[jira] [Created] (GEODE-5402) Disk recovery sometimes hangs when no member is recognized as having the most recent data

["Lynn Gallinat (JIRA)" <ji...@apache.org>]

Lynn Gallinat created GEODE-5402:
------------------------------------

             Summary: Disk recovery sometimes hangs when no member is recognized as having the most recent data 
                 Key: GEODE-5402
                 URL: https://issues.apache.org/jira/browse/GEODE-5402
             Project: Geode
          Issue Type: Bug
          Components: persistence
            Reporter: Lynn Gallinat


During disk recovery, sometimes a the system will hang because every member is waiting for others to initialize first, and nobody is recognized as having the most recent data.

This can happen when a member persists its most recent view of another member to disk, then that member is killed before it can remove the older view of the other member from the op log files. When starting the cluster from disk, the member then has two views of the other member, one recent and the other stale. The member then decides is needs to wait for the stale view of the other member which causes the hang.



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)