0

As read here Can a raw Lucene index be loaded by Solr? Lucene indexes can be imported into Solr. This works well when the Solr server is not running (creating a Solr core folder structure in the data folder with all the needed configuration files) but it does not work when the Solr server is up and running.

Is there any call (via rest endpoint or java api) to tell Solr to re-scan the data folder?

Community
  • 1
  • 1
Fabrizio Fortino
  • 1,479
  • 1
  • 14
  • 22

1 Answers1

2

You want to generate an index with lucene (outsite solr) and insert this to solr without restart.

You must not change the index-folder directly. But you can create a new core which point to the already build index folder and switch/swap the core with the (outdated) old one. Or you can merge the new index-Folder in the old core.

All this can be done by the solrj admin api.

e.g. create:

    CoreAdminRequest.Create req = new CoreAdminRequest.Create();
    req.setConfigName(configName);
    req.setSchemaName(schemaName);
    req.setDataDir(dataDir);
    req.setCoreName(coreName);
    req.setInstanceDir(instanceDir);
    req.setIsTransient(true);
    req.setIsLoadOnStartup(false); // <= unless its productive core.
    return req.process(adminServer);

e.g. the swap:

    CoreAdminRequest request = new CoreAdminRequest();
    request.setAction(CoreAdminAction.SWAP);
    request.setCoreName(coreName1);
    request.setOtherCoreName(coreName2);
    request.process(solrClient);

For SolrCloud use the first "create" approach with the collections api and use alias instead of swap.

e.g. the alias:

    CollectionAdminRequest.CreateAlias req = new CollectionAdminRequest.CreateAlias();
    req.setAliasedCollections(coreName);
    req.setAliasName(aliasName);
    return req.process(solrClient);
Karsten R.
  • 1,628
  • 12
  • 14