Index Status

Administrators open this window using Tools menu ➔ Index Status... The menu command is shown when the user's role has Collections/Rules administrator marked.

The Index Status tool shows the status of indexed items (documents, parts, change forms) and file attachments.

There are two index functions, with different purposes:

• A SQL Server database table index determines how fast a data table can be scanned to find a specific data record, and how fast it's linked to related records in other tables. These table indexes are used for item searches; making new items; opening existing records; running reports and transforms; and anything else that gets data from the SQL Server database.

• The search engine index selects documents, parts, change forms and file attachments in response to an Item Explorer search request. This index only affects what text can be found using the Item Explorer; it has no impact on how fast it can be found. This index has no performance effect on creating or viewing items; processing change forms; running reports; or using transforms.

Although both indexes are automatically managed, you can use the Re-index button to rebuild the database table indexes and then update the search engine index.

After you click the Re-index button, you can close the Index Status tool – the indexing process continues to run on the server computer until it's finished.

A typical computer indexes about 2000 items per minute, but indexing can be much faster or slower depending on many factors.

Database table index §

• Maintain database performance by re-indexing at least once a year. More frequent re-indexing is normally not needed.

• The Batch Importer tool can fragment database table indexes, particularly when importing items that have custom attributes. Re-index after you finish a large import project.

• If frequent re-indexing noticeably improves performance, consider scheduling a task to regularly de-fragment the table indexes. If you're comfortable using Transact-SQL, search the web for sql server reorganize rebuild fragmented index

Search engine index §

Item index §

• The search index is updated by the server computer's PDXpert Server service. The index service selects a new group of records from the index queue every few seconds.

• When you save a record, it's added to the service's indexing queue. If there's nothing else in the queue, then indexing starts on the next index cycle.

  • After making or changing a data record, the record and all of its direct relationships are re-indexed; the indexed results may not be instantly available.

  • After changing a collection member, every item using that member must be re-indexed. If you edit a frequently-used member (for example, an item lifecycle name or unit of measure), then most items in the database may need re-indexing.

• Server hardware, storage connection bandwidth, SQL Server edition, and transaction load have a big effect on index speed.

• Item indexing is first-in, first-out. If the queue is working on other records, these are completed before new items are started. If a group of items is added to the queue, some items become searchable before others. A search may return a few — but probably not all — matching items, and then more may appear in a new search, until all have been indexed.

• If the same item is added to the queue several times, then the system indexes the item once, and deletes the duplicate entries.

• An item's indexed terms may include information about related items. For example, a change form may mention a part number that it released, and the change form may appear in that part's search results.

• Adding, changing or removing a type's custom attribute definition adds all related items to the queue. Unlike standard attributes, a custom attribute cannot be optimized and needs more time to index. Indexing speed is directly related to the number of custom attributes multiplied by the number of items with those attributes.

File index §

• File attachments are indexed using software components ("IFilters") installed on the server computer. There may be a noticeable delay before file indexing starts, and the filter service runs at a much slower speed than item indexing.

• Adding a new file to an item automatically schedules the file for indexing. In normal operation, there's never a need to use the Re-index button.

• Restoring a database PDXZ file automatically indexes all files. When you restore a database .BAK and separate library folder, and do not restore the contents of the Data\Filter folder, use the Re-index button.

• Indexing speed and quality depend on file size, language/culture settings, source application, and IFilter capabilities. These capabilities may be affected by Microsoft Windows and .NET updates, as well as third-party plug-ins installed on the server computer.

• Most IFilters are designed to extract normal text from files; only specialized IFilters can convert scanned images to usable text characters (OCR).

• After installing or upgrading an IFilter, restart the server computer.

• For simpler IFilters, such as the File Properties filter, results may be directly saved in the item index, not in the Data\Filter folder. If you assign a new IFilter to (for example) CAD files that previously used the File Properties filter, use the Re-index button to re-index these files.

• An IFilter produces the same results each time a file is indexed. Previously-indexed results that have been saved in the Data\Filter folder are not re-indexed. In general, older files should be re-indexed only when an IFilter is changed or upgraded; for example, if the Windows default Reader Search Handler is replaced with a different PDF filter.

• In the special case when a complete re-index of all library files is required, select one of these options to delete the Data\Filter folder contents:

  Option A: Preferred for all systems.

  1. On the PDXpert Server window, note the location of the current data directory.

     .

  2. Close the PDXpert Application Server window and all PDXpert clients.

  3. Stop the PDXpert Server service and, if asked, the PDXpert Filter Service.

  4. Delete all files in the server data directory's Data\Filter folder.

     Do not make any changes in other data directory folders.

  5. Restart the server computer. Do not simply restart the PDXpert Server service.

  6. After Windows restarts, open a PDXpert client to re-index all files using the Re-index button.

  Option B: Acceptable for smaller systems using the .PDXZ backup format.

  Resetting the system deletes the content of all folders, including the Data\Filter folder.

  1. Back up the database and library.

  2. Reset the database and library.

  3. Restore the backup.

1096