none
FAST ESP 5.3 Upgrade to FSIA

    Pergunta

  • Can someone provide a URL or documentation with the specific steps to upgrade an existing FAST ESP 5.3 installation in place to FSIA?

    Thanks, Tom

    sexta-feira, 10 de dezembro de 2010 21:30

Todas as Respostas

  • Hi Tom,

    ESP 5.3 and FSIA are actually the same product.  In case you were referring to the Fast Search for Internet Sites product (FSIS), I am including instructions on how to perform the upgrade from ESP 5.3 to FSIS here.

    To upgrade ESP 5.3 to FSIS, all you need to do is install the Feeding Overlay.  The FSIS FeedingOverlay for ESP can technically be installed at any point before you actually try and feed data into ESP via the processes.  The overlay consists of two FSIS specific pipeline stages that are added to the ESP installation.  Below is some conceptual information about the install process:

    Installing the ESP Feeding overlay
    The ESP feeding overlay is available as a self-extracting .exe as part of the FSIS installation package.

    It is installed by copying the contents of the etc and lib folders contained in the overlay archive into the corresponding etc and lib folders of the ESP installation directory.

    After restarting ESP, you will find two new FSIS specific stages are now available in the document processing section of the ESP admin GUI: CtsParser and CtsAnnotationsImporter.

    CtsParser creates document attributes based on XML data from CTS.

    Parse XML data from the input attribute and create one new document attribute per record field found in the parsed data.  If the original record field is a textual field the content of the created document attribute is a TextChunk, preparing the document attribute for tokenization in the same way the FastHtmlParser does.

    CtsParser has to be invoked directly after the DocInit stage.

    Configuration parameters:
    ·         InputAttribute: name of the document attribute that holds XML data from CTS

    ·         CtsParagraphAnnotation: name of paragraph annotations in the CTS system

    ·         AttributeFilter: space-separated list of the document attributes that should be created. Document attributes will be generated for CTS record fields whose names are listed in this parameter. All other CTS record fields will be ignored. If this parameter is left blank document attributes will be generated for every CTS record field.

    CtsAnnotationsImporter reads annotations from CTS parsed data and add them to the corresponding document attribute.

    This stage depends on information added by the CtsParser stage and the Tokenizer, so it needs to placed in the pipeline after these two stages.

    Configuration Parameters:
    ·         AnnotationFilter: space-separated list of the annotations that should be created. Only annotations whose names are in this parameter will be added to document attributes. All other annotations will be ignored. If this parameter is left blank all annotation will be imported.

    ·         AnnotationOutput: space-separated list of annotation output document attributes. Each element in the list should be in the format annotationName:attributeName, indicating that the surface form of all annotations with the name annotationName should be added to a document attribute with the name annotationName. Annotations will be concatenated using the separator specified in the Separator parameter.

    Additional notes
    The CtsAnnotationsImporter stage imports the annotations from the Content Transformation engine.

    By default, the imported annotations are not added to any document field in ESP. To add the imported annotations to document fields, it is necessary to update the AnnotationOutput parameter in the CtsAnnotationsImporter stage either by editing the CtsImporter.xml file or by using the ESP admin GUI.

    The following is an example of the CtsAnnotationImporter stage configuration to add personname annotations to a field called personnames and location annotations to a field called locations, as well as company annotations to a field called companies:

    <processor>

    <load module="processors.CtsImporter" class="CtsAnnotationsImporter"/>

    <config>

    <param name="AnnotationFilter" value="" type="str"/>

    <param name="AnnotationOutput" value="personname:personnames location:locations company:companies" type="str"/>

    <param name="Separator" value=";" type="str"/>

    </config>

    </processor>

    Instead of directly modifying the CtsImporter.xml file, it is recommended you create a custom pipeline stage which is based on the CtsAnnotationsImporter stage by creating a copy of the original stage using the admin GUI of ESP. The advantage of this approach is that the default pipeline stage is not modified.

     ·         Separator: separator used to concatenate annotations

    Installation Steps

    The steps to upgrade ESP 5.3 to FSIS would be similar to the below:
    1) Make sure ESP 5.3 is patched to SP3 or SP3.

    2) Navigate to the FeedingOverlay.exe self extracting archive

    3) Double click on FeedingOverlay.exe to launch the WinZip Self-Extractor, and unzip them to an installation folder

    4) Through Windows Explorer access the newly extracted ‘C:\FeedingOverlay’ directory

    5)  Hold the Ctrl key and select both directories.  Then drag them and drop them in the %FASTSEARCH% directory.

    6)  When prompted with the Confirm Folder Replace menu for ‘etc’ click ‘Yes’.

    7)  When prompted with the Confirm Folder Replace menu for ‘lib’ click ‘Yes’.

    8)    The overlay is now installed, close Windows Explorer.

    9)   Though the overlay is now installed in ESP, you will still need to restart the Document Processor before they are available for inclusion in a document pipeline.

    After completing the install, you can still use ESP independently of FSIS, and vice versa.  Let me know if you have more questions about the install process.

    Thanks!
    Rob Vazzana | Microsoft | Enterprise Search Group | Senior Support Engineer | http://www.microsoft.com/enterprisesearch


    segunda-feira, 13 de dezembro de 2010 22:03