ElasticSearch：更新紧跟查询（Java API）

Question

We are using elasticsearch. Essentially, I need to insert a record and then query against the data and reliably return the recently inserted row. (This is an import tool that imports hierarchical data, and as the entities are inserted into ES, the tool needs to check for existing nodes that new nodes refer to).

I have tried a few things.

First, if I delete and rebuild the index on every insert, it works (obviously can't do this in the real world).

Second, I have tried setting the refresh_interval to -1:

 settings_map.put("index.refresh_interval", "-1");
      org.elasticsearch.common.settings.Settings settings = ImmutableSettings.settingsBuilder().put(settings_map)
          .build();

      UpdateSettingsRequestBuilder usrb = es.admin().indices()
          .prepareUpdateSettings();
      usrb.setIndices("pm");
      usrb.setSettings(settings);
      usrb.execute().actionGet();

Third, I have tried settings the threading option to false:

tc.prepareUpdate(domain, type, id)).setListenerThreaded(false)...

I've also seen the setOperationThreaded setting, but that does not appear to be available on prepareUpdate().

---

None of these seem to have the desired effect.

The desired effect being: After inserting a record into ElasticSearch, when immediately running a query that should return that record, reliably have the record in the results.

Answer 1

Setting the refresh_interval at -1 actually disabled automatic refresh.

You can achieve the desired effect by setting the refresh parameter to true like this :

client.prepareUpdate(...).setRefresh(true).execute().actionGet();

This way, a refresh operation will be performed immediately after your update and the updated document will be searchable as you wish.

Beware : the refresh operation is costly and should not be performed while indexing lots of documents.