如何从 Spark 结构化流作业中的每个微批次中的相同起始偏移量读取？

Question

I am using spark structured streaming. Is it possible to reset Kafka offset after every batch execution so that every batch read from same starting offset instead of only newly discovered events?

Quoting description of startingOffsets from spark Kafka integration documentation here

For streaming queries, this only applies when a new query is started, and that resuming will always pick up from where the query left off. Newly discovered partitions during a query will start at earliest.

Right now I am doing it by creating a static dataframe from Kafka inside for each batch loop and using a dummy streaming dataset with format as "rate". Wondering if there is a better way to do it

Answer 1

For Structured Streaming can set startingOffsets to earliest so that every time you consume from the earliest available offset. The following will do the trick

.option("startingOffsets", "earliest")

---

However note that this is effective just for newly created queries:

startingOffsets

The start point when a query is started, either "earliest" which is from the earliest offsets, "latest" which is just from the latest offsets, or a json string specifying a starting offset for each TopicPartition. In the json, -2 as an offset can be used to refer to earliest, -1 to latest. Note: For batch queries, latest (either implicitly or by using -1 in json) is not allowed. For streaming queries, this only applies when a new query is started, and that resuming will always pick up from where the query left off . Newly discovered partitions during a query will start at earliest.

---

Alternatively, you might also choose to change the consumer group every time:

.option("kafka.group.id", "newGroupID")