Wie kann Python Änderungen am Oplog von Mongodb beobachten?

Ich habe vor einiger Zeit ein inkrementelles Backup-Tool für MongoDB in Python geschrieben. Das Tool überwacht Datenänderungen durch Tailing des oplog . Hier ist der relevante Teil des Codes.

Aktualisierte Antwort, MongDB 3.6+

Wie datdinhquoc in den Kommentaren unten geschickt anmerkt, gibt es für MongoDB 3.6 und höher Change Streams .

Aktualisierte Antwort, Pymongo 3

from time import sleep

from pymongo import MongoClient, ASCENDING
from pymongo.cursor import CursorType
from pymongo.errors import AutoReconnect

# Time to wait for data or connection.
_SLEEP = 1.0

if __name__ == '__main__':
    oplog = MongoClient().local.oplog.rs
    stamp = oplog.find().sort('$natural', ASCENDING).limit(-1).next()['ts']

    while True:
        kw = {}

        kw['filter'] = {'ts': {'$gt': stamp}}
        kw['cursor_type'] = CursorType.TAILABLE_AWAIT
        kw['oplog_replay'] = True

        cursor = oplog.find(**kw)

        try:
            while cursor.alive:
                for doc in cursor:
                    stamp = doc['ts']

                    print(doc)  # Do something with doc.

                sleep(_SLEEP)

        except AutoReconnect:
            sleep(_SLEEP)

Siehe auch http://api.mongodb.com/python/current/examples /tailable.html .

Ursprüngliche Antwort, Pymongo 2

from time import sleep

from pymongo import MongoClient
from pymongo.cursor import _QUERY_OPTIONS
from pymongo.errors import AutoReconnect
from bson.timestamp import Timestamp

# Tailable cursor options.
_TAIL_OPTS = {'tailable': True, 'await_data': True}

# Time to wait for data or connection.
_SLEEP = 10

if __name__ == '__main__':
    db = MongoClient().local

    while True:
        query = {'ts': {'$gt': Timestamp(some_timestamp, 0)}}  # Replace with your query.
        cursor = db.oplog.rs.find(query, **_TAIL_OPTS)

        cursor.add_option(_QUERY_OPTIONS['oplog_replay'])

        try:
            while cursor.alive:
                try:
                    doc = next(cursor)

                    # Do something with doc.

                except (AutoReconnect, StopIteration):
                    sleep(_SLEEP)

        finally:
            cursor.close()