[
https://issues.apache.org/jira/browse/TINKERPOP-1358?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15362890#comment-15362890
]
Marko A. Rodriguez commented on TINKERPOP-1358:
-----------------------------------------------
Spark tests do not require a standalone SparkServer setup. Simple `mvn clean
install` should work. Perhaps you didn't `mvn clean install` from root first as
`SparkGraphComputer` relies on `HadoopGraph` and service loaders. ?
> spark-gremlin tests require special setup
> -----------------------------------------
>
> Key: TINKERPOP-1358
> URL: https://issues.apache.org/jira/browse/TINKERPOP-1358
> Project: TinkerPop
> Issue Type: Test
> Components: hadoop
> Affects Versions: 3.2.0-incubating
> Environment: Linux
> No spark, hadoop installed, configured, or running.
> No spark, hadoop env vars set.
> Reporter: Robert Dale
> Labels: spark
> Fix For: 3.2.1
>
>
> cd spark-gremlin; mvn clean install
> Most tests fail with:
> java.net.BindException: Cannot assign requested address: Service
> 'sparkDriver' failed after 16 retries!
> Eventually leads to resource starvation: OutOfMemoryError: unable to create
> new native thread. Presumably because exceptions are not caught and
> SparkContexts closed.
> If tests require external server, perhaps they would be best served as
> integration tests. Otherwise, there's some undocumented special sauce
> missing.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
