更新添加新字段时，弹性搜索不遵循映射

Question

Using python update API I add a string field to my ES indexed documents. Even my mapping has this:

"mappings" : {
    "_default_": {
            "dynamic_templates": [
                { "notanalyzed": {
                      "match":              "*", 
                      "match_mapping_type": "string",
                      "mapping": {
                          "type":        "string",
                          "index":       "not_analyzed"
                      }
                   }
                }
              ]
    },

the newly added field is still analyzed. Anybody noticed the same problem?

Here a piece of python code that adds the new column:

res = es.search(index=ind, body=st, size=2000) for rec in res['hits']['hits']: es.update(index=rec['_index'],doc_type=rec['_type'],id=rec['_id'], body={"doc": {"srcSite": sS}})

So the newly added "srcSite" is still somehow "analyzed". I have tried to add the new field explicitly into the template but that did not help either.

Answer 1

Ah, I found what was the problem. It appears that the template somehow gets fixed at the time a document was created and changing the template later will change nothing - even if one updates the document it won't use the changed template but an original one...