[
https://issues.apache.org/jira/browse/NIFI-5224?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16487327#comment-16487327
]
Bryan Bende commented on NIFI-5224:
-----------------------------------
I like the idea of a client service for Solr, but I think the interface should
just be one method:
{code:java}
SolrClient getSolrClient();{code}
The implementation would be a controller service that had all the properties
for configuring the Solr Type, URL, timeout params, security settings, etc.
This may be what [~jope] meant when creating this ticket, but just wanted to be
sure since it mentioned "CRUD functions", and I don't think there really needs
to be other methods on the interface since processors can do whatever they want
with the client.
Also, we should have at least one (maybe two?) overlapping release where all
the Solr processors keep all their existing properties, but also have a
property for client service. If client service is populated use it, otherwise
fallback to current approach. This way people have a chance to migrate existing
flows before being forced to.
> Add SolrClientService
> ---------------------
>
> Key: NIFI-5224
> URL: https://issues.apache.org/jira/browse/NIFI-5224
> Project: Apache NiFi
> Issue Type: Improvement
> Reporter: Johannes Peter
> Assignee: Johannes Peter
> Priority: Major
>
> The Solr CRUD functions that are currently included in SolrUtils should be
> moved to a controller service.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
