enable multi/cross index store setup via wildcard

[giaconiamark]

A lot of people create multiple indices of the same objects to partition the data nicely over time or space. For instance, if I index twitter all the time, but my majority use case is only to query the last three hours... I can create an index for each day or week or something, and only query a subset of the data most of the time instead of hitting a massive uber index. Currently, in Elastic, we can use the * at the end of a prefix, and this will enable query and aggregations across any indices beginning with the prefix. This would be great to be able to apply to ElasticGeo store setup.

[sjudeng]

Just merged in commit that might help with this. It includes a new store configuration parameter, search_indices. The original parameter, index_name, defines the specific index used to read the type, and the new parameter, search_indices, is used for reading features.

[giaconiamark]

nice... I'll give it a try thanks!

[giaconiamark]

I was able to build and deploy to GeoServer, and i input my search indexes no problem to create a store. I also created a layer and published it, but i cannot preview via GeoServer's "preview layer" tool in Openlayers. Not sure if it matters, but my index is of mixed geometry types (geo_shape)

[sjudeng]

Was it working before? Any error messages in the logs?

[sjudeng]

Are you able to see data when you click GML on the Layer Preview page? If so, try copying the OpenLayers link into another tab and then add "&maxFeatures=1000" to the end.

[giaconiamark]

this is what I got...

error:java.lang.RuntimeException: java.io.IOException: Error executing query search java.lang.RuntimeException: java.io.IOException: Error executing query search java.io.IOException: Error executing query search Error executing query search

[giaconiamark]

that was for the GML output

[sjudeng]

I just pushed an update that adds more logging around the search. To get this in your logs, you'll have to increase the logging level to FINE. If using Tomcat, edit TOMCAT_ROOT/conf/logging.properties and add the below line.

mil.nga.giat.level=FINE

Then try hitting GML again. This should show you the request being executed (see example below) and the error stack trace.

May 01, 2015 4:35:28 PM mil.nga.giat.data.elasticsearch.ElasticFeatureSource prepareSearchRequest
FINE: {
  "from" : 0,
  "size" : 50,
  "query" : {
    "match_all" : { }
  },
  "post_filter" : {
    "match_all" : { }
  },
  "_source" : {
    "includes" : [ "_id", "_index", "_type", "_score", "_relative_score", "status_s", "id", "security_ss", "standard_ss", "speed_is", "geo3", "modem_b", "installed_tdt", "vendor_s" ],
    "excludes" : [ ]
  },
  "sort" : [ {
    "_uid" : {
      "order" : "asc"
    }
  } ]
}

[giaconiamark]

I'll try to get back to this next week, so far so good. But as I look at the code, and the fact that I have mixed geometry types, I am tinkering with the idea of being able to register an actual query as a datasource definition, rather than an index or index alias. Kind of like a "view" in a traditional database.