Tika

Remote streaming lets Solr fetch a file from a URL and extract its text, instead of you uploading the bytes. Two settings have to be present, and the second one is the one everybody misses.

01 · The symptom

You are indexing remote PDFs with stream.url and Solr answers:

Remote Streaming is disabled.

even though enableRemoteStreaming="true" is already set inside <requestDispatcher>. That setting is required, but on its own it is not sufficient.

02 · Both settings must exist

In solrconfig.xml, inside <requestDispatcher>:

<requestDispatcher handleSelect="true">
  <requestParsers enableRemoteStreaming="true" multipartUploadLimitInKB="2048000" />
</requestDispatcher>

And inside the extraction handler itself:

<requestHandler name="/update/extract" class="solr.extraction.ExtractingRequestHandler">
  <lst name="initParams">
    <bool name="allowRemoteStreaming">true</bool>
  </lst>
</requestHandler>

This second block is the part that is usually forgotten. Without it Solr rejects every stream.url request with the error above, and says nothing about why.

03 · Upload, reload, test

curl "https://YOUR_HOSTNAME.opensolr.com/solr/YOUR_INDEX_NAME/update/extract?stream.url=https://example.com/sample.pdf&literal.id=test-doc&commit=true"

04 · Related reading

Tika extraction failing on your index, from Drupal or from your own code? Four things to check, in the order that resolves most cases.

01 · Try a different Solr version

Before reading logs, create a new Opensolr index on Solr 9, or on 7, 6 or 5. Avoid Solr 8.x: its Jetty version has known issues that show up as Tika failures. Create a new Opensolr index and point your integration at it to confirm.

02 · Make sure Tika is enabled in your region

Tika is not present in every environment. If you suspect it is missing, contact Opensolr support and we will enable it for your index and region.

03 · Confirm the handler in solrconfig.xml

Your solrconfig.xml needs the extraction request handler defined:

<requestHandler name="/extract/tika" class="org.apache.solr.handler.extraction.ExtractingRequestHandler" startup="lazy">
    <lst name="defaults">
    </lst>
    <!-- This path only extracts - never updates -->
    <lst name="invariants">
        <bool name="extractOnly">true</bool>
    </lst>
</requestHandler>

With extractOnly set as an invariant, this path extracts text and never writes documents, so a misconfigured client cannot update your index by accident.

04 · Still failing