Can a raw Lucene index be loaded by Solr?

问题:

Some colleagues of mine have a large Java web app that uses a search system built with Lucene Java. What I'd like to do is have a nice HTTP-based API to access those existing search indexes. I've used Nutch before and really liked how simple the OpenSearch implementation made it to grab results as RSS.

I've tried setting Solr's dataDir in solrconfig.xml, hoping it would happily pick up the existing index files, but it seems to just ignore them.

My main question is:

Can Solr be used to access Lucene indexes created elsewhere? Or might there be a better solution?

回答1:

I have never tried this, but you would have to adjust the schema.xml to include all the fields of the documents that are in your Lucene index, because Solr won't allow you to search for a field if it is not defined in schema.xml.

The adjustment to schema.xml should also include defining the query-time analyzers to properly search in your field, especially if the field where indexed using custom analyzers.

In solrconfig.xml you may have to change settings in the indexDefaults and the mainIndex sections.

But I'd be happy to read answers from people who actually did it.

回答2:

Success! With Pascal's suggestion of changes to schema.xml I got it working in no time. Thanks!

Here are my complete steps for anyone interested:

1. Downloaded Solr and copied dist/apache-solr-1.4.0.war to tomcat/webapps
2. Copied example/solr/conf to /usr/local/solr/
3. Copied pre-existing Lucene index files to /usr/local/solr/data/index
4. Set solr.home to /usr/local/solr
5. In solrconfig.xml, changed dataDir to /usr/local/solr/data (Solr looks for the index directory inside)
6. Loaded my Lucene indexes into Luke for browsing (awesome tool)
7. In the example schema.xml, removed all fields and field types except for "string"
8. In the example schema.xml, added 14 field definitions corresponding to the 14 fields shown in Luke. Example: <field name="docId" type="string" indexed="true" stored="true"/>
9. In the example schema.xml, changed uniqueKey to the field in my index that seemed to be a document id
10. In the example schema.xml, changed defaultSearchField to the field in my index that seemed to contain terms
11. Started tomcat, saw no exceptions finally, and successfully ran some queries in localhost:8080/solr/admin

This is just proof for me that it can work. Obviously there's a lot more configuration to be done.

回答3:

Three steps in the end:

1. Change schema.xml or (managed-schema)
2. Change <dataDir> in solrconfig.xml
3. Restart Solr

I have my study notes here for those who are new to Solr, like me :)
To generate some lucene indexes yourself, you can use my code here.

public class LuceneIndex {
    private static Directory directory;

    public static void main(String[] args) throws IOException {
        long startTime = System.currentTimeMillis();

        // open
        Path path = Paths.get("/tmp/myindex/index");
        directory = new SimpleFSDirectory(path);
        IndexWriter writer = getWriter();

        // index
        int documentCount = 10000000;
        List<String> fieldNames = Arrays.asList("id", "manu");

        FieldType myFieldType = new FieldType();
        myFieldType.setIndexOptions(IndexOptions.DOCS_AND_FREQS);
        myFieldType.setOmitNorms(true);
        myFieldType.setStored(true);
        myFieldType.setTokenized(true);
        myFieldType.freeze();

        for (int i = 0; i < documentCount; i++) {
            Document doc = new Document();
            for (int j = 0; j < fieldNames.size(); j++) {
                doc.add(new Field(fieldNames.get(j), fieldNames.get(j) + Integer.toString(i), myFieldType));
            }
            writer.addDocument(doc);
        }
        // close
        writer.close();
        System.out.println("Finished Indexing");
        long estimatedTime = System.currentTimeMillis() - startTime;
        System.out.println(estimatedTime);
    }
    private static IndexWriter getWriter() throws IOException {
        return new IndexWriter(directory, new IndexWriterConfig(new WhitespaceAnalyzer()));
    }
}

回答4:

I am trying the same steps with HDF as the home directory and locktype as HDFS but no luck. I see the below error

labs_shard1_replica1: org.apache.solr.common.SolrException:org.apache.solr.common.SolrException: Index dir 'hdfs://127.0.0.1/user/solr/labs/core_node1/data/index/' of core 'labs_shard1_replica1' is already locked. The most likely cause is another Solr server (or another solr core in this server) also configured to use this directory; other possible causes may be specific to lockType: hdfs

solar dir config

<directoryFactory name="DirectoryFactory"

class="${solr.directoryFactory:solr.NRTCachingDirectoryFactory}">

but not with HDFS as below

<directoryFactory name="DirectoryFactory" class="solr.HdfsDirectoryFactory">
                <str name="solr.hdfs.home">hdfs://127.0.0.1/user/solr</str>
                <bool name="solr.hdfs.blockcache.enabled">true</bool>
                <int name="solr.hdfs.blockcache.slab.count">1</int>
                <bool name="solr.hdfs.blockcache.direct.memory.allocation">false</bool>
                <int name="solr.hdfs.blockcache.blocksperbank">16384</int>
                <bool name="solr.hdfs.blockcache.read.enabled">true</bool>
                <bool name="solr.hdfs.blockcache.write.enabled">false</bool>
                <bool name="solr.hdfs.nrtcachingdirectory.enable">true</bool>
                <int name="solr.hdfs.nrtcachingdirectory.maxmergesizemb">16</int>
                <int name="solr.hdfs.nrtcachingdirectory.maxcachedmb">192</int>
            </directoryFactory>

Lock type hdfs