[
https://issues.apache.org/jira/browse/SOLR-6894?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14262648#comment-14262648
]
Alan Woodward commented on SOLR-6894:
-------------------------------------
I've rethought this slightly, what I want to do now is introduce another
abstraction, SolrNodeClient, which extends SolrClient and adds two methods:
* SolrClient forCore(String core)
* SolrClient forCollection(String collection)
HttpSolrClient and CloudSolrClient both extend SolrNodeClient. You create a
SolrNodeClient to point at a particular node (or zk-discovered cluster, in the
case of CloudSolrClient), which you can then use for admin requests. If you
want to execute requests against a particular collection, call
#forCollection(String collection), and you get a specialised SolrClient that
shares its resources (HttpClient, ZkClient, etc) with the parent
SolrNodeClient. Calling #forCore() on an HttpSolrClient has the same effect as
calling #forCollection(), but on a CloudSolrClient it will return a client that
talks to a specific core in the cluster, with a default parameter of
distrib=false, which is useful for debugging.
This means that your API when communicating with an empty node would be
something like:
{code}
SolrNodeClient client = new HttpSolrClient(pathToMultiCoreServer);
CoreAdminRequest.Create createReq = new Create();
createReq.setCoreName("myCore");
createReq.setConfigSet("myConfigSet");
client.request(createReq);
SolrClient myCoreClient = client.forCore("myCore")
{code}
and for SolrCloud:
{code}
CloudSolrClient clusterClient = new CloudSolrClient(zkPath);
CollectionAdminRequest.Create createReq = new CollectionAdminRequest.Create();
createReq.setName("myCollection")
createReq.setConfig("myConfig")
clusterClient.execute(createReq);
SolrClient collectionClient = client.forCollection("myCollection");
{code}
> Add MultiCoreSolrServer
> ------------------------
>
> Key: SOLR-6894
> URL: https://issues.apache.org/jira/browse/SOLR-6894
> Project: Solr
> Issue Type: New Feature
> Components: SolrJ
> Affects Versions: 5.0, Trunk
> Reporter: Alan Woodward
> Assignee: Alan Woodward
> Priority: Minor
> Attachments: SOLR-6894.patch
>
>
> At the moment, it isn't possible to use a single SolrServer instance to
> create new cores via CoreAdmin on an empty node, and then query those nodes.
> MultiCoreSolrServer is a utility class that does just that, by allowing you
> to create child SolrServer instances for individual cores that share the
> underlying HttpClient.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
