beam-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Kenneth Knowles (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (BEAM-335) beam_PostCommit_RunnableOnService_GoogleCloudDataflow broken since #514
Date Mon, 13 Jun 2016 01:38:20 GMT

    [ https://issues.apache.org/jira/browse/BEAM-335?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15326732#comment-15326732
] 

Kenneth Knowles commented on BEAM-335:
--------------------------------------

Probably should run a PAssert on each runner in the presubmit. Could be by throwing it into
and e2e test like we run today, or by a small selection of other tests.

Another though is that the TestDataflowPipelineRunner could possibly be a more generic TestPipelineRunner(<actual
runner>) in which case we can get coverage via local Flink & Spark endpoints more cheaply.

> beam_PostCommit_RunnableOnService_GoogleCloudDataflow  broken since #514
> ------------------------------------------------------------------------
>
>                 Key: BEAM-335
>                 URL: https://issues.apache.org/jira/browse/BEAM-335
>             Project: Beam
>          Issue Type: Bug
>          Components: testing
>            Reporter: Daniel Halperin
>            Assignee: Kenneth Knowles
>            Priority: Critical
>
> The RunnableOnService+Dataflow postsubmit has been a little flaky recently, but started
failing 143/143 tests at build #514. [Changes]([https://builds.apache.org/job/beam_PostCommit_RunnableOnService_GoogleCloudDataflow/514]):
> Base PAssert on GBK instead of side inputs (commit: 810ffeb2785bf996001c8fadb992410d1f9409c6)
— klk / detail
> Fix TriggerExampleTest (commit: 77aa0938f75f9f3d18a4fa79a5ffe6159167f4d5) — klk / detail
> Switch Spark streaming tests to custom assertions (commit: f222df109e773f23e56f9e830454356893989a15)
— klk / detail
> Spark runner: Assign windows when re-windowing into global window (commit: d53e96a0d1f5f26ad0e3efc90dc9f7b53135443b)
— klk / detail
> My best guess is the PAssert change. Kenn, can you please investigate?



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Mime
View raw message