Details
-
Bug
-
Status: Closed
-
Major
-
Resolution: Fixed
-
1.13.0, 1.13.1, 1.13.2, 1.13.3, 1.14.0, 1.15.0
Description
GEODE-7921 in 1.13 introduced this issue. When processing a message used by entry idle expiration to find out when an entry was last accessed on any member of the cluster it now checks for a null cache, region, and entry. If any of them are null the message simply returns. The problem is that the expiration thread that sent this message waits forever for a reply. So unless the member that found a null leaves the cluster, the expiration thread will be hung waiting for a reply that will never be sent.
The fix for this is easy. We just need to send a reply of zero when we find a null.
Attachments
Issue Links
- is caused by
-
GEODE-7921 NullPointerExceptions logged during auto-reconnect
- Closed
- links to