[
https://issues.apache.org/jira/browse/BEAM-5953?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16684315#comment-16684315
]
Mark Liu commented on BEAM-5953:
--------------------------------
Turns out that the pipeline.pb file staging to gcs is corrupted and cause
getPipelineFromEnv in DataflowWorkerHarnessHelper.java failed to parse the
Pipeline proto. The pipeline proto is serialized and uploaded to staging
location via storage_v1_client which is built on top of google-apitools
library. I don't have much insight of that library but this could be a
potential bug in Python 3.
There is a discussion to move off google-apitools from SDK in
https://issues.apache.org/jira/browse/BEAM-4850.
> Support DataflowRunner on Python 3
> ----------------------------------
>
> Key: BEAM-5953
> URL: https://issues.apache.org/jira/browse/BEAM-5953
> Project: Beam
> Issue Type: Sub-task
> Components: sdk-py-core
> Reporter: Mark Liu
> Assignee: Mark Liu
> Priority: Major
> Time Spent: 0.5h
> Remaining Estimate: 0h
>
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
