[
https://issues.apache.org/jira/browse/SOLR-14660?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17480178#comment-17480178
]
Kevin Risden commented on SOLR-14660:
-------------------------------------
The HDFS integration isn't THAT complicated. 99% of the crap to deal with is
the integration tests. The directory factory, block cache, and other classes -
are reasonably simple to understand. Most of the integration tests suck due to
the dependencies, configs, and how heavy the HDFS namenode/datanode processes
are. So conceptually it works really well for read so querying against a
relatively static index works - indexing is questionable from a performance
standpoint but again its just a simple directory -> hdfs filesystem.
The module quality is an interesting question - I personally haven't tried the
module, but since its just a move it should be the same as above. Nothing
conceptually changed just the file locations and build configs.
So yea even though the HDFS code module hasn't gotten much attention - the core
functionality hasn't really been touched since its "complete" - its mostly in
the integration test code.
> Migrating HDFS into a module
> ----------------------------
>
> Key: SOLR-14660
> URL: https://issues.apache.org/jira/browse/SOLR-14660
> Project: Solr
> Issue Type: Improvement
> Components: Hadoop Integration, Plugin system
> Reporter: Ishan Chattopadhyaya
> Priority: Blocker
> Labels: package, packagemanager
> Fix For: 9.0
>
> Time Spent: 3h 10m
> Remaining Estimate: 0h
>
> Following up on the deprecation of HDFS (SOLR-14021), we need to work on
> isolating it away from Solr core and making a package for this. This issue is
> to track the efforts for that.
--
This message was sent by Atlassian Jira
(v8.20.1#820001)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
