[
https://issues.apache.org/jira/browse/FLINK-18649?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17161840#comment-17161840
]
Aljoscha Krettek commented on FLINK-18649:
------------------------------------------
I think this could be implemented either like the Cassandra connector, where we
use a write-ahead log, or potentially like the ES connector where don't use a
log but instead rely on idempotent writes and overwrite in case of recovery.
> Add a MongoDB Connector with Exactly-Once Semantics
> ---------------------------------------------------
>
> Key: FLINK-18649
> URL: https://issues.apache.org/jira/browse/FLINK-18649
> Project: Flink
> Issue Type: Wish
> Components: Connectors / Common
> Reporter: Eric Holsinger
> Priority: Minor
>
> Before taking the Flink Plunge, and per the following recommendation, I'm
> opening a Jira ticket to see if someone can provide me with a formal
> recommendation for obtaining exactly-once semantics for MongoDB:
>
> [https://stackoverflow.com/questions/35158683/kafka-flink-datastream-mongodb]
>
> FYI, we cannot use Kafka or any other framework other than Flink and MongoDB,
> and we have constraints as to what can be installed in production.
>
> Thanks in advance.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
