使用SolrNet启动多个Solr词典索引

Question

I'm currently in the process of implementing a search using Solrnet. My indexes are very large, and because we're a global company they are translated. My initial thought was to have multiple records with all translations in one index, however this ended up being an issue because the indexes became to large for one index, so we split up the indexes by language. For instance I created an English index called SearchEnglish and a French index called SearchFrench.

To initiate Solrnet, I use something like:

Startup.Init<Dictionary<string, object>>(SearchIndexUrl);

I use dictionaries because my Solr indexes contain dynamic facets. The issue is that my code base is going to need to initiate all indexes. So I have no way of distinguishing one index from another. What would be your recommendation to handling the initiation of multiple dictionary Solr indexes using Solrnet? I don't see anything in the documentation regarding this.

Thanks, Paul

Answer 1

I've figured out how to do this by using SolrNet/Windsor to initialize my Solr instance. I didn't find a ton of documentation on how to do this, so I wanted to share.

Here's some of my code.

In the Global.asax.cs I have the following

    public static WindsorContainer _WindsorContainer { get; set; }

    protected void Application_Start()
    {
        InitiateSolr();
    }

    protected void Application_End()
    {
        _WindsorContainer.Dispose();
    }

    /// <summary>
    /// Initialized Misc Solr Indexes
    /// </summary>
    protected void InitiateSolr() {
        var reader = ApplicationConfig.GetResourceReader("~/Settings/AppSettings.resx");
        InitiateSolrFacetedIndex(reader);
    }

    /// <summary>
    /// Initializes The Faceted Search Indexes
    /// </summary>
    protected void InitiateSolrFacetedIndex(ResourceReader reader) {
        Data d = new Data();
        _WindsorContainer = new WindsorContainer();

        var solrFacility = new SolrNetFacility(reader.ResourceCollection["Url.SolrIndexPath"] + "EN");

        foreach (var item in d.GetLanguages())
        {
            solrFacility.AddCore("ProductSpecIndex" + item.LanguageCode.ToString(), typeof(Dictionary<string, object>), reader.ResourceCollection["Url.SolrIndexPath"] + item.LanguageCode.ToString());
        }

        _WindsorContainer.AddFacility("solr", solrFacility);

        Models.Solr.SolrWindsorContainer c = new Models.Solr.SolrWindsorContainer();
        c.SetContainer(_WindsorContainer);
    }

I also created an extension static class to hold the WindsorContainer object.

public class SolrWindsorContainer
{
    public static WindsorContainer Container { get; set; }

    public void SetContainer(WindsorContainer container){
        Container = container;
    }
    public WindsorContainer GetContainer(){
        return Container;
    }
}

Then in my application, I just call that static object to get my Windsor container

Models.Solr.SolrWindsorContainer c = new Models.Solr.SolrWindsorContainer();
ISolrOperations<Dictionary<string, object>> solr = container.Resolve<ISolrOperations<Dictionary<string, object>>>("ProductSpecIndex" + languageCode);

var results = solr.Query("*:*");

If you have any questions about this, you can read about solrnet and Windsor initialization at the links below.

https://github.com/mausch/SolrNet/blob/master/Documentation/Initialization.md

https://github.com/mausch/SolrNet/blob/master/Documentation/Multi-core-instance.md