[
https://issues.apache.org/jira/browse/SPARK-3655?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14706817#comment-14706817
]
Koert Kuipers commented on SPARK-3655:
--------------------------------------
hey nick,
i believe your problem sounds like a good fit for GroupSorted. you can
process the sessions per machine sorted by timestamp as an iterator using
mapStreamByKey, keeping state as you iterate to detect status-code changes
and then assign some identifier to break up sessions accordingly.
the only tricky part is keeping state as you iterate. i am not sure of my
suggested way below is the best.
pseudo scala code:
rdd[(MachineId, Session)].groupSort(numPartitions,
timestampOrdering).mapStreamByKey{ iterator =>
var uuid = UUID.randomUUID.toString
iterator.map{ session =>
if (session.statusCode == 123) uuid = UUID.randomUUID.toString
(uuid, session)
}
}
the way to do this without a mutable state would be something like
Iterator.scanLeft
> Support sorting of values in addition to keys (i.e. secondary sort)
> -------------------------------------------------------------------
>
> Key: SPARK-3655
> URL: https://issues.apache.org/jira/browse/SPARK-3655
> Project: Spark
> Issue Type: New Feature
> Components: Spark Core
> Affects Versions: 1.1.0, 1.2.0
> Reporter: koert kuipers
> Assignee: Koert Kuipers
>
> Now that spark has a sort based shuffle, can we expect a secondary sort soon?
> There are some use cases where getting a sorted iterator of values per key is
> helpful.
--
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
