[
https://issues.apache.org/jira/browse/SOLR-8081?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14904250#comment-14904250
]
Jan Høydahl commented on SOLR-8081:
-----------------------------------
Well, that's up to the IT guys that provision the servers by choosing a
sutitable RAID level or other similar technology to present a large single,
virtual volume to the application. Thus we do not need to worry how Solr
utilize single disks, that is taken care of at OS level.
> When creating a collection, we need a way to utilize multiple disks available
> on a node.
> ----------------------------------------------------------------------------------------
>
> Key: SOLR-8081
> URL: https://issues.apache.org/jira/browse/SOLR-8081
> Project: Solr
> Issue Type: Improvement
> Components: SolrCloud
> Reporter: Timothy Potter
>
> Currently, if I want to change the dataDir for a core (such as to utilize
> multiple disks on a Solr node), I need to either setup a symlink or change
> the dataDir property in core.properties and restart the Solr node. For
> instance, if I have a 4-node SolrCloud cluster and want to create a
> collection with 4 shards with rf=2, then 8 Solr cores will be created across
> the cluster, 2 per node. If I want to have each core use a separate disk,
> then I have to do that after the fact. I'm aware that I could create the
> collection with rf=1 and then use AddReplica to add additional replicas with
> a different dataDir set, but that feels cumbersome as well.
> What would be nice is to have a way for me to specify available disks and
> have Solr use that information when provisioning cores on the node.
> [~anshumg] mentioned this might be best accomplished with a replica placement
> strategy.
--
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
