beam-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Stas Levin (JIRA)" <j...@apache.org>
Subject [jira] [Created] (BEAM-2074) SourceDStream's rate control mechanism may not work
Date Tue, 25 Apr 2017 09:28:04 GMT
Stas Levin created BEAM-2074:
--------------------------------

             Summary: SourceDStream's rate control mechanism may not work
                 Key: BEAM-2074
                 URL: https://issues.apache.org/jira/browse/BEAM-2074
             Project: Beam
          Issue Type: Bug
          Components: runner-spark
            Reporter: Stas Levin
            Assignee: Stas Levin


If the {{boundMaxRecords}} for {{SourceDStream}} is not set (as a result of users choosing
not to specify {{SparkPipelineOptions.setMaxRecordsPerBatch()}}), {{SourceDStream}} is designed
to use {{rateControlledMaxRecords()}} to determine the max number of records it should be
reading.

However, at the moment {{SourceDStream}} only consults  {{rateControlledMaxRecords()}} once,
when it is created, thus following
instantiations of {{MicrobatchSource}} will be using the now-set-in-stone {{boundMaxRecords}}
value without consulting with {{rateControlledMaxRecords()}}, making the rate controlling
mechanism sub-optimal.



--
This message was sent by Atlassian JIRA
(v6.3.15#6346)

Mime
View raw message