Upgrade to ECM SM 5.2.0.1 disables schedule for Database Cleanup

After upgrading to 5.2.0.1, the scheduled execution of Database Cleanup is disabled.

Solution

Open the Database Cleanup Configuration editor and review the settings on the Scheduling tab. Leave the editor by pressing Apply to make sure that the schedule is correctly enabled again.

Import / Export function for agent key file will not be implemented in full installer.

At the moment an agent must be accepted in case it is connected to a new server. That leads to the issue, that all passwords must be redefined. Especially that might happen in customer environments where an ECM SM Server is replaced. Since the full installer is already released, this issue cannot be fixed any more. For further implementations we want to use another implementation (not based on Install Anywhere).

Solution

If a second server should be installed with imported keyfile, first define it as "secondary server" and select the old server as primary server to get its keyfile. Click Next to import the key >. The key file is imported now to the new server. Don't proceed with the installation after this step, but click the "Back" to the step, where you can change your installation to be a "primary server". Proceed with your installation as usual.

Upgrade To New Default Roles Does Not Leave Potentially Existing Roles Untouched

If a role exists that is named as one of the default roles created by an upgrade the existing role will get the roles-rights assignments of the new default role, but will not remove already defined roles-rights. The performed DB changes are logged into the initdb.0.log, when installation logging is set to TRACE.

Solution

Check the user's guide for the names of the to-be-created default roles before installation and, if possible and necessary rename existing roles.

Logging In Internal Installer Configured By Product Debug Switch Instead Of Installer Debug Switch

The Java based parts of the installer only write tracing information, if the debug switch for server and GUI is activated in the installer. The installer's own debug switch has no effect on that component's tracing behaviour.

Solution

To trace the actions of all installer components activate the product's debug switch also and deactivate tracing in product after installation manually, if not needed.

Timestamp columns no longer displayed in customized Event List View after upgrade to 5.2

In the case the user has customized the EventList view, the timestamp columns will no longer be displayed after the upgrade to 5.2. This is caused by changes in the representation of event timestamps and the resulting new timestamp column.

Solution

Re-add the timestamp column to the Event List View after the upgrade. Users who have not customized the Event List View will automatically see the new timestamp column.

Timestamp handling is limited in Rules Engine Script Editor

The timestamps in the TIMESTAMP_UPDATED and TIMESTAMP_INITIAL are only available in "milliseconds since 1970" format. This means it is hard to implement time dependent requests. Furthermore the information in which timezone the agent from which the event was sent is not available inside script context.

Solution

Java Script supports a Date object, which at least can transform the millisecond format to a human readable format. It is recommended to look up tutorials on the internet.

Login dialog does not appear in Firefox 29

The RAP GUI does not work with Firefox 29. The login dialog does not appear.

Solution

Keep Firefox 28, change to a newer version of Firefox or use one of the other supported browsers (see list in the Hardware and Software Guide).

Whitespace in JDBC driver path does not work (UNIX / Linux only)

Product installation on UNIX / Linux systems fails if the required JDBC driver is located in a directory that contains whitespace in its name.

Solution

Provide the required JDBC driver in a directory without whitespace.

Handling of Corrupt Agent Configuration Files

In case an agent configuration file, e.g. fnet40_sys_env.sh, is corrupt, the user will be informed about the incident by a message dialog.

Solution

There is no workaround as there is no reliable possibility for fixing such configurations without any user interaction. But to prevent major data loss, before saving new versions of the agent configuration files, automatically a backup will be created for each of these files. The backup files will be placed in parallel to the normal/new configuration files and will be named like <filename>_yyyy.MM.dd.hh.mm.ss.bak. So each backup file will have a timestamp in its name indicating the time it was created. The backup files will not be removed or altered later on by the agent configurator. So from time to time the operator should delete no longer used/needed files.

Java Webstart tools (TeMA, MoMA, Installer) cannot save into Windows Explorer Symbolic Link structures

Due to a Java limitation it is not possible to store files from one of the Java Webstart Tools (Task Execution Manager, Monitoring Manager, Installer) into a directory that is based on the Windows symbolic link technology. This applies to Windows 7 and newer Windows systems. Example: At the end of an executed task by the TaskExecution Manager you are able to save the output as text file to a directory of your choice. Since Windows 7 and later personal directory structure "Documents and Setting" (or the corresponding language specific name) is just a symbolic link. The Java Webstart application cannot browse through this directory via the symbolic link and is therefore not able to save it at this location.

Solution

Instead of saving the files into one of the folders represented by symbolic links use a real directory to store the file

Precheck not executable on Windows machines when having chosen another IATEMPDIR due to not enough disk space

After choosing a different IA installer temp dir (IATEMPDIR) due to unsufficient disk space on C:/ while installing IBM ECM SM 5.1.0 FP3 under Windows, the precheck aborts with a "The system cannot find the file specified..." error, referring to the "precheck.cmd" file, that has obviously not been copied into the IA installer temp directory.Several return codes can occur in this case, for example 2 or 4.

Solution

When setting the IATEMPDIR to a directory with enough disk space before starting the installation (like "E:/tmp"), the installer precheck runs through without an error returning the "0" code.

Number of MBeans vary between classic JMX MBean request (over RMI) and over ECM WebService in Oracle WebLogic environments.

The classic JMX MBean request method via RMI returns more MBeans than the Webservice connection in Oracle WebLogic environments. So the View CE Status Task and the AE Status task won't run successful.

Solution

Use the classic JMX Mbean request (RMI connection) method to connect to the WebLogic servers, if you want to have full MBean support. Since it also might be a policy configuration issue, a WebLogic admin might be able to configure the webservice application in a way to grant full MBean access in webservice mode too.

Multiple selected items in the KnowledgeBase list view are not deletable

The user can select several items, but only the first element in the list of selected elements is deleted by selecting "Delete".

Solution

There is no workaround. The user must delete items one by one.

Leave full screen menu entry always visible

The Leave full screen menu entry is alway shown, although the view is not in full screen.

Solution

There is currently no workaround as it is technically not possible to disable or hide this entry based upon the view mode (full screen or framed).

InstallAnywhere installer sometimes fails in re-checks under IBM AIX

The InstallAnywhere installer has problems under AIX to jump back in case a necessary check fails and the user should re-enter something or must execute a specific task to proceed with the installation.

Example: The /etc/cenit directory must be writable for the installer. If the directory does not exist or is not writable, the installer stops and informs the user about that issue. The user must now create the directory or change the permission of it. After that the installer re-tries the writeability of the directory. In case the test fails again the user should be informed about it, again. Under AIX the back-jump in case the test fails again is broken, so a still non-existent or non-writable directory will not be signaled to the user, but the installer will proceed without any further warnings.

Solution

There is no workaround at the moment.

Large tooltips may be cropped when they are close to the borders of the GUI

Large tooltips, e.g. the ones for the message field in the event list view, will be cropped at the borders of the GUI if they are too wide or high. The RAP framework does this automatically, it cannot be influenced at the moment.

Solution

No real workarounds are possible. Suggestions are to move the part of the GUI, the tooltips belong to away from the borders or to increase the size of the GUI window.

MS Windows Eventlog Entry for a SideBySide Error

During the installation there can be a so called SideBySide error log entry in the event log in case the installation was made under MS Windows. The message can look like the following example:

Activation context generation failed for "C:\Program Files\MyServerRoot\Uninstall_CALA_REX_Agent\resource\IconChng.exe". Dependent Assembly Microsoft.VC90.MFC,processorArchitecture="x86",publicKeyToken="1fc8b3b9a1e18e3b",type="win32",version="9.0.21022.8" could not be found. Please use sxstrace.exe for detailed diagnosis.

This indicates a missing MS Visual C++ Redistributable package in the system. It does not affect the functionality of the event server or the GUI.

Solution

To get rid of it the MS Visual C++ Redistributable package has to be installed. The correct version is the one mentioned in the log message itself. The entry point for the download are

"http://www.microsoft.com/en-us/download/details.aspx?id=5555" and "http://www.microsoft.com/en-us/download/details.aspx?id=30679". But

you should search for the version you really need and that is provided in the log message; in the example it would be 9.0.21022.8.

Accumulators and their statistics aka Durations normally not shown in PCH events

Due to some incompatibilities between the Listener implementation and the needs of a monitor, the limitation for Accumulators and their statistic values still exists. It is of no value for a monitor to report the statistic values of a duration like "average", "min", "max", and "standard deviation" as the time base for the calculation of these values for a monitor run must be set to a value too short to create any meaningful data. Unfortunately the period can not be made to last longer as the monitor would run too long for any customer environment.

Solution

Currently there is no workaround for this issue as the Listener API is limited here and not fully compatible with the needs of a monitor.

JPS Config is destroyed by missing path

When a JPS monitor is configured (PCH, JMX, WMI, ESX) the monitor configuration is stored as xml config file. If the config file is reopened in the JPS editor another time to change the settings and one of the specified paths is missing at this time, the configuration is destroyed in this moment, because the path, which is defined in the config file cannot be found. If "Ok" is clicked and so the configuration is stored again, the configuration will be damaged and invalid. This will lead to an incomplete request of the monitor values. It is recommended not to save such configurations. They can be recognized by only one element ("Structure") in the filter tree (left hand of the editor).

Solution

If there are paths, where there is the possibility that they are not available at any time, it is recommended to copy the jps xml monitor configurations, which are stored at "$CENIT_ROOT/cala/monitors/pam". If you recognize such changes, in an opened JPS editor ("Structure" element at left hand tree) do not exit the configurator via "Ok", but "Cancel" and do not save the cala monitor configuration.

Empty hosts in SYSTEM-Tree

On systems, where there are several internal IP addresses, which are mapped to one outgoing IP address, the event is sent with a "adapter host" and "adapter IP address", which may differ from the "source host" and the "source IP address". This leads to two hosts in the SYSTEM-Tree in the gui. One with the "adapter host" name, which contains the events and another with the "source host" name, which does not contain any events. But the events are related to the "source host" as well. If you delete the empty "source host" from the SYSTEM-Tree, the related events will be deleted too (which are seen in the "adapter host"). So in this constellation, there will be an empty host seen in the tree.

Solution

None

Concurrency problems, when starting WebStart tools, when already a PDF document is opened in the browser.

When a PDF document (Any users guide of the product) is opened and then a WebStart tool is opened, the browser windows showing the PDF document will be closed automatically and the RAP gui will disappear. An error message is shown in the browser instead.

Solution

There are three workarounds 1. Use Firefox or Chrome as browser, the problem does not occur with these browsers. 2. Save the documentation as local copy and use external Acrobat Reader to open the document. 3. Open the client tools before opening a pdf document with the browser.

PeInfo Monitors need at least Java 5

The IBM FileNet ProcessEngine 4.0 FP 2 for Solaris and HP-UX are using a Java2 1.4 JRE, which breaks the PeInfo monitors as these require at least Java 5.

The PeInfo monitors are bound to use the JRE shipped with the PE that is to be monitored. For technical reasons the PeInfo monitor JARs are compiled with Java 5 which leads to the restriction they cannot be used with a Java version lower than 5.0.

Solution

There is no workaround for this issue.

Server restart required after changing the suppression_key

After changing the suppression_key and pressing "Apply changes", the changes are saved but active.

Solution

After a restart of the event server, the changes are in use.

stderr and stdout logs on windows grow infinitely

The log files server.stdout.log, server.stderr.log, gui.stdout.log and gui.stderr.log are enabled by default when the services are installed on Windows. The log files grow infinitely and cannot be deleted, while the server is running.

Solution

As long as teh log level is on the default log level (ERROR) the file won't grow much. Only when logging runs on a finer log level, this is a problem. It is recommended to delete the files manually after the errors were resolved.

The state of the CALA_REX clients is not updated in the clientconsole on Unix based installations

The state (online/offline) of the CALA_REX clients is notupdated in the GUI in case the product is installed on a Unix(TM) orLinux(TM) system.The Task Execution Manager and the Monitoring Manager tools are not affected by this issue. They will work as expected.

Solution

There is no workaround at the moment.

GUI of task Configure Datacap Database Settings is not updated correctly

Under certain circumstances, the GUI for the task Configure Datacap Database Settings is not updated correctly in CalaTEMa.

All cases are related to changing the content of the comboboxes for Datacap Application Name and Database Parameter Set by typing in them instead of selecting an existing item:

This happens due to restrictions in the GUI handling inside CalaTEMa.

Solution

In the first case, select the existing item from the combobox instead of typing in the name.

In the second case, press the Reload list button after defining the first Database Parameter Set of the new Datacap Application to update the Database Parameter Set combobox with the correct settings.

Host name 'localhost' not supported by several monitors

Users of the product should not use the host name 'localhost', because several monitors don't support this commonly unsupported host name

Solution

Change the host name of the system to a real name that fits host name naming conventions of your enterprise

Don't use a "suppression_key" in the SMTP sink email suppression that exists in the event duplicate detection

The email suppression in the SMTP sink has a bug that overwrites duplicate detection entries in the event duplicate detection.

Solution

Prevent using a "suppresion_key" that exists in the event duplicate detection

Incorrect 3rd Party log entry in GUI log file

The following 3rd party log entry is written into the GUI logfiles gui*.log: Warning: Detected recursive attempt by part de.cenit.eb.sm.finca.gui.knowledgebaseentryview.view to create itself (this is probably, but not necessarily, a bug)

Due to limitations this log file entry, which doesn't describe an error in this case, cannot be suppressed.

Solution

No workaround available and necessary, ignore this message.

FP Call 3681: MBean-Attributes of the type javax.management.openmbean.CompositeData are not supported

The MBean-Attributes with the name javax.management.openmbean.CompositeData are not supported by the monitoring components. So they are filtered out to avoid configuring monitors with these values.

Solution

There is no workaround necesarry for this problem.

Default user for automatic task execution does not work when using LDAP authentication

With 4.5.0 FP1, a default user "cala_rex" and password is defined in the finca-cfg.xml for use with automatic and scheduled task execution. This user works only if ECM SM internal authentication is used. If LDAP authentication is configured, the user must be an LDAP user or the user "admin".

Solution

The LDAP user and password for automatic task execution must be manually changed in the finca-cfg.xml file, section cala_rex. The password can be encrypted using the Monitoring Manager tool "Password dialog". The change must be repeated after each fixpack or interim fix installation.

XML Configuration Files are not transferred to other clients

If a monitor configuration containing JPS monitors is copied to another client via menu 'File > Save monitor configuration to client...', the JPS related XML files will not be copied.

Solution

Copy the XML files manually from one client to another.

Different web-browsers provide different search results in online help

If you search for a specific key word in the web application's online help, the help provides different search results (chapters), depending on the web browser, which is used. Since the online help is not implemented by us, but provided by the framework, we assume, that this will be fixed in RAP version 1.4.

Solution

The PDF documentation contains the same help, as the online help. The text search of the PDF viewer can be used to find more specific results.

/usr/bin/tr does not work as expected onSolaris 64 bit

/usr/bin/tr does not work as expectedon Solaris 64 bit machines if the LC_* environment variables are setto UTF-8. This causes some tasks and monitors to fail.

Solution

Use the tr command from/usr/xpg4/bin:

Directory tools contains wrong cosst binary on Solaris 64 bitclients

The tools directory on Solaris 64 bit clients contains aversion of the cosst tool that only returns 32bit processes correctlywhen executing cosst procinfo -l. This leads to incorrect monitorresults when monitoring 64bit processes, e.g. in the CALAProcessesmonitor.A corrected version of the binary is available in the cala directoryafter installing the client but it is not copied automatically to thetools directory.

Solution

Copy the binary manually to the correct location.

In addition, you must replace the the CALARexProcesses monitoron the Solaris 64 bit client with the ProcessInstances monitor to makesure that monitoring of the CALA_REX process workscorrectly:

cosst.exe fails on older Windows 2003 and Windows XP versions

The command line tool cosst.exe fails on Windows 2003 that doesn't have at least Windows 2003 SP1 applied and on Windows XP without applied XP SP2 or newer.

Solution

Update the Windows system to the minimum supported Service Pack or use cosst.exe from version 4.0.1

swapSpace monitor on AIX requires root

The Swapspace monitor doesn't run on non-root AIX CALA installations.

Solution

Run the monitor as root or add the s bit to the pstat binary e.g. chmod u+s /usr/sbin/pstat In addition to the s bit the file /dev/mem needs read right for others e.g. chmod o+r /dev/mem

CALA's logctlcmd command on Windows 2008 does only work when run under the same user as the CALA service.

When running the CALA command logctlcmd manually on Windows 2008, the command must be run under the same user as the service. It won't work when started under an other user and will also break subsequent calls of logctlcmd under the service user. (You will need to restart the CALA service to make it work again.)

Solution

None

It is not possible to monitor a service where the display namecontains German umlaute.

It is not possible to monitor a service where the display namecontains German umlaute, e.g. "Dienst für virtuelleDatenträger (VDS)".The (windows shell) awk command fails to search for the string.

Solution

Replace the German umlaute with regular expressionwildcards, e.g.. (dot): "Dienst f.r virtuelle Datentr.ger (VDS)"

File jmx_classpaths.prop is overwritten during client update

The $CENIT_ROOT/cala/monitors/pam/jmx_classpaths.prop file is overwritten by CALA client installer if the option "Reconfigure only" is not selected. Changes made via task "Configure JMX Classpath" or manually to the file are overwritten.

This causes JMX monitors to fail if the classpath was extended before to make them work.

Solution

Create a backup of the file before updating the CALA client. After the update, merge your changes into the new file created by the installer.

Monitoring Manager saves escalation table content and name, even if the Cancel button is pressed

If the name of the escalation file is changed or the contents of the specified escalation file is changed within the Monitoring Manager, the changes are done even if the edited monitor instance is closed without saving (pressing the X or Cancel button).

This behaviour cannot be changed in this version of the Monitoring Manager but will be changed in the next implementation of this administrative GUI.

Solution

Use unique escalation files names in the case of copying monitor instances within the Monitoring manager to minimize the risk of this behaviour.

When specifying custom execution hours over midnight the monitor stops running at midnight.

When specifying execution hours crossing midnight, e.g. 16-8 if the monitor should run from 16:00 till 8:59, the monitor stops working at midnight.

Solution

Specify execution hours in this format: 16-23,0-8

Oracle variable TNS_ADMIN not evaluated by Oracle monitors, connect fails

When starting CALA out of a user's (fnsw) crontab, the .profile of this user is not sourced. If the .profile contains the user variable TNS_ADMIN pointing to a different location than $ORACLE_HOME/network/admin the tnsnames.ora will not be found and the connect to the database will fail with error "ORA-12154: TNS:could not resolve service name".

The search order for tnsnames.ora is:

Solution

Create a special CALA start script containing this varaible and use this instead of cala.sh inside the crontab.

After startup / restart of CALA, monitors can return -30 or ERROR_installation if the system has got a very high load.

After startup / restart of CALA, monitors can return -30 or ERROR_installation if the system has got a very high load. Since FSM 4.0.1 FP2 a new functionality is implemented that limits the number of parallel executed IBM FileNet and Database command line tools to one instance each. If more monitors are running that need to execute the same command line tool, the monitor waits up to 150 seconds to proceed. If this period is not long enough, this normally indicates an system issue.

Solution

Wait for the next monitoring schedule or change the execution period of some monitors to reduce the number of parallel running monitors.

Linux: cosst procinfo -l only returns 4096 characters per process

Due to a system limitation on Linux systems, cosst procinfo -l can only show the first 4096 characters from the command line of a process.

The following monitors and tasks may be affected:

IBM CM v8, CMOD, CSx and IICE archives:

IBM FileNet P8 4.x archives:

IBM FileNet P8 PE 3.x, PA 3.x archives:

IBM FileNet Image Manager 3.x and 4.x archives:

Standard archives:

Solution

None

HP-UX 10.20 and 11: cosst procinfo -l only returns 1024 characters per process

Due to a system limitation on HP-UX 10.20 and 11, cosst procinfo -l can only show the first 1024 characters from the command line of a process.

The following monitors and tasks may be affected:

IBM CM v8, CMOD, CSx and IICE archives:

IBM FileNet P8 4.x archives:

IBM FileNet P8 PE 3.x, PA 3.x archives:

IBM FileNet Image Manager 3.x and 4.x archives:

Standard archives:

Solution

None

Monitor Publishing Queue Entries does not work for object stores containing blanks

Due to restrictions in the FileNet Java API, the monitor Publishing Queue Entries does not work for object stores that contain blanks in the name.

Solution

None

Reading ibm_cm8_eventlog events only works with standard port

Reading ibm_cm8_eventlog events from the tables ICMSTSYSADMEVENTS and ICMSTITEMEVENTS only works if the DB2 RDBMS uses the standard port 50000.

Solution

If the DB2 RDBMS uses a different port than 50000, the CALA configuration must be adjusted manually after each reinstall.

Windows security policy may block access to directories containing a dot

On some Windows installations, the security policy does not allow access to directories that contain a dot. In this case, the ECM SM Java Webstart tools do not work because some of the required files are located in directories containing a dot (e.g. de.cenit).

Solution

Rename the affected directories (e.g. rename de.cenit to de_cenit) and adjust the jnlp files accordingly.

At least one logfile must be activated on ECM SM client

If only monitors are activated for a ECM SM client, the installation fails with an error message.

Solution

In the client configuration, select at least one logfile or the Windows eventlog.

Custom script monitors should not contain blank in path or file name

When defining custom script monitors with blanks in the path or the monitor file name itself (for example C:/Program Files/scripts/custmon1.cmd) the monitoring engine calamon will fail for this monitoring instance. Additionally the monitor definition table from the system can no longer be loaded by the Monitoring Manager GUI (CalaMoMa).

Solution

If it is required to define a monitor with blanks in the file or path name, the DOS name (sometimes called short name) needs to be used, for instance C:/PROGRA~1/scripts/custmon1.cmd.

CALA sometimes fails because files are in use

On some systems, CALA installation fails because files are in use by the running CALA processes and cannot be replaced by the installer.

Solution

Stop CALA before re-installing the client.

CMOD log entries from the database have wrong timestamp

Content Manager On Demand log entries from the table SL2 in the CMOD database are always sent with the current time instead of the timestamp stored in the database.

Solution

None

Monitor OracleNextExtent works for dictionary-managed tablespaces only

The OracleNextExtent monitor can only be used for dictionary-managed tablespaces. The system table entries for locally managed tablespaces do not provide information about extents. Starting with version 9.2, Oracle Corporation recommends usage of locally managed tablespaces.

See Oracle 9i Database Concepts and Oracle 9i SQL Reference for details about tablespaces and extent management.

Solution

None

Antivirus / Spyware alert during ECM SM install

During ECM SM install some AntiVirus / Spyware tool alert a Tool-Wget Virus. For example McAfee detected the file C:\Program Files\IBM\ECMSM\shell\wget3048.rra when installing the Windows Client (CALA_REX).

Solution

Stop the AntiVirus agent / Service during ECM SM client install and restart it after successful installation.

MS Windows: CALA update fails if service is selected in Services Control Panel

The ECM SM CALA Service (cala_srv) cannot be registered during update while the existing service entry is selected in the Services Control Panel. The installation logfile caladist.log shows one of the following messages:

1072 The specified service has been marked for deletion.

1075 The dependency service does not exist or has been marked for deletion.

Solution

Do not select the ECM SM CALA Service in the Windows Control Panel during CALA update.

User names or password with ';' or '#' are not supported.

User names or password of used OS or database accounts that contain semicolon (';') or hashes ('#') are not supported with ECM SM.

Solution

Another account or password should be used to configure ECM SM.

Name of start / stop script of Application server must not contain blanks

In the configuration plugin for Process Engine 3.5 and older, a start and a stop script can be given for an Application server. The filenames specified in these fields must not contain blanks, otherwise configuration will fail.

Solution

Use the short name (DOS name) of the file.

To determine the short name of a file, you can open a DOS box and execute the command <path_to_fsm_installation>/tools/cosst.exe shortname "<very long path/that contains blanks/and points to script>".

MS Windows: logfiles cannot be deleted while ascfileread reads from them

Logfiles which are opened by ascfileread cannot be deleted on Windows platforms until ascfileread has been shut down.

Solution

Stop CALA, delete the file and restart CALA.