solr2es 0.7

Python solr/elasticsearch migration script

solr2es

Migration script from solr to elasticsearch via a queue, that could be either redis or postgresql.

CLI

Here are the option to use as a command line :

• -m | –migrate : to migrate from a solr index to an elasticsearch index
• -r | –resume : to resume from a given queue to an elasticsearch index. By default, the queue will be redis. If the parameter “postgresqldsn” is set, the queue will be postgresql.
• -d | –dump : to dump from a solr index into a queue. By default, the queue will be redis. If the parameter “postgresqldsn” is set, the queue will be postgresql.
• -t | –test : to test the solr and elasticsearch connections
• -a | –async : to use python 3 asyncio
• –solrhost : to set solr host (by default: ‘solr’)
• –solrfq: to set solr filter query (by default: ‘*’)
• –solrid: to set solr id field name (by default: ‘id’)
• –core: to set solr core name (by default: ‘solr2es’)
• –index: to set index name for solr and elasticsearch (by default: solr core name, see –core parameter)
• –redishost: to set redis host (by default: ‘redis’)
• –postgresqldsn: to set postgresql Data Source Name (by default: None, by example: ‘dbname=solr2es user=test password=test host=postgresql’)
• –eshost: to set elasticsearch host (by default: ‘elasticsearch’)
• –translationmap: dict string or file path (starting with @) to translate fields from queue into elasticsearch (by default: None, by example: ‘{“postgresql_field”: {“name”: “es_field”}}’)
• –esmapping: dict string or file path (starting with @) to set elasticsearch mapping (by default: None)
• –essetting: dict string or file path (starting with @) to set elasticsearch setting (by default: None)

Use

translation_map

The purpose of a translation_map is to create a mapping between the fields coming from the queue (either Redis or Postgresql) to the ones inserted to Elasticsearch.

1. If a field from the queue doesn’t exist in the translation_map, it will be inserted as it is into Elasticsearch.
2. Use the property name to rename a field in Elasticsearch :

{"queue_name": {"name": "elasticsearch_name"}}

3. Use the property default if you want to set a default value into a field in Elasticsearch.

If the field exists into the queue and has a value, it won’t be changed by the translation_map. Otherwise a field queue_name willl be added to Elasticsearch with value john doe.

{"queue_name": {"default": "john doe"}}

4. Use the property name with some . in it, to create a nested field in Elasticsearch.

If the queue record has a field nested_a_b, the Elasticsearch record will get a field nested, that will have a nested field a, that will have a nested field b that will get the content of nested_a_b.

{"nested_a_b": {"name": "nested.a.b"}}

5. Use the property name with some regex groups capture to rename a bulk of queue fields in Elasticsearch by adding [regexp] at the beginning of the field. This will rename all the fields prefixed by queue_ into elasticsearch_.

{"[regexp]queue_(.*)": {"name": "elasticsearch_\\1"}}

6. Use the property ignore at true to ignore some fields from the queue to Elasticsearch.

{"ignored_field": {"ignore": true}}

7. Use the property routing_field set to true to use one field for routing in elasticsearch. An exception will be raised if several fields are set to true.

{"my_root_doc": {"routing_field": true}}

8. Use the property multivalued set to false to ignore multi valued array field. Get the first value instead. By default the array is copied.

{"my_array": {"multivalued": false}}

execution

1. Execute a dump from Solr into Postgresql specifying the Solr host, the Solr core, the Solr id and the Postgresql DSN

solr2es --postgresqldsn 'dbname=solr2es user=test password=test host=localhost' --solrhost 127.0.0.1 --core test_core --solrid solr_id -d -a

2. Execute a resume from Postgresql into Elasticsearch specifying the Postgresql DSN, the Elasticsearch index, the Elasticsearch mapping, the Elasticsearch settings and the translation map

solr2es --postgresqldsn 'dbname=solr2es user=test password=test host=localhost' --index es-index --translationmap @examples/translation-map.json --esmapping @examples/datashare_index_mappings.json --essetting @examples/datashare_index_settings.json -r -a

Develop

To build and run tests you can make :

virtualenv --python=python3.6 venv
source venv/bin/activate
python setup.py develop
python setup.py test

To release :

python setup.py  sdist bdist_egg upload

Misc

Some features are not implemented yet :

• Resume from the redis queue to elasticsearch in asynchronous mode (function aioresume_from_redis)
• Resume from the redis queue to elasticsearch in synchronous mode (function resume_from_redis)
• Resume from the postgresql queue to elasticsearch in synchronous mode (function resume_from_postgresql)

Changes

v. 0.7

• multivalued field : flatten the array if it has one value
• multivalued field : ignore multi valuated field in translation map
• multivalued field : copy the array into elasticsearch

v. 0.6

• error handling : logs ids that have failed when resuming from postgresql
• adds a the possibility to specify a routing field in the translation map

v. 0.5

• adds postgresql resume
• elasticsearch : adds mappings and settings support
• better logs and progress marks
• doc : README
• translation map : support for empty default list
• adds postgresql blocking queue
• translation map : ignore field
• translation map : default value

v. 0.4

• [solr2es] wildcard support in translation_map
• [solr2es] nested fields support in translation_map
• [solr2es] adds solrid parameter to change sort field
• [solr2es] adds solrfq parameter to parallelize solr reading

v. 0.3

• [solr2es] adds translation map for fields
• [solr2es] adds elasticsearch mapping for index creation
• [test] compatible with 6.6.0

v. 0.2

• [log] adds logger and progression feedbacks
• [cli] exit if no args

v. 0.1

• [solr2es] initial version