why new cursor begin with oplog_start, not _last_optime

[heroWang]

def _sync_oplog(self, oplog_start):
        """ Replay oplog.
        """
        self._last_optime = oplog_start

        n_total = 0
        n_skip = 0

        while True:
            # try to get cursor until success
            try:
                start_optime_valid = False
                host, port = self._src.client().address
                log.info('try to sync oplog from %s on %s:%d' % (self._last_optime, host, port))
                # set codec options to guarantee the order of keys in command
                coll = self._src.client()['local'].get_collection('oplog.rs',
                                                                  codec_options=bson.codec_options.CodecOptions(document_class=bson.son.SON))
                cursor = coll.find({'ts': {'$gte': oplog_start}},
                                   cursor_type=pymongo.cursor.CursorType.TAILABLE_AWAIT,
                                   no_cursor_timeout=True)

                # New in version 3.2
                # src_version = mongo_utils.get_version(self._src.client())
                # if mongo_utils.version_higher_or_equal(src_version, '3.2.0'):
                #     cursor.max_await_time_ms(1000)

                while True:
                    try:
                        if not cursor.alive:
                            log.error('cursor is dead')
                            raise pymongo.errors.AutoReconnect
....

Maybe I am wrong. But what I am confused is that when you sync the oplog and create new cursor, Why dont you create it from self._last_optime ? that make the oplog sync progress more efficient under bad network env

[caosiyang]

You're right. It's a mistake. Thanks : )