[
https://issues.apache.org/jira/browse/SOLR-7499?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Varun Thacker updated SOLR-7499:
--------------------------------
Attachment: SOLR-7499.patch
Same patch as before but adds a deprecated message in the response as well if
the "name" param has been specified in the ADDREPLICA call.
Summarizing:
- Added checks to make sure the "name" specified is unique for that collection
- Added deprecated message when the "name" param is used. We can remove it in
the future.
If that patch is okay I'll commit this tomorrow.
> Deprecate the "name" parameter from the ADDREPLICA Collection API call
> ----------------------------------------------------------------------
>
> Key: SOLR-7499
> URL: https://issues.apache.org/jira/browse/SOLR-7499
> Project: Solr
> Issue Type: Bug
> Reporter: Varun Thacker
> Assignee: Varun Thacker
> Priority: Minor
> Fix For: Trunk, 5.2
>
> Attachments: SOLR-7499.patch, SOLR-7499.patch
>
>
> Right now we take a "name" parameter in the ADDREPLICA call. We use that as
> the core name for the replica. Are there any use cases where specifying the
> name of the core for the replica is useful?
> Here are the disadvantages of doing so -
> 1. We don't verify if the name is unique in the collection. So if a
> conflicting name ends up in the same node then the call will fail.
> 2. If it core is created on some other node, it will fail with
> legacyCloud=false as that checks for uniqueness in core names.
> https://cwiki.apache.org/confluence/display/solr/Collections+API#CollectionsAPI-api_addreplica
> - The ref guide has never documented the "name" parameter.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org
For additional commands, e-mail: dev-h...@lucene.apache.org
