[ https://issues.jboss.org/browse/ISPN-4584?page=com.atlassian.jira.plugin.... ] Sanne Grinovero edited comment on ISPN-4584 at 9/18/14 12:16 PM: ----------------------------------------------------------------- {quote}Another check that would be useful is that the triple locking/metadata/data caches is not shared, e.g cache indexedA using lockingA, metadataA and dataA, while cache indexedB uses lockingA, metadataA and dataA - I've accidentally configured it that way (as in my test only indexedB should have been used) and stumbled again on ISPN-4577.{quote} Good point, if we don't accept that configuration we should validate against it. Although, there is the use case in which people actually want to share the same index across Caches or even applications; a better solution would be to finish the CacheManager level Query Engine, so that different caches can actually forward work to the same IndexWriter rather than locking each other out. But I agree, if we don't do that timely we should validate this. was (Author: sannegrinovero): {quote}Another check that would be useful is that the triple locking/metadata/data caches is not shared, e.g cache indexedA using lockingA, metadataA and dataA, while cache indexedB uses lockingA, metadataA and dataA - I've accidentally configured it that way (as in my test only indexedB should have been used) and stumbled again on ISPN-4577.{quote} Good point, if we don't accept that configuration we should validate against it. Although, there is the use case in which people actually want to share the same index across Caches or even applications; a better solution however would be to finish the CacheManager level Query Engine, so that different caches can actually forward work to the same IndexWriter rather than locking each other out. But I agree, if we don't do that timely we should validate this. -- This message was sent by Atlassian JIRA (v6.3.1#6329)