP QC ID 4065) Solution: Workaround Microsoft has released a fix for this error. Need to rebuild perfmon counters at registry level by running the following command from the command line: Lodctr.exe /r Precise for J2EEUIData of non-Latin language may be displayed as ???? Data of non-Latin language may be displayed as ????. (PR1004841) Solution: Workaround To change the regional settings of the browser's machine: Open the settings dialog: Start>Settings>Control Panel>Regional and Language Options. Click the Advance tab and select the correct language (Hebrew/Turkish/Japanese) in the drop down menu to be the language of non-Unicode programs. Restart the machine. Statistics ParametersNull value for Data Source name When adding DataSource metrics for Tomcat 5.5 we may get a 'null' value for the Data Source name. The problem appeared in Tomcat 5.5 installations on Windows. (QCID 4186) Solution: Workaround After adding the desired attributes for the DataSource metric (using the Add Statistics in the Statistics Administration dialog box), edit the following file: <i3_root >\products\j2ee\etc\metrics\TOMCAT\5.5\ metrics-collector.xml Each <metric> that is related to the DataSource value, we can identify by its name: <name>DataSource.XXX</name> Change key-property from: <value/> to <value>name</value> For example: BEFORE: <metric> <name>DataSource.maxWait</name> <connector-name>DefaultConnector</connector-name> <class>com.precise.javaperf.metrics.plugin.metric.PassThruMetric</class> <schema-name>fraction</schema-name> <unit-type>Number</unit-type> <description>Introspected attribute maxWait</description> <property> <name>MBeanType</name> <value>DataSource</value> </property> <property> <name>attribute</name> <value>maxWait</value> </property> <property> <name>key-property</name> <value/> </property> </metric> AFTER: <metric> <name>DataSource.maxWait</name> <connector-name>DefaultConnector</connector-name> <class>com.precise.javaperf.metrics.plugin.metric.PassThruMetric</class> <schema-name>fraction</schema-name> <unit-type>Number</unit-type> <description>Introspected attribute maxWait</description> <property> <name>MBeanType</name> <value>DataSource</value> </property> <property> <name>attribute</name> <value>maxWait</value> </property> <property> <name>key-property</name> <value>name</value> </property> </metric> InstrumentationJVM fails to start after configuring Precise VM arguments for it. After adding the following arguments and starting the JVM, the load task is never completed: -Xbootclasspath/p:<i3_root>\products\j2ee\etc\apl\jvms\[JVM-ID]\lib\rt.jar -Xbootclasspath/p:<i3_root>\products\j2ee\lib\indepthj2eeboot.jar -Xbootclasspath/p:<i3_root>\products\j2ee\classes -Dindepth.j2ee.server=<i3_root>\products\j2ee\config\<jvm_id> The problem is that the hooking mechanism fails to receive an entry point to the JVM. (QCID 7975) Solution: Workaround Usually, the arguments you have to use after the installation are the following: -Xbootclasspath/p:<i3_root>\products\j2ee\etc\apl\jvms\<jvm_id>\lib\rt.jar -Xbootclasspath/p:<i3_root>\products\j2ee\lib\indepthj2eeboot.jar -Xbootclasspath/p:<i3_root>\products\j2ee\classes -Dindepth.j2ee.server=<i3_root>\products\j2ee\config\<jvm_id> If using java 1.5 or higher Instead of the above, add the following arguments to the JVM: -javaagent:<i3_root>\products\j2ee\lib\indepthj2ee-agent.jar=com.precise.javaperf.j avaagent.agents.DefaultJavaAgent -Dindepth.j2ee.server=<i3_root>\products\j2ee\config\<jvm_id> For WebSphere JVM use the following entry instead: -javaagent:<i3_root>\products\j2ee\lib\indepthj2ee-agent.jar=com.precise.javaperf.j avaagent.agents.IBMJavaAgent -Dindepth.j2ee.server=<i3_root>\products\j2ee\config\<jvm_id> If using JRockit 1.4 Instead of the above, add the following arguments to the JVM: -Xbootclasspath/p:<i3_root>\products\j2ee\lib\indepthj2eeboot.jar -Xbootclasspath/p:<i3_root>\products\j2ee\classes -Xmanagement:class=com.precise.javaperf.extensions.weblogic.JRockitClassPreProcesso r -Dindepth.j2ee.server=<i3_root>\products\j2ee\config\<jvm_id> Instrumenter Out of Bounds exception when WebLogic Diagnostic Framework is enabled When the WebLogic Diagnostic Framework is enabled, on rare occasions, the instrumenter can throw an ArrayOutOfBounds exception when performing instrumentation on select classes. (PR1056216) Solution: Workaround Ignore the class or disable the WebLogic Diagnostic Framework by editing the /domains/ServerName/config/diagnostics/ServerNameWLDF.xml file and setting to false. CollectorNo access to JMX metrics if Websphere security is enabled On WebSphere, if the internal security is turned on, the mechanism for internal JMX collection does not work and Metrics are not collected. If only Java security is on, the procedure works fine. (Spira IN 33912) (QC ID 2574) Solution: Workaround Revert back to the previous method using an external connector that requires setting of a username and a password. Follow this procedure: Go to the <i3_root>\products\j2ee\etc\metrics\WEBSPHERE\<version> directory Open the metrics-collector.xml file Under the defaultConnector tag, change the default class from LicalJMXManager to JMXManager Enter the username and password under the Statistics Configuration screen in the Monitor Settings dialog box: In Precise for J2EE, go to Settings>Monitor Settings>Statistics Configuration. Mark the Password Required check box and enter the user name and password for your WAS. You should also enter the host/port information that matches the SOAP_CONNECTOR_ADDRESS host/port from the WebSphere Advanced Administration Console's Ports page. To get to the Ports page, select: Application Servers><SERVER_NAME>>Communications> Ports. The default values (blank) are appropriate if security is disabled. GeneralMonitoring by search results saves the pattern with the default custom type instead of the new name From the Monitoring Configuration dialog box, search for a pattern when the method is "*", or undefined. Click Monitor Search results and insert a new type. The pattern type is saved as Custom instead of the new name. (QCID 7886) Solution: Workaround Click on Manage Patterns…, find the pattern that was added and select it from the drilldown list. JVM deadlocks when SmartLink trace increased SmartLink Timers thread deadlocks with main thread when formatting a message to the log. This only happens when com.precise.javaperf.smartlink is added to the trace.xml for the JVM. Important: This trace option is off by default. (IN 34265) Solution: Workaround Remove com.precise.smartlink from the JVM's trace xml. Precise for Microsoft .NETFocalPointUnexpected SQL statement shown when launching from Report Manager to .NET When launching from Report Manager into .NET, an unexpected SQL statement appears. This occurs because we do not have the connection string (which database server). When there is more than one database server, the details will not necessarily be correct for the launched SQL statement. The launch in this case will display information on the first database server he finds. Solution: By Design UIRoot Callers/Callers views for a recursively called method show incorrect data In the Activity workspace, when you select the root-callers or callers view of a method that is called recursively, the numbers shown in each row for the different caller / root-callers are incorrect. Solution: None Data of non-Latin language may be displayed as ???? Data of non-Latin language may be displayed as ????. (PR1004841) Solution: Workaround To change the regional settings of the browser's machine: Open the settings dialog: Start>Settings>Control Panel>Regional and Language Options. Click the Advance tab and select the correct language (Hebrew/Turkish/Japanese) in the drop down menu to be the language of non-Unicode programs. Restart the machine. Incorrect metrics counters The metrics counters are incorrect when creating two .Net instances for the same IIS (using Application Pools). The graphs are not correct (on CPU and memory). The following Memory and CPU metrics are not sampled correctly: Memory : Free Heap Size. Memory : Used Heap Size. Memory : Total Heap Size. Memory : Maximum allowed Heap Size. Garbage Collection : # of Bytes Collected. Garbage Collection : Execute Time (non GC time). Garbage Collection : Garbage Collection Time. Garbage Collection : Garbage Collection Count. CPU : CPU Consumption %. Solution: None Activity CollectorOutOfMemory exceptions on a Web server On environments that do not limit the Web server memory usage and that activate many application domains at the same time, an out of MemoryException may occur when Precise for Microsoft .NET is installed with the default configuration. Each application domain that is monitored uses a default buffer of 1MB that is created by Precise for Microsoft .NET and another 20MB (max size) for the queue that is responsible for data transfer. On environments that have no memory control or have many application domains, the above configuration can cause excessive memory usage and crash the IIS with an outOfMemory exception. (PR1051089) Solution: Workaround Changing the configuration to a smaller buffer and a smaller queue will reduce memory usage and prevent the outOfMemory exception. The buffer size and queue size (both in KBs) are changed in the Collector.xml file: <collector-config> . . <activity-collector> . . <tracker> . . <events-buffer-size> <events-queue-buffer-size> The Collector.xml file can be found under <i3_root>\products\dotnet\config. Excessive disk space usage by Instrumentation Cache Every time a DLL is instrumented by Precise for Microsoft .NET, it is saved in the cache directory. The cache should delete all files that are older than a month. This feature is not working in version 8 and will be addressed with a hot fix upon request. (PR1060615) Solution: Workaround Delete files from the cache manually. The cache directory can be found under <i3_root>\products\dotnet\cache. No support for two Collectors on the same machine When installing two Precise for Microsoft .NET Collectors on the same server, the application will not work properly. (PR1058967) Solution: None IIS crashes with CannotUnloadAppDomainException On busy servers, the AppDomain recycle takes too much time and not all of the tracker threads abort on time. When a tracker thread stays up when the updomain is trying to recycle, the recycle operation fails and the IIS crashes with CannotUnloadAppDomainException. Solution: Workaround If the above problem happens we can configure the amount of time we supply the tracker threads to abort. Default time is one second and it is recommended to raise it to 5 seconds if the server suffers from high loads. The wait time (in milliseconds) is changed in the Collector.xml file: <collector-config> . . <activity-collector> . . <tracker> . . <exit-delay> The Collector.xml file can be found under <i3_root>\products\dotnet\config. psdn_dncol_ctrl.exe process crashes On some installation, you may encounter occasional crashes of the psdn_dncol_ctrl.exe process. The crash does not affect anything on the computer and Precise functions properly and can be ignored. (PR994803) Solution: None Failure to perform live configuration updates on 64-bit environments On 64-bit environments the Collector is using our 64-bit APIs. The communication agent between the FocalPoint and the Collector is not supporting connection with a 64-bit process and will therefore fail alerting the Collector on configuration changes in real time. (IN 35683) Solution: Workaround Restart the Collector service after a configuration change in the FocalPoint for the changes to be applied. Statistics Collector.NET CLR Interop performance counters show higher values The .NET CLR Interop performance counters and remoting counters are always indicating higher values than the real application has, because the probing mechanism is using both technologies. Solution: None Inaccurate performance results if the same instance is run concurrently from different locations Precise for Microsoft .NET samples Microsoft performance counters to build the statistics information (Statistics workspace). The performance counter mechanism does not provide a way to distinguish between processes with the same executable file name but with different path. As a result, trying to measure performance statistics on the same executable running from two different locations will result in inaccurate results. This is a known issue of Microsoft performance counters. Other tools that sample Microsoft performance counters (for example: Microsoft PerfMon) have the same limitation. Solution: None InstrumentationIf you decide to transfer all columns to the Invisible list, you can no longer restore column visibility through the product user interface The Precise for Microsoft .Net Instrumentation Settings screen allows all columns to be set to invisible. When all columns are transferred to the invisible list, there is no way to return any columns to be visible again through the product user interface. (QCID 8221) Solution: Workaround Perform the following steps: Go to the <i3_root>\products\gui\website\webapps\i3\shared\userDB\DN.xml file. Find the relevant table column node <column> elements) and change their "visible="false"" attribute value to "true". Clear the browser cache. Cannot instrument mscorlib.dll and system.dll Precise for Microsoft .NET may experience problems when instrumenting specific internal classes in mscorelib.dll and system.dll. You must not include these DLLs in the DLL section of the instrumentation.xml file. Solution: None Changes in instrumentation configuration not activated After modifying the instrumentation configuration (editing the instrumentation.xml file or adding new DLLs via the UI), the appropriate application processes must be restarted to activate the changes (for example: an IISRESET for the ASP.NET application). Solution: None Initial instrumentation overhead A DLL is being instrumented the first time is it loaded by the .NET application. An initial instrumentation overhead occurs. The overhead is discussed in the Precise Administration Guide. The approximate memory consumption of this instrumentation process is about 20 times the size of the original DLL. An extreme example: An instrumentation of a huge 30MB DLL uses 700MB RAM and takes 45 minutes to instrument. It is recommended to run the instrumentation driver utility, that performs instrumentation off-line. For more information see the Precise Administration Guide. Solution: None Instrumentation is not working when path for a dll is used Under the instrumentation.xml file, if a dll is chosen for instrumentation it may only be entered without the full path. (IN 188832) Solution: Workaround Enter the dll name without the path. Precise for SybaseUITotal number of objects mismatch In the SQL workspace, Operations tab under Plan view, the total number of objects in the “Major Operations” of the “Other” Major Operation type does not add up to the total number of objects displayed in the “Operations” grid of the Major Type “Other”. Solution: None SQL error during Explain process When you try to explain a batch or statement with text that contains objects with names in languages other than English in the SQL Workspace, an error is displayed, such as: SQL error received during explain process: [DataDirect][ODBC Sybase Wire Protocol driver][SQL Server]Gr not found. Specify owner.objectname or use sp_help to check whether the object exists (sp_help may produce lots of output). Solution: Workaround In the Precise FocalPoint server, add <pss_use_charset>YES</pss_use_charset> to the file <i3_root>\products\i3fp\registry\products\indepth-sybase\<server name>.xml under the instance’s entry and restart the Precise for Sybase FocalPoint agent. Objects with names in languages other than English, are not displayed correctly Objects with names in languages other than English, are not displayed correctly in Objects Workspace. Solution: Workaround In the Precise FocalPoint server, add <pss_use_charset>YES</pss_use_charset> to the file <i3_root>\products\i3fp\registry\products\indepth-sybase\<server name>.xml under the instance’s entry and restart the Precise for Sybase FocalPoint agent. New estimated cost value of Statement/Batch plan does not show The Cost overtime graph in the History tab of the SQL workspace does not show the new estimated cost value of the Statement/Batch Plan, even though there is a new cost for the Plan. This problem occurs because Precise for Sybase only saves new costs in cases when the Access Plan has also been changed. Solution: None Access Plan tree displayed in the SQL workspace does not match the Statement text The Access Plan tree displayed in the SQL workspace does not match the statement text. For example, the Plan tree displays the SELECT statement while the statement text is an IF command. This problem occurs when the statement ID returned by Precise for Sybase during the statement run does not match the statement ID returned by the SET SHOWPLAN ON for the same statement. Solution: None Temporary table in Plan In the Objects tab in the SQL workspace is a user defined table which is considered as a temporary table in the tempdb. This problem occurs when you defined table name contains the word “Worktable”. The Precise for Sybase Explain process cannot make a distinction between Sybase worktables and user defined tables containing the word “Worktable”. Therefore, the user defined table is considered as a Sybase worktable. Solution: None The text of a batch not fully loaded in the Activity workspace In some cases the text of the batch is not fully loaded while performing a Save As or Edit in the Activity workspace. Solution: None Cannot communicate with server message When you are in the Objects workspace and try to drill down into a database with a name like “[[ }}]]|/\ @!#*()&^(' </s>”, you will get a “Cannot communicate with server” message. The problem is an internal Sybase problem. The reason is that all installations use the same registry path. This path is overridden when another installation is done. If the need to split occurs, you can install some agents on a remote computer. Solution: None Mismatch of statement execution value In some cases, the number of Statement Executions (Summed) for the selected entity in the Main area is smaller than the number of Statement Executions in the Association area. This happens when the number of Statement Executions is zero because for example a session was opened, but did not execute any statement. It is then displayed as 1. Therefore, if you sum the number of Statement Executions for all the rows in the Association area, the result is one number higher than the one displayed in the Main area. For example, if the number of Statement Executions is zero for the first three items in the Association area table, each of these rows displays the number 1. In the Main area, you would expect a summed execution count of 3, but instead the number displayed is 1 because it sums 0 + 0 + 0 = 0, then turning the 0 into 1. Solution: None Access plan changes are not updated in the History screen After the access plan has changed, you can see a new row in the table, but the cost graph does not reflect this with a new bar and cost. Solution: None Graphs display future data In some graphs, when you select a time range that is later than the current date, the last value is displayed until the end of the selected range. (PR 3204) Solution: None Launch to SQL Advantage from the SQL workspace is not working The activation of SQL Advantage from the SQL workspace is not working. (PR 2910) Solution: Workaround Add the directory containing sqladv.exe to the Path in the Server Variables on the server you are working on. Data of non-Latin language may be displayed as ???? Data of non-Latin language may be displayed as????. (PR1004841) Solution: Workaround To change the regional settings of the browser's machine: Open the settings dialog: Start>Settings>Control Panel>Regional and Language Options. Click the Advance tab and select the correct language (Hebrew/Turkish/Japanese) in the drop down menu to be the language of non-Unicode programs. Restart the machine. The number of partitions that are displayed is not accurate In the Objects workspace, when changing the table view to observe the partition, its number can sometimes be incorrect. (PR1063394) Solution: None The partition pie for the table partition is empty In the Objects workspace, when choosing to view the table partition pie, the graph can be empty if the table exists in a database other than the 'master.' (PR1063399) Solution: None No filter exists for database devices on a specific database for Sybase 15 In the Objects workspace, when associating to database devices, you always see that all database devices reside on the instance, even when you filter on seeing database devices that belong to specific database. This only happens on Sybase 15. (PR1061732) Solution: None Parallel indicator is not set for Sybase ASE version 15 In the SQL workspace, Statistics tab, the "Parallel" column is always empty. For Table Scan and Index Scan steps, parallel execution messages are available only in the ToolTip on the Plan tree. (PR 1112138) Solution: None Execution Plan tree is empty in some of the Stored Procedures automatically generated by Sybase In the SQL workspace, the Execution Plan is empty for statements that are part of the Stored Procedure definition. This occurs in cases where a label appears before the "create procedure" command in the corresponding batch text. (PR 1135861) Solution: Workaround The problem occurs when launching from the Activity workspace with statement in context. For workaround, associate the statement with its batch and launch to the SQL workspace with batch in context. Partitions supported only for tables including Sybase 15 Sybase did not support partition information prior to version 15 except for the round-robin partition type. In version 15, Sybase began to support partition information (list, hash and range) for tables and indexes. Currently Precise for Sybase displays partition information only for tables and not for indexes. (PR 1068474) Solution: By design The following message is displayed instead of the Execution Plan tree: Parse show plan output failed. Cannot display Execution Plan tree, contact Precise Technical Support In several cases, the Execution Plan generated using the "show plan on" setting differs significantly from the best plan generated using the "show_best_plan" option. In this case an error message is displayed instead of a Plan tree. (PR 1127440) Solution: Workaround Press the Sybase Plan button to examine both plans. When parallel execution is enabled, some Estimated Subtree fields in the ToolTip of the Execution Plan tree have zero values When parallel execution is enabled, Estimated Subtree Logical IO, Estimated Subtree Physical IO, and Estimated CPU Cost fields in the Execution Plan tree ToolTip have zero values. This occurs because the data is missing in the output of the “show_best_plan” option generated by the Sybase ASE optimizer. Estimated Logical IO, Estimated Physical IO, and CPU Cost fields contain correct values. (PR 1144609) Solution: None Receiving negative values in the Replication Server Alerts, DSEI-Exec General Behavior metric Receiving negative values in the Replication Server Alerts, DSEI-Exec General Behavior metric. (IN 35818) Solution: None Incorrect launch from Replication Server Alerts to Precise for Sybase Receiving No Data Found when launching from Replication Server Alerts, DSEI-Exec General Behavior metric to Precise for Sybase. (IN 32224) Solution: Workaround Do the following: Select Home in the Statistics workspace Select the ASE that was chosen in alert In the Details Area, select Replication Outbound In AdminPoint>Agents, when selecting Sybase Replication Sever and clicking the Verify button, an error message may be displayed The error message Unable to connect to Sybase Monitor server. Original Message from Sybase:%message% is displayed when the Replication Server is down. (IN 35195) Solution: None There is no correlation between the counters and their ToolTip in the Replication Server Alerts, DSEI-Exec General behavior metric, History tab There is no correlation between the counters in the graph and their ToolTip. (IN 35733) Solution: None Incorrect launch from the Dashboard workspace, All instances, Replication tab to the Statistics workspace When launching from Dashboard workspace>All instances>Replication tab: Outbound Load to Statistics workspace, we get to Overview instead of Replication Outbound Inbound Load to Statistics workspace, we get to Overview instead of Replication Inbound (IN 31252) Solution: Workaround In the Details area, select Replication Outbound/Replication Inbound Incorrect launch from Replication Server Alerts to Precise for Sybase Displaying No Data Found when launching from Replication Server Alerts, DSEI-Exec General Behavior metric to Precise for Sybase. (IN 31761) Solution: Workaround Select Home in the Statistics workspace, select the ASE that was chosen in Alerts, and in the Details area select Replication Outbound. Error displayed when adding a new statement to the SQL workspace and the Database or User contains special characters The following error is displayed in the SQL workspace: Sybase received an error while %p1%: %p0%' or 'Sybase received an error while executing the following error The problem occurs in a different language when adding a new statement to the SQL workspace and the Database or User contains special characters. (IN 35034) Solution: None In the Dashboard and Statistics workspaces, the SQM\SQMR counters of the replication server are empty In the warm standby configuration, SQM/SQMR counters are not sampled by design (because the replication server Collector does not know how to parse the logical connection name). In the Dashboard workspace, Replication tab, Outbound Load SQM\SQMR counters are empty. In the Statistics workspace, Replication Outbound view, SQM processing graph, Commands written is empty. In the Statistics workspace, Replication Servers and Replication Databases entities, the counter of Commands Applied is empty. (IN 34729) Solution: By design LaunchingUser interface error when you launch from a table in the SQL workspace, Objects tab to Objects workspace The launch from a table in the SQL workspace, Objects tab to the Objects Workspace may result in a user interface error. This happens when the table you try to launch from is deleted and it does not exist in the Objects workspace any more. (PR 3015) Solution: None Wrong execution plan tree in SQL workspace The launch into the SQL workspace from a statement caught by the Collector may result in the display of a wrong execution plan tree. This can happen because Precise for Sybase reports on the statement ID in the batch and line. There have been cases in which there is no correlation between the statement ID and the line number. (PR 5053, PR 5052, PR 5049, PR 5047) Solution: Workaround After the launch, save the statement in the SQL workspace. CollectorText in batches contains items starting with RPC When you view text in batches, they sometimes contain something like the following: “RPC UpdateItem”, “RPC AddItem”, and so on. This happens when the monProcessSQLText table reports that a session is executing a Stored Procedure, but the monProcessProcedures table reports that it is not. The text appearing after “RPC” is the name of the stored procedure as reported by monProcessSQLText. Solution: Workaround If this happens at a high frequency, contact Precise Technical Support. ProcessesMany simultaneously running pssy_tuner.exe processes cause a high CPU usage When Sybase reaches the limit of 25 user connections, the next connection will cause a high CPU usage until there is an available connection. (PR1084422) Solution: Workaround Enlarge the number of user connections or drop existing connections. ExplainExplain fails with a Memory failed message The Explain statement process fails with the following error message: Attempt to allocate 286 bytes of memory failed It happens when there is a long batch with a long access plan. (PR 3166) Solution: Workaround Contact Precise Technical Support to disable the Explain of the long batch (updating SYSM and SYBA tables). Incorrect Explain process of statement with a remark If there is a remark in the statement, the Explain tree is shown incorrectly. Solution: Workaround Edit the statement and remove the remarks. Explain fails with an “Incorrect syntax near the keyword ‘end’” message Explain fails with an Incorrect syntax near the keyword 'end' message if the batch contains an empty BEGIN END block or when the BEGIN END block only contains comments. The problem occurs when the BEGIN END block only contains a DECLARE CURSOR statement. In this case, before explaining the batch, we replace the DECLARE CURSOR with the following SQL statement /*Cursor ignored for SAPlan*/, otherwise the Explain will fail with message that DECLARE CURSOR must be the only statement in the batch. (PR 4330) Solution: Workaround Open the Batch in the SQL workspace. Select Edit and remove the BEGIN END block that surrounds the DECLARE CURSOR statement. Explain fails with the error Invalid character value for cast specification Explain fails for batches with more than 1000 operators in the execution plan. (PR 4915) Solution: None Failed to explain batches during the Explain Statements process If the Explain statement process recognizes the following error for a specific batch: "Run subquery in union that do not appears in the first part of union is not supported", for the current run, the rest of the batches from this point onwards will fail with the same error message. (PR1071894) Solution: None Instance StatisticsStatistics workspace shows no data in some of the time slices The problem that the Statistics workspace shows no data in some of the time slices occurs after the sp_sysmon stored procedure has run on the production server. (PR 3107) Solution: Workaround Do not run the sp_sysmon stored procedure or add the noclear option to the run on version 12.5.3 and later. Deadlocks on Statistics Process log Deadlocks which are reported on the Statistics Process log, are caused by a system table query and causes the current run of the process to fail. (PR 2367) Solution: Workaround This should not happen regularly. If it does, contact Precise Technical Support. Deadlocks in the Instance Statistics process When few Precise installations monitor the same Sybase instance, we may get a deadlock in the process that collects the instance statistics. (PR812165) Solution: Workaround Change the scheduling of the Collect Instance statistics process so they will not run at the same time. User Tables in Database Master metric (Sybase) does not exclude the mon tables User Tables in Database Master metric displays tables created by the user in the master database. Although they are system tables, mon tables are displayed in the metric. This occurs when the Collectors are not installed on same server as the Precise for SQL Server FocalPoint. (PR1001476) Solution: Workaround Copy the ignoreTables.txt file from the <i3_root>\products\sybase\<instance>\etc folder on the Collector's server to the <i3_root>\products\sybase\<instance>\etc folder on the Precise for SQL Server FocalPoint server. Some of the Instance Statistics counters may be unavailable due to minus values returned by the Sybase sysmonitors table There are cases when some of the counters displayed in the Instance Statistics workspace are not available due to a Sybase bug. The problem occurs when the counters in the "sysmonitors" table exceed 2147483647 and the value becomes negative. Since the Precise for Sybase "Collect Instance Statistics" process ignores negative values, the counters are not displayed in the Instance Statistics workspace. (Sybase case ID 11408334) Solution: None Precise for DB2FocalPointStatements do not appear after the time slice If the "Top-N Statements" default is changed and set to be high (around 2000) the statements are loaded and seen only after the next time slice. Note that the higher the number, the higher the risk of occurrence. (ID 31285) Solution: Workaround Follow these steps: In the AdminPoint click Warehouse>Processes Select the DB2 "Load Performance Data" process and press the Scheduling button. Select the relevant instance(s) and change the number to "4" before the text "minutes after the slice". After Roles/Permissions definition, it can take up to 10 minutes (in federated environments) for the Precise for DB2 FocalPoint to be refreshed with the new definitions. Meanwhile, the user may not see his changes in Precise for DB2 or may not see any instances in the Dashboard (only the "All" row is displayed). (IN 35170) Solution: By design Wait 10 minutes for the refresh process to complete. Performance Data LoadFrom time to time the error message Connection-pool attribute not received may appear in the Load Performance Data warehouse process From time to time the error message "Connection-pool attribute not received" may appear in the Load Performance Data warehouse process. This message is displayed when the load performance data process has not completed and a new one is executed. (IN 35221) Solution: Workaround Wait for the previous load to complete. CollectorCollector does not create the precise.cfg file when it is started from the services console The collector fails to start through the services console when the precise.cfg file does not exist. Starting the agent from the AdminPoint works fine. The problem can be identified by the following message within the trace file: ID : SEVERE-5002803 Time : 2009-01-12 19:14:47.771 GMT+0200 i3 Time : 2009-01-12 17:13:20.547 GMT Process : pss_service.exe (ID:6108) Src : getParm.c:1197 Logger : com.precise.DB2.collector.log.qasds01.UD.DB2CTLSV.SATCTLDB.0 Msg : Parameter pss_db2_database is mandatory. Try restarting Indepth for DB2 Focal Point and after that restart Indepth for DB2 Agent. ======================================== ID : SEVERE-5002004 Time : 2009-01-12 19:14:47.786 GMT+0200 i3 Time : 2009-01-12 17:13:20.562 GMT Process : pss_service.exe (ID:6108) Src : usrMsgUtil.c:572 Logger : com.precise.DB2.collector.log.qasds01.UD.DB2CTLSV.SATCTLDB.0 Msg : Precise/i3 Indepth for DB2 collector was shutdown due to errors. See previous messages in db2.collector.trc file for further details. ======================================== (IN 35397) Solution: Workaround If the precise.cfg file is missing, the collector should be activated through the AdminPoint Database or Precise for Storage collector failed to start on Windows environment The collector fails to start as a result of low system resources. The problem can be determined by the following example message within the trace file: ======================================== ID : SEVERE-5002233 Time : 2009-01-22 15:34:45.528 GMT+0200 i3 Time : 2009-01-22 13:34:45.927 GMT Process : pss_symmetrix.63.exe symmetrx -START (ID:5020) Src : iSocketNt.c:206 Logger : com.precise.oracle.collector.log.storage-srv-1.OR.symmetrx Msg : Operation create on a Socket(unix)/Pipe(win) failed SysErr : 231:All pipe instances are busy. ======================================== (IN 35582) Solution: Workaround Change the system and desktop heaps based on this document: http://support.microsoft.com/kb/126962 DB2 collector fails to start on UNIX environments Using manual startup returns the following error message due to new version of ksh shell: ./pss_start.sh[24]: .[268]: uname: not found [No such file or directory] ./pss_start.sh[24]: .[308]: get_pid_from_file[242]: cat: not found [No such file or directory] ./pss_start.sh[24]: .[339]: readVar[75]: sed: not found [No such file or directory] ./pss_start.sh[24]: .[349]: uname: not found [No such file or directory] ./pss_start.sh[24]: .[349]: [: argument expected (IN 35657) Solution: Workaround Do the following: Open the {i3_HOME}\products\db2\bin\precvars.sh file for editing. Locate the command addExport PATH ./db2bin (line 268) Alter the above line with set $PATH=$PATH:./db2bin Go to the Precise AdminPoint and restart the collector. DB2 collector report tablespace name concatenated with bufferpool name in Japanese environment when the tablespace name is in Japanese characters Precise for DB2 Collector is calculating the maximum tablespace length as 18 characters (18 bytes). In Japanese, each Character encoding is calculated in "double-byte" or "multi-byte" method, meaning each character can be two bytes or more. This makes the 18 bytes buffer insufficient for the more than 9 characters tablespace name making it overflow on the bufferpool name section during collector processing. (IN 35712) Solution: None Will be fixed as a hotfix. GeneralRecommend is not disabled when a plan is not available for statement When the explain fails for statement, the recommend button should be disabled. Since it is not disabled, the user gets a wrong error message when he clicks the Recommend button. (IN 34219) Solution: None The buffer pool hit ratio may get negative value The buffer pool hit ratio is displayed in several places: Tablespace activity, Bufferpool activity in the Activity workspace, and in the Statistics workspace. The hit ratio is calculated as the ratio between physical reads and logical reads: Formula: 1 - (Physical Reads / Logical Reads) In some cases, DB2 reports that the number of physical reads is greater than the number of logical reads. This is probably due to the prefetchers activity that read more pages than actually needed by the application. When getting a negative ratio, it should be considered as a 0% ratio. (IN 34595) Solution: None DB2 database on HP is not supported when defined to use roman8 charset (cp1051) FocalPoint cannot establish JDBC connection to the database when the database is defined with default charset roman8. Load process does not support client sessions connected to the database using roman8 charset. (IN 35435) Solution: None The problem is due to an IBM issue with JDBC connectivity. For more information, see: http://www-1.ibm.com/support/docview.wss?uid=swg21114928 Wrong display in the Objects workspace for partitioned database When the database is partitioned, the following could happen: When associating from partition group to partitions – all the database partitions appear in the list When the tablespace is not defined on partition 0, we get “no data found” on the details area In association to tablespaces, for tablespaces that exist in more than one partition, the tablespace is duplicated in the list as many times as the number of partitions it is defined in. (IN 35969, 35966, 35868, 35971) Solution: None Precise TPM DashboardGeneralSelecting the Users tab sometimes causes an error message to be displayed When there is a lack of Client/Server correlation for Web users, an error message is displayed upon selection of the Users tab. In this case, User data is not visible. This happens only in applications configured as Page Title or Page URL groupings. (QCID 9099) Solution: Patch Install Hotfix IS_ISFP_870_P100_HOT_FIX_001_BU When there is no client-server correlation in Precise for Web, Precise for Web reports only server-side activity to SmartLink When there is no client-server correlation in Precise for Web, Precise for Web reports only server-side activity to SmartLink. The client-side data (Page/Page title) is available in Precise for Web (TPM Dashboard’s data source) but not in SmartLink. (QCID 9086) Solution: None LaunchingWhen launching from the Precise TPM Dashboard to Precise for Web/SmartLink the Time Frame may be different When launching from the TPM Dashboard to Precise for Web/SmartLink, the Time Frame may be different by up to one time slice. (QCID 8921) Solution: None Precise for TPMMonitoring Siebel on IISSiebel on IIS - Missing application user-name and business transaction name When monitoring Siebel on IIS with p4web, Application user is missing from the GUI.In addition, transactions do not have a clear business name, they just show up as URLs. This is because post parameters on the Siebel virtual directories are not monitored by p4web. (QC ID 17108) Solution: Workaround To monitor (collect) post parameters on your Siebel applications do the following: Identify which “virtual directories” (application's prefix - /callcenter_enu), you want to monitor: Launch the IIS manager GUI. Expand the tree of the web site called Default web site to view all the “virtual directories”. For each “virtual directories” you want to monitor do the following: Open a command line window. cd to <precise-root> on the monitored IIS machine. Run the following: products\www\install\psww_insert_isapi_handlers.bat <precise_root> <instance_id> <site_name>/<virtual-directory-name> add-handler PreciseWebHandler <filter_path> where: <filter_path> is the path to the instance bin folder concatenated with "IISFilter", without the _32.dll or _64.dll extension. For example, c:\precise\products\www\instances\2643\bin\IISFilter. Example for running the command: products\www\install\psww_insert_isapi_handlers.bat E:\Nightly95 2807 "Default Web Site/callcenter_ENU" add-handler PreciseWebHandler E:\Nightly95\products\www\instances\2807\bin\IISFilter. UICurrent time slice lost When you move from the TPM screen to Web or .NET, TPM does keep the current time frame in the following scenarios: From the TPM screen go to the Web or .Net screens. Change the time frame. Go back to the TPM screen using the breadcrumbs. The time frame in TPM is the same as it was before going to the Web or .Net screens. Solution: Workaround Reselect the correct time frame. |