Wednesday Jun 24, 2015

Keeping Your WebCenter Content File System Clean

If you have configured WebCenter Content (WCC) to store your content in the database you are probably not expecting it to be storing any content on the file system. You may choose to go with a minimal file system to hold the vault and weblayout directories. You need to be aware that WCC will store content on the file system for two reasons:

  1. During the check in process WCC needs to store the native file in the vault directory so it is available for full-text indexing and for rendition processing. Any webviewable files are also stored on the file system after they are generated by the Inbound Refinery (IBR).
  2. Whenever a content item is retrieved, WCC will store the file on the file system after retrieving it from the database. The file system is used as a cache so that if the content item is retrieved again it can be retrieved from the file system instead of the database.

For #1 this can have an impact when you are either loading a large amount of content into WCC or rebuilding your full-text index. When doing a full-text index rebuild every file in WCC will be retrieved from the database and stored on the file system.

If you want the files on the file system removed after the content item has been indexed you can add this configuration setting:


If you want to keep the file system as small as possible you can also add these configuration settings to minimize the amount of files keep on the file system as a cache:


Monday Dec 03, 2012

Migrating from Oracle Content DB to WebCenter Content

Someone asked about migrating from Oracle Content DB (CDB) to WebCenter Content (WCC)

There is no product offering that supports migrating content from CDB to WCC. Oracle Consulting has done some migrations for customers and has some internal tools that are available to assist with the migration. If you need assistance migrating from CDB to WCC you should contact Oracle Consulting.

Wednesday Oct 17, 2012

A very useful custom component

Whenever I am debugging a problem in WebCenter Content (WCC) I often find it useful to see the contents of the internal data binder used by WCC when executing a service. I want to know the value of all parameters passed in by the caller, either a user in the web GUI or from an application calling the service via RIDC or web services. I also want to the know the value of binder variables calculated by WCC as it processes a service. What defaults has it applied based on configuration settings or profile rules? What values has it derived based on the user input?

To help with this I created a  component that uses a java filter to dump out the contents of the internal data binder to the WCC trace file. It dumps the binder contents using the toString() method. You can register this filter code using many different filter hooks to see how the binder is updated as WCC processes the service.

By default, it uses the validateStandard filter hook which is useful during a CHECKIN service. It uses the system trace section, so make sure that trace section is enabled before looking for the output from this component.

Here is some sample output

>system/6    10.09 09:57:40.648    IdcServer-1    filter: postParseDataForServiceRequest, binder start --
 system/6    10.09 09:57:40.698    IdcServer-1    *** LocalData ***
 system/6    10.09 09:57:40.698    IdcServer-1    (10 keys + 0 defaults)
 system/6    10.09 09:57:40.698    IdcServer-1    ClientEncoding=UTF-8
 system/6    10.09 09:57:40.698    IdcServer-1    IdcService=CHECKIN_UNIVERSAL
 system/6    10.09 09:57:40.698    IdcServer-1    NoHttpHeaders=0
 system/6    10.09 09:57:40.698    IdcServer-1    UserDateFormat=iso8601
 system/6    10.09 09:57:40.698    IdcServer-1    UserTimeZone=UTC
 system/6    10.09 09:57:40.698    IdcServer-1    dDocTitle=Check in from RIDC using Framework Folder
 system/6    10.09 09:57:40.698    IdcServer-1    dDocType=Document
 system/6    10.09 09:57:40.698    IdcServer-1    dSecurityGroup=Public
 system/6    10.09 09:57:40.698    IdcServer-1    parentFolderPath=/folder1/folder2
 system/6    10.09 09:57:40.698    IdcServer-1    primaryFile=testfile5.bin   
 system/6    10.09 09:57:40.698    IdcServer-1    ***  RESULT SETS  ***
>system/6    10.09 09:57:40.698    IdcServer-1    binder end --------------------------------------------

See the readme included in the component for more details.

You can download the component from here.

Tuesday Oct 16, 2012

Framework Folders and Duplicate File Names

I have been working with Framework folders a little bit in the past few days and found one unexpected behavior that is different from Contribution Folders (Folders_g). If you try and check a file into a Framework Folder that already exists in the folder it will allow it and rename the file for you.

In Folders_g this would have generated an error and prevented you from checking in the file.

A quick check of the Framework Folder configuration settings in the Application Administrator’s Guide for Content Server does not show a configuration parameter to control this. I'm still thinking about this and not sure if I like this new behavior or not. I guess from a user perspective this more closely aligns Framework Folders to how Windows handle duplicate file names, but if you are migrating from Folders_g and expect a duplicate file name to be rejected, this might cause you some problems.

Friday Oct 05, 2012

Solving Inbound Refinery PDF Conversion Issues, Part 1

Working with Inbound Refinery (IBR)  and PDF Conversion can be very frustrating. When everything is working smoothly you kind of forgot it is even there. Documents are cheeked into WebCenter Content (WCC), sent to IBR for conversion, converted to PDF, returned to WCC, and viola your Office documents have a nice PDF rendition available for viewing.

Then a user checks in a bunch of password protected Word files, the conversions fail, your IBR queue starts backing up, users start calling asking why their document have not been released yet, and your spend a frustrating afternoon trying to recover and get things back running properly again.

Password protected documents are one cause of PDF conversion failures, and I will cover those in a future blog post, but there are many other problems that can cause conversions to fail, especially when working with the WinNativeConverter and using the native applications, e.g. Word, to convert a document to PDF. There are other conversion options like PDFExportConverter which uses Oracle OutsideIn to convert documents directly to PDF without the need for the native applications. However, to get the best fidelity to the original document the native applications must be used. Many customers have tried PDFExportConverter, but have stayed with the native applications for conversion since the conversion results from PDFExportConverter were not as good as when the native applications are used.

Of course it is important to mention that most IBR "problems" are not really problems with the IBR itself, but more likely problems with the documents being converted or problems with the third-party applications IBR is using to convert the documents to PDF.

One problem I ran into recently, that at least has a easy solution, are Word documents that display a Show Repairs dialog when the document is opened. If you open the problem document yourself you will see this dialog.

This will cause the conversion to time out. Any time the native application displays a dialog that requires user input the conversion will time out. The solution is to set add a setting for BulletProofOnCorruption to the registry for the user running Word on the IBR server. See this support note from Microsoft for details. The support note says to set the registry key under HKEY_CURRENT_USER, but since we are running IBR as a service the correct location is under HKEY_USERS\.DEFAULT. Also since in our environment we were using Office 2007, the correct registry key to use was:


Once you have done this restart the IBR managed server and resubmit your problem document. It should now be converted successfully.

For more details on IBR see the Oracle® WebCenter Content Administrator's Guide for Conversion.

Wednesday Sep 26, 2012

How to run RCU from the command line

When I was trying to figure out how to run RCU on 64-bit Linux I found this post. It shows how to run RCU from the command line. It didn't actually work for me, so you can see my post on how to run RCU on 64-bit Linux.

But, seeing how to run RCU from the command got me started thinking about running RCU from the command line to create the schema for WebCenter Content. That post got me part of the way there since it shows how run RCU silently from the command line, but to do this you need to know the name of the RCU component for WebCenter Content. I poked around in the RCU files and found the component name for WCC is CONTENTSERVER11. There is a contentserver11 directory in rcuHome/rcu/integration and when you look at the contentserver11.xml file you will see

<RepositoryConfig COMP_ID="CONTENTSERVER11">

With the component name for WCC in hand I was able to use this command line to run RCU and create the schema for WCC.

.../rcuHome/bin/rcu -silent -createRepository -databaseType ORACLE -connectString localhost:1521:orcl1 -dbUser sys -dbRole sysdba -schemaPrefix TEST -component CONTENTSERVER11 -f <rcu_passwords.txt

To make the silent part work and not have it prompt you for the passwords needed (sys password and password for each schema) you use the -f option and specify a file containing the passwords, one per line, in the order the components are listed on the -component argument. Here is the output from rcu when I ran the above command.

Processing command line ....
Repository Creation Utility - Checking Prerequisites
Checking Global Prerequisites

Repository Creation Utility - Checking Prerequisites
Checking Component Prerequisites
Repository Creation Utility - Creating Tablespaces
Validating and Creating Tablespaces
Repository Creation Utility - Create
Repository Create in progress.
Percent Complete: 0
Percent Complete: 100
Repository Creation Utility: Create - Completion Summary
Database details:
Host Name              : localhost
Port                   : 1521
Service Name           : ORCL1
Connected As           : sys
Prefix for (prefixable) Schema Owners : TEST
RCU Logfile            : /u01/app/oracle/logdir.2012-09-26_07-53/rcu.log
Component schemas created:
Component                            Status  Logfile
Oracle Content Server 11g - Complete Success /u01/app/oracle/logdir.2012-09-26_07-53/contentserver11.log

Repository Creation Utility - Create : Operation Completed

This works fine if you want to use the default tablespace sizes and options, but there does not seem to be a way to specify the tablespace options on the command line. You can specify the name of the tablespace and temp tablespace, but they must already exist in the database before running RCU. I guess you can always create the tablespaces first using your desired sizes and options and then run RCU and specify the tablespaces you created.

When looking up the command line options in the RCU doc I found it has the list of components for each product that it supports. See Appendix B in the RCU User's Guide.

WebCenter Content (WCC) Trace Sections

Kyle has a good post on how to modify the size and number of WebCenter Content (WCC) trace files. His post reminded me I have been meaning to write a post on WCC trace sections for a while.

searchcache - Tells you if you query was found in the WCC search cache.

searchquery - Shows the processing of the query as it is converted form what the user submitted to the end query that will be sent to the database. Shows conversion from the universal query syntax to the syntax specific to the search solution WCC is configured to use.

services (verbose) - Lists the filters that are called for each service. This will let you know what filters are available for each service and will also tell you what filters are used by WCC add-on components and any custom components you have installed. The How To Component Sample has a list of filters, but it has not been updated since 7.5, so it is a little outdated now. With each new release WCC adds more filters. If you have a filter that has no code attached to it you will see output like this:

services/6    09.25 06:40:26.270    IdcServer-423    Called filter event computeDocName with no filter plugins registered
When a WCC add-on or custom component uses a filter you will see trace output like this:
services/6    09.25 06:40:26.275    IdcServer-423    Calling filter event postValidateCheckinData on class collections.CollectionValidateCheckinData with parameter postValidateCheckinData
services/6    09.25 06:40:26.275    IdcServer-423    Calling filter event postValidateCheckinData on class collections.CollectionFilters with parameter postValidateCheckinData
As you can see from this sample output it is possible to have multiple code points using the same filter.

systemdatabase - Dumps the database call AFTER it executes. This can be somewhat troublesome if you are trying to track down some weird database problems. We had a problem where WCC was getting into a deadlock situation. We turned on the systemdatabase trace section and thought we had the problem database call, but it turned out since it printed out the database call after it was executed we were looking at the database call BEFORE the one causing the deadlock. We ended up having to turn on tracing at the database level to see the database call WCC was making that was causing the deadlock.

socketrequests (verbose) - dumps the actual messages received and sent over the socket connection by WCC for a service. If you have gzip enabled you will see junk on the response coming back from WCC. For debugging disable the gzip of the WCC response.Here is an example of the dump of the request for a GET_SEARCH_RESULTS service call.

socketrequests/6 09.25 06:46:02.501 IdcServer-6 request: REMOTE_USER=sysadmin.USER-AGENT=Java;.Stel
socketrequests/6 09.25 06:46:02.501 IdcServer-6 request: lent.CIS.11g.CONTENT_TYPE=text/html.HEADER
socketrequests/6 09.25 06:46:02.501 IdcServer-6 request: _ENCODING=UTF-8.REQUEST_METHOD=POST.CONTEN
socketrequests/6 09.25 06:46:02.501 IdcServer-6 request: T_LENGTH=270.HTTP_HOST=CIS.$$$$.NoHttpHead
socketrequests/6 09.25 06:46:02.501 IdcServer-6 request: ers=0.IsJava=1.IdcService=GET_SEARCH_RESUL
socketrequests/6 09.25 06:46:02.501 IdcServer-6 request: TS...?hda.jcharset=UTF-8?...@Properties.Lo
socketrequests/6 09.25 06:46:02.501 IdcServer-6 request: calData.SortField=dDocName.ClientEncoding=
socketrequests/6 09.25 06:46:02.501 IdcServer-6 request: UTF-8.IdcService=GET_SEARCH_RESULTS.UserTi
socketrequests/6 09.25 06:46:02.501 IdcServer-6 request: meZone=UTC.UserDateFormat=iso8601.SortDesc
socketrequests/6 09.25 06:46:02.501 IdcServer-6 request: =ASC.QueryText=dDocType..matches..`Documen
socketrequests/6 09.25 06:46:02.501 IdcServer-6 request: t`.@end.

userstorage, jps - Provides trace details for user authentication and authorization. Includes information on the determination of what roles and accounts a user has access to. In 11g a new trace section, jps, was added with the addition of the JpsUserProvider to communicate with WebLogic Server.

The WCC developers decide when to use the verbose option for their trace output, so sometime you need to try verbose to see what different information you get.

One of the things I would always have liked to see if the ability to turn on verbose output selectively for individual trace sections. When you turn on verbose output you get it for all trace sections you have enabled. This can quickly fill up your trace files with a lot of information if you have the socket trace section turned on.

How to run Repository Creation Utility (RCU) on 64-bit Linux

I was setting up WebCenter Content (WCC) on a new virtual box running 64-bit Linux and ran into a problem when I tried to run the Repository Creation Utility (RCU). I saw this error when trying to start RCU

.../rcuHome/jdk/jre/bin/java: /lib/ bad ELF interpreter: No such file or directory

I think I remember running into this before and reading something about RCU only being supported on 32-bit Linux. I decided to try and see if I could get it to run on 64-bit Linux.

I saw it was using it's own copy of java (.../rcuHome/jdk/jre/bin/java), so I decided to try and get it to use the 64-bit JRockit I had already installed. I edited the rcu script in rcuHome/bin and replaced




Sure enough that fixed it. I was able to run RCU and create the WCC schema.

Tuesday Sep 25, 2012

A Generic RIDC Test Program

Many times I have found it useful to use a java program that communicates with WebCenter Content (WCC) using RIDC for testing. I might not have access to the web GUI or need to test a service running as a specific user. In the past I had created a number of "one off" programs that submitted specific services, e.g GET_SEARCH_RESULTS, DOCINFO, etc. Recently I decided to create a generic RIDC test program that could submit any service with the desired parameters based on a configuration file.

The programs gets the following information from the configuration file:

  • WCC connection information (host, port)
  • User to use to run service
  • Service to run
  • Any parameters for the service

The program will make a connection to the WCC server, send the service request, and print the results of the service call using the getResponseAsString() method.

Here is a sample configuration file:
idcservice.QueryText=dDocType <matches> `Document`

There is a readme file included in the zip with instructions for how to configure and run the program. The program takes one command line argument, the configuration file name. The configuration file name is optional and defaults to

If you have any suggestions for improvements let me know. Right now it only submits a single service call each time you run it. One enhancement I have already thought about would be to allow you to specify multiple services to tun in the configuration file. You can do that with the current program by having multiple configuration files and running the program multiple times, each with a different configuration file.

You can download the program here.

Friday Sep 21, 2012

Always disable the 8.3 name creation on Windows before installing WebCenter Content or WebLogic Server

You should always disable the 8.3 name creation feature when installing WebCenter Content on a Windows platform. The installs will normally work without it disabled, but you will find the weird 8.3 file and directory names in all the config files. Disabling it can also improve performance.

On Windows XP and Windows Server 2003 and above you can do it with this command:

fsutil.exe behavior set disable8dot3 1

To make sure it is disabled you can run this command to check:

fsutil.exe behavior query disable8dot3

If the 8.3 file name creation is disabled you will see the following output from the command:

The registry state of NtfsDisable8dot3NameCreation is 1 (Disable 8dot3 name creation on all volumes).

Here is a Microsoft note on how to do this on Windows 2000 and Windows NT.

How to Disable the 8.3 Name Creation on NTFS Partitions

I see now they have added this to the WebCenter Content Install Guide. They suggest doing it using a Windows registry settings.

Wednesday Sep 19, 2012

When is the default storage rule not really the default storage rule?

In 11g WebCenter Content (WCC) introduced dispersion rules in the vault and weblayout directory paths to better distribute content across the directories. The dispersion rule was based on dRevClassID. The only problem with this is that dRevClassID did not remain the same when you copied content from one WCC instance to another using Archiver like in a contribution-consumption scenario. This could cause problems because the web-viewable path would not be the same between the contribution and consumption instances.

In the PS5 ( release of WCC they addressed this by configuring the File Store Provider (FSP) so that all new content would use a storage rule with a dispersion rule based on dDocName, which would stay the same when content was copied to another WCC instance. To support migration from older versions of WCC they left the default storage rule unchanged and created a new storage rule called DispByContentId and made that the default storage rule for all new content.

I only stumbled upon this a while back when I was trying to change the FSP configuration so that all content used a webless storage rule. I changed the default storage rule, restarted WCC, and checked in a new content item. To my surprise the new content was not created as webless. I struggled with this for a while until I noticed there were multiple storage rules defined in the FSP configuration. When I looked at the default value for the xStorageRule field in Configuration Manager, sure enough it was no longer default, but was now DispByContentId.

Once I updated the DispByContentId storage rule to webless and restarted WCC all my new content was now created using the webless storage rule, just like I wanted.

I noticed when I was creating this blog post that the default storage rule is also listed on the File Store Provider Information page, but I guess I didn't see that when I originally did this.

Wednesday Jul 25, 2012

ScriptDebugTrace is now IsPageDebug in 11g

I keep forgetting what they changed this option to in 11g. Luckily Kyle has a nice blog post on it.

Friday May 27, 2011

WebDAV in UCM 11g

The format for the WebDAV URL has changed in 11g. It is now


If you are using OHS in front of UCM it will be


and you will have to add _dav location to themod_wl_ohs.conf file

   <Location /_dav>

P.S. If you are creating a connection from the Desktop Integration Suite to WebCenter Content make sure to include the / at the end of the URL. I left it off by mistake and it would not connect.

Friday May 20, 2011

More on File Store Provider

In PS4 ( you can now fully edit the FSP storage rule using the web interface instead of having to edit the hda files. I think this was actually added in a patch for PS3 ( This makes it easier to manage FSP storage rules.

When I was researching the FSP behavior I found this in the Admin User's Guide.

4.3.2 FileStoreProvider Upgrade

The FileStoreProvider component is installed, enabled, and upgraded by default for a new Oracle Content Server instance (with no documents in it). The upgrade includes creation of metadata fields with default values for the file store system (DefaultFileStore). Earlier versions of the Oracle Content Server software did not automatically upgrade the file store system, and sites could choose to not use the file store system, to not upgrade to use FileStoreProvider, and to uninstall the FileStoreProvider component and metadata fields. If you do not want to upgrade FileStoreProvider from your current settings, prior to installation you must enter the configuration variable FsAutoConfigure=false in the Additional Configuration Variables field on the General Configuration page of the Oracle Content Server Admin Server page.

I'm not exactly sure how you get to the Admin Server before you actually do the install. That would be quite a trick if you could do it. What I think they mean is that you need to add it to the config.cfg file that exists before you start the install, or actually before you create the WebLogic domain or add UCM to an existing domain. I have not tried this myself, but I would suspect that you would add it to the following file:


The other place you might want to try adding it is to the config.cfg that is created in the ucm/cs/config directory AFTER you add UCM to a WebLogic domain, but BEFORE you start it for the first time. I suspect that the FSP upgrade is done as part of the initial start up of UCM.

You would do this if you were installing a new 11g UCM instance and wanted it to work that same way as a 10g instance that was not using  the FSP.

Wednesday Mar 30, 2011

IBR 11g does not support 64-bit Ghostscript

I recently installed IBR 11g on Windows Server 2008 running on a 64-bit machine. When downloading Ghostscript there was a 64-bit version available so I thought I would use that.


gs901w32.exe, GPL Ghostscript 9.01 for 32-bit Windows (the common variety).

gs901w64.exe, GPL Ghostscript 9.01 for 64-bit Windows (x86_64).



When I tried to configure the Distiller engine in IBR I received this message:


Unable to load configuration options. The path to Ghostscript is not valid.

File 'E:/gs/gs9.01//bin/gsdll32.dll' does not exist.


It looks like IBR is hard coded to reference the 32-bit dll. I had to uninstall the 64-bit version of Ghostscript and install the 32-bit version. Then I was able to configure the distiller engine successfully.

I guess my first clue that the 64-bit version of Ghostscript was not going to work should have been the Win32 Ghostscript Settings heading on the Distiller Engine Options page.




Hopefully they will support 64-bit Ghostscript in a future release or patch.


Wednesday Feb 09, 2011

Changes to File Store Provider in UCM PS3

In the recent PS3 release of UCM ( there are some significant changes to the File Store Provider (FSP) configuration. For new PS3 installs (not upgrades from PS2) the FSP default storage rule includes a dispersion rule that will change the web-layout and vault paths by adding dispersion directories to the paths to limit the number of files in the vault and web-layout directories. What that means is that if you install a new PS3 UCM instance and migrate content in from a previous version of UCM the web URL will change. That is a critical problem for web sites and just general document management. Read the rest of this post where I describe the issue in detaill and provide instructions for how to modify a PS3 instance to use the old format for the web-layout path.[Read More]

Kevin Smith is a Technical Director in Oracle Consulting's WebCenter practice. He has been working with content management products since 2004 when he joined Stellent.


« June 2016