beam-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "ASF GitHub Bot (JIRA)" <j...@apache.org>
Subject [jira] [Work logged] (BEAM-5455) Don't info log for every bundle in the python sdk
Date Fri, 21 Sep 2018 18:38:00 GMT

     [ https://issues.apache.org/jira/browse/BEAM-5455?focusedWorklogId=146546&page=com.atlassian.jira.plugin.system.issuetabpanels:worklog-tabpanel#worklog-146546
]

ASF GitHub Bot logged work on BEAM-5455:
----------------------------------------

                Author: ASF GitHub Bot
            Created on: 21/Sep/18 18:37
            Start Date: 21/Sep/18 18:37
    Worklog Time Spent: 10m 
      Work Description: tweise commented on issue #6459: [BEAM-5455] Log per-bundle messages
at debug in python sdk
URL: https://github.com/apache/beam/pull/6459#issuecomment-423633284
 
 
   I checked that `Slf4jLogWriter` receives INFO log entries but no DEBUG log entries. So
there isn't any gRPC overhead, but there also does not seem to be a way to capture the SDK
debug logging, even when running the job server at debug level. How is the SDK harness logging
controlled?

----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
users@infra.apache.org


Issue Time Tracking
-------------------

    Worklog Id:     (was: 146546)
    Time Spent: 0.5h  (was: 20m)

> Don't info log for every bundle in the python sdk
> -------------------------------------------------
>
>                 Key: BEAM-5455
>                 URL: https://issues.apache.org/jira/browse/BEAM-5455
>             Project: Beam
>          Issue Type: Improvement
>          Components: sdk-py-harness
>            Reporter: Micah Wylde
>            Assignee: Micah Wylde
>            Priority: Minor
>              Labels: portability-flink
>          Time Spent: 0.5h
>  Remaining Estimate: 0h
>
> Currently the python sdk logs a bunch of messages at INFO level for every bundle. For
the portable flink runner, which uses a bundle size of 1 in streaming mode, this produces
huge log spew:
> {code:java}
> [grpc-default-executor-1] INFO sdk_worker.run - Got work 4
> [grpc-default-executor-1] INFO bundle_processor.process_bundle - start <DataOutputOperation
>
> [grpc-default-executor-1] INFO bundle_processor.process_bundle - start <DoOperation
window output_tags=['out']>
> [grpc-default-executor-1] INFO bundle_processor.process_bundle - start <DoOperation
pair_with_one output_tags=['out']>
> [grpc-default-executor-1] INFO bundle_processor.process_bundle - start <DoOperation
timestamp output_tags=['out']>
> [grpc-default-executor-1] INFO bundle_processor.process_bundle - start <DoOperation
decode output_tags=['out']>
> [grpc-default-executor-1] INFO bundle_processor.process_bundle - start <DataInputOperation
receivers=[ConsumerSet[.out0, coder=WindowedValueCoder[BytesCoder], len(consumers)=1]]>
> [grpc-default-executor-1] INFO bundle_processor.process_bundle - finish <DataInputOperation
receivers=[ConsumerSet[.out0, coder=WindowedValueCoder[BytesCoder], len(consumers)=1]]>
> [grpc-default-executor-1] INFO bundle_processor.process_bundle - finish <DoOperation
decode output_tags=['out'], receivers=[ConsumerSet[decode.out0, coder=WindowedValueCoder[FastPrimitivesCoder],
len(consumers)=1]]>
> [grpc-default-executor-1] INFO bundle_processor.process_bundle - finish <DoOperation
timestamp output_tags=['out'], receivers=[ConsumerSet[timestamp.out0, coder=WindowedValueCoder[FastPrimitivesCoder],
len(consumers)=1]]>
> [grpc-default-executor-1] INFO bundle_processor.process_bundle - finish <DoOperation
pair_with_one output_tags=['out'], receivers=[ConsumerSet[pair_with_one.out0, coder=WindowedValueCoder[TupleCoder[FastPrimitivesCoder,
FastPrimitivesCoder]], len(consumers)=1]]>
> [grpc-default-executor-1] INFO bundle_processor.process_bundle - finish <DoOperation
window output_tags=['out'], receivers=[ConsumerSet[window.out0, coder=WindowedValueCoder[TupleCoder[FastPrimitivesCoder,
FastPrimitivesCoder]], len(consumers)=1]]>
> [grpc-default-executor-1] INFO bundle_processor.process_bundle - finish <DataOutputOperation
>
> {code}
>  
> These log messages should probably be at debug instead.



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)

Mime
View raw message