[ https://issues.jboss.org/browse/ISPN-3214?page=com.atlassian.jira.plugin.... ] William Burns commented on ISPN-3214: ------------------------------------- Actually looking into this, it would require locking the L1 cache before the remote invocation and after. This may be okay? With the changes I have for ISPN-3197, we only update the L1 cache on a read if it is able to acquire the lock immediately. But this could cause a L1 update to miss if a conditional write occurs and fails since it would no longer update the L1 but held the lock - acceptable? Also this would be a similar regression to how locking would work in that it would acquire the local lock first before the remote. But I believe this should be fine since the issue prior was if it acquired the remote owner locks in a different order, causing a deadlock. Now we only acquire the primary owner lock instead. -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators For more information on JIRA, see: http://www.atlassian.com/software/jira