[
https://issues.apache.org/jira/browse/SOLR-8999?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15243849#comment-15243849
]
Hoss Man commented on SOLR-8999:
--------------------------------
inspired by this thread...
*
http://mail-archives.apache.org/mod_mbox/lucene-solr-user/201604.mbox/%3c7f050c0f16dd934f8599eb5dae2690350160ca5...@wayne.lan.gossinteractive.com%3E
*
http://mail-archives.apache.org/mod_mbox/lucene-solr-user/201604.mbox/%[email protected]%3E
> MiniSolrCloudCluster should support starting a cluster w/existing zk data
> -------------------------------------------------------------------------
>
> Key: SOLR-8999
> URL: https://issues.apache.org/jira/browse/SOLR-8999
> Project: Solr
> Issue Type: Improvement
> Reporter: Hoss Man
>
> In Solr tests, we always start with a completely pristine tmp directories for
> each test to use, and all of our existing MiniSolrCloudClusters expect that
> and MiniSolrCloudCluster creates everything it needs.
> But for Solr clients, that want to write tests using MiniSolrCloudCluster to
> test their own code, MiniSolrCloudCluster can currently fail in confusing
> ways if they re-use the "baseDir" in multiple tests runs.
> At a minimum, if MiniSolrCloudCluster is going to expect baseDir to always be
> empty on construction, then we should document this and assert it in the
> constructor.
> But it seems to me that it might also be useful if it were _possible_ (even
> if not the default behavior) to spin up a MiniSolrCloudCluster pointing at an
> existing baseDir that already had zk server data and existing solr
> instanceDirs and verify that everything could be started up after a previous
> total shutdown.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
