Lucene indexes: automatic corrections?

[simon]

Hi,

We are looking how we could build a backup system for our Alfresco installation and data. After reading the documents in the WIKI I do understand that it's advised to save the Lucene indexes but you could rebuild them from scratch if needed.

Now, when I add a document to the content repository it gets added to the Lucene indexes, if I remove a document it's removed (I suppose).

But what if a document is missing? Database inconsistence (pointer in database but the document is not on the file system or vice versa) may be possible and how does the Lucene indexes recover from these sorts of problems?

Will Lucene detect this and if so in what timeframe?

Thanks!

[andy]

Hi

The enterprise build has index consistency checks and index rebuilding.

If content is missing from the file system it will not be found.
There are some failure codes that will be put in the index for things like missing content and conversion to text failures.

    public static final String NOT_INDEXED_NO_TRANSFORMATION = "nint";

    public static final String NOT_INDEXED_TRANSFORMATION_FAILED = "nitf";

    public static final String NOT_INDEXED_CONTENT_MISSING = "nicm";
‍‍‍‍‍‍‍

If the index is lost/corrupted/etc the enterprise version will detect and fix this.
The time frame is configurable.

Hope this helps,

Regards

Andy