[ 
https://issues.apache.org/jira/browse/HDDS-702?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16657474#comment-16657474
 ] 

Bharat Viswanadham edited comment on HDDS-702 at 10/19/18 9:47 PM:
-------------------------------------------------------------------

[~elek]

Thanks for the info, it makes sense to have parent project still as 
hadoop-project.

One more observation I have seen is when I tried to build with below command

mvn clean install -T 6 -Pdist -Phdds -DskipTests -Dmaven.javadoc.skip=true -am 
-pl :hadoop-ozone-dist

I see below warning, I am able to successfully build it though.

[WARNING] The requested profile “hdds” could not be activated because it does 
not exist.

[~arpitagarwal]

The scenario you have mentioned should not happen, as this in pom.xml we have 
provided snapshot repo's from where we should get the Hadoop snapshot jars.

 


was (Author: bharatviswa):
[~elek]

Thanks for the info, it makes sense to have parent project still as 
hadoop-project.

[~arpitagarwal]

The scenario you have mentioned should not happen, as this in pom.xml we have 
provided snapshot repo's from where we should get the Hadoop snapshot jars.

 

> Used fixed/external version from hadoop jars in hdds/ozone projects
> -------------------------------------------------------------------
>
>                 Key: HDDS-702
>                 URL: https://issues.apache.org/jira/browse/HDDS-702
>             Project: Hadoop Distributed Data Store
>          Issue Type: New Feature
>            Reporter: Elek, Marton
>            Assignee: Elek, Marton
>            Priority: Major
>         Attachments: HDDS-702.001.patch
>
>
> In the current form of the project ozone uses the in-tree snapshot version of 
> the hadoop (hadoop 3.3.0-SNAPSHOT as of now).
> I propose to use a fixed version from the hadoop jars which could be 
> independent from the in-tree hadoop.
> 1. With using already released hadoop (such as hadoop-3.1) we can upload the 
> ozone jar files to the maven repository without pseudo-releasing the hadoop 
> snapshot dependencies. (In the current form it's not possible without also 
> uploading a custom, ozone flavour of hadoop-common/hadoop-hdfs)
> 2. With using fixed version of hadoop the build could be faster and the yetus 
> builds could be simplified (it's very easy to identify the projects which 
> should be checked/tested if only the hdds/ozone projects are part of the 
> build: we can do full build/tests all the time).
> After the previous work it's possible to switch to fixed hadoop version, 
> because:
> 1) we have no more proto file dependency between hdds and hdfs (HDDS-378, and 
> previous works by Mukul and Nanda)
> 2) we don't need to depend on the in-tree hadoop-project-dist (HDDS-447)
>  



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

---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to