FAST Search index servers out of disk space after or during a crawl

Symptom

When adding new content to the FAST Search backend, you find that the %FASTSEARCH%\data\data_index size has expanded to all available disk space, and your crawled content may not be fully searchable or indexed.

Cause

FAST Search Server keeps a read-only binary index file set to serve queries while building the next index file set. The worst-case disk space usage for index data is approximately 2.5 times the size of a single index file set. There may not be enough space on the current storage solution for indexing to complete.

Resolution

If additional storage can be added to the server, the entire %FASTSEARCH%\data directory can be moved to a new location with the same permissions ("Full Control", granted to the FASTSearchAdministrators local group), and connected back to the installation via a junction point. To do so, follow the steps below on each FAST index server:
  1. Stop the FAST Search for SharePoint service.
  2. Stop the FAST Search for SharePoint Monitoring service.
  3. Move %FASTSEARCH%\data to the larger storage you have added.
  4. Run the following in a command prompt: mklink /j %FASTSEARCH%\data %NEW_LOCATION%\data
  5. Start the FAST Search for SharePoint Service.
Please note that while other methods of relocating the index outside of the %FASTSEARCH% parent directory are not supported, the entire partent directory can also be moved to a new physical location without using a junction point (skipping step #4 above) if the drive letter, path, and permissions remain identical.

More Information

If the data_index directory has many incomplete index directories within each numeric subdirectory (e.g. in the "2" directory), you may choose to force a clean rebuild, to have a better idea of the storage requirements going forward and to ensure that space remains available. To do this, before starting FAST Search in step #5 above, delete the data_index directory. Then, start the FAST Search service, and indexing will pick up all previously crawled content stored in data_fixml.

Please note that it is not supported to have separate junction points data_fixml and data_index on an indexer-node instead of %FASTSEARCH%\data.

Rekvizīti

Raksta ID: 2506015. Pēdējo reizi pārskatīts: 2013. gada 12. nov.. Pārskatījums: 1

Atsauksmes