elasticsearch-indexes-cleanup.py

import requests
import re
import datetime

elasticsearch_url = 'http://elasticsearch'

req = requests.get(elasticsearch_url + '/_aliases?pretty=true')

indexes_json = req.json()

patt = re.compile(r'-\d{4}\.\d{2}\.\d{2}')

logs_indexes = []

for index in indexes_json:
    m = patt.findall(index)
    if m:
        prefix = index.replace(m[0], '')
        str_date = m[0].replace('-', '')
        d = datetime.datetime.strptime(str_date, "%Y.%m.%d").date()
        logs_indexes.append({'date': d, 'index': index})

sorted_indexes = sorted(logs_indexes, key=lambda k: k['date'], reverse=True)
if len(sorted_indexes) > 60:
    sorted_indexes = sorted_indexes[0:60]

sorted_indexes_names = [index['index'] for index in sorted_indexes]
logs_indexes_name = [index['index'] for index in logs_indexes]

to_remove = []
for index in logs_indexes_name:
    if index not in sorted_indexes_names:
        to_remove.append(index)

if to_remove:
    for x in to_remove:
        print('Removing "' + x + '"...')

        try:
            requests.delete(elasticsearch_url + '/' + x)
        except:
            pass