If Tika fails win an OutOfMemory, the document is still indexed in Solr, but with only a minimum of meta-data. As fields derived from (w)arc-names og server headers are not handled by Tika, these should not suffer from Tika OOM. It seems that there is too broad a try-catch somewhere.
If Tika fails win an OutOfMemory, the document is still indexed in Solr, but with only a minimum of meta-data. As fields derived from (w)arc-names og server headers are not handled by Tika, these should not suffer from Tika OOM. It seems that there is too broad a try-catch somewhere.