Allows to easily migrate Elasticsearch indexes

migrate_es.py

#!/usr/bin/env python

import sys
from argparse import ArgumentParser
from elasticsearch.helpers import reindex
from elasticsearch import Elasticsearch


class MigrateException(Exception):
    pass


def migrate_es_data(source_index, dest_index, query=None, host='localhost'):
    es = Elasticsearch(host)

    if source_index == dest_index:
        msg = 'Can not migrate to index with same name'
        raise MigrateException(msg)

    if not es.indices.exists(source_index):
        msg = 'Source index {0} does not exist'.format(source_index)
        raise MigrateException(msg)

    if es.indices.exists(dest_index):
        msg = 'Destination index {0} already exists'.format(dest_index)
        raise MigrateException(msg)

    es.indices.create(dest_index)
    reindex(es, source_index, dest_index, query=query)
    es.indices.refresh(dest_index)

    source_count = es.indices.stats(source_index)
    source_count = source_count['_all']['primaries']['docs']['count']

    dest_count = es.indices.stats(dest_index)
    dest_count = dest_count['_all']['primaries']['docs']['count']

    return source_count, dest_count


def parse_arguments():
    # TODO: Grab query form CLI so we can migrate a subset of the data.
    parser = ArgumentParser()
    parser.add_argument('--elasticsearch', '-e', type=str, default='localhost')
    parser.add_argument('sourceIndex', type=str)
    parser.add_argument('destIndex', type=str)
    return parser.parse_args()


def main():
    args = parse_arguments()

    try:
        source_count, dest_count = migrate_es_data(args.sourceIndex,
                                                   args.destIndex,
                                                   host=args.elasticsearch)
        print('Migrated {0} of {1} docs form {2} to {3}'.format(
            dest_count, source_count, args.sourceIndex, args.destIndex
        ))
        sys.exit(0)
    except MigrateException as e:
        print(e)
        sys.exit(1)


if __name__ == '__main__':
    main()