[
https://issues.apache.org/jira/browse/METRON-1302?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16311719#comment-16311719
]
ASF GitHub Bot commented on METRON-1302:
----------------------------------------
Github user cestella commented on the issue:
https://github.com/apache/metron/pull/831
Ok, I thought the general consensus was different. As per my comment:
```
That sounds good. Which one would you guys prefer:
* display names in the mpack of "Elasticsearch" and "HDFS"
* variable prefixes of es and hdfs
I think I'd prefer the first as I really have been trying to set us up to a
world where we can make things pluggable (this is the first baby step).
```
Nobody commented on that, so I went with adjusting the display names rather
than variable prefixes. The reason why I preferred to do that was that when we
do a better job of including Solr or other indices, prefixing the variables
`elasticsearch` will be confusing. The flux files are generic to the writer
type for the random access indices. I can understand the confusion for the
user given we only really support HDFS and elasticsearch right now, which is
why I changed the display names.
Can we recap the reasons against this approach?
> Split up Indexing Topology into batch and random access sections
> ----------------------------------------------------------------
>
> Key: METRON-1302
> URL: https://issues.apache.org/jira/browse/METRON-1302
> Project: Metron
> Issue Type: Improvement
> Reporter: Casey Stella
>
> Currently we have the indexing topology handle writing to both random access
> indices (e.g. elasticsearch) as well as batch write indices (e.g. hdfs). We
> should split these up and configure them separately.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
