• [+] expand all

Procedure: Change from attribute to index

Changing between index and attribute is a common field change operation to optimize performance. Use the reindexing feature to safely migrate data to/from index structures.

Changing from attribute to index can be seen as "drop attribute" and "add index". When the attribute aspect of a field is removed, the field's data is not queryable after deployment. The reindexing process will populate the field's index structure, but this takes time, depending on corpus size.

Another approach is hence to run with both attribute and index in the transition, keeping data available for queries. The gist of this procedure is hence to add index, run a reindex - then remove attribute aspect:

# field configuration at start
field artist type string {
    indexing: summary | attribute
}

->

# intermediate step to populate index structure, keeping the data in the attribute
field artist type string {
    indexing: summary | attribute | index
    match: word
    stemming: none
}

->

# final configuration, migrated to index
field artist type string {
    indexing: summary | index
    match: word
    stemming: none
}

Procedure

  1. Test this using the quick-start, changing the artist field to an attribute before running. Also add a validation override in src/main/application/validation-overrides.xml:

    <validation-overrides>
        <allow until="2021-08-30">indexing-change</allow>
    </validation-overrides>
    
  2. Run the quick start, stop after feeding documents. Run a query to validate data can be queried:

    $ curl "http://localhost:8080/search/?ranking=rank_albums&yql=select%20%2A%20from%20sources%20%2A%20where%20artist%20contains%20%22Coldplay%22%3B"
    

    One can also dump current index structures, see artist as an attribute.

  3. Add index aspect and match/stemming settings to the field, deploy and observe output

    field artist type string {
        indexing: summary | attribute | index
            match    : word
            stemming : none
    }
    
    $ (cd src/main/application && zip -r - .) |   curl --header Content-Type:application/zip --data-binary @- \
      localhost:19071/application/v2/tenant/default/prepareandactivate
    
    {
      "log": [
        {
          "time": 1628239290150,
          "level": "WARNING",
          "message": "Change(s) between active and new application that may require re-index:\nindexing-change:
                      Consider re-indexing document type 'music' in cluster 'music' because:\n
                      1) Document type 'music': Field 'artist' changed: add index aspect,
                      indexing script: '{ input artist | summary artist | attribute artist; }' ->
                                       '{ input artist | exact | summary artist | attribute artist | index artist; }'\n"
        }
      ],
      "tenant": "default",
      "url": "http://localhost:19071/application/v2/tenant/default/application/default/environment/prod/region/default/instance/default",
      "message": "Session 3 for tenant 'default' prepared and activated.",
      "configChangeActions": {
        "restart": [],
        "refeed": [],
        "reindex": [
          {
            "name": "indexing-change",
            "documentType": "music",
            "clusterName": "music",
            "messages": [
              "Document type 'music': Field 'artist' changed:
                 add index aspect, indexing script:
                   '{ input artist | summary artist | attribute artist; }'
                 ->
                   '{ input artist | exact | summary artist | attribute artist | index artist; }'"
            ],
            "services": [
              {
                "serviceName": "searchnode",
                "serviceType": "searchnode",
                "configId": "music/search/cluster.music/0",
                "hostName": "vespa-container"
              }
            ]
          }
        ]
      }
    }
    
  4. Wait for the new configuration generation to be activated on the config server(s) - this is normally quite immediate. After that, allow up to 3 minutes for the config servers to set reindexing ready, track this using the reindexing endpoint:

    $ while true; do
        curl http://localhost:19071/application/v2/tenant/default/application/default/environment/prod/region/default/instance/default/reindexing | jq .
        sleep 10
      done
    
    {
      "enabled": true,
      "clusters": {
        "music": {
          "pending": {
            "music": 3
          },
          "ready": {
            "music": {}
          }
        }
      }
    }
    
    ...
    
    {
      "enabled": true,
      "clusters": {
        "music": {
          "pending": {},
          "ready": {
            "music": {
              "readyMillis": 1628665589516
            }
          }
        }
      }
    }
    
  5. When ready, deploy again to start reindexing, wait for it to complete (use the loop in previous step):

    $ (cd src/main/application && zip -r - .) |   curl --header Content-Type:application/zip --data-binary @- \
      localhost:19071/application/v2/tenant/default/prepareandactivate
    
    ...
    
    {
      "enabled": true,
      "clusters": {
        "music": {
          "pending": {},
          "ready": {
            "music": {
              "readyMillis": 1628665589516
            }
          }
        }
      }
    }
    
    ...
    
    {
      "enabled": true,
      "clusters": {
        "music": {
          "pending": {},
          "ready": {
            "music": {
              "readyMillis": 1628665589516,
              "startedMillis": 1628668739973,
              "endedMillis": 1628668740536,
              "state": "successful"
            }
          }
        }
      }
    }
    
  6. Dumping the index structures now shows artist both in index and attribute, and there is an entry in vespa.log. Verify the query still works:

    $ docker exec vespa /usr/bin/sh -c '/opt/vespa/bin/vespa-logfmt | grep Reindexer'
    [2021-08-11 07:59:00.535] INFO    : container-clustercontroller Container.ai.vespa.reindexing.Reindexer
                                        Completed reindexing of datatype music (code: 1412693671) after PT0.558683S
    
    $ curl "http://localhost:8080/search/?ranking=rank_albums&yql=select%20%2A%20from%20sources%20%2A%20where%20artist%20contains%20%22Coldplay%22%3B"
    
  7. As data is now reindexed into the index data structures, deploy without attribute. (Observe changes to index files, "artist" is now in index only). Test query after restart:

    field artist type string {
        indexing: summary | index
            match    : word
            stemming : none
    }
    
    $ (cd src/main/application && zip -r - .) |   curl --header Content-Type:application/zip --data-binary @- \
      localhost:19071/application/v2/tenant/default/prepareandactivate
    
    $ curl "http://localhost:8080/search/?ranking=rank_albums&yql=select%20%2A%20from%20sources%20%2A%20where%20artist%20contains%20%22Coldplay%22%3B"
    
  8. Optional: restart Vespa - a restart will reclaim memory from the attribute:

    $ docker exec vespa /usr/bin/sh -c '/opt/vespa/bin/vespa-stop-services && /opt/vespa/bin/vespa-start-services'
    

Notes:

  • The match/stemming settings above are set to the same at default attribute settings

Appendix

To inspect attribute and index data (can be useful when troubleshooting), use vespa-proton-cmd, then list files:

$ docker exec vespa /usr/bin/sh -c \
  '/opt/vespa/bin/vespa-proton-cmd --local triggerFlush' && \
  docker exec vespa /usr/bin/sh -c \
  '/usr/bin/find /opt/vespa/var/db/vespa/search/cluster.music/n0/documents/music/0.ready'