[
https://issues.apache.org/jira/browse/SPARK-15406?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15553613#comment-15553613
]
Ofir Manor commented on SPARK-15406:
------------------------------------
I see three somewhat-related issues:
1. More flexibility is generally needed in stating offsets when using a Kafka
source - that moved to SPARK-17812.
2. Regarding exactly-once, what I'm missing (given a transactional / idempotent
sink) is the ability to restart a failed structured streaming job (after it
crashed or was killed) by submitting a new one and asking to "copy" the context
(offset and internal state) of another job / query.
The programing guide hints that it is possible, but doesn't give an example
(and I couldn't find a relevant method):
"In case of a failure or intentional shutdown, you can recover the previous
progress and state of a previous query, and continue where it left off".
If that was possible, it would remove the need to manually monitor / manage
source offsets for exactly-once.
However, this is not specific to Kafka source - it is relevant for all
fault-tolerant sources.
3. Another related issue (though out-of-scope of this JIRA) is adding an
"exactly-once" Kafka sink.
Since in Kafka we can't commit a batch of messages together (like in a Foreach
sink), the open of [version,partition] can't just return a boolean - it should
likely return the messages within a [version,partition] that were already
written so only those will be filtered out (or otherwise filter within the
partition before process() is called).
That again is not Kafka-specific, will be useful in other non-transactional
sinks,
> Structured streaming support for consuming from Kafka
> -----------------------------------------------------
>
> Key: SPARK-15406
> URL: https://issues.apache.org/jira/browse/SPARK-15406
> Project: Spark
> Issue Type: New Feature
> Reporter: Cody Koeninger
>
> This is the parent JIRA to track all the work for the building a Kafka source
> for Structured Streaming. Here is the design doc for an initial version of
> the Kafka Source.
> https://docs.google.com/document/d/19t2rWe51x7tq2e5AOfrsM9qb8_m7BRuv9fel9i0PqR8/edit?usp=sharing
> ================== Old description =========================
> Structured streaming doesn't have support for kafka yet. I personally feel
> like time based indexing would make for a much better interface, but it's
> been pushed back to kafka 0.10.1
> https://cwiki.apache.org/confluence/display/KAFKA/KIP-33+-+Add+a+time+based+log+index
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org
For additional commands, e-mail: issues-h...@spark.apache.org
