You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@storm.apache.org by "Stig Rohde Døssing (JIRA)" <ji...@apache.org> on 2017/08/20 07:37:00 UTC

[jira] [Updated] (STORM-2689) storm-kafka-examples and storm-kafka-client-examples are difficult for new users to run

     [ https://issues.apache.org/jira/browse/STORM-2689?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Stig Rohde Døssing updated STORM-2689:
--------------------------------------
    Description: 
The storm-kafka-examples and storm-kafka-client-examples projects configure their dependencies in a way that makes them difficult to run for a new user. The other example projects set up a provided dependency on storm-client, and otherwise include all their dependencies in a shaded jar. 

storm-kafka(-client) by default produce jars without several necessary dependencies, e.g. the Kafka client libraries. The provided.scope Maven parameter was intended to be used to allow users to produce a shaded jar with all dependencies, but if provided scope is set to compile, the resulting jar will also contain storm-client. This prevents the jar from running on a real cluster.

While users can work around this by producing the slim jar and using --artifacts when submitting the topology, this is unnecessarily tedious. We should just produce a fat jar by default, then mention in the example documentation that --artifacts is there for users that want to make slimmer jars.

Edit:
This issue now includes simplifying storm-kafka-examples and storm-kafka-client-examples in general. The examples demonstrate use of State and DRPC when the focus should be on how to use storm-kafka(-client). It also causes the modules to have some undesirable dependencies, e.g. they both depend on storm-starter. 

  was:
The storm-kafka-examples and storm-kafka-client-examples projects configure their dependencies in a way that makes them difficult to run for a new user. The other example projects set up a provided dependency on storm-client, and otherwise include all their dependencies in a shaded jar. 

storm-kafka(-client) by default produce jars without several necessary dependencies, e.g. the Kafka client libraries. The provided.scope Maven parameter was intended to be used to allow users to produce a shaded jar with all dependencies, but if provided scope is set to compile, the resulting jar will also contain storm-client. This prevents the jar from running on a real cluster.

While users can work around this by producing the slim jar and using --artifacts when submitting the topology, this is unnecessarily tedious. We should just produce a fat jar by default, then mention in the example documentation that --artifacts is there for users that want to make slimmer jars.


> storm-kafka-examples and storm-kafka-client-examples are difficult for new users to run
> ---------------------------------------------------------------------------------------
>
>                 Key: STORM-2689
>                 URL: https://issues.apache.org/jira/browse/STORM-2689
>             Project: Apache Storm
>          Issue Type: Bug
>          Components: examples
>    Affects Versions: 2.0.0, 1.1.1
>            Reporter: Stig Rohde Døssing
>            Assignee: Stig Rohde Døssing
>          Time Spent: 1h 50m
>  Remaining Estimate: 0h
>
> The storm-kafka-examples and storm-kafka-client-examples projects configure their dependencies in a way that makes them difficult to run for a new user. The other example projects set up a provided dependency on storm-client, and otherwise include all their dependencies in a shaded jar. 
> storm-kafka(-client) by default produce jars without several necessary dependencies, e.g. the Kafka client libraries. The provided.scope Maven parameter was intended to be used to allow users to produce a shaded jar with all dependencies, but if provided scope is set to compile, the resulting jar will also contain storm-client. This prevents the jar from running on a real cluster.
> While users can work around this by producing the slim jar and using --artifacts when submitting the topology, this is unnecessarily tedious. We should just produce a fat jar by default, then mention in the example documentation that --artifacts is there for users that want to make slimmer jars.
> Edit:
> This issue now includes simplifying storm-kafka-examples and storm-kafka-client-examples in general. The examples demonstrate use of State and DRPC when the focus should be on how to use storm-kafka(-client). It also causes the modules to have some undesirable dependencies, e.g. they both depend on storm-starter. 



--
This message was sent by Atlassian JIRA
(v6.4.14#64029)