Re: Questions about using Flink MongoDB CDC

2024-11-17 Thread Jiabao Sun
Hi, The Flink SQL Planner uses the ChangelogNormalize operator to cache all incoming data for upsert type Changelog in order to complete the pre-image values, which results in additional state overhead. When the MongoDB version is below 6.0, the oplog does not contain Pre-Images of changed record

Re: [External] Re: Flink table materialization

2024-11-17 Thread Feng Jin
Hi Jacob. Based on your latest error message, it seems you're using an outdated parameter 'connector.type' = 'jdbc'. You should use 'connector' = 'jdbc' instead. Please refer to the following documentation[1] [1]. https://nightlies.apache.org/flink/flink-docs-release-1.17/docs/connectors/table/jd

Re: flink connector hive support collect metrics

2024-11-17 Thread Feng Jin
Hi Leilinee, numRecordsIn should be a metric[1] reported by the operator/task, and it doesn't require support from the connector. You should be able to access it directly. [1]. https://nightlies.apache.org/flink/flink-docs-master/docs/ops/metrics/#io Best, Feng On Thu, Nov 7, 2024 at 2:57 PM

Re: Order data with OVER AGGREGATION functions.

2024-11-17 Thread Feng Jin
Hi Guillermo > "When is this ordering done, and until when?" Assuming the current watermark is 10:00 1. Currently, data before 10:00 will be sorted. 2. If data after 10:00 arrives at this time, the record will be stored in the state, waiting for the watermark to become 10:01 before sorting and ou