While a new index is being populated, it will miss out any nodes that are created during the population. It will also miss any nodes that are created between the population finishing and the swap of the indexes.

Should "unlive" indexes also receive events to ensure that they are also kept up-to-date?

The method for retrieving a node's url is not consistent across content and media. The default method is to use the NiceUrlProvider, but this only works for content. A workaround until this is fixed is to override the UrlFor() method of the MediaIndexService.

protected override string UrlFor(IMedia contentInstance)
{
  var helper = new UmbracoHelper(UmbracoContext.Current);
  var media = helper.TypedMedia(contentInstance.Id);
  return media.Url();
}

The package under the hood is using content service, does it index both published and unpublished items?

Implement support for creating and managing search index via an index alias.

I'm not positive, but after quickly looking at some code it seems you might be using UmbracoHelper as a singleton in your IndexService. If your Index service class is request scoped or transient than its usage is fine but otherwise you'll run into problems because the UmbracoContext is a request scoped object which is used to create the helper.

Just found out this project and am really excited. I came in a perfect time as I am working on a project with tons of content nodes, a need for geolocation searches, faceted searches, and some advanced lucene features that do not exist on 2.9.4.

Just a quick question - why cant I rebuild the indices once are active? It seems to me that this is a common task I would like to carry out through dev, or when deploying a new feature.

Thanks in advance
Doron

The ability to request and process an indexing request for an individual node by its Id.

  var nodeId = 1234;
  var result = something.Index(nodeId);
  if(!result.Success) {
    // report something
  }

Phil,

How do i set an analyser during the index create / indexing process? Nothing is set at the moment and I am guessing the default is keyword. I want to add lowercase filter.

Regards

Ismail

Is it possible to get copy of db for sample site so we can run it?

Regards

Ismail

Investigate whether it is feasible to abstract the NEST dependencies into a facade layer that will allow the core plugin to remove its nuget package dependency on the NEST library.

Doing this will however change the plugin from being an Elasticsearch indexing plugin to a non-descriptive plugin that has hooks for acting on nodes during admin events.

I'm not convinced this is a valid approach as it will introduce the need to provide additional packages to implement the abstracted interfaces.

Leaving issue here for review/discussion.

Results in error message:

Search: Unable to index content : 'node name' => value cannt be null. Parameter name: object

Suspect this is related to getting a urlname for the content on first publish, the url is not available at initial publishing and has to be retrieved after the publishing has passed through the cache.

Hi Phil
Will this package work with Umbraco v8?

Being caused by null reference at line 91 of SearchApplicationEventHandler_Base.cs

Hi Phil,
Not an issue as such more a question. Any tips on setting up scheduled job to rebuild the index? Would we need to create and end point to do it and then call that using umbraco scheduler?

Regards

Ismail

https://our.umbraco.org/projects/collaboration/umbracoelasticsearch/umbracoelasticsearch-plugin/83658-cannot-index

Support will be dropped for both Elasticsearch v1 and v2.

This error appears when a hosted Elasticsearch cloud service (ES Cloud) doesn't respond with the expected JSON structure for a index info request when no indices exists.

Will need to ensure that in this case the initialisation process can continue without halting Umbraco

This is a breaking change and will require alter the requirements on users of this plugin.

Currently the plugin supports Umbraco v7.3+

This upgrade is expected to move that minimum Umbraco version requirement to v7.5+ due to the downlevel dependency graph primarily relating to the use of Json.NET in both Umbraco and NEST.

When hitting the "Save and Publish" button an error message is received stating there is no active index, this is also occuring when there is in fact a valid, active index configured.

This issue is happen due to an active index check that is not executed on Umbraco startup, but only when visiting the index management tab within the developer section.

Workaround

Visit the developer/Elasticsearch tab at least once after Umbraco startup