[ https://issues.apache.org/jira/browse/SOLR-9598?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15595712#comment-15595712 ]
Michael Sun commented on SOLR-9598: ----------------------------------- It seems Solr sometimes doesn't reopen the searcher once collection is restored. Therefore a query soon after restoring sometimes doesn't return the correct result. Currently it requires to restart Solr before using restored collection to make sure query returns correct result. > Solr RESTORE api doesn't wait for the restored collection to be fully ready > for usage > ------------------------------------------------------------------------------------- > > Key: SOLR-9598 > URL: https://issues.apache.org/jira/browse/SOLR-9598 > Project: Solr > Issue Type: Bug > Security Level: Public(Default Security Level. Issues are Public) > Affects Versions: 6.2 > Reporter: Hrishikesh Gadre > > As part of the RESTORE operation, Solr creates a new collection and adds > necessary number of replicas to each shard. The problem is that this > operation doesn't wait for this new collection to be fully ready for usage > (e.g. querying and indexing). This requires extra checks on the client side > to make sure that the recovery is complete and reflected in cluster status > stored in Zookeeper. e.g. refer to the backup/restore unit test for this > check, > https://github.com/apache/lucene-solr/blob/722e82712435ecf46c9868137d885484152f749b/solr/core/src/test/org/apache/solr/cloud/AbstractCloudBackupRestoreTestCase.java#L234 > Ideally this check should be implemented in the RESTORE operation itself. -- 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